Gerhard Mercator Universität Gesamthochschule …Gerhard-Mercator-Universität Gesamthochschule...
Transcript of Gerhard Mercator Universität Gesamthochschule …Gerhard-Mercator-Universität Gesamthochschule...
Gerhard MercatorUniversitätGesamthochschuleDuisburg
Annual Report 1996/97
Department ofOptoelectronics
Gerhard-Mercator-UniversitätGesamthochschule Duisburg
Fachbereich ElektrotechnikFachgebiet Optoelektronik
ZHOLotharstr. 55
D - 47057 DuisburgGermany
Prof. Dr. rer. nat. D. Jäger
+49-203-379-2340
+49-203-379-2409
http://www.oe.uni-duisburg.de
Editor: R. Buß
Head:
Tel:
Fax:
Email:
URL:
Pictures of the Center for Solid-State Electronics and Optoelectronics taken in Dec. 1997
4
The Center for Solid-State Electronics and Optoelectronis (ZHO)
After only nine month of construction the topping-out-ceremony of the new Center for Solid-State
Electronics and Optoelectronics (Zentrum für Halbleitertechnik und Optoelektronik, ZHO) has been
celebrated on June, 4th, 1997.
The center is planned to be the new home of the Department of Optoelectronics and the Solid-
State Electronics Department in the beginning of October 1998. It consitst of two parts: The clean-
room building with an area of approx. 470 m2 and the building for the offices and laboratories with an
area of approx. 1200 m2.
Table of Contents
1 Foreword
2 Members of the Department
3 Research
3.1 Optical Networks3.1.1 High-speed, high-power travelling-wave photodetectors
3.1.2 Fabrication and characterization of a travelling-wave photodetector
3.1.3 Simulation of the microwave generation of a travelling-wave photodetector
3.1.4 Determination of RF-equivalent circuit elements of travelling-wave photode-
tectors using network analysis
3.1.5 Polarization insensitive waveguide modulators on InP
3.2 Optical Interconnects and Processors3.2.1 Neurotechnology: Retina Implant
3.2.2 Analysis of the optical energy and signal transfer module for an artificial vision
prosthesis
3.2.3 Development of an optical signal and energy transmission system
3.2.4 Infrared data link for rotating display-systems
3.2.5 Nonlinear hybrid GaAs/AlGaAs multilayer-heterostructures for high-speed
information processing
3.2.6 8 x 8 LED arrays integrated with 64 channel Si-driver circuits
3.3 Millimeterwave Electronics3.3.1 Picosecond pulse generation on monolithic nonlinear transmission lines using
high-speed InP-HFET diodes
3.3.2 Millimeter wave power generation nonlinear transmission lines
3.3.3 Nonlinear RTD circuits for high-speed A/D conversion
3.4 Optical Sensor Systems3.4.1 MQW-Electroabsorption-Modulator for application in a fiberoptic fieldsensor
3.4.2 Photovoltaic cells for fiber optic EMC - Sensor power supply
3.4.3 Time- and frequency-domain electro-optic field mapping of nonlinear trans-
mission lines
7
9
11
1111
14
17
20
25
2929
32
35
38
41
48
5151
55
59
6262
64
66
5
3.4.4 Characterization of monolithic microwave integrated circuits by heterodyne
electro-optic sampling
3.4.5 Development of an experimental setup for field probe measurements on
nonlinear transmission lines
3.5 Technologies for Optoelectronic Components and Systems3.5.1 Development of a measurement system for the optical characterization of
full-colour-LED-displays
3.5.2 Opinion poll on the evaluation of the legibility of LED-based displays
3.5.3 Evaluation of possible improvements to enhance the UV-power efficiency of
a xenon flashlamp system
3.5.4 Construction of a flip chip device for bonding integrated circuits
4 Teaching activities
4.1 Lectures, excercises, and practical studies
4.2 Seminars and colloquia
4.3 Doctoral, Diploma, and Graduate theses
5 Publications and presentations
6 Guide to the Department of Optoelectronics
TABLE OF CONTENTS6
71
74
7777
80
83
89
93
93
95
100
103
107
1 Foreword
The Fachgebiet Optoelektronik at the Ger-
hard-Mercator-Universität Duisburg has a tradi-
tion of research and teaching excellence dating
back to its establishment in 1989/90. This Re-
port provides a summary of major research in-
volvements and teaching activities reviewing
also publications and presentations by the mem-
bers of the institute .
The last two years were characterized by a
further-broadening of our scientific networks and
additional projects funded by various external
institutions. A first key research area is micro-
wave photonics with special emphasis on trav-
elling-wave photodetectors, electro-absorption
modulators and nonlinear optoelectronic devic-
es, where system aspects played a continuous-
ly increasing role. Special emphasis has further
been laid upon the two-dimensional electro-op-
tical characterization of monolithic microwave
integrated circuits and high-speed devices. As
a third topic, nonlinear optics and optoelectron-
ics in III-V-heterostructures and pulse compres-
sion in nonlinear MMICs were studied in detail.
Additionally to our activities within the Sonder-
forschungsbereich 254, mayor funding has been
provided by the Retina Implant project (EPI-
RET) and the collaborative programme on the
development of an EMC-field sensor, where our
Fachgebiet acts as the coordinator. We are al-
ready proud of looking back to a relevant exhi-
bition during the Laser 97 Fair in Munich. An-
other remarkable event was the International
Topical Meeting on Microwave Photonics
(MWP) held at the moated castle Schloß Hu-
genpoet in September 1997 which was orga-
nized by our Fachgebiet, D. Jäger being simul-
taneously the Chair of the International Steering
Committee on MWP (details on next page).
7
As a result of their remarkable research work,
Dr.-Ing. G. David received a fellowship of the
Alexander von Humboldt-Stiftung, funding a two-
years stay at the University of Michigan. More-
over, Dr.-Ing. A. Stöhr was awarded a grant to
carry out research work at the Communications
Research Laboratories, Ministry of Posts & Tele-
communications in Tokyo. Further, D. Jäger re-
ceived the title Professor Onorific from the
University of Brasov/Romania and became the
Chair of the German IEEE/LEOS Chapter.
Besides the usual and obligatory courses, the
Fachgebiet Optoelektronik offered in 1997 a new
lecture Einführung in die Multimediatechnik -
Technologien, Systeme, Anwendungen. More-
over, the Institute was involved in the Duisburg
Summerschool for Women, the Tag der Fors-
chung and the organisation of research activi-
ties on photonic bandgap materials in the frame-
work of the Forum Materialforschung in our
university. Finally, we note with great pleasure
that T. Alder and D. Kalinowski have been
awarded the University Price for excellent diplo-
ma theses.
I wish to thank all friends inside and outside
the university for their continuous encourage-
ment and assistance. Also, I would like to ex-
press my sincere thanks to all members of the
Institute for their efforts and contributions to our
success in optoelectronics.
Duisburg, September 1998
8
he 1997 International Topical Meeting
on Microwave Photonics (MWP97)
has been held from September 3 through 5
in the historical buildings of the majestic 17th
century moated Castle Schloß Hugenpoet,
situated in the beautiful Ruhrtal countryside
in the south of Essen near Duisburg.
This 7th Topical Meeting in the series on this
subject followed those in Cernay-la-Ville, France
(1994), Keystone, U.S.A. (1995), and Kyoto,
Japan (1996). It was the first one held in Ger-
many and has been organized by the Gerhard-
Mercator-Universität-Duisburg. On September
3, the Meeting started with a Workshop entitled
Photonic technologies for phased array anten-
nas where 6 invited papers addressed recent
results in this continously growing field of re-
search. In the Plenary Session on September 4,
3 invited speakers, Dr. R. Heidemann, Alcatel
SEL AG, Stuttgart, Germany, Dr. M.J. Wale,
GEC-Marconi Materials Technology Ltd,
Northampton, U.K., and Dr. D. Novak, The Uni-
versity of Melbourne, Australia, presented lec-
tures on the topic Microwave Photonics:
Present and Future. The regular conference
program consisted of 7 sessions on topics such
as Optical generation of microwave signals,
Optoelectronic modulators, mixers, and receiv-
ers, Microwave photonic systems, Fibre radio
networks, Modelling in microwave
photonics, and Microwave photonics
for measurements. Each session has
been opened by invited senior tech-
nologists from France, Hong Kong,
Japan, U.K., U.S.A., and Germany
having provided additional impetus to
this multi-disciplinary research area
of microwave photonics.
The whole program included fur-
theron a video session via internet with KDD in
Tokyo, Japan, a poster session and was com-
pleted by a postdeadline session. More than 100
papers have been submitted from 13 countries
showing the increasing interest of scientists and
engineers in this area. After careful evaluation
the Technical Program Committee has recom-
mended 69 papers - 54 oral and 15 poster - for
presentation. Additionally, 4 papers have been
selected during the conference for the postdead-
line session. 140 scientists and engineers have
registered for the Topical Meeting. In addition
to the technical program, a Partners Program,
a Welcome/Barbecue Party and a Gala Dinner
have been organized.
The Meeting has been sponsored by the
Deutsche Forschungsgemeinschaft (Bonn, Ger-
many), Hewlett-Packard GmbH (Ratingen, Ger-
many), Institut für Mobil- und Satellitenfunktech-
nik GmbH (Kamp-Lintfort, Germany), Lucent
Technologies (Allentown, P.A., U.S.A.), and the
Gerhard-Mercator-Universität Duisburg. More-
over it has been cooperatively sponsored by the
IEEE MTT-S and LEOS including the German
Chapters.
The organizers of MWP97 look back to a very
fruitful conference and look forward to the up-
coming meetings MWP98 in Princeton, N.J.,
U.S.A., and MWP99 in Melbourne, Australia.
T
2 Members of the Department
Department of OptoelectronicsZHO, Lotharstr. 55
47057 Duisburg, Germany
fon: +49 203 379-2340
fax: +49 203 379-2409
Head of the DepartmentJäger, Dieter Prof. Dr. rer. nat.
SecretaryGappa, Ulrike Optoelectronics
Tempel, Karin SFB 254
ScientistsAlles, Martin Dipl.-Ing.
Alder, Thomas Dipl. Ing.
Braasch, Thorsten Dipl.-Phys.
Buß, Rüdiger Dipl.-Ing.
David, Gerhard Dr.-Ing.
Groß, Matthias Dipl.-Phys.
Heinzelmann, Robert Dipl.-Ing.
Hülsewede, Ralf Dipl.-Phys.
Jäger, Irina Ph. D.
Kalinowski, Dirk Dipl.-Ing.
Knigge, Steffen Dr.-Ing.
Kremer, Ralf Dr.-Ing.
Redlich, Stefan Dipl.-Ing.
Schmidt, Manuel Dipl.-Phys.
Stöhr, Andreas Dr.-Ing.
Wingen, Georg Dipl.-Phys.
Zumkley, Stefan Dr. rer. nat.
Guest ScientistsDragoman, Mircea Prof. Dr.
Johnson, Roger Dipl.-Ing.
Lee, Chi Prof. Dr.
Mezentsev, Vladimir Ph. D.
Wendrix, Veronique Dipl.-Ing.
TechniciansMang, Sabine
Schedwill, Veronique
Slomka, Heinz Ing. grad.
StudentsAppenrodt, Nils
Balci, Senay
Baumeister, Thomas
Berger, Oliver
Boscher, Guido
Brings, Ludger
Bussek, Peter
Christoffers, Niels
Einweck, Michaela
Engel, Thomas
Ervens, Jutta
Hedtke, Ralph
Heinzdorf, Michael
Jabs, Mirco
Kampermann, Claus
Kreuder, Andreas
Lüdeke, André
Lotz, Oliver
Manh-Duc, Ngo
Meininger, Mark
Moeck, Jens-Peter
Neuhaus, Birgit
Ponellis, Bernd
Reintjes, Stefanie
Rogall, Michael
Spiegeler, Britta
Wenning, Michael
Weimann, Uwe
9
3 RESEARCH10
3 Research
3.1 Optical Networks
3.1.1 High-speed, high-power travel-ling-wave photodetectors
M. ALLES
ecently, new communication sys-
tems combining the advantages of
wireless transmission and fiber optics have
been proposed. Applications are transmis-
sion of traffic information, multimedia, or In-
ternet access. These systems operate
usually at an optical wavelength of 1.55µm
at frequencies up to 60GHz. Since the pho-
todetector should generate as much electri-
cal power as possible, the travelling-wave
photodetector under investigation has to ful-
fill these requirements.
IntroductionNovel wireless millimeterwave communica-
tion systems have been proposed by several
groups, see for example [1-4]. These systems
use fiber optics to transmit the signals over large
distances. Fiber optic cables have very low at-
tenuation of about 0.8dB per km, which is also
independent from the signal frequency, where-
as loss of coaxial cables is about 100dB per km.
An optical source generates a heterodyne
signal at 1.55µm, where the difference frequen-
cy of the two optical carriers corresponds to the
electrical millimeterwave frequency. The data
signal is modulated on one optical carrier. The
optical heterodyne signal is distributed to an
antenna station using usual fibre optic compo-
nents. At this antenna station, the photodetec-
tor converts the optical signal to a millimeter-
wave, which is amplified and transmitted to a
remote station. These communication systems
should be used for traffic information, digital vid-
eo, multimedia, or Internet access.
This system approach leads to some impor-
tant requirements for the photodetector. The
photodetector has to work at an optical wave-
length of standard fiber optics, i.e. at 1.3/1.55µm.
In a further point, the photodetector should be
capable to operate in the high frequency regime
to generate electrical signals in the millimeter-
wave regime. Additionally, the photodetector has
to generate as much electrical output power as
possible to reduce the requirements of the
electrical amplifier used in the antenna station.
Usually, high-speed photodetectors are fab-
ricated as lumped devices which are RC-time
limited. This means that high bandwidths can
only be reached if the device size is scaled down
to the micrometer regime. Dimensions of photo-
detectors operating at 100GHz are about 10µm2
[5]. Due to the small device size the photode-
tectors can only operate at low optical input pow-
ers to avoid saturation effects in the small vol-
ume. This limitation can be overcome if the
travelling-wave concept is considered for the
development of high-speed photodetectors [6].
High-speed travelling-wave photodetectorsThe travelling-wave photodetector is fabricat-
ed as an optical waveguide which is coupled to
an electrical waveguide due to an optical ab-
sorption layer. The capacitance of the electrical
waveguide is compensated by the inductance
of the transmission line resulting in an electrical
bandwidth which is not RC-time limited. This
concept avoids scaling down the device dimen-
sions. In contrast, the travelling-wave photode-
tector can be fabricated as a distributed device
in order to reduce optical saturation effects.
3.1 Optical Networks 11
R
Fig. 1: Sketch of the travelling-wave photodetector. Popt and Pel are the input optical and output electri-
cal powers,
The structure of the fabricated travelling-wave
photodetector is depicted in Fig. 1. The device
consists of an active region and a taper at the
end of the structure, used for hybrid integration
with electrical millimeterwave amplifiers. The
photodetector is MBE-grown on a semi-insulat-
ing InP-wafer for operation in the 1.3/1.55µm
regime. An InGaAlAs layer is used as an optical
waveguide. An InGaAs absorbing layer, leak-
age coupled to the waveguide, generates elec-
tron-hole pairs. Finally, an InAlAs-layer as a clad-
ding layer and an InGaAs/InGaAlAs superlattice
as a Schottky-barrier enhancement layer are
grown.
Electrical waveguiding is achieved using a
coplanar transmission line. The outer conduc-
tors form ohmic contacts to the n+ doped region
of the optical waveguide, whereas the center
conductor is fabricated as a Schottky contact.
The depletion layer underneath the center con-
ductor separates the photo-generated electron-
hole pairs in the absorption region.
Since the travelling-wave photodetector uses
the interaction of optical and electrical waves,
the optical and the electrical phase velocity
should be equal (phase-matching condition).
This can be achieved due to the fact, that the
Schottky-contact generates slow-wave effects
on the electrical transmission line [7].
The efficiency of the travelling-wave photo-
detector can be calculated numerically using a
distributed equivalent circuit model for genera-
tion and propagation of electrical waves on the
coplanar transmission lines (see Calculation of
the electrical millimeterwave generation of a trav-
elling-wave photodetector in this annual report).
A distributed current source describes the im-
pressed photocurrent per unit length due to elec-
tron-hole generation in the depletion layer. The
numerical calculation of the output power leads
3 RESEARCH12
to an electrical millimeterwave power of -
18.4dBm at 40GHz for a travelling-wave photo-
detector with an active length of 1mm. The opti-
cal wavelength is 1.3µm and the input power is
0dBm per optical carrier.
The electrical output power of the travelling-
wave photodetector has been measured using
an optical heterodyne setup. Two tunable 1.3µm
Nd:YAG-lasers generate an optical heterodyne
signal with beating frequencies in the millime-
terwave regime. The generated electrical sig-
nal is measured using a coplanar on-wafer probe
and a spectrum analyzer.
The fabricated travelling-wave photodetector
leads to an electrical output power of -19.7dBm
at a reverse bias voltage of 12V and a frequen-
cy of 40GHz using the heterodyne measurement
setup, which is in good agreement with the the-
oretically determined value.
Since the capacitance of this device is 1.2pF,
the resulting RC -time constant in a 50W sys-
tem would lead to a 3dB-frequency of about 2.5
GHz, which is far below the measured frequen-
cy. This shows the validity of the travelling-wave
concept to overcome RC-time limitation.
ConclusionFuture wireless millimeterwave communica-
tion systems using optical heterodyne tech-
niques are described. The requirements for high-
speed photodetectors are discussed. Since high
bandwidth and large electrical output power are
needed simultaneously, travelling-wave photo-
detectors are under investigation. Up to now,
an electrical output power of -19.7dBm at a fre-
quency of 40 GHz could be measured.
AcknowledgmentThe author would like to thank U. Auer (Fach-
gebiet Halbleitertechnik/-technologie) for grow-
ing the epilayer and for fabrication of the travel-
ling-wave photodetector.
References[1] R. Heidemann, R. Hofstetter, H. Schmuck,
60GHz fibre-optic distribution technology for traf-
fic information and multimedia, IEEE, MTT-S
and LEOS Topical Meeting on Optical Microwave
Interactions, Proc. pp. 133-136, Abbaye des
Vaux de Cernay, 1994
[2] J. Park, K.Y. Lau, Millimetre-wave (39GHz) fi-
bre-wireless transmission of broadband multi-
channel compressed digital video, Electron.
Lett., pp. 474-476, Vol. 32, 1996
[3] D. Wake, C.R. Lima, P.A. Davies, Transmission
of 60-GHz signals over 100km of optical fibre
using a dual-mode semiconductor laser
sources, IEEE Photon. Techn. Lett., pp. 578-
580, Vol. 8, 1996
[4] E. Boch, High bandwidth mm-wave indoor lo-
cal area networks, Microwave Journal, pp. 152-
158, 1996
[5] K. Kato, A. Kozen, Y. Muramoto, Y. Itaya,
T. Nagatsuma, M. Yaita, 110-GHz, 50%-effi-
ciency mushroom-mesa waveguide p-i-n photo-
diode for a 1.55-µm wavelength, IEEE Photon.
Techn. Lett., pp. 719-721, vol. 6, 1994
[6] D. Jäger, Optical Information technology, ed.
S.D. Smith and R.F. Neale, Springer-Verlag, pp.
328-333, 1193
[7] D. Jäger, Slow-wave propagation along variable
Schottky contact microstrip line, IEEE Trans. Mi-
crowave Theory and Techn., pp. 566-573, vol.
24, 1976
3.1 Optical Networks 13
InP
InGaAlAs : Si
InGaAs SQW/MQWInAlAs
Fig. 1: Travelling-wave photodetector layer structure.
3.1.2 Fabrication and characteriza-tion of a travelling-wave photodetec-tor
V. WENDRIX AND M. ALLES
ecently, high-speed travelling-wave
photodetectors are under investiga-
tion as optoelectronic power converters used
in future communication systems. In this
work the fabrication of high-speed travelling-
wave photodetectors is described. The fab-
ricated devices are characterized using
standard measurement techniques. The elec-
trical millimeterwave generation is deter-
mined using an optical heterodyne setup with
two 1.3µm Nd:YAG lasers.
IntroductionFor future communication systems combin-
ing fiber optic links with wireless transmission
techniques, high-speed photodetectors are
needed. These photodetectors should be used
for hybrid integration with millimeterwave am-
plifiers. A taper at the output end of the photo-
detector facilitates flipchip- or wire-bonding with
additional devices. The fabrication of travelling-
wave photodetectors with tapers is described in
this work [1].
Fabrication of travelling-wavephotodetectors
For the fabrication of a travelling-wave pho-
todetector several etch steps, a polyimide step
and metalization steps are needed [2]. The MBE-
grown wafers, which contain the necessary lay-
ers are depicted in Fig. 1 are fabricated in the
Department of Optoelectronics. The travelling-
wave photodetector contains in general three
different layers grown on a semi-insulating InP
wafer.
An InGaAlAs layer acts as optical waveguide,
an InGaAs quantum well layer provides optical
absorption, and, finally, an InAlAs layer is used
as a cladding layer. The metalization of the trav-
elling-wave photodetector is fabricated as an
electrical coplanar waveguide. The center con-
ductor forms a Schottky contact to the InAlAs -
layer. The outer metalization is evaporated on
the InGaAlAs-layer. This metalization is alloyed
in order to form ohmic contacts to the n-doped
semiconductor. The taper is fabricated at the
output end of the travelling-wave photodetec-
tor. In order to reduce millimeterwave attenua-
tion and for a characteristic impedance of 50W,the metalization of the taper has to be fabricat-
ed directly on the semi-insulating InP-wafer.
Therefore, an insulation of the edge of the me-
sas is necessary to prevent a short circuit be-
tween the center conductor and the outer met-
alization.
The processing of the travelling-
wave photodetector starts with two
etch steps. The etching defines the
lateral dimension of the optical
waveguide and of the absorbing lay-
er, cf. Fig. 2.
3 RESEARCH14
R
InP (b)(a)
Fig. 2: Etching of the two mesas, (a) cross-section, (b)
top-view.
InP (a) (b)
Fig. 3: Polyimide step, (a) cross-section, (b) top-view.
InP (a) (b)
Fig. 4: Evaporation of the metalization for the ohmic
contacts, (a) cross-section, (b) top-view.
InP (a) (b)
Fig. 5: Processing of center conductor and taper, (a)
cross-section, (b) top-view.
The fabrication of the mesa structure is done
using wet chemical etching with a liquid etchant
consisting of H3PO4:H2O2:H2O (1:1:40). This
etch system has almost no effect on InP.
The insulation of the mesa edge is done us-
ing a non-conducting polyimide step (Probimid
408), Fig. 3. The polyimide is processed
as a negative light sensitive resist and de-
veloped with a polyimide developer. A hard
bake process at 350°C makes the polyim-
ide resistant for the following process
steps.
In the next step, the metalization for the
ohmic contacts is evaporated on top of the
n-doped InGaAlAs-layer, Fig. 4. The met-
alization consists of Ge (30nm), Ni (5nm),
and Au (300nm). The metallic layers are
alloyed at about 550°C. This leads to ohm-
ic contacts with low impedances between
the metalization and the semiconductor.
The fabrication of the travelling-wave
photodetector finishes with the metaliza-
tion of the center conductor and the taper,
Fig. 5. Since the center conductor should
form a Schottky contact to the semicon-
ductor, it is necessary to evaporate Pt/Ti/
Pt/Au or Cr/Au.
MeasurementsThe characterization of the travelling-
wave photodetector is first done using cur-
rent-voltage and capacity voltage mea-
surements. The current-voltage
measurement gives information about the
functionality of the Schottky-contact diode
formed between the center conductor and
the outer metalization. As can be seen
from Fig. 6, the fabricated devices show
the typical current-voltage characteristic of a di-
ode.
In forward direction the current raises expo-
nentially with increasing voltage. With reversed
bias, the diode shows a high dark current which
increases with the voltage applied to the device.
The build-in voltage of this diode is about 0.6V.
3.2 Optical Networks 15
U (V)-2 -1 0 1 2
0.001
0.01
0.1
1
10
100
Fig. 6: Current-voltage measurement of a travelling-
Spektrum-analyzer
lens DUT
on-waferprobe
Multi-meter
Laser 2
Laser 1
lens
Bias-Tee
Fig. 7: Heterodyne measurement setup using monomode fibers.
The capacitance-voltage measurements al-
lows the determination of the doping level of the
fabricated structures. The capacitance C of a
depletion layer is given by:
C AN
U UkT
q
r D
B
=− − −
ε ε0
2
with the area of the depletion layer A, the
dielectric constant er e0, the density of do-
nor dopants ND, the build-in voltage UB, the
applied bias voltage U, Boltzmanns con-
stant k, the temperature T, and the elec-
tron charge q.
If the capacity is known, it is possible to
calculate the density of donor dopants:
( )Nq A
U
CD
r
= − 2
102 2ε ε
d
d
The capacity-voltage measurement
leads to a doping level of about 1018cm-3
in the InGaAlAs-layer and less than
1017cm-3 in the InGaAs and the InAlAs-lay-
er.
Optical heterodyne setupFor the measurement of the optoelectronic
conversion efficiency, a heterodyne setup with
two Nd:YAG-lasers operating at 1.3µm is used.
The wavelength of the lasers is adjustable by
detuning the temperature of the laser head. Up
to now, the two lasers illuminate the travelling-
wave photodetector via free space and a micro-
scope lens. To facilitate
the optical coupling to the
photodetector the use of
optical fibers has been
investigated. This mea-
surement setup is shown
in Fig. 7.
Each laser is coupled
to a monomode fiber.
Both fibers are coupled
to a third monomode fi-
ber using a GRIN-lens.
Finally, this fiber is direct-
3 RESEARCH16
L’
R’
C’ I’
G’C’
Vdc
m
G’ C’L
bb
hl
R’
prlz rlz
Fig. 1: Equivalent circuit model of the travel-
ling-wave photodetector. All elements are per
unit length.
I
ly coupled to the device under test (DUT).
The electrical measurement of the optoelec-
tronically generated millimeterwave is achieved
using a coplanar on-wafer probe, a bias-tee to
separate the high-frequency and the dc-signals,
a multimeter for photocurrent, and an spectrum
analyzer to measure the amplitude of the milli-
meterwave in frequency domain.
ConclusionIn this report, the fabrication of travelling-wave
photodetectors is described. Photodetectors
have been processed successfully. Character-
ization of these devices is done using current-
voltage and capacity-voltage measurements.
Finally, the heterodyne measurement setup has
been investigated in order to facilitate optical
coupling to the photodetectors using monomode
fibers.
References[1] V. Wendrix, Fabricatie en karakterisatie van een
Traveling-Wave Photodetector, Diploma thesis,
Department Toegepaste Natuurkunde, Vrije
Universiteit Brussel in cooperation with
Fachgebiet Optoelektronik, Gerhard-Mercator-
Universität Duisburg, 1996
[2] R. Haupt Experimentelle Untersuchungen zur
Integration von Schottky-Kontakt-Varaktordioden
für den Einsatz in periodischen Leitungs-
strukturen, Graduate thesis, Fachgebiet
Optoelektronik, Gerhard.Mercator-Universität
Duisburg, 1995
3.1.3 Simulation of the microwavegeneration of a travelling-wave pho-todetector
A. LUEDEKE AND M. ALLES
n this report the photoelectronic micro-
wave generation of a coplanar In-
GaAlAs/InP Schottky-contact travelling-wave
photodetector (TWPD) is analyzed by numer-
ical solution of the wave-equations. Compu-
tational simulations of the electrical behavior
of the travelling-wave photodetector are car-
ried out using an equivalent circuit model.
IntroductionThe electrical behavior of the travelling-wave
photodetector can be described using the equiv-
alent circuit model of Fig. 1 [1]. The travelling-
wave photodetector is fabricated as an electri-
cal millimeter waveguide. Therefore all elements
are per unit length.
3.1 Optical Networks 17
∂
u3
i B
I z dzp’ ( )
Y dz3’
u1
u2Y dz2’
I z dzp + ’ ( )i A
Y dz1’
i A
W dz’i
u u uz
+ ∂∂
i iz
dz+ ∂
Fig. 2: Summarized equivalent circuit model of the travelling-wave photodetector. All elements are per
unit length.
The impedances Rm and Rhl describe the
longitudinal ohmic losses in the metalization and
the semiconductor, respectively. The inductance
of the electrical waveguide is taken into account
with the inductance L. The conductance Grlzand the capacity Crlz consider the Schottky-con-
tact depletion layer. The two elements Cb and
Gb describe the behavior of the bulk material.
An additional capacity CL is introduced for the
electric field in air above the photodetector. The
impressed current source IPh describes the opto-
electric conversion in the absorbing quantum
well layer.
Numerical solutionThe wave-equations of the travelling-wave
photodetector are derived by using a summa-
rized equivalent circuit model, shown in Fig. 2,
which based on the model above. The equation
of the complex voltage amplitude along the
transmission line can be shown as
∂∂
2
21 2 1 3 2 3
1 2
u
zu W
Y Y Y Y Y Y
Y Y= ⋅ ⋅ ⋅ + ⋅ + ⋅
+’
’ ’ ’ ’ ’ ’
’ ’
2
1 2
I zW Y
Y Yp+ ⋅ ⋅
+’
’ ’
’ ’( ) (1)
where W ist the impedance and Y1, Y2 and Y3the admittances of the transmission line. Ip(z)
is given by
I zq R P
hpopt opt opt’ ( )
( ) =
⋅ ⋅ − ⋅ ⋅⋅
η αν
1
j zopt opt( )e⋅ − + ⋅α β
(2)
where hopt is the internal quantum efficiency, aopt
is the optical absorption coefficient, hn is the
photon energy, Popt is the incident light power,
R is the reflection of interface device/air and bopt
is the optical phase coefficient.
3 RESEARCH18
0 0,2 0,4 0,6 0,8 1z (mm)
0
5
10
15
20
25
30
35
40
Vo
ltage
(m
V)
Fig. 3: Voltage distribution along z direction for a simplified
(interrupted line) and the fully equivalent circuit model.
Using the finite differences method [2], the
numerical solution of equation (1) is given by
u
u u h I zW Y
Y Y
h WY Y Y Y Y Y
Y Y
k
k k p k
=+ − ⋅ ⋅ ⋅
+
+ ⋅ ⋅ ⋅ + ⋅ + ⋅+
+ −1 12 2
1 2
2 1 2 1 3 2 3
1 2
2
’’ ’
’ ’
’’ ’ ’ ’ ’ ’
’ ’
( )
(3)
with
hb a
N = −
(4)
where a is the beginning and b the end of the
transmission line. N is the number of discrete
points, where the voltage is calculated
(0 £ k £ N).
The loads at the two ends of transmission line
are defined as Z1and Z2, the characteristic re-
sistance of the transmission line is Z. According
to microwave theory there exist reflections r1 and
r2 at the two ends z = 0 and z = l of the device.
The boundary condition for this problem is de-
termined by the current voltage relation at the
specific load resistance.
u z u( )= =0 0
[ ]Z
W
r
r hu u’= ⋅
+−
⋅ ⋅ − −1
11
21
11 1
(5)
u z l uN( )= =
[ ]Z
W
r
r hu uN N’= − ⋅
+−
⋅ ⋅ −+ − 1
11
22
21 1
(6)
The solution of equation (3) is computation-
ally calculated.
SimulationFor the simulation of photoelectrical micro-
wave generation a frequency of 40GHz is tak-
en. The wavelength of the optical sources is
1.3µm and the light power of the two beams are
both 1mW. In the following calcu-
lation the quantum efficiency hopt
of photoelectrical conversion is
assumed to be 1 and the incident
light energy is fully and uniformly
coupled to the active layer, so the
optical reflection R is 0.
Fig. 3 shows the voltage dis-
tribution along z direction. The
reflection factor at z = 0mm is 1
and at z = 1mm the reflection is
0.
The dashed line shows the re-
sult by using a simplified equiva-
lent circuit model, in which Cb, Gband CL are neglected. An exist-
ing simulation program, which
calculates the solution analytical-
ly, is based on this model. The
3.1 Optical Networks 19
0 0,1 0,2 0,3 0,4 0,5z (mm)
0
10
20
30
40
50
60
Vo
ltag
e (m
V)
Fig. 4: Voltage disribution along z direction for Z2 = 50W
(dashed line) and Z2 = Z (solid line) for a device length of
comparison with the numerical simulation, which
is based on the complete equivalent circuit mod-
el, shows, that the efficiency is reduced by 10%.
One possibility to optimize the device rela-
tive to the output-voltage is to reduce the length
of the transmission line. In the following simula-
tion the length of the device is set to 0.5mm.
Fig. 4 shows the results by using Z 2 = Z (solid
line) and Z2 = 50W (dashed line). In case of ad-
aptation the efficiency is raised by 44% relative
to a device length of 1mm. In case of mismatch-
ing there is a high voltage at the end of the de-
vice, but the efficiency has not raised because
of higher load at the device´s end.
Another possibility of optimization is to reduce
the value of Rm. Simulations have proved that
the efficiency is raised by 20%, if Rm is set at
0W (device length is 1mm).
ConclusionThis report presents the nu-
merical solution of the photoelec-
trical microwave generation of a
coplanar travelling-wave photo-
detector. The wave equations are
solved by the finite differences
method. The results of the simu-
lations have shown, that the effi-
ciency is reduced by 10% relative
to using a simplified equivalent
circuit model and that a further
optimization is possible.
References[1] D. Jäger, R. Kremer, Trav-
elling-wave optoelectronic devices
for microwave applications, Proc.
IEEE, MTT-S and LEOS Topical
Meeting on Optical Microwave Interactions, pp.
11-14, 1994, France
[2] D. Marsal, Finite Differenzen und Elemente:
numerische Lösung von Variationsproblemen
und partiellen Differentialgleichungen, Springer-
Verlag, Berlin/Heidelberg, 1989
3.1.4 Determination of RF-equivalentcircuit elements of travelling-wavephotodetectors using network analy-sis
O. BERGER AND M. ALLES
n this report, the determination of the
elements of the equivalent circuit model
of coplanar waveguides is described. Since
the longitudinal and the transverse complex
3 RESEARCH20
I
Fig. 1: Equivalent circuit model of the travelling-wave photode-
tector.
impedance of the equivalent circuit model
can be calculated from the characteristic im-
pedance and the propagation coefficient
measured with a network analyzer, it is pos-
sible to compute the equivalent circuit
straight forward. This method has been used
to determine the equivalent circuit model of
travelling-wave photodetectors. Results and
comparison to theory and other measure-
ments are shown.
Introduction
Recently, 60GHz-travelling-wave photodetec-
tors are under development in the Fachgebiet
Optoelektronik. For characterization and further
optimization of the device, network analyzer RF-
measurements are analyzed in a new way in
order to determine the high-frequency equiva-
lent circuit elements of the device. The imple-
mented method determines the equivalent cir-
cuit model directly from network analyser mea-
surements.
Equivalent Circuit ModelThe coplanar waveguide structure of the trav-
elling-wave photodetector can be described us-
ing the distributed equivalent circuit model
shown in Fig.1 [1]. Note that all elements are
per unit length. The impedances RM and R
describe the longitudinal ohmic losses in the
metalization and the semiconductor, respective-
ly. The inductance of the electrical waveguide
is considered by the inductance L. The deple-
tion layer of the Schottky-contact is taken into
account with the conductance G and the ca-
pacitance C while GB and CB characterize the
bulk material of the semiconductor. The addi-
tional capacitance CL is associated with the
electrical field in air above the device. Finally,
the impressed current source IPh is introduced
to describe the optoelectronic conversion with-
in the absorbing layer.
To determine the equivalent
circuit elements it is necessary to
make some simplifications. Usu-
ally, network analyser measure-
ments take place without optical
illumination of the device, there-
fore IPh can be neglected. The
two capacitances CL and CBhave much less influence on the
behavior of the device in compar-
ison to the Schottky-capacitance
C and can be neglected.
One can conclude, that in the
longitudinal and in the transverse
part of the equivalent circuit three
elements are to be considered:
two real impedances and one
imaginary one. The longitudinal
3.1 Optical Networks 21
theory low-frequency high-frequencymeasurements measurements
RmmM’Ω
6.58 6.21 13.8
Rmm
’Ω
707 - 887
GS
mm’
6.7×10-14 750×10-12 1.3×10-3
GS
mmB’
874 1.1 2.54
LnHmm
’
0.466 - 0.62
CpF
mm’
1.39 1.17 0.91
Tab. 1: Values for the elements of the equivalent circuit model
of a travelling-wave photodetector.
and the transverse part of the
equivalent circuit can be de-
scribed separately with
WR R j L
R R j LM
M’
’( ’ ’)
’ ’ ’= +
+ +ωω
and
.Y
G G j C
G G j CB
B’
’ ( ’ ’)’ ’ ’
= ++ +
ωω
Using separate Smith charts
for the longitudinal and the trans-
versal part of the equivalent cir-
cuit to display the frequency de-
pendence of W and Y,
semicircles with two intersections
with the real axis for ω = 0 and
ω → ∞ can be determined. The
intersections with the real axis are
ZR R
R RM
M’( )
’ ’’ ’
ω = =+
0 ,
Z R’( ) ’ω → ∞ =
for the longitudinal part and
YG G
G GB
B’( )
’ ’’ ’
ω = =+
0 ,
Y G B’( ) ’ω → ∞ =
for the transverse part.
Determination of the equivalent circuitelements
The network analyser measures S-parame-
ters from a device under test (DUT). The char-
acteristic impedance Z and the propagation co-
efficient g are determined from this data.
The new method calculates the impedance
of the longitudinal part and the admittance of
the transverse part using the relations W = g Zand Y = g / Z. The frequency dependence of
both parts is displayed in separate Smith charts.
An statistic-based algorithm fits a semicircle to
the measured data. The intersections with the
real axis are used to determine the real elements
of the equivalent circuit model. With these re-
sults, it is possible to calculate the imaginary
elements L and C for each frequency. This
method has been implemented to an easy-to-
handle windows-program with graphic features.
The calculation results are displayed instantly
on-screen.
ResultsThe equivalent circuit elements calculated
using this method have been compared with
results of low-frequency measurements and the-
oretical determined values [2], Tab. 1. As can
3 RESEARCH22
0 10 20 30 40
frequency (GHz)
40
30
20
10
0
−10
−20
−30
reZ
imZ
Fig. 2: Characteristic impedance of a travelling-wave
photodetector.
4
3
2
1
0
4
3
2
1
00 10 20 30 40
frequency (GHz)
Fig. 3: Phase coefficient and attenuation coefficient of a
travelling-wave photodetector.
be seen from this table, the measure-
ments are in good agreement with the-
oretically determined values. Only the
two admittances show different values.
In case of G the network analyzer mea-
surement leads to a higher value. An
explanation is, that the accuracy of the
network analyzer makes it impossible
to measure these low admittances.
The theoretical value of the bulk ad-
mittance GR is much larger than the
measured values. A reason is, that the
metal-semiconductor resistance is ne-
glected in the derivation of GB.
Taking measurements at various
bias voltages one can see that only the
elements C and G regarding the de-
pletion layer show major bias depen-
dence while the other ones keep con-
stant over the bias voltage.
Fig.2 shows the characteristic im-
pedance, Fig. 3 the phase and attenu-
ation coefficient calculated from the
measured data. The real part of the
characteristic impedance rises at fre-
quencies above 35GHz indicating that
probably the major part of the electri-
cal quasi-TEM wave on the structure
changes from TEM to TM.
The smithcharts with the measured
data and the semicircle calculated as
an approximation for the longitudinal
part and the transversal part of the
equivalent circuit are shown in Fig. 4.
As is visible, the frequency dependence
of the measured data leads to semicir-
cles, indicating that the device under
test can be described with the simpli-
fied equivalent circuit model described
above.
3.1 Optical Networks 23
(a)frequency
(b)
frequency
Fig. 4: Smithcharts for the longitudinal part (a) and for the transversal part (b).
frequency (GHz)
0 10 20 30 40
1.0
0.8
0.6
0.4
0.2
0.0
1.0
0.8
0.6
0.4
0.2
0.0
Fig. 5: Inductance and capacitance versus frequency.
The frequency dependence of the inductance
L and the capacitance C, shown in Fig. 5, can
be determined directly with the program. Due to
a suggested change of the quasi-TEM-Mode to
a TM-mode, the inductance rises and the ca-
pacitance decreases with higher frequencies.
ConclusionsA method to determine the RF-equivalent el-
ements directly from network analyser measure-
ments has been developed. This method ap-
proximates the frequency dependence of the
measurement data in a Smith chart graphically
without any further knowledge about the values
to be determined. The measured results fit well
with both theoretically determined and low-fre-
quency measured values. It is also possible to
examine the frequency dependence of the in-
ductance and the capacitance.
AcknowledgementThe author would like to thank U. Auer (Fach-
gebiet Halbleitertechnik/-technologie) for grow-
ing the epilayer and for fabrication of the travel-
ling-wave photodetector.
References[1] M. Alles, T. Braasch, D. Jäger,
High-speed coplanar Schottky trav-
elling-wave photodetectors, Int. Conf.
on Integrated Photonics Research,
Proc. pp. 380-383, Boston, USA, 1996
[2] O. Berger, Bestimmung der HF-
Ersatzschaltbildelemente von Photo-
detektoren mit Hilfe der
Netzwerkanalyse, Graduate thesis,
Fachgebiet Optoelektronik, Gerhard-
Mercator-Universität Duisburg, 1997
3 RESEARCH24
3.1.5 Polarization insensitivewaveguide modulators on InP
T. ALDER AND R. HEINZELMANN
lectroabsorption modulators using
strained multiple quantum well (MQW)
structure have been designed, fabricated and
characterized. Utilizing the Quantum Con-
fined Stark Effect (QCSE) due to high elec-
tric field underneath a Schottky-electrode,
the absorption coefficient of the optical
waveguide can be changed. The use of
strained quantum wells enables an operation
of the device, with almost no sensitivity to
different polarisation. With this device an on/
off-ratio of 18.5dB has been achieved.
IntroductionIn general, the absorption change in a MQW
structure is strongly polarization dependent [1].
From the viewpoint of system applications, a
polarization insensitive or at least polarization
independent modulator is desirable. Appropri-
ate structures can be designed using quantum
wells with tensile strain [2-3]. In this paper elec-
troabsorption waveguide modulators using a
strained InGaAs/InAlAs MQW structure in the
electrooptical active region will be presented.
Device structure and principle of operationA schematic diagram of the modulator struc-
ture is shown in Fig. 1. The modulators investi-
gated utilize a nin-structure containing Si-doped
InAlAs top and bottom cladding layers with thick-
ness of 570nm and 1120nm, respectively.
3.1 Optical Networks 25
Fig. 1: Schematic diagram and cross section of the modulator.
E
0V -2V -4V -6V -8V -10V
position position position position position position
Fig. 2: Nearfield-pattern and lateral profile of
the optical waveguide mode at different
reverse biases.
-10 -8 -6 -4 -2voltage [V]
0
20
40
60
80
100
tran
smis
sion
[%]
= 1,2mm
w = 14µm
Fig. 3: Transmission as a function of different
reverse biases.
The doping concentration of the bottom clad-
ding layer is ND = 1×1017cm-3, whereas that of
the top cladding layer is ND = 1×1016cm-3. The
non intentionally doped guide consists of 19 ×
6nm thick InGaAs MQWs separated by
19 × 7.7nm thick InAlAs barriers. The structure
was grown using the MBE machine of the De-
partment of Optoelectronics. To examine the
dependence of the device behaviour on contact
geometry, a number of devices were fabricated
with chrome-gold-Schottky-electrodes of differ-
ent width, ranging from 8µm to 16µm. The Schot-
tky-electrodes were manufactured by thermal
evaporation of chrome and gold in ultra high
vacuum. The waveguide structure was formed
using wet chemical etching after evaporation of
the Schottky contacts. The second contact,
shown in Fig. 1 is carried out as Ohmic contact
and was manufactured by thermal evaporation
of germanium, nickel and gold in ultra high vac-
uum.
As a reverse bias is applied to the Schottky-
electrode, there will be a high electric field un-
derneath the Schottky-contact within the deple-
tion region. This increases the absorption
coefficient of the guide due to the quantum con-
fined Stark effect (QCSE). In this way, the opti-
cal output power can be controlled electrically.
Experimental resultsFig. 2 shows the nearfield-pattern and the
lateral profile of the optical waveguide mode at
different reverse biases. From this figure, it can
be seen, that the output becomes weaker as
the applied reverse bias is increased. This be-
havior is due to the increased absorption coeffi-
cient in the optically guiding region.
As the previous result shows, it is possible to
control the optical output power by a reverse
bias. In the following, systematic results on trans-
mission changes will be presented. In Fig. 3 the
transmission is plotted as a function of different
reverse biases. From this figure, it can be seen,
that within the range from -4.3V to -7.4V the
transmission changes almost linear with the
applied bias. The ratio between maximum and
minimum transmission is 18.5dB. Furthermore
it can be seen, that the change in transmission
from 0V to about -4V is very low. This indicates
that quantum wells were grown smaller than they
were designed.
Fig. 4 shows the transmission as a function
of reverse bias for TE- and TM-polarization. It is
evident from the figure, that for TE- and TM-po-
3 RESEARCH26
7 9 11 13 15 17 contact width [µm]
0
4
8
12
16
20
max
imum
tra
nsm
issi
on c
hang
e [d
B]
TE - polarisation
TM - polarisation
Fig. 5: Maximum transmission change as a function of different
reverse biases for TE- and TM-polarization.
-10 -8 -6 -4 -2voltage [V]
0
0,2
0,4
0,6
0,8
1
tran
smis
sion
[a.
u.]
18,5 dB
17,2 dB
TE - polarisation
TM - polarisation
Fig. 4: Transmission as a function of different reverse biases for
TE- and TM-polarisation.
larization the change in transmis-
sion is almost equal. While for TE-
polarization an on/off-ratio of
18.5dB could be measured a
slightly smaller value of 17.2dB
appeared for TM-polarized light.
Considering only the linear
range (the voltage range be-
tween -4.3V and -7.4V bias volt-
age) of the transmission charac-
teristic, for TE- as well as
TM-polarization an on/off-ratio of
8.2dB is measured.
To examine the dependence
on device dimensions modulators
with different Schottky-electrode
widths from 8µm to 16µm were
investigated. The results are
shown in Fig. 5. As can be seen,
there is no recognizable influence from contact
width on the maximum transmission change, nei-
ther for TE-polarization nor for TM-polarization.
This is of major importance, as with a change in
the contact width the propagation
properties of the electrical
waveguide can be fit to those of
the optical waveguide, for high-
speed operation the travelling-
wave concept [4] can be applied.
ConclusionsElectroabsorption waveguide
modulators based on strained In-
GaAs/InAlAs-MQW have been
designed, fabricated and charac-
terized. A maximum on/off-ratio of
18.7dB has been achieved. It
could be shown, that the polari-
sation influence on the transmis-
sion behaviour was small, due to
the influence of the strain in the
quantum well region. Additionally
3.1 Optical Networks 27
no influence of the contact dimensions on the
transmission was observed.
References[1] T. Aizawa, K. G. Ravikumar, R. Yamauchi,
Polarisation Independent Refractive Index
Change In InGaAs/InGaAsP Tensile Strained
Quantum Well , Electronics Letters, Vol. 29, No.
1, pp. 21 - 22, January 1993
[2] H. W. Wan, T. C. Chong, S. J. Chua, Consider-
ations For Polarisation Insensitive Optical Switch-
ing and Modulation Using Strained InGaAs/
InAlAs Quantum Well Structure, IEEE
Photonics. Techn. Lett., Vol. 3, No. 8, pp. 730 -
732, August 1991
[3] J. Shimizu, T. Hiroshima, A. Ajisawa, M.
Sugimoto, Y. Ohta, Measurement of the
polarisation dependence of field induced refrac-
tive index change in GaAs/AlAs multiple quan-
tum well structures, Appl. Phys. Lett., Vol. 53,
No. 2, pp. 86 - 88, 1988
[4] D. Jäger, R. Kremer, and A. Stöhr, Travelling-
wave optoelectronic devices for microwave ap-
plications, IEEE MTT-S 1995 International Mi-
crowave Symposium, Vol. 1, pp. 163-166, 1995
(invited paper)
3 RESEARCH28
Fig. 1: Signal and energy transmission into the eye
3.2 Optical Interconnects andProcessors
3.2.1 Neurotechnology: Retina Im-plant
M. GROSS AND R. BUSS
he Department of Optoelectronics is
a member of a consortium of 14 Ger-
man expert groups, working on the project
EPI-RET: Retina Implant. This interdiscipli-
nary project, funded by the Federal Ministry
for Education, Science, Research and Tech-
nology (BMBF) in Germany, is developing a
retina implant.
This device is a neural prosthesis, designed
for patients blinded by a disease where the
outer retinal layer degenerates (retinitis pig-
mentosa or macula degeneration). It consists
of three parts: a so-called retina encoder (RE)
outside the eye, simulating the function of
the retina, the retina stimulator (RS), a mi-
crochip placed on the retina with electrodes
stimulating the ganglion cells in the outer
retina layer, and a wireless signal and ener-
gy transfer from RE to RS.
The task of the Department of Optoelectron-
ics in this project is the development of a
device for optoelectronic signal and energy
transfer into the eye. To achieve this, a pro-
totype consisting of a laserdiode as trans-
mitter and a receiver consisting of a
monolithically integrated photovoltaic cell
array and a photodiode, together with driv-
ing and receiving electronics, was manufac-
tured.
IntroductionThe signal and energy transmission line de-
scribed here is part of a technical system func-
tioning as a vision aid for blind people who have
lost their vision due to retinal degenerations,
especially retinitis pigmentosa [1]. An often ap-
pearing kind of blindness is the partially degen-
eration of the retina, e.g. the disease retinitis
pigmentosa, which is leading to blindness
through following steps: The typical begin is the
loss of the rod photoreceptors, causing night
blindness. Next the cone photoreceptors are
3.2 Optical Interconnects and Processors 29
T
dying off, beginning at the outer perimeter of vi-
sion. This leads to a tunnel vision and finally to
total blindness, when the cones in the fovea are
lost.
However, while the photoreceptors are dying
off, the nerve cells in the retina and subsequent
parts of the central visual system are remaining
mostly intact [1]. This leads to the possibility of
developing a visual prosthesis with the ability to
replace main parts of the retina that gives sight
back to the visually impaired [2].
System description The whole system sketched in Fig. 1 con-
sists of three main parts:
> a retina encoder (RE), consisting of a CMOS
camera and an artificial neural network (en-
coder) for image data processing,
> a retina stimulator (RS), a flexible chip, epiret-
inal affixed, with µ-electrodes on the back,
> a wireless signal and energy transmission line
from the retina encoder to the retina stimula-
tor.
The system works as follows: First the high
dynamic range CMOS camera generates a pic-
ture. This dataset is then reduced by an artifi-
cial neural network and transformed into digital-
ly coded pulse trains (nerve signals), to which
the ganglion cells can react. This corresponds
to the data reduction from 120 million photore-
ceptors to 1 million ganglion cells by the human
retina. The dataset is then optically transmitted
at a rate of 1 Mbit/s to a microcontact foil on the
retina (receiver), where eye movements of up
to +/- 15°, measured from looking straight ahead
have to be compensated. Together with this in-
formation transfer an optical bias is transmitted,
supplying the driving circuit with 5 mW electri-
cal power. The retina stimulator is a soft micro-
contact foil which is implanted adjacent to the
ganglion cell layer on the outer retinal limit. The
µ-electrodes are stimulating the ganglion cells
of the retina, thus transmitting the signals via
the optic nerve to the visual cortex in the brain.
ResultsThe signal and energy transmission has been
realized in a first prototype using a laser diode
as transmitter and a photovoltaic cell array to-
gether with a photodiode as receivers for ener-
gy and signals, respectively. In the first step we
designed the parts for the optical transmission,
considering the boundary conditions given by
the human eye and the technical demands of
the whole device. For surgical reasons optical
fibres cannot be used to connect transmitter and
receiver directly. Therefore, a light source (e.g.
a pigtailed laser diode) has to be fixed in front of
the eye, transmitting the signal and energy onto
the retina, using free space optics. A micro-lens
system was developed mapping the light homo-
geneously on the retina in a spot with a diame-
ter of about 5 mm. This assures that the receiv-
er is illuminated for eye movements of up to +/
- 15°. The material used for the receiver is GaAs,
mainly to achieve high conversion efficiencies
with the photovoltaic cell array [2]. This has sev-
eral advantages:
1. The fibre has an gaussian beam profile, while
laser diodes have strong astigmatism that has
to be corrected with a microoptic in front of
the eye.
2. The heat that the laser diode produces is led
away very easily.
3. The high frequency modulation of the laser
diode for the signal transmission is made far
away of the eye avoiding problems with elec-
tromagnetic compliance.
Latest results are shown in Fig. 2 and Fig.3:
Fig. 2 schematically depicts the system design.
3 RESEARCH30
Fig. 2: System design
In Fig. 3(a) the I-V characteristics of a single pho-
tovoltaic cell and of an array with 5 cells con-
nected in series are plotted. At a wavelength of
l = 800 nm this array delivers up to 5 mW elec-
trical power with a conversion efficiency of about
23%, which turns out to be a great improvement
as compared with the results published in [3]. It
should be noted, however, that this efficiency
was obtained without any antireflection coating.
Experiments have shown that the efficiency can
be increased up to almost 31% by encapsulat-
ing the cell array with a biocompatible antire-
flection coating consisting of SiO2/Si3N4 multi-
layers. Moreover, Fig. 3(b) shows
measurements of the signal transmission: The
signals at the output of the encoder, curve (1),
are transmitted optically into the eye at a rate of
1 Mbit/s. The output of the receiver is plotted in
curve (2) of Fig. 3(b) together with the recovered
clock, curve (3), in Fig. 3(b).
ConclusionIn this report the progress in the work for an
optoelectronic signal- and energy transmission
line for use in a visual prosthesis is presented.
A concept is developed and a prototype is de-
scribed. This prototype currently has the capa-
bility of delivering 5 mW electrical power together
with digitally coded signals at a rate of 1 Mbit/s
simultaneously, thus meeting the current sys-
313.2 Optical Interconnects and Processors
Fig. 3: (a) I-V characteristic of photovoltaic cells (PVCs), (b) digitally coded signals before (1) and
after (2) transmission, and (3) recovered clock signal.
tem requirements. However, the optical link de-
scribed here is capable of transmission rates up
to 1 Gbit/s, suitable for optically powered high-
speed data links.
AcknowledgementThe authors would like to thank the Federal
Ministry for Education, Science, Research and
Technology for financial support and all mem-
bers of the EPI-RET team for fruitful discussions.
References[1] R. Eckmiller Retina implants with adaptive retina
encoders, Proc. of the 1996 RESNA Research
Symp., Salt Lake City, pp. 21-24, 1996
[2] M. Groß, T. Alder, R. Buß, R. Heinzelmann, M.
Meininger, and D. Jäger, Micro Photovoltaic Cell
Array for Energy Transmission into the Human
Eye, Proc. of the 14th European Photovoltaic So-
lar Energy Conference, Barcelona, Spain, vol.
1, pp. 1165-67, 1997
[3] J. Rizzo, J. Wyatt, Silicon retinal implant to aid
patients suffering from certain forms of blind-
ness, Proc. of the 1996 RESNA Research
Symp., Salt Lake City, pp. 1-3, 1996
3.2.2 Analysis of the optical energyand signal transfer module for anartificial vision prosthesis
T. BAUMEISTER, M. GROSS, AND R. BUß
ithin the scope of the Retina Im-
plant Project supported by the Ger-
man government the possibility of a wireless
transfer of signal and energy into the eye of
human patients was analyzed.
IntroductionThe Retina Implant project was founded in
1995 as part of a young and interdisciplinary
area of research: Neurotechnology.
The goal of this ten year project is the devel-
opment of both an artificial eye implant (retina
stimulator), stimulating the ganglion cells of the
human retina from patients, who lost their eye-
sight due retinitis pigmentosa or macula degen-
eration and an encoder transforming the signals
coming from a video system like human retina
does.
One of the coming tasks is to develop a sys-
tem transporting the electrical power for this
implant and the signals from the output of the
encoder into the eye. This system analysis
shows the technical preferences for this trans-
port by IR-rays.
Criteria catalogueThe base of each scientific analysis is a cri-
teria catalogue being a decisive help for the
evaluation of possible alternatives. The follow-
ing criteria were found:
1. The fundamental criteria are the dimensions
of the implant. Due to surgical reasons the
maximum length is limited to 1.5 mm.
2. The Efficiency of the power transmission is a
criterion of great importance for any implant-
ed system, because of the absence of possi-
bility to cool any part of system inside the eye.
3. The reliability is fundamental too, because it
is nearly impossible to repair any failure and
the exchange of the whole system is more
dangerous for the patient as the first time im-
plantation.
4. The biocompatibility is one more very impor-
tant criterion for the long time function of any
3 RESEARCH32
W
implant. This may be given by a biocompati-
ble coating of any material, but there is the
risk of damage during the implantation and
fixation of the innerocular part. Further, we
should be aware of the influence of overgrow-
ing the receiving part of the signal and ener-
gy transport system.
5. The receiver can´t be placed at all places on
the retina, because the fixation of it may dam-
age axons of stimulated ganglion cells.
6. The technical availability and the need of
development of parts of the system are im-
portant due to the cost and the time to mar-
ket of the system.
7. The possibility of extension, especially the
number of stimulating electrodes, is an im-
portant criterion for the future.
8. The acceptance of the whole system by the
patient and his milieu is quiet important too.
System analysisFirst the need of bandwidth and power for the
stimulation of a given amount of stimulating elec-
trodes were analyzed. The first version of the
retina stimulator will consist of an
array of twenty stimulating elec-
trodes.
The bandwidth needed for the
stimulation with twenty electrodes
is nearly 100kbit/s, for 400 elec-
trodes we found a bandwidth of
approximately 25Mbit/s.
This calculation of the band-
width includes the scheme shown
in Fig. 1, the number of bits need-
ed to address the electrodes, the
number of bits needed for encod-
ing the stimulating pulseform, the
rate of neuro impulses, and fac-
tor needed for encoding the signal by wireless
transmisson.
The power needed using a single amplitude
modulated IR-laser for simultaneous transport
of energy and signals is 190mW of optical pow-
er in a worst case analysis with twenty electrodes
stimulating in one time frame. The worst case is
defined here by a constant electrical stimula-
tion power of 750µW at the electrodes.
Next, the possibilities of signal encoding were
analyzed. The whole signal of amplitude modu-
lated laser beam includes an AC signal, the en-
coded signal for stimulation and addressing the
electrodes, and a DC offset for the energy trans-
fer.
The CMI-code (see Fig.2) also known as 2-
AMI-1-code or modified FSK-code is the best
compromise between the technical expense of
signal and clock recovering and the need of a
DC free signal in this application. This coding is
mentioned with a factor of two in analysis of
bandwidth [1]. Considering this we found the
best way to transfer the energy and signals is to
3.2 Optical Interconnects and Processors 33
Fig. 1: Stimulation scheme
use a single amplitude modulated IR-laser di-
ode.
Analysis of the opticsTo avoid a complicated, heavy and cost in-
tensive eyetracking system focussing the laser-
beam exactly on the position of the energy and
signal receiving photodiode, we decided to illu-
minate an area on the retina great enough to
compensate eyemovements about 10° in each
direction. The position of this area is vertical
ahead the macula. At this position the beam is
least influenced by the eye lid and the greatest
eye movement possible. The best way to lower
the reflection of the beam at the iris and thereby
to expand the possible angle of movement is to
focus the beam in the hole of the iris. For a nearly
uniform illumination of area, needed for an uni-
form supply of power of the stimulation electron-
ic, we have developed a ring-shaped focus of
the beam. Thus, we solved the problem how to
3 RESEARCH34
Fig. 2: Calculation of the power needed
Fig. 3: Example of a DC-free encoded signal
added with an offset for transmision of power
get a uniform illumination on the screen of a good
imaging optical system i.e. human eye. To solve
this problem we simulated the whole optical path
with the ray tracing program ZEMAX SE for hu-
man eye and additional for the rabbit eye, see
Fig. 5. For the simulation of the human eye we
used a slightly modified model of the eye from
Helmholtz. We completed this model with the
axes of movement [2]. For the eye of the rabbit
we used a similar model [3].
The optics consists of a two lens beam ex-
pander and one complex collecting lens to get
the above mentioned ring shaped focus (see
Fig.5). The optics is not modular, i.e. it is not
possible to use parts of the optics designed for
the human eye in the optical system for the rab-
bit eye.
The optical source can be the end of an opti-
cal fiber or a laser diode with a micro lens that
corrects the astigmatism of the laser. The lens-
es used in our model are all in the range of to-
days industrial optics including one standard
micro lens.
ConclusionHere is shown, that, principally, it is possible
to transfer enough power and a signal with suf-
ficient bandwidth into the eye by using an IR
light beam. Further it is shown, that it is prefera-
ble to use a single amplitude modulated laser
diode according to the above mentioned cata-
logue of criteria. This paper shows that a trans-
port of power and signal from the signal pro-
cessing unit outside the eye to the implanted
microelectronics by optical means can be real-
ized without the need of an eye tracking sys-
tem.
3.2.3 Development of an optical sig-nal and energy transmission system
R. HEDTKE AND M. GROSS
n optical transmission system to pow-
er and provide a retinal implant with
energy and digitally coded information is de-
veloped.
IntroductionThe EPI-RET: Retina
Implant project is part of the
Neurotechnology Program
of the Federal Ministry for
Education, Science, Re-
search and Techno-
logy (BMBF). The implant
system is evolved in co-op-
eration with several interdis-
ciplinary project partners as
a vision aid for people who
3.2 Optical Interconnects and Processors 35
Fig. 4: Example of CMI encoding
Rabbit eye Human eye
Fig. 5: Optical analysis of the rabbit eye and the human eye
A
are suffering from retinal degenerative defects
like retinitis pigmentosa and macula degenera-
tion. With the help of this optical transmission
system the retinal implant is provided with the
needed information and energy.
PrincipleFor the transmission to the ret-
inal implant the information is
modulated onto the laser light,
which transmits the energy. The
kind of modulation that is used is
shown in Fig. 1. A continuos en-
ergy transmission into the implant
is guaranteed by modulating the
driving current around a mean
operating point (OP). The driving
current and the optical output-
power for the laser diode is
sketched for an stimulation with
a rectangular-signal.
TransmitterThe driving current, as shown
in Fig.2, is regulated by a current
control unit to secure a constant
optical output power of the laser diode. This con-
trol unit compares the preset current value with
the actual and provides the control signal for the
driving unit. To generate the actual current val-
ue a photodiode integrated into the laser diode
[1] (alternatively: an external shunt-resistor) is
used. With the help of a signal conversion the
measured signal is processed for the following
current control unit. The modulation of the driv-
ing current occurs in the previously described
way.
ReceiverThe modulated laser light hits the photodiode
and the photovoltaic cell array (see Fig.3). The
photodiode generates the detector signal which
is transformed to TTL-level by a signal process-
ing unit. A clock recovery unit is used to gener-
ate the clock signal. The other part of the laser
3 RESEARCH36
DRIVING CURRENT (I )I th
TIME
TIM
EOP
TIC
AL
OU
TP
UT-
PO
WE
R (
PO
PT )
I
PO
PT
23
Fig. 1: Principle of the modulation of the
laser diode
LASERDIODE
CURRENT ADJUSTMENTCURRENTCONTROL
DRIVINGUNIT
OPTICAL OUTPUT-POWER
PHOTODIODE
AND
SIGNALCONVERSION
SIGNAL MODU-LATOR
DR
IVIN
G C
UR
RE
NT
+
ALTERNATIVE GENERATION OFACTUAL CURRENT VALUE
RESISTORAND
SIGNALCONVERSION
MO
DU
LAT
ION
CU
RR
EN
T
ACTUAL CURRENT VALUE
Fig. 2: Block diagram of the transmitter
light is converted into electrical energy by the
photovoltaic cell array [2].
Measured resultsFig. 4 shows the driving current of the laser-
diode for a stimulation with a 1 MHz rectangular
wave. The used mean operating point is 400
mA [3] or rather 190 mW optical power. To code
the signal the Manchester code [4] is used. The
main advantage of this coding are the well-bal-
anced relation of logical high an low states (im-
portant for a constant energy transmission) and
the easy recovery of the clock signal. The trans-
mitted and the received Manchester coded sig-
nal and the recovered clock are shown in Fig. 5. ConclusionAt the Fachgebiet Optoelektronik (Gerhard-
Mercator-Universität Duisburg) an optical signal
and energy transmission system was developed
to provide a retinal implant with digitally coded
signals. To realize this aim a transmission and
a receiving unit were designed.
References[1] SDL, Laser Diode Operators Manual & Techni-
cal Notes, SDL Inc., San Jose/USA, 1994
[2] M. Meininger, Entwicklung photovoltaischer
Zellen zur Energieversorgung einer künstlichen
Sehprothese (Retina Implantat) Diploma thesis,
3.2 Optical Interconnects and Processors 37
1 2 3 4 50
200
400
600
800
t/µs0
I/mA
Fig. 4: Modulated driving current for the
laserdiode
PHOTO-VOLTAIC
CELLARRAY
SIGNALPHOTODIODE
SIGNAL PRO-
CESSING
CLOCKRE-
COVERVY
CLOCK
MODULATED LASER LIGHTDETECTOR
SIGNAL
ENERGY
Fig. 3: Block diagram of the receiver
0 5 10 15 20
CLOCK
RECEIVED SIGNAL
TRANSMITTED SIGNAL
t/µs
Fig. 5: In- and output signals of the transmis-
sion system
Fachgebiet Optoelektronik, Universität Duisburg,
1997
[3] T. Baumeister, Systemanalyse des optischen
Energie- und Signalübertragungsmoduls für eine
künstliche Sehprothese (Retina Implantat) Di-
ploma thesis, Fachgebiet Optoelektronik,
Universität Duisburg, 1996
[4] R. Mäusl, Digitale Modulationsverfahren,
Hüthing, Heidelberg, 1991
3.2.4 Infrared data link for rotatingdisplay-systems
U. WEIMANN AND R. BUß
he magicball is a recently designed
rotating LED display system for texts
and graphics. In this report two methods of
wireless programming of the magicball by
using either an infrared remote control or a
notebook-PC are presented. In the first in-
stance, a polymethylmethacrylate (PMMA)-
body is designed and a new software is
developed allowing the programming via an
IR-remote control unit. An infrared interface
(dongle) for notebook-PCs without an built-
in infrared interface and a special data trans-
mission software protocol is described in the
second part of the report.
IntroductionThe magicball display-system is used as an
eye-catcher in stores or at exhibitions amongst
others. Its design and operation principles are
shown in Fig. 1. The 16 LEDs are fixed at the
end of the rotating arm. The microcontroller cre-
ates moving titles on the surface of the mag-
icball by switching the LEDs on and off individ-
ually. The microcontroller and the LEDs on the
rotating carrier and arm are powered by a gen-
erator situated in the socket of the magicball.
The data transmission was formerly achieved
by a serial PC cable and rubbing contacts. By
designing a wireless infrared data link between
the programming unit and the display, program-
ming of the magicball is simplified and allows
customizing. In addition, the life-span of the
product is increased while running costs de-
crease.
IR Data Link: Remote Control => MagicballThe data transmission via in-
frared remote control unit is uni-
directional from the programming
unit to the display. The infrared
signal of the remote control is bi-
phase coded , similar to the stan-
dard RC 5 code, and consists of
a pre-signal and the main signal
[1]. A photodetector - with an in-
tegrated preamplifier, a demodu-
lator and a filter - is used for de-
tecting the infrared signal of the
remote control unit [2]. Since the
receiver module is placed on the
3 RESEARCH38
Fig. 1: Side and top view of the magicball-display-system
T
rotating parts of the magicball,
an optical system (e.g. a special
mirror) is necessary to ensure an
uninterrupted connection be-
tween transmitter and receiver.
This is why a dynamically bal-
anced body made of polymethyl-
methacrylate (PMMA) has been
developed.
Fig. 2 shows the path of an ex-
emplary infrared ray from the re-
mote control unit through the
PMMA-body onto the detector. The loss of in-
tensity caused by the reflection and transmis-
sion involving the PMMA-body is less than 10%.
Tests show that the configuration consisting of
the remote control unit, PMMA-body and pho-
todetector enables a transmission range of up
to 10 m. In addition to the hardware components,
specific software has been developed to pro-
cess the incoming infrared bi-phase coded sig-
nal. After sampling the signal the microcontrol-
ler reassembles the keycode of the remote
control unit and displays the new character on
the surface of the magicball.
IR Data Link: Notebook <=> MagicballFor the data transmission to the magicball
and vice versa , the notebook-PC needs an in-
frared interface (built-in or additional) which op-
erates according to the SIR (Serial Infrared) stan-
dard of the IrDA (Infrared Data Association) [3].
The SIR standard enables a data transmission
range of up to 3 m, at angles of up to 15° and at
a data rate of 115.2 kbit/s. For notebook-PCs
without a built-in infrared interface a dongle fit-
ting to the serial port of a PC has been designed.
The encoder/decoder chip and the transceiver
module with an additional IR-LED for wider trans-
mission ranges support the infrared data trans-
mission based on the SIR standard. The con-
verters allow a power supply of the dongle with
the serial RS232-port. The principle block dia-
gram of the dongle is shown in Fig.3.
In contrast to the remote control unit configu-
ration, the notebook - magicball infrared data
link is bidirectional. Because of the bidirection-
ality and the lower radiation intensity using the
SIR standard, the PMMA-body cannot be used
here. Therefore, a simple transmitter-receiver
system has been chosen, omitting the optical
system used before. This configuration is illus-
3.2 Optical Interconnects and Processors 39
15 m
m
15 mm
10 mm
IR-Detektor
epoxy glue
IR-beam
PMMA-body
remotecontrol
Fig. 2: Exemplary infrared signal beam
Fig. 3: Block diagramm of the developed infrared dongle
trated in Fig.4 . Because of the symmetry of the
infrared data link, transmitter and receiver as
shown on the illustration are interchangable. The
bi-phased character of the data link remains in
place.
Due to the rotation of the detector and be-
cause of the angle of detection/emission, we
obtain a periodically recurring optical contact
between transmitter and receiver. Because of a
rotation of about 3600 rpm and a detection an-
gle of 15° the optical contact time (time window)
is nearly 1 ms, while the non-contact time is 15
ms. The software we have developed takes ad-
vantage of the resulting time window, thus en-
abling the data exchange between the notebook
and magicball. The software program first par-
titions the information into blocks of data with
equal size depending on the time window. After
being received, the single data blocks are reas-
sembled. The data security of the transmission
is safeguarded by parity bits on the one hand
and by feedback of magicball on the other hand
( acknowledge / non-acknowledge). For exam-
ple, a bad data block effects a non-acknowledge
response and the block will be transmitted again.
The software for this type of data transmission
and the infrared dongle have been successfully
tested for functionality.
ConclusionIn this report, two
methods of the
wireless program-
ming of a mag-
icball via infrared
data transmission
are discussed. In
the first instance,
programming is
done with an IR remote control using a PMMA
body with integrated photodetector as an opti-
cal medium on top of the rotating arm. In the
second instance, a customized program has
been developed to enable a bidirectional IR data
link between a notebook-PC and the magicball
without an additional body. Many tests have
shown the possibilty of infrared data transmis-
sion on such a rotating receiver system.
In terms of practicability, the remote control
allows a wider transmission range and a lower
level of noise interference. On the other hand
the comfortable editor for the notebook-PC is a
big advantage for the second solution, especially
if large ammounts of text has to be transmitted.
AcknowledgementWe would like to thank the LUMINO Licht Ele-
ktronik GmbH for the possibilty of working on
the display system magicball and the support
during this work.
References[1] Adaptive Micro Systems Infrared Communica-
tion Theory of Operations Abstract, 13.06.88
[2] TEMIC Semiconductors TFMx IR Detector
Photomodules Design Guide, June 1996
[3] St. Williams, I. Millar The IrDA Platform HP
Labaratories, Bristol, 1996
3 RESEARCH40
Sendewinkel
receiver path
receiver
transmitter
detection
angle
emission angle
reception area
Fig. 4: Transmitter - rotating receiver - configuration
3.2.5 Nonlinear hybrid GaAs/AlGaAsmultilayer-heterostructures for high-speed information processing
C. KAMPERMANN, A. KREUDER, AND S. REDLICH
n this report we present theoretical and
experimental results on the nonlinear
optical, electrical, electrooptical and opto-
electronic properties of hybrid GaAs/AlGaAs
multilayer heterostructures. These struc-
tures exhibit fast nonlinear properties and
high sensitivity which can be used for high-
speed information processing in microwave-
photonics.
IntroductionIn recent years optical nonlinearity and bista-
bility in multilayer heterostructures (MLHS) have
received increasing attention because of their
potential use in all-optical high-speed informa-
tion processing systems. Applications are fore-
seen in the areas of photodetectors and modu-
lators with internal amplification as well as fast
optical switching and memory devices. In 1991
He et.al. [3] achieved all-optical bistability in a
30 period GaAs/AlAs structure at an optical in-
tensity of 10kW/cm² [1]. Switching intensities in
the range of kW/cm², however, are orders of
magnitude too high for optical information pro-
cessing. In the presence of an applied electric
field perpendicular to the layers of a MLHS, the
Franz-Keldysh effect together with the accumu-
lation of photocarriers in the GaAs layers and
the voltage dependence of the current through
the device are used to decrease the switching
intensities by five orders of magnitude [2]. These
kind of hybrid MLHS exhibit the lowest switch-
ing intensities in comparison to other device
concepts. Appropriate designed MLHS also ex-
hibit s-shaped negativ differential conductivity
(SNDC), based on bistability between tunneling
and thermionic emission across the heterobar-
riers. Calculations and experiments have shown
that selfsustained voltage oscillations up to 100
GHz occur, if the MLHS is driven in an external
resonator. In this report we present theoretical
and experimental results concerning these novel
kind of devices.
Device StructureFig. 1 shows the cross section of the device
containing a periodical GaAs/ Al0.45Ga0.55As
MLHS. The MLHS consists of 20 bilayers with
nominal thicknesses of 58 nm (GaAs) and 69
nm (AlGaAs). The layers are grown by usual
3.2 Optical Interconnects and Processors 41
Fig. 1: Sketch of the device
I
MBE on s.i. GaAs substrates. A 1000nm n+-
GaAs contact layer is introduced for sufficient
high values of the bulk conductance. A number
of different process techniques such as wet etch-
ing and evaporation are used to form the de-
vice. Conventional photolithography is applied
to pattern structures on the wafer. The trans-
mission line is made of a TiPtAu multilayer met-
alization, which is applied to the wafer by evap-
oration. To prevent a short circuit between the
center conductor of the transmission line and
the contact layer at the bottom of the MLHS the
edges of the mesas are coated with polyimide.
A voltage can be applied to the coplanar
transmission line to get a high electric field con-
centrated in the MLHS. With coplanar transmis-
sion lines used as electrical contacts, operation
up to millimeterwave frequencies is possible.
The optical input and output ports are defined
by a via hole in the center conductor of the trans-
mission line.
TheoryTo simulate the device it is necessary to anal-
yse the optical wave propagation as well as the
transport and the accumulation of charge carri-
ers in the MLHS. Additionally, the electrooptical
and the optoelectronic interactions between the
optical and the electrical subsystems, namely
the Franz-Keldysh effect and the generation of
photocarriers in the GaAs layers have to be con-
sidered.
Optical propertiesFor an optical wave propagating in a period-
ically layered medium like a MLHS it has been
shown that, there exists an optical resonance
effect when the optical wavelength is close to
the optical stopband. This means that the inten-
sity distribution in the MLHS depends on the
optical wavelength and strongly increases when
approaching the resonance. This effect is es-
sential for the behaviour of the device and we
had to take this into consideration for our simu-
lations. In the linear case one can use the trans-
fer-matrix method (TMM) to calculate the inten-
sity distribution in the layered medium. In the
nonlinear case of the MLHS the standard TMM
cannot be applied because of the intensity de-
pendent refractive index of the GaAs layers. To
overcome this problem we used a generalised
3 RESEARCH42
Fig. 2: (a) Linear reflectivity spectrum of a InGaAlAs/InAlAs MLHS and average optical intensity in the
InGaAlAs layers of the structure.( b) Reflectivity over incident optical intensity of the same MLHS for
two different wavelength.
form of the TMM [3] where the GaAs layers are
divided into a number of sublayers. Assuming
that the optical intensity in each sublayer is con-
stant one can determine the intensity dependent
refractive indices by using the boundary condi-
tions of the wave amplitudes in two adjacent lay-
ers. Then, specifying the field at the end of the
structure one can apply the standard TMM to
calculate the nonlinear characteristics of the
MLHS. Figure 2(a) shows the linear reflectivity
spectrum of a MLHS consisting of 40 pairs of
InGaAlAs/InAlAs bilayers, lattice matched to InP.
As can be seen at a glance, by using the In-
GaAlAs/InP system one can shift the operation
wavelength of the device towards 1.5µm, an
important wavelength for applications in optical
communication systems. Like demonstrated for
the AlGaAs/GaAs system by He et.al. all-opti-
cal bistability based on the nonlinear refractive
coefficient n2 without an applied electric field can
also be observed. The nonlinear reflectivity-ver-
sus-intensity characteristics of the structure,
calculated with the above method, are shown in
Fig.2(b). The curves are calculated for two dif-
ferent wavelengths at the long-wavelength side
of the stopband. Both are Z-shaped and exhibit
a bistable hysteresis loop. As mentioned above
we have experimentally shown that by applying
an electric field perpendicular to the layers of
the MLHS switching intensities below 10mW/cm²
can be achieved. Therefore, besides the optical
properties, the electronic, optoelectronic and
electrooptical properties of the MLHS are from
special interest.
Electrical propertiesOur model of the transport and the accumu-
lation of charge carriers in the MLHS is based
on an analytical model of a heterostructure hot
electron diode (HHED) by Wacker et.al.[4]. The
HHED shows S-shaped negative differential re-
sistance (NDR) or differential gain and consists
of two undoped adjacent heterolayers (GaAs/
AlGaAs) with ohmic contacts. In this structure
two conduction mechanisms exist: At low fields
the current is limited by tunneling through the
AlGaAs layer (low conductance state on
theFig.3(a)). At higher fields the charge carriers
are heated up to sufficiently high energies so
that thermionic emission over the barrier be-
comes dominant (high conductance state on the
Fig. 3(b)). The extremely fast transition between
these conduction modes leads to NDR or differ-
ential gain. We extended the model of Wacker
et.al. to calculate the electronic properties of
MLHS. The physical processes of the charge
transport in the MLHS are sketched in Fig. 4(a).
As an additional effect, the cooling of the charge
carriers, which means the capture by the GaAs
wells is included. The numerically obtained cur-
rent-density-voltage characteristics (see Fig.
4(b)) are in good agreement with the results of
Monte-Carlo simulations published by Reklaitis
[5]. Fig. 4(b) further elucidates a pronounced S-
shaped NDR for (GaAs 100nm / AlGaAs 70nm)
and for thicknesses used in our device struc-
ture merely a preliminary stage of NDR.
3.2 Optical Interconnects and Processors 43
GaAs AlGaAs
W
GaAs AlGaAs
W
b)a)
Fig. 3: Schematic conduction band structure
of a GaAs/AlGaAs heterostructure with a
perpendicular electric field. The two possible
conduction states are shown.
InteractionFig.5 shows schematically the system of a
hybrid MLHS [5]. The optical and the electrical
subsystems are coupled by two interaction
mechanisms, the generation of photocarriers
(optoelectronic) and the Franz-Keldysh effect
(electrooptical). The photocarriers are generat-
ed by optical absorption in the GaAs layers. Due
to the nonlinear electrical properties of the MLHS
a small photocurrent leads to a strong variation
of the internal voltage distribution and, by the
Franz-Keldysh effect (FKE), to a large change
of the refractive indi-
ces in the GaAs lay-
ers. It has been shown
that the same change
of the refractive index
can be achieved at
much lower optical in-
tensities as for the in-
trinsic optical nonlin-
earity. Including these
mechanisms the feed-
back, which deter-
mines the optical bi-
stability of the hybrid MLHS can be described
as follows: An incident optical intensity leads to
an optical intensity in the MLHS, where a part of
the light will be absorped. The generated carri-
ers give rise to a photocurrent. This in turn leads
to a change of the voltage drop across the GaAs
layers and a change of the refractive indices of
this material. This variation finally changes the
reflectivity of the hybrid MLHS and in turn the
absorped optical intensity. Thus, a feedback loop
exists. The interaction mechanisms and the
feedback loop are also implemented in our mod-
el of the hybrid MLHS. Thus, the experimentelly
observed device characteristics including the op-
tical, electrooptical, optoelectronic and optically
induced electrical bistability of a hybrid MLHS
could be verified (Fig. 6).
Experimental resultsComprehensive measurements of the opto-
electronic properties of hybrid MLHS have
shown a high optical sensitivity of these devic-
es. At a reverse bias of V=30V we have mea-
sured photocurrents of around I=1mA and dark
currents of merely 20nA (see Fig. 7).
A bias voltage also changes the reflectivity,
as shown on theFig. 8(a) , where the electroop-
3 RESEARCH44
CoolingEmission
Heating
WC
TOTALCURRENT
2,0 3,0 4,0 5,0 6,0
0
400
800
1200
1600
curr
ent
dens
ity in
kA
/c
voltage in Va) b)
GaAs 100nm / Al0.45Ga0.55As 70nmGaAs 58nm / Al0.45Ga0.55As 69nm
Fig. 4: (a) Schematic view of the conduction band structure of a MLHS with
a perpendicular electric field and the physical processes of the charge
transport. (b) Current density-vs.-voltage characteristic of a heterostructure
for two different layer thicknesses.
ZV
i
0
0
Fig. 5: Cross section of the MLHS with I the
current flow and P the optical wave propagat-
ing through the device.
3.2 Optical Interconnects and Processors 45
Fig. 6: (a) Measured current-voltage characteristic of a hybrid MLHS under
illumination. (b) Calculated I-V curve of the same device and operation pa-
rameters.
voltage in V
-30 -20 -10 0 10 20 30
10-8
10-710-610-510
-410
-310
-2dark
curr
ent i
n A Popt = 1mW
Fig. 7: Measured current-voltage character-
istics in the dark and illuminated case.
cut-off frequencies of up to 420 MHz, measured
from MLHS devices based on another contact
geometry. The frequency response of these de-
vices is RC limited, therefore higher cut-off fre-
quencies should be reached by using the trans-
mission line design. The coupling of the
electrooptical modulation and the optoelectron-
ic properties of the device leads to optical bista-
bility, which has been measured at optical in-
tensities below 10mW/cm² (Fig. 9).
ConclusionIn this report new theoretical and experimen-
tal results concerning hybrid MLHS are present-
-8
-4
0
4
105 106 107 108
frequency in Hz
A = (600µm)²
rel.
mod
ulat
ion
in d
B
wavelength in nm860 880 900
-2
0
2
4
6 -10 V -20 V
mod
ulat
ion
cont
rast
in
Fig. 8: (a) Modulation contrast over wavelength at different reverse biases. (b) Relative modulation
characteristic as a function of frequency.
tical modulation
near the band-
gap wavelength
is due to the
Franz-Keldysh
effect. A modula-
tion contrast of
about 6dB could
be reached at a
voltage change
of 20V. Time and
frequency do-
main measure-
ments were carried out to investigate the dynamic
properties of the MLHS. We have determined
3 RESEARCH46
3.2 Optical Interconnects and Processors 47
above 1 MHz are to be expected. In compari-
son, the results differ only slightly from those
obtained with commercially available superlu-
minescence diodes.
64 Channel Silicon Driver CircuitThe TTL-compatible silicon chip was de-
signed to power the above mentioned array of
8 x 8 LEDs, with each LED driven independent-
ly by an output current adjustable in the range
of 0 - 10 mA. Based on the requirement of a
maximum input current for the IC of 1 mA, a
current amplification with a factor of 10 must be
achieved. The realized circuit consists of the
following main components: (i) two 3 to 8 de-
multiplexers for binary coded x- and y-selection
of each LED driver, (ii) a current mirror to tune
the gain of 10 and to shorten the input Iin if Izero
is set to zero, and (iii) 64 independently select-
able LED drivers containing xy-selection units
and a capacitor providing a constant output cur-
rent Iout during regeneration cycles. In Fig. 3 the
layout of the chip, consisting of 64 identical
amplifier cells, bondpads, and demultiplexing
circuits, is sketched. Each cell requires an area
of 200 * 200 µm², leading to a total square sur-
face of 1.6 * 1.6 mm² and consequently to a pixel
density of 127 DPI. Together with control cir-
cuitry and pads for external wire bonding, a to-
tal chip dimension of only 2.3 * 1.6 mm² is
achieved.
The electrical characterization of the silicon
circuit by measuring the pulse response with a
sampling oscilloscope was leading to rise and
fall times less than 250 ns. This results in a cut-
off frequency of fc ³ 1.45 MHz. Due to the fact
that the I-V characteristic referred to the input of
the circuit shows strong non-linear behaviour,
an interface circuit (voltage driven current
source) was applied, leading to a
decrease of the cut-off frequen-
cy. Together with a D/A convert-
er computer board the system
shown in Fig. 4 was built, provid-
ing a good linearity between in-
put voltage and output current.
Hybrid integration In Fig. 5 the final device con-
sisting of the silicon driver IC
bonded to the LED array is de-
picted. A composition of almost
Fig. 2: Photograph of the array with one LED
illuminated .
Fig. 3: Layout of silicon integrated circuit with detail of LED
drivers.
3 RESEARCH48
eutectic solder (60 wt % Sn, 40 wt % Pb) is evap-
orated onto the metal contacts of the LED ar-
ray. After the reflow process, at
200° C for 10 seconds, both the
array and silicon driver IC are ad-
justed and bonded together.
Since the PbSn layer thickness is
much smaller than electroplated
PbSn due to the evaporation pro-
cess, this bonding technique is a
mixture of thermocompression
and soldering. Following the flip-
chip process the ground contact
for the LED array, together with
connections for packaging of the
Fig. 4: Computer controlled silicon driver circuit.
Fig. 5: Cross-sectional view of the silicon
driver circuit bonded to the LED array.
photonic IC, are established using
wire bonding technique.
ApplicationsWith this system presented
here several applications can be
realized. One example is a spe-
cial kind of vision aid for blind
persons with a blurred cornea. In
Fig. 6 one possible realization of
this vision aid is sketched. Under
various circumstances (accidents
where the cornea is damaged, e.g
in explosions or by erosion due to acid) a num-
ber of people loose their sight although their ret-
ina is fully intact. A photodetector array converts
images into digital information wirelessly trans-
mitted to a miniature display like our proposed
model, implanted into the lens. This display
projects a very simple image onto the retina, of-
fering a primitive vision.
ConclusionExperimental investigations of both the silicon
circuit and the LED array show cut-off frequen-
Fig. 6: Vision aid for people with blurred cornea.
3.2 Optical Interconnects and Processors 49
cies beyond 1 MHz, leading to the conclusion that
this hybrid integrated circuit is a promising sub-
system not only for parallel optical information
processing systems but also for a novel applica-
tion of photonic integrated circuits in the field of
neurotechnology.
AcknowledgementsThis work was financially supported by the
Federal Ministry for Education, Science, Re-
search, and Technology (BMBF) in the frame of
the EPI-RET: Retina Implant project under
contract number 01 IN 501 G. The authors would
like to thank G. Sixt (TEMIC Telefunken, Heil-
bronn) for providing the GaAsP/GaP wafer and
R. Klinke (Fraunhofer Institut-IMS, Duisburg) for
helping designing the silicon chip. Thanks goes
also to K. Heimann (Uni-Augenklinik, Köln) for
giving an insight into several ophtalmological
problems.
References[1] H.F. Bare et al., IEEE Photon. Technol. Lett., 5,
2, pp.172, 93
[2] G.W. Turner et al., IEEE Photon. Technol. Lett.,
3, 8, pp.761, 91
[3] A.J. Moseley et al., Electron. Lett., 27, 17,
pp.1566, 91
[4] K. Werner, IEEE Spectrum, pp.30-39, Jul. 94
[5] W.R. Imler, et al., IEEE Trans. Compon., Hybr.
Manufact. Technol., 15, 6, pp.977, 92
[6] Y. Nitta et al., IEEE Photon. Technol. Lett., 4, 3,
pp.247, 92
[7] H. Yonezu et al., Electron. Lett., 25, 10, pp.670,
89
[8] M.A. Brooke et al., Optics & Photonics News,
pp.26, Jun. 93
[9] M. Wale et al., IEEE Circuits & Devices, pp.25,
Nov. 92
3 RESEARCH50
DD
DD
Fig. 1: Monolithic NLTL with periodic array of
Schottky diodes
C(V)
L/2 R/2
V
L/2R/2Ik
Gk
Fig. 2: Equivalent circuit for one element of
the NLTL
3.3 Millimeterwave Electronics
3.3.1 Picosecond pulse generationon monolithic nonlinear transmis-sion lines using high-speed InP-HFET diodes
R. HÜLSEWEDE
lectrical pulses with transients less
than 5 ps are generated and com-
pressed on monolithic InP-HFET diode non-
linear transmission lines. The transients are
measured by time domain electro-optic sam-
pling technique and the waveforms show
good agreement with numerical results. Ad-
ditionally, frequency domain measurements
and numerical simulations reveal that the
nonlinearities work with frequencies higher
than 400GHz for 20µm x 20µm InP-HFET di-
odes. Instead of costly ion implantation tech-
nology a chemical recess is used to isolate
the active structures.
IntroductionMonolithic nonlinear transmission lines
(NLTL) are circuits with an alternating arrange-
ment of coplanar waveguides and Schottky di-
odes as shown in Fig. 1.
The capacitance-voltage characteristic of the
Schottky diodes in combination with the low pass
filter characteristic of the periodic structure leads
to the generation of shock waves and the for-
mation of pulses with (sub-) picosecond tran-
sients [1,2]. Therefore, these circuits are impor-
tant for novel measurement and characterization
methods for new high-speed devices.
For numerical simulations the equivalent cir-
cuit as shown in Fig.2 is used leading to a dif-
ference equation for current and voltage at each
element of the NLTL.
Applying a transition to a differential equa-
tion one obtains the following wave equation
which considers separately the influences of the
nonlinearity of the diodes, the periodic structure
and the losses of the transmission line (for de-
tails see [3,4]):
∂∂
∂∂
Vx
C VC
Vt= −
+1
0
( )
∂∂
∂∂
L C Vt
RL V
CG
Vt
+⋅
− +120
3
30
2
2 (1)
Here the nonlinearity of the diodes is consid-
ered by the normalized capacitance-voltage
dependence C(V)/C0 , where C0 is the capaci-
tance at the operating point of InP-HFET diode
In order to improve the nonlinearity of the
Schottky diodes in NLTLs d-doped diodes based
3.3 Millimeterwave Electronics 51
E
on InP-HFET layer structures are used ( see
Fig.3 and [5]). High electron concentration in the
d-doped layer (4.9 1012cm-2), maximum mobili-
ty (10900 Vs/cm-2), and the 2 dimensional elec-
tron gas (2-DEG) in the InGaAs channel are
special features of this layer structure at T=300K.
The strong nonlinearity of the HFET-layer struc-
ture is shown in Fig. 4 where the normalized ca-
pacitance-voltage characteristic of an InP-HFET
diode is sketched. Over the 0.5V bias range
around the working-point a 2200% change of
the capacitance is achieved. This is a 20x great-
er nonlinearity than d-doped GaAs Schottky di-
odes used in NLTLs described in [6]. One rea-
son for this strong nonlinear behaviour is the
depletion of the 2-DEG underneath the nega-
tive biased Schottky contact. Using InP-HFET
diodes in NLTLs the nonlinear interaction of the
propagating waves is increased and thus the
line-losses are decreased due to shortening of
the line length. Another advantage is the appli-
cation of a C4H6O4,H2O2,NH3 recess [7] for elec-
trical isolation of the InP-HFET diodes in NLTLs.
Thus, no costly ion implantation process is need-
ed and no preparation of an accelerator is re-
quired.
InP-HFET NLTLIn a first step a 10 diode periodic InP-HFET
NLTL was fabricated in order to verify experi-
mental and numerical results. For that purpose
frequency domain measurements along the cen-
ter conductor are shown in Fig. 5 (see also [8]
and P. Bussek et al., Time- and frequency do-
main electro-optic field mapping of nonlinear
transmission lines, in this annual report).The
electro-optic signal of the excited input wave
(15GHz, 25dBm) decreases from input to out-
put of the NLTL, whereas the generated har-
monic signals at 30GHz, 45GHz and 60GHz in-
crease. Using the nonlinearity of InP-HFET
diodes (Fig.4) and a FFT the nonlinear wave
propagation on this NLTL is simulated. The re-
sult is shown by the grey lines in Fig. 5. With
respect to the -128dB noise level and the +/-5dB
accuracy of the sampling signal both results are
in good agreement. This agreement and the
numerical value for C0/G = 3,5 10-13s indicates
that the nonlinearity works with frequencies high-
er than 400GHz for the 20µm ´ 20µm InP-HFET
diodes.
Thereupon different NLTLs are simulated based
on these perceptions in order to generate one
Schottky contactOhm contactInGaAs-channelInAlAsInGaAlAsInPδ-Si
Fig. 3: Schematic profile of an InP-HFET
diode (for details see [5])
0-1-2-3
10
1
0.1
0.01
Bias voltage(V)
C(V
)/C
o
Fig. 4: Normalized capacitance-voltage
characteristic of InP-HFET diodes
3 RESEARCH52
single pulse per period of the sinusoidal input sig-
nal. The simulation in Fig.6a demonstrates the
generation and compression of single pulses out
of the exiting input signal (6.5GHz, 2.5V) on a
graded NLTL with increasing values of L and C0
in direction of the propagating microwave. The
transient with minimum 10-90% fall time of 4ps
and an amplitude of 1.8V is shown in Fig. 6b.
After fabrication of this graded InP-
HFET NLTL using self aligned op-
tical contact lithography process-
es [7] the electro-optic sampling
set-up was modified making time
domain measurements (see P.
Bussek et al., Time- and frequen-
cy domain electro-optic field map-
ping of nonlinear transmission
lines, in this annual report). In
Fig. 7 a top view of the graded
InP-HFET NLTL is figured includ-
ing the four points of measurement
(a)-(d). The frequency of the input
signal is 6.5GHz with an amplitude
of 3.5V (measured at 50W). Clear-
ly the steeping of a shock wave
(k=11) and the generation of a sin-
gle pulse (k=31) with FWHM of
10ps and a fall time of about 5ps
is observed. Thus, picosecond
pulse generation on NLTL using
high speed InP-HFET diodes is
shown for the first time. Addition-
ally, the numerical results (Fig.6)
at the corresponding points of
measurements are plotted in Fig. 7
(grey lines). The agreement of
waveforms is satisfying demon-
strating that the fundamental
mechanisms of nonlinear wave
propagation on NLTLs have been considered in
equation (1).
ConclusionIn this work the generation and compression
of picosecond pulses on InP-HFET NLTL is dem-
onstrated. Advances have been achieved by
applying high speed InP-HFET diodes exhibit-
0 500 1000 1500 2000 2500
x (µm)
Sig
nal (
dBm
)
-90
-100
-110
-120
-130
-140
-80
Input signal: 15GHz,27dBmexperimentsimulation
15GHz
30GHz
45GHz
60GHz
Fig. 5: Generation of harmonic signals on periodic InP-HFET
NLTL. The structure of NLTL is sketched at top of this figure.
(Data of simulation: L pH=120 , C0 = 1.6 pF,
R L s/ = ⋅ −2 1010 1, C G s0133 5 10/ .= ⋅ − )
Signal voltage
x
t
1V
10ps(a)
Sig
nal v
olta
ge
(V)
Time (ps)
1
-2
0
-1
0 40 12080
(b)
(b)
Fig. 6: Simulation of pulse generation on a graded InP-HFET
NLTL; (a) development of a sinusoidal signal along the NLTL,
(b) transient with minimum fall time (Data of simulation:
L pH= 960 , C0 = 1.76 pF, R L s/ = ⋅ −2 1010 1,
C G s0133 5 10/ .= ⋅ − , grading: 0 89. αx , α = ⋅ −8 4 1010 1. s )
3.3 Millimeterwave Electronics 53
ing strong nonlinearities. The numerical simula-
tion has been improved by considering sepa-
rately the influence of nonlinearity, periodic struc-
ture and losses to nonlinear wave propagation
on NLTLs.
AcknowledgementThe author would like to thank Dipl.-Phys.
U. Auer (Fachgebiet Halbleitertechnik / -technol-
ogie) for processing of the transmission lines and
Dr. D. v.d.Weide (at that time: Max-Planck-In-
stitut für Festkörperforschung, Stuttgart) for lend-
ing a suitable mask for optical contact lithogra-
phy processes.
References[1] M.J.W. Rodwell et al, Active and nonlinear wave
propagation devices in ultrafast electronics and
optoelectronics, Proc. IEEE, Vol. 82, No. 7, 1994,
pp. 1037-1059
[2] D. Jäger, Characteristics of travelling waves
along nonlinear transmission lines for monolithic
integrated circuits: A review, Int. J. Electron.,
Vol. 58, 1985, pp. 649-669
[3] D. Jäger, Pulse generation and compression on
nonlinear transmission lines, workshop on Pi-
cosecond and Femtosecond Electromagnetic
Pulses: Analysis and Applications, MTT-S Symp.
Dig., 1993, pp. 37-57
[4] R. Hülsewede et al, CAD of pulse compression
on nonlinear transmission lines, Proc. MIOP 95,
Sindelfingen, 1995, pp. 511-515
[5] U. Auer et al, InP based HFETs with high qual-
ity short period InAlAs/InGaAs Superlattice
Channel Layers, J. o. Crystal growth, vol. 146,
1995
simulationexperiment
(a)
(d)(c)(b)
(a)(b)
(c)
(d)
k = 31
k = 7
k = 15k = 11
input
output time (ps)
sign
al v
olta
ge (
a.u.
)
-80 -40 0 40
-80 -40 0 40si
gnal
vol
tage
(a.u
.)
sign
al v
olta
ge (
a.u.
)
sign
al v
olta
ge (a
.u.)
-80 -40 0 40-80 -40 0 40time (ps)time (ps)time (ps)
Fig. 7: Pulse compression on graded InP-HFET NLTL (top view of the processed NLTL in the upper
left side of this figure, (a)-(d): points of electro-optic measurements)
3 RESEARCH54
[6] D.W. van der Weide, Delta-doped Schottky di-
ode nonlinear trans-mission lines for 480-fs, 3.5-
V transients, Appl. Phys. Lett. Vol. 65, No. 7,
1994, pp.881-883
[7] C. Heedt et al, On the Optimisation and Reli-
ability of Ohmic- and Schottky Contacts to InAlAs/
InGaAs HFET, Proc. 4th InP & Related Materi-
als Conference, Newport, USA, 1992
[8] Report on the Special Collaborative Programm
SFB 254, 1993-1995, Gerhard-Mercator-
Universität - GH - Duisburg, 1995
3.3.2 Millimeter wave power genera-tion on nonlinear transmission lines
R. HÜLSEWEDE, V. K. MEZENTSEV, AND
I. V. RYJENKOVA
n this paper nonlinear transmission
lines are described which are used to
generate millimeterwave signals with high
efficiencies. In particular, arrays of monolith-
ic varactor diodes loading a coplanar
waveguide are studied which can be applied
for travelling wave harmonic generation
where special phase matching and filter
structures give rise to high conversion effi-
ciencies. A second transmission line consist-
ing of any array of resonant tunneling diodes
is used as a distributed active device which
can generate millimeterwave power at fre-
quencies as determined by a resonance con-
dition of the resonator structure under study.
In this paper theoretical and numerical re-
sults are presented based upon experimen-
tal data.
Introduction:The generation of millimeter waves by har-
monic frequency generation and active wave
propagation along nonlinear transmission lines
(NLTLs) has recently become a subject of ma-
jor research activities [1-4]. However, the pow-
er efficiencies achieved so far are small because
millimeterwave power is converted into undes-
ired frequency components when the dispersion
and filter characteristics of the NLTL are not
designed in a suitable way. In this paper, firstly
we describe the bi-modal NLTL which uses con-
cepts of nonlinear optics aiming towards achiev-
ing phase matching condition between the fre-
quency components under study [5,6]. In
particular, we study a bi-modal NLTL where, as
an example, the phase velocity of the second or
third harmonic equals that of the fundamental
wave and where other components are sup-
pressed by a suitable filter structure leading to
a distinct cut-off frequency [7-9]. Secondly, we
discuss the characteristics of a travelling-wave
tunneling-diode transmission line resonator ca-
pable of generating high power millimeter wave
signals [7-9].
In Fig. 1, the basic structure of an NLTL is
sketched consisting of a suitable array of non-
linear devices D in a passive coplanar wave-
guide [2,3]. As nonlinear elements, we have
studied Schottky diodes, quantum barrier var-
DD
DD
Fig. 1: Sketch of a nonlinear transmission
line
3.3 Millimeterwave Electronics 55
I
actor structures (QBV), as well as resonant tun-
neling diodes (RTDs).
The second characteristic feature of the cir-
cuit in Fig. 1 is the dispersion which is mainly
determined by the arrangement of the diodes,
which can be periodic, bi-periodic, graded etc.
The dispersion itself controls the phase veloci-
ties of different spectral components and hence
the strength of interaction and the superposi-
tion in the time domain. Reflections at input and
output ports further determine the resonance
behavior of the whole structure.
Harmonic frequency genera-tion:
We have studied nonlinear
wave propagation along NLTLs of
Fig. 2. In a first example the pa-
rameters used are those of an
experimental device on InP-HFET
substrate as discussed in [10].
Input frequency and power are
100GHz and 11dBm as delivered
into a small-signal characteristic
impedance of 50W. As a numeri-
cal tool we have used a continu-
um approximation on the basis of
a corresponding nonlinear evolu-
tion equation as described in
[10,11] and compared the results
with a CAD model based on a dis-
crete representation of the NLTL
, cf.[10].
The results of our numerical
calculations are plotted in Fig. 3
showing the spatial distributions
of the amplitudes of the funda-
mental and second harmonic
wave. As can be seen, the ampli-
tudes of the two waves at the in-
put are comparable, which leads to power effi-
ciencies > 70% for second harmonic genera-
tion (SHG), here at 200GHz.
In a second numerical experiment we have
studied third harmonic generation (THG) on a
bi-modal NLTL of Fig. 2(b) assuming special
quantum barrier varactor diodes [12] with a sym-
metric capacitance voltage relationship. Fig. 4
presents the numerical results. Note that in this
case of THG phase matching is achieved be-
tween frequencies f1 = 72GHz and f3 = 216GHz.
As can be seen from Fig. 4, the 72GHz input
signal is converted into millimeter wave power
RTDRTD
RTDRTD
InP 350 µm
InGaAs 40 nm
InGaAs 500 nm n+
InGaAs 4.3 nm
InGaAs 40 nm undoped
InGaAs 500 nm n+
undoped
undoped
undoped
undoped7.2 nmInAlAs
InAlAs 7.2 nm
o
o
C
CC
CInP 350 µm
InGaAs 250 nm
InGaAs 500 nm n+
InGaAs 25 nm
InGaAs 25 nm
InGaAs 250 nm n
InGaAs 250 nm n+
undoped
undoped
n
InAlAs 25 nm undoped
o
o
C
CC
C
InP 350 µm
InGaAs
300 nm
InGaAs
50 nm n
n
n
n50 nm
InAlAs
800 nm
-
-
+
+
InGaAs
(a)
(b)
(c)
Fig. 2: Monolithic NLTL on InP substrate for millimeter wave
generation. (a) Bi-modal NLTL with Schottky varactor diodes for
SHG, (b) bi-modal NLTL with QBV for THG, and (c) RTD-NLTL
for a distributed oscillator.
3 RESEARCH56
at 216GHz with an efficiency of about 25%. Again,
the third harmonic is available at the input of the
NLTL because the propagation characteristic is
that of a backward wave [5,6].
Tunneling diode NLTL:Very recently, another type of NLTL has be-
come known where resonant tunneling diodes
are used as nonlinear elements [3]. However,
such RTD-NLTLs can also be used for nonlin-
ear active wave propagation ef-
fects leading to a travelling wave
oscillator, when a transmission
line with limited length, provided
by short circuits at input and out-
put ports, for example, is used
[7,8]. In a numerical experiment
we have studied the generation
of millimeter waves in a RTD-
NLTL resonator. The results are
plotted in Fig.5 revealing self-gen-
erated oscillation at 170 GHz af-
ter about 700 ps which is the
switch-on time.
In a further numerical example
we have studied the relationship
between the length N of the os-
cillator and the frequency f(n, N) of the self-gen-
erated oscillations where n = 1, 2, , N defines
the mode. Fig. 6 shows the results where the
dots represent the results of the simulations.
Clearly, a decreasing N leads to an increasing
f(n,N) because the wavelength, as given by
2 x N, decreases. In Fig. 6 a comparison with
analytical results is additionally carried out where
f(n, N) is given by
0
0.2
0.4
0.6
0.8
1
1.2
0 20 40 60 80 100 120 140 160number of elements, k
f1
f = 2f2 1
Fig. 3: Spatial distribution of amplitudes at
frequencies f1 and f2 for the NLTL of Fig.2(a).
0 20 40 60 800
0.1
0.2
0.3
0.4
0.5
0.6 f1
f3
100 120 140 160number of elements, k
Fig. 4: Spatial distribution of amplitudes at
frequencies f1 and f3=3f1 for the NLTL of
Fig.2(b)
0 100 200 300 400 500 600 700 800
-0.4
0
0.4
0.8
1.2
1.6
time, ps
volta
ge, V
0 100 200 300
ampl
itude
400frequency, GHz
Fig. 5: Generation of a 170 GHz signal on a tunneling diode
NLTL. The spectrum is shown in the inset.
3.3 Millimeterwave Electronics 57
f n NLC
nN( , ) sin( )= 1 1
2ππ , n = 1,2...,N
(1)
as calculated from the dispersion relation for a
cascaded LC - chain neglecting losses.
ConclusionIn this paper, specific NLTLs are presented
which are capable to generate millimeter waves
with high conversion efficiencies. The NLTLs are
compact, easily fabricated using standard InP
technology, suitable for monolithic integration,
and can provide high output powers. We there-
fore conclude that the travelling wave concept
under study can provide a solution to the prob-
lem of realizing efficient millimeter wave signal
sources.
References[1] A. Scott, Active and nonlinear wave
propagation in electronics, John Wiley &
Sons, New York, 1970
[2] D. Jäger, Characteristics of travelling
waves along nonlinear transmission lines for
monolithic integrated circuits: A review, Int.
J. Electron. 58, 649-669 (1985) (invited pa-
per)
[3] M.J.W. Rodwell, S.T. Allen, R.Y.Y. Yu,
M.G. Case, U. Bhattacharya, M.Reddy, E.
Carman, M. Kamegawa, Y. Konishi, J. Pusl,
R. Pullela, Active and nonlinear wave propa-
gation devices in ultrafast electronics and op-
toelectronics, IEEE Proc., vol. 82, no. 7, pp.
1037-1059, 1994 (invited paper)
[4] E. Carman, M. Case, M. Kamegawa, R. Yu, K.
Giboney, and M.J.W. Rodwell, V-band and W-
band broadband, monolithic distributed frequency
multipliers, in: 1992 IEEE MTT-S Digest, pp.
819-822, 1992
[5] B. Wedding and D. Jäger, Phase-matched sec-
ond harmonic generation and parametric mixing
on nonlinear transmission lines, Electron. Lett.
17, 76-77 (1981)
[6] D. Jäger, Nonlinear slow-wave propagation on
periodic Schottky coplanar lines, IEEE Micro-
wave and Millimeter-Wave Monolithic Circuits
Symposium, St. Louis 1985, Symp. Dig., 15-17
(1985)
[7] I. V. Ryjenkova, V. K. Mezentsev, S.L.Musher,
S. K. Turitsyn, R. Hülsewede, and D. Jäger, Mil-
limeter Wave Generation on Nonlineat Transmis-
sion Lines, Proc.1996 International Workshop
on Millimeter Waves, April 11-12, Orvieto, Italy,
1996
[8] V. K. Mezentsev, S.L.Musher, I. V. Ryjenkova, S.
K. Turitsyn, R. Hülsewede, D. Jäger, Travelling
wave generation of millimeter waves in bi-modal
N0 5 10 15 20 25 30
0
100
200
300
400
500
600
numerictheory
n = 1
Fig. 7: Oscillation frequency vs. number N of elements
of the RTD-NLTL, theory according to eq. (1)
3 RESEARCH58
NLTLs, Proc. 26th European Microwave Con-
ference 1996, Prague
[9] I. V. Ryjenkova, V. K. Mezentsev, S.L.Musher, S.
K. Turitsyn, R. Hülsewede, and D. Jäger, Milli-
meter Wave Generation on Nonlineat Transmis-
sion Lines, Ann. des Telecomm., Special Issue
(submitted).
[10] R. Hülsewede, U. Effing, I. Wolff, and D. Jäger,
CAD of pulse compression on nonlinear trans-
mission lines , Proc. MIOP 95, Sindelfingen, pp.
511-515
[11] M. Dragoman, R. Kremer, and D. Jäger, Pulse
generation and compression on a travelling-wave
MMIC Schottky diode array, in: Ultra-Wideband,
Short-Pulse Electromagnetics, H.L. Bertoni, L.
Carin, and L.B. Felsen, eds., Plenum Press, New
York, pp. 67-74, 1993
[12] M.A. Frerking, J.R. East, Novel heterojunction
varactors, IEEE Proc., vol. 80, no. 11, pp. 1853-
1860, 1992
3.3.3 Nonlinear RTD circuits forhigh-speed A/D conversion
I. JÄGER
n this report a novel nonlinear MMIC
structure based upon monostable reso-
nant tunneling diodes (RTDs) is studied. For
the first time, it is shown that an input signal
can be converted into a set of output spikes
to be used for GHz A/D conversion.
IntroductionA huge amount of work has recently been
dedicated to the study of resonant tunneling di-
odes (RTDs) which can provide gain and can
directly be used as the key components for os-
cillator circuits approaching the THz frequency
range [1]. The underlying characteristic is a non-
linear N-shaped current voltage relationship even
at millimeterwaves. The lack of these very inter-
esting devices, however, is the low power con-
version efficiency and the small output power
levels [2]. Up to now the only solution to the latter
problem which has become known is the use of
a series i.e. distributed connection of several
RTDs using MMIC technology [3,4]. Such a RTD
nonlinear transmission line (NLTL) can further
provide the basis of very interesting microwave
signal processing devices as has been predicted
by Crane already in 1962 [5].
In this paper, we discuss first the fundamental
concept of nonlinear active wave propagation
effects along monostable RTD-NLTLs utilized to
generate a set of spikes from anelectrical input.
The idea of such a transmission line, where loss-
es are exactly compensated by distributed am-
plification, dates back to the so called neuristor´
[5] as a line-analog of axons in the nervous sys-
tem, where the information of an input signal is
converted into a number of output spikes, travel-
ling in a stationary way for arbitrarily long dis-
tances. In a second step, we describe an electri-
cal circuit, where the monostable RTD-NLTL is
the main building block to realize a n-bit A/D con-
RTD
L GRTD
L G
Fig. 1: Sketch of a nonlinear array of
monostable resonant tunneling diodes in a
coplanar transmission line
3.3 Millimeterwave Electronics 59
I
verter at GHz rates, similar to the lumped RTD
A/D converter described in [7-9].
RTD-circuitThe array of monostable RTDs is sketched
in Fig.1. One can see a coplanar transmission
line which is periodically loaded, cf.[3,4], with
RTDs shunted by LG circuits -here air bridges- in
order to provide monostable behavior, see [6].
The cross section of the MMIC structure in Fig. 1
has been described in [3,4].
The simulation carried out in this paper is
based upon a suitable equivalent circuit, as
shown in Fig.2. Each section consists of an T-
equivalent representation of the transmission
line. The nonlinear element in Fig.2 is deter-
mined by the RTD current voltage relationship
approximated by , where an external bias cur-
rent and V1,V2 > 0 have been assumed.
The basic idea of the circuit in Fig. 2 is roughly
the following. An input current source charges
the capacitance C up to a threshold value given
by J(V) of the RTD. A switching up occurs which,
however, will be inverted due to the LG time
constant. As a result, a spike is produced and
after an RC time constant another switching
occurs. Hence the spiking period is determined
by the amplitude of the input current. The trans-
C
R
GV
J(V) L
Fig. 2: Equivalent circuit of a
monostable RTD-NLTL
mission line itself ensures the generation and
propagation of identical pulses - such as solito-
ns - formed after a few diodes.
ResultsFig.3 shows a numerical result for an input
sinusoidal wave of 25 GHz. As can be seen, the
monostable RTD-NLTL produces a set of 6 puls-
es per period. When the input frequency or the
input amplitude are changed, the number,
phase, and position of the spikes are altered in
a characteristic way.
Fig.4 shows an example where the width and
amplitude of a rectangular input signal have been
changed. As a result, the generated spiking as
obvious from the regions with different shadings
is a characteristic pattern for the input signal. In
particular, we observe that the number of spikes
per time depends linearly on the applied current
amplitude providing a linear voltage-frequency
conversion.
Such a NLTL can be used to realise a high
speed n-bit A/D converter similar to the lumped
version in [7-9]. Correspondingly, we propose
10 20 30 40 50 60 70
-1,0
-0,5
0,0
0,5
1,0
Time
Fig. 3: Generation of five pulses
per period of sinusoidal input wave (dashed
line)
3 RESEARCH60
to realise a common n-channel (for n bits) copla-
nar signal devider to provide input amplitudes by
powers of 2. Hence each channel delivers a spike
train to the output array establishing a Gray code
as in Ref.[7]. In the present case, the LC time
constant will determine the bandwidth which ex-
ceeds 100 GHz in the device under test.
ConclusionIn conclusion, a novel monostable RTD-NLTL
in MMIC technology is proposed which can gen-
erate a characteristic pulse pattern for a given
input signal. The application of such a NLTL for
an ultrafast A/D conversion is discussed in a
second step. The use of RTDs in the presented
circuit is expected to yield a bandwidth in ex-
cess of 100 GHz.
References[1] E.R.Brown, J.R.Söderström, and T.C.McGill,
Oscillations up to 712 GHz in InAs/AlSb
Resonant-Tunneling Diodes, Appl.Phys.Lett.,
vol.58, no. 20, pp. 2291-2293, 1991
[2] R.Sun, O.Boric-Lubecke, D.-S.Pan, and T.Itoh,
Considerations and Simulations of
Subfrequency Excitation of Series Integrated
Resonant Tunneling Diodes Oscillator,
IEEE Trans. Microwave Theory Techn., vol. MTT
- 43, no.10, pp. 2478-2485, 1995
[3] I.V.Ryjenkova, V. K.Mezentsev, S.L.Musher,
S.K.Turitsyn, R.Hülsewede, and D.Jäger, Milli-
meter Wave Generation on Nonlineat Transmis-
sion Lines, Proc.1996 International Workshop
on Millimeter Waves, April 11-12, Orvieto, Italy,
1996
[4] [4] I.V.Ryjenkova, V.K.Mezentsev, S.L.Musher,
S.K.Turitsyn, R.Hülsewede, and D.Jäger, Milli-
meter Wave Generation on Nonlineat Transmis-
sion Lines, Ann. Telecomm., Special Issue,
vol.52, No 3-4, pp. 134-139, 1997
[5] H.D.Crane, Neuristor - A Novel Device and Sys-
tem Concept, Proc. IRE, vol.50, pp. 2048-
2060, 1962
[6] J.Nagumo, S.Arimoto, and S.Yoshizawa, An Ac-
tive Pulse Transmission Line Simulating Nerve
Axon, Proc. of the IRE, vol.50, p.2061, 1962
[7] T.-H..Kuo, H.C.Lin, R.C.Potter, D.Shupe, A
Novel A/D Converter Using Resonant Tunnel-
ing Diodes, IEEE Journal of Solid-State Circuits,
vol.26, No.2, pp.145-149, 1991
[8] S.-J.Wei, H.C.Lin, R.C.Potter, D.Shupe, Dy-
namic Hysteresis of the RTD Folding Circuit and
ist Limitation on the A/D Converter, IEEE Trans-
action on Circuits and Systems II: Analog and
Digital Signal Processing, vol.39, No.4, pp.247-
251, 1992
[9] S.-J.Wei, H.C.Lin, R.C.Potter, D.Shupe, A Self-
Latching A/D Converter Using Resonant Tunnel-
ing Diodes, IEEE Journal of Solid-State Circuits,
vol.28, No.6, pp.697-700, 1993
3.3 Millimeterwave Electronics 61
2.00 3.00 4.00 5.00 6.00 7.00 8.002.00
3.00
4.00
5.00
6.00
7.00
8.00
9.00
10.00
Fig. 4: Contour plot of generated
number of spikes (see text)
3.4 Optical Sensor Systems
3.4.1 MQW-Electroabsorption-Modu-lator for Application in a fiberopticfieldsensor
M. SCHMIDT, R. HEINZELMANN, AND A. STÖHR
n this report we present electroabsorp-
tion waveguide-modulators for an op-
eration wavelength of 1.55µm. These devices
are fabricated for an application in a fiberop-
tical E-field sensor system [1], [2]. In this
system the task of the modulator is to con-
vert electrical signals with frequencies up to
6 GHz into optical signals.
IntroductionIn recent years there has been an increasing
interest in electrooptical modulators. The main
application of these devices is in fiberoptical
communication systems for the external modu-
lation of laserdiodes. Electroptical modulators
have been realised in lithium niobate as well as
in semiconductors using the Franz-Keldish-ef-
fect in bulk materials and the quantum confined
starck effect in multiple quantum well structures.
As MQW structures exhibit the strongest elec-
trooptic effect they allow the use of smaller elec-
trodes than the other mentioned modulator prin-
ciples. The resulting lower capacitance has the
advantage of higher cut off frequencies. Further-
more MQW modulators can be made insensi-
tive to the polarisation of the modulated light by
introducing tensile strain to the quantum wells.
This avoids the need for expensive polarisation
maintaining fibers.
InGaAs/InAlAs - MQW
nid InAlAs
n InAlAs
s. i. InP
+
n InAlAs-
nid InAlAs
contactLayer structure:
Fig. 1: Sketch of the electroabsorption waveguide modulator
3 RESEARCH62
I
Device structureIn Fig. 1 a sketch of the device is shown. The
modulators are grown by MBE in the ternary
material system InGaAs/InAlAs on InP subtrates.
The device structure is n+in -. The ridge
waveguide is formed by wet chemical etching
with cytric acid down to the n+ layer. Afterwards
the electrical contact are produced by vacuum
coating. The ohmic contact on the n+ layer is
realised in GeNiAu, for the Schottky contact on
the n- layer we use CrAu.
The design of the device structure was sup-
ported by BPM-Simulations and by calculations
of the electrooptical behaviour. For the BPM-
Simulations we used the comercial BPM-Soft-
ware BPM-Cad. The main aim of this simulation
was to determine the optical confinement fac-
tor, i. e. which part of the guided modes over-
laps with the absorbing MWQ layer. We calcu-
lated a confinement factor of 14 %. For the
calculation of the absorption coefficient of the
MQW-material in dependence of the electrical
field we used a transfer matrix method. From
the obtained results we calculated the optical
absorption of the modulator in dependence of
the applied electric field as shown
in Fig. 2.
Experimental resultsThe epitaxial wafers were char-
acterized by photoluminescence
measurements. The point of in-
terest was the spectral position of
the excitonic peak of the MQW
material, which indicates the po-
sition of the absorption edge. A
comparison between the experi-
mental results and the calcula-
tions is shown in Fig. 3.
The optical transmission of the
modulator is characterized by
coupling the light of an erbium
doped fiber laser into the
waveguide and detecting the
transmitted light on the other fac-
0 %
20 %
40 %
60 %
80 %
100 %0 5 10 15
Field (10 6 V / m)
Abs
orpt
ion
0 2 4 6Voltage [V]
Slope: 0,38 / V
Wavelength: 1.55µmDevice lenght: 500 µm
Fig. 2: Calculated electrooptical behaviour
of a modulator device.
6 7 8 9 101340
1360
1380
1400
1420
1440
1460
1480
1500
1520
1540
1560
1580
Mod12Shg 08
Mod 08 Mod 13 Shg 03
Mod 07Mod 05 Shg 07
Shg 04Shg 05
Mod 09
Exc
itoni
c w
avel
engt
h [n
m]
Thickness of quantum-wells [nm]
PL-Measurement at 12 KSimulation at 12 KSimulation at 300 K
Shg 06
Fig. 3: Excitonic wavelength of the MQW material determined
by photoluminescence measurements compared with calculated
values.
3.4 Optical Sensor Systems 63
et of the waveguide. The optical tranmission is
measured in dependence of the applied voltage
as shown in Fig. 4.
ConclusionsAn electrooptical MQW waveguide modula-
tor has been designed. The epitaxial layers have
been grown by MBE and characterised by pho-
toluminescence measurements. The position of
the measured exciton peaks was in good agree-
ment with the calculated values. Modulator de-
vices have been processed from these wafers
by wet chemical etching and vacuum coating of
the electrical contacts. The devices have been
characterized by optical transmisson measure-
ments.
References[1] Stöhr, R. Heinzelmann, T. Alder, M. Schmidt, M.
Groß, and D. Jäger, Integrated Optical E-Field
Sensors using TW EA-Modulators,
Interational Topical Workshop on Contemporary
Photonic Technologies CPT98, Technical Di-
gest, Tokyo, January 1998
[2] Heinzelmann, A. Stöhr, M. Groß, D. Kalinowski,
T. Alder, M. Schmidt, and D. Jäger, Optically
Powered Remote Optical Field Sensor Sys-
tem using an Electroabsorption-Modulator,
IEEE MTT-S International Microwave Sympo-
sium, Conference Proceedings, Baltimore, June
1998
3.4.2 Photovoltaic cells for fiber opticEMC - Sensor power supply
D. KALINOWSKI
hotovoltaic cells play an important
role in power supply of hybrid sen-
sors. A photovoltaic cell array is under con-
struction to supply an active fiber optic hybrid
sensor head. A prototype with first
experimental results will be shown.
IntroductionDue to more and more restrictive laws regu-
lating the electromagnetic compatibility (EMC)
of electronic equipment the necessity of devel-
oping precise and reliable sensors to measure
electromagnetic fields steadily increases. One
request for such sensors is non invasiveness.
Hence, our approach to reach this goal is to
develop a hybrid fiber optic fiels sensor. This
concept takes advantage of the fact, that opti-
cal fibers do not interfere with the electromag-
netic field that is to measure, but that they are
capable to transmit optical information. By this
distortion of the E-field is minimized. The photo-
voltaic cell array (PVC) described in this article
is part of this optical E-field sensor which is
shown in Fig. 1.
0 -1 -2 -3 -40,0
0,1
0,2
0,3
0,4
0,5
Mod
ulat
ion
[a.u
.]
Voltage [V]
λ = 1550 nm
Fig. 4: Modulation of the MQW modulator
versus applied voltage.
3 RESEARCH64
P
DeviceOne requirement to be matched by the PVC
is a high efficient conversion of optical into elec-
trical power. Therefore, special effort has to be
laid upon the layout and the composition of the
heterostructures. Since cheap and powerful la-
ser diodes are available in the 800 to 850 nm
wavelength regime and since GaAs has its opti-
mum photovoltaic response at about 800 nm,
the PVCs are designed as AlGaAs/GaAs pin-
diodes. Fig. 2 shows a photograph of a cell ar-
ray consisting of 4 cells. The diameter of the
active region is 600
µm, i.e. the diameter of
the core of the multi-
mode fiber used. Thus,
each quarter of this
array, i.e. each PVC, is
illuminated uniformly
leading to a maximum
generation of electrical
power by this configu-
ration [1].
The GaAs and AlGaAs layers are MBE grown
on semi-insulating GaAs substrate. The layer
structure, illustrated in Fig. 3, consists of an 100
nm n+-AlGaAs contact layer, a 2 µm i-GaAs ab-
sorption layer and a 100 nm p+-AlGaAs window
layer. A 10 nm p+-GaAs contact layer offers an
aluminium protection from oxidation.
The metallic contacts act as ohmic contacts.
GeNiAu is used for the n-contact and PtTiPtAu
for the p-contact.
Fig. 1: Sketch of the optically powered integrated optical field sensor
Fig. 2: Perspective view of the photovoltaic
cell array consisting of 4 cells
Fig. 3: Cross section of the photovoltaic cells
3.4 Optical Sensor Systems 65
ResultsThe cell array is illuminated by a laser with a
emission wavelength of 800 nm. Measurements
of the max. efficiency show results up to 28%
depending on the optical input power (Fig. 4).
ConclusionThis report presents a photovoltaic cell array
for power supply of our hybrid EMC - Sensor.
The design and a first result are shown.
References[1] M. B. Spitzer, et. al., Monolithic series-con-
nected gallium arsenide converter development,
Proc. 22nd IEEE Photovoltaic Specialists Confer-
ence, Las Vegas, USA, 1991
3.4.3 Time- and frequency-domainelectro-optic field mapping of nonlin-ear transmission lines
P. BUSSEK, TH. BRAASCH, AND G. DAVID
e report on the measurements of
electric field distributions in mono-
lithic microwave integrated circuits (MMICs)
using electro-optic probing techniques. In
1996 the activities have been focused on the
analysis of microwave propagation in non-
linear transmission lines (NLTLs). The mea-
surements have been performed in frequency
domain as well as in time domain as they
have been done in one dimension as well as
in two dimensions. As an example, in this
documentation we present experimental
results of periodic NLTLs demonstrating the
generation of higher harmonics on these
devices and the formation of shock waves.
IntroductionIn recent years, the complexity of monolithic
microwave integrated circuits (MMICs) expand-
ed necessitating the development of measure-
ment techniques which keep abreast of the in-
creased demands of an appropriate
characterization of these devices. So far, net-
work-analyzers (NWA) are mostly used for on-
wafer microwave characterization of MMICs.
This measurement technique is well established
but its application is limited due to the fact that
the on-wafer probes needed for this technique
only allow the access to external ports. Thus,
no circuit-internal measurement or local failure
test of the device nore the observation of wave
propagation effects is possible using NWAs.
In contrast, electro-optic sampling has be-
come a sophisticated technique to study quan-
Fig. 4: Photovoltaic cell efficiency
3 RESEARCH66
W
titatively field distributions and wave propagation
effects insight microwave and millimeter-wave
devices [1-3]. This technique can be performed
in frequency as in time domain enabling the
detection of the amplitude and phase of a
microwave signal as of the temporal evolution
of this signal. The spatial resolution of this meth-
od is measured to be down to less than 0.5 µm
[4]. Hence, each MMIC component can be test-
ed and evaluated noninvasively up to millime-
ter-wave frequencies. By combining the direct
electro-optic probing with 2D scanning of the
laser beam two-dimensional field mappings of
the device under test (DUT) are possible [5].
Thus, wave propagation effects can be studied
which gain a growing interest by circuit design-
ers for two reasons. On one hand, these effects
influence the electrical behaviour of MMICs re-
sulting in a limitation of their bandwidth or the
generation of unwanted modes [3]. On the other
hand, novel types of integrated circuits such as
nonlinear transmission lines (NLTLs) can be
designed which make use of these effects, e.g.
to generate short electrical pulses or to excite
higher harmonics. This has been observed par-
ticularly in periodic NLTLs and will be shown in
this report.
Experimental setupThe experimental setups used in this project
are sketched in Fig. 1. An actively modelocked
Nd:YAG laser (wavelength = 1064 nm, pulse
repetition rate = 82 MHz) combined with a fiber-
grating pulse compressor provides short pulses
of 5 ps FWHM (full width at half maximum) cor-
responding to a bandwidth of the setup in ex-
cess of 80 GHz. The device under test (DUT) is
illuminated from the backside, i.e. the direct elec-
Fig. 1: Experimental setup, (a) for the frequency domain measurements, (b) for the time domain
measurements
3.4 Optical Sensor Systems 67
tro-optic sampling is applied since the
linear electro-optic effect in the sub-
strate itself is used for the modulation
of the polarization. To convert this po-
larization modulation into an intensity
modulation, polarizers and a quarter-
wave plate are implemented in the op-
tical pathway. The reflected intensity is
detected by a small area photodiode.
Due to the combination of a small area
photodiode and the confocal arrange-
ment of the setup out-of-focus-light is
suppressed improving the spatial res-
olution of the measurement system
down to less than 0.5 µm [4]. For 2D
scans the probe stage is movable in
the x- and y-direction.
Fig 1(a) illustrates the experimental setup for
frequency domain measurements. Here, a spec-
trum analyzer working as a tunable bandpass is
used for the detection of the signal amplitude.
The intermediate frequency is set to several
MHz, since in this regime the high speed ava-
lanche photodiode used exhibits a maximum
sensitivity. The microwave synthesizer, the mod-
elocker synthesizer of the laser system and the
spectrum analyzer are phase stabilized via a
phase locked loop (PLL). For the phase mea-
surements the spectrum analyzer is replaced by
a lock-in amplifier (not shown in Fig. 1(a)). In a
second mode this setup is used to receive an
optical image of the measurement region by sim-
ply detecting its front surface reflectivity. Thus,
the electro-optical signal can be normalized to
the particular reflectivity of the device, and the
absolute value of the voltage between the de-
vices top and bottom surface can be determined
[6].
For the measurements performed in time
domain some modifications of the experimental
setup are needed. As depicted in Fig. 1(b), the
optical pulses themselves generate the electri-
cal signal in order to establish a phase locking
between the probe pulses and the electrical mi-
crowave signal. A small part of the output beam
of the Nd:YAG laser is separated via a beam
splitter and chopped at about 4 kHz. The photo-
current of a fast photodiode detecting this out-
put signal then traverses a mechanical delay line
that periodically shifts the phase of the signal
while the observation point is kept constant. The
photodiode has to be changed by a slow Ge-
diode to apply lock-in techniques at the chop-
ping frequency.
Frequency domain measurementsThe results presented here all have been
done with periodic nonlinear transmission lines.
For a more detailed description of the examined
samples see R. Hülsewede, Investigations of
0 500-140
-130
-120
-110
-100
-90
-80
1500 2500propagation distance (µm)
15 GHz
30 GHz
45 GHz60 GHz
Fig. 2: Electro-optic signal of the fundamental microwave
at 15 GHz and of its higher harmonics along the center
conductor of a periodic NLTL from input to output.
3 RESEARCH68
pulse compression on nonlinear transmission
lines, in this annual report. Fig. 2 depicts the
spatial distribution of the incident fundamental
electrical signal at 15 GHz and the amplitudes
of the second, third, and fourth harmonic with
frequencies up to 60 GHz. As can be seen the
amplitude of the fundamental signal decreases
in the direction of propagation whereas the am-
plitudes of the
higher harmon-
ics increase in-
dicating that
they are
g e n e r a t e d
along the trans-
mission line.
The obvious
standing wave
patterns are
caused by an
i m p e d a n c e
mismatch at
the end of the
line and phase
mismatching of
the harmonics.
Two-d imen-
sional field
mappings of
an NLTL are
shown in figs.
3. Here, the
frequency of
the fundamen-
tal is 6 GHz,
and the metal-
lization struc-
ture of the de-
vice, the electro-optic signal of the fundamental,
the second harmonic at 12 GHz and the third
harmonic at 18 GHz are presented in Figs. 3(a)
- (d), respectively. These Figs. show the decrease
of the fundamental signal and the increase of the
harmonics while propagating along the NLTL as
Fig. 2 does, but additionally they reveal an un-
symmetrical distribution of the electro-optic signal
Fig. 3: Nonlinear transmission line; (a) metallization structure; results of 2D field
mappings (b) at the fundamental at 6 GHz, (c) at the second harmonuc at 12 GHz
and (d) at the third harmonic at 18 GHz.
3.4 Optical Sensor Systems 69
that can not be detected with one-dimensional
linescans as is the case in Fig. 2. We contribute
this behaviour to the excitation of parasitic prop-
agation modes [3].
Time domain measurementsIn time domain measurements there is a fixed
phase relation between each particular mea-
surement point. Thus, the evolution of a period-
ic signal can be observed as is elucidated in
Figs. 4(a) to 4(d) for the development of a sinu-
soidal electrical signal of 6 GHz propagating
along a periodic NLTL at the 1st, the 5th, the
10th and the 14th diode, respectively. As can
be seen, shock waves are generated with fall
times down to 5 ps due to the interaction of the
higher harmonics. In the
time domain, this forma-
tion of a shock wave is the
counterpart to the
generation of harmonics in
the frequency domain.
The presented results
validate that the electro-
optic probing technique is
capable of studying and
demonstrating this effect
as well.
ConclusionIn summary, electro-
optic measurement tech-
niques have been used
to internally investigate
wave propagation effects
along periodic nonlinear
transmission lines en-
abling circuit-designers to
get an insight into the in-
circuit electrical characteristics of complex mi-
crowave devices. The generation of harmonics
and the formation of shock waves have been
demonstrated showing, that this method is suit-
able to examine internal field distributions in
MMICs in both, frequency domain and time do-
main.
References[1] K.J. Weingarten, M.J.W. Rodwell, and D. Bloom,
Picosecond optical sampling of GaAs integrated
circuits, IEEE J. Quantum Electron., vol. QE-
24, (1988), pp. 198-220
[2] G. David, S. Redlich, W. Mertin, R.M. Bertenburg,
S. Kosslowski, F.J. Tegude, E. Kubalek, and D.
Jäger (1993), Two-dimensional direct electro-
-80 -40 0 40 80
k = 1
(a)
-80 -40 0 40 80
k = 5
(b)
-80 -40 0 40 80
k = 10
(c)
-40 0 40 80-80
k = 14
(d)
time (ps) time (ps)
time (ps) time (ps)
Fig. 4: Electro-optic signal of the fundamental microwave at 15 GHz and
of its higher harmonics along the center conductor of a periodic NLTL
from input to output.
3 RESEARCH70
optic field mapping in a monolithic integrated
GaAs amplifier, Proc. 23rd EuMC 1993, Madrid,
Spain, 1993, pp. 497-499
[3] G. David, R. Tempel, I. Wolff, and D. Jäger,
Analysis of microwave propagation effects us-
ing 2D electro-optic field mapping techniques,
Optical and Quantum Electronics, Special Issue
on Optical Probing of Ultrafast Devices and In-
tegrated Circuits, vol. 28, 1996, pp. 919-931
[4] G. David, P. Bussek, U. Auer, F.J. Tegude, and
D. Jäger, Electro-optic probing of RF signals in
submicrometre MMIC devices, Electron. Lett.,
1995, Vol. 31, No. 25, pp. 2188-2189
[5] M.G. Li, E.A. Chauchard, C.H. Lee, and H.-L.A.
Hung, Two-dimensional field mapping of GaAs
microstrip circuit by electrooptic sensing, Proc.
OSA Int. Top. Meeting `Picosecond Electronics
and Optoelectronics`, March 13-15, 1991, Salt
Lake City, USA, pp. 54-58
[6] G. David, W. Schröder, D. Jäger, and I. Wolff,
2D electro-optic probing combined with field
theory based multimode wave amplitude extrac-
tion: a new approach to on-wafer measurement,
Symposium Digest 1995 IEEE MTT-S Interna-
tional Symposium, May 15 -19, 1995, Orlando,
USA, pp. 1049-1052
3.4.4 Characterization of monolithicmicrowave integrated circuits byheterodyne electro-optic sampling
TH. BRAASCH
he propagation of electric signals in
the millimeter- and microwave regime
along monolithic microwave integrated cir-
cuits (MMICs) can be studied by electro-op-
tic measurement techniques. In this paper,
we describe the implementation of a hetero-
dyne electro-optic measurement setup and
present first experimental results.
IntroductionUsing common network analyzer methodes
(NWA) for the characterization of MMICs, the
device under test (DUT) is measured as an in-
tegral device and no in-circuit measurements are
possible [1]. The increasing working frequencies
of integrated circuits up to the millimeter- and
microwave region [2-4] necessitate measure-
ment techniques that allow an insight into the
device. In recent years, electro-optic sampling
(EOS) has become a sophisticated technique
to observe field distributions in MMICs with a
spatial resolution down to less than 0.5µm [5,
6]. Quantitative characterizations have been
carried out, and one- as two-dimensional mea-
surements are possible in time-domain as in fre-
quency domain [7, 8]. Thus, circuit-designers get
a knowledge of circuit-internal parameters, which
is of increasing importance since the complexi-
ty of the devices expands. Nevertheless, so far
the EOS has been mostly performed with a
pulsed laser. Here, to convert the microwave
down to frequencies, where spectrum analyzer,
lock-in amplifier and the photodiode used are
able to detect the electro-optic signal, the n-th
harmonic of the repetition frequency of the
pulsed laser is used to interact with the electric
signal. The electric bandwidth of these setups
is limited by the pulse width of the laser pulses.
In frequency domain, measurements on nonlin-
ear transmission lines up to 100 GHz are re-
ported [9, 10], and fall times down to 1.5 ps have
been measured in time domain [11]. However,
utilizing the n-th harmonic of the repetition fre-
quency of the pulsed laser for the down conver-
sion of the electric signal leads to a reduction of
the signal to noise ratio of the electro-optic sig-
3.4 Optical Sensor Systems 71
T
nal since the phase noise of the setup increases
with the measurement frequency. To circumvent
this restriction two cw lasers can be used where
the second laser acts as a local oscillator. Here,
phase noise of the setup only depends on the
stability of the two lasers [12]. In this paper, we
present our measurement setup and show first
experimental results.
Experimental setupWe operate with two identical Er-doped fiber
lasers exhibiting a linewidth < 10 kHz. Both la-
sers are continuously tunable between 1530 nm
and 1560 nm leading to beat frequencies up to
4 THz. They deliver > 20 mW optical output and
show almost no mode-hops once they reached
thermal equilibrium. The degree of polarization
is > 99%. Fig. 1 demonstrates the configuration
of our setup. A small percentage of both lasers
is coupled into a Fabry-Perot optical spectrum
analyzer with a finesse >150 for the detection
of the beat fre-
quency between
the two lasers.
After passing a l/
4 polarization
control the light of
the first laser is
back-side cou-
pled into the de-
vice under test
where the stray
field of the micro-
wave in the sub-
strate interacts
with the laser
light via the Pock-
els-effect. The
reflected light is
again coupled
into the fiber and traverses the circulator and a
second polarization control. Thus, the polariza-
tion modulation due to the Pockels-effect is
converted into an intensity modulation directly cor-
related to the strength of the electric field at the
particular point of measurement. The DUT is
placed on a translational stage enabling two-
dimensional field mappings of the field distribu-
tion. Due to the confocal arrangement of the setup
it can also be used as an optical microscope. In
this mode, the front surface reflectivity of the
device can be detected and afterwards the elec-
tro-optic signal can be normalized to the particu-
lar reflectivity at each particular measuring point.
As a consequence, the absolute value of the
voltage between the devices top and bottom side
can be determined [8]. Via a fiber coupler the
local oscillator, i.e. the second laser, is
superposed to the reflected light from the DUT
carrying the information of the microwave signal
applied to the DUT. The intermediate frequency
Fig. 1: Sketch of the heterodyne electro-optic measurement setup
3 RESEARCH72
between the second laser and one sideband of
the first laser, i.e. f1 ± fm with f1 the frequency of
the first laser and fm the microwave frequency,
can now be adjusted by the frequency f2 of the
second laser. This intermediate frequency is de-
tected by a fast travelling-wave photodetector
[3]. Hence, any microwave frequency within the
tuning range of the two lasers can be converted
to some MHz or GHz only affected by the inherent
phase noise of the two lasers but independently
of the frequency.
ResultsA coplanar waveguide structure (CPW) was
used to demonstrate the feasibility of the setup
as a scanning microscope. Fig. 2 depicts the
surface reflectivity of the CPW. The difference
frequency between the lasers was arbitrarily set
to 8.5 GHz since at this value they worked ex-
tremely stable and the detected signal of the
spectrum analyzer was about 50 dB larger than
the noise floor. In the next step, a microwave
has now to be applied to the device and the elec-
tro-optic signal has to be detected as in [6], [8]
or [10] with the pulsed laser system.
ConclusionIn summary, to bypass the phase noise re-
strictions of a pulsed electro-optic measurement
setup two narrow linewidth tunable cw lasers
have been implemented in the configuration.
Thus, the phase noise only depends on the char-
acteristics of the lasers but is not affected by
the measurement frequency. Owing to the tun-
ing range > 30 nm of the Er-doped fieber lasers
used heterodyne detections of electric signals
up to 4 THz should be possible. As a first result,
the surface reflectivity of a coplanar waveguide
structure detected at 8.5 GHz difference fre-
quency is presented.
References[1] D.J. Bannister and M. Perkins, Tracebility for on-
wafer s-parameter measurements, IEE Proc. A,
vol. 139, 5, 1992, pp. 232-233
Fig. 2: (a) Sketch of a coplanar waveguide structure, (b) reflected intensity measured with the hetero-
dyne setup at 8.5 GHz.
3.4 Optical Sensor Systems 73
[2] M.J.W. Rodwell, S.T. Allen, R.Y. Yu, M.G. Case,
U. Bhattacharya, M. Reddy, E. Carman, M.
Kamegawa, Y. Konishi, J. Pusl, and R. Pullela,
Active and nonlinear wave propagation devices
in ultrafast electronics and optoelectronics, Proc.
IEEE, vol. 82, 7, 1994, pp. 1037-1060
[3] M. Alles, Th. Braasch, R. Heinzelmann, A. Stöhr,
and D. Jäger, Optoelectronic devices for micro-
wave and millimeterwave optical links, Proc.
MIKON96, Workshop Optoelectronics in Micro-
wave Technology, Warsaw, Poland, 1996 (in-
vited)
[4] I.V. Ryjenkova, V.K. Mezentsev, S.L. Musher,
S.K. Turitsyn, R. Hülsewede, and D. Jäger, Mil-
limeter wave generation on nonlinear transmis-
sion lines, Ann. des Telecomm., Special Issue,
1996 (invited)
[5] K.J. Weingarten, M.J.W. Rodwell, and D. Bloom,
Picosecond optical sampling of GaAs integrated
circuits, IEEE J. Quantum Electron., 1988, QE-
24, pp. 198-220
[6] G. David, P. Bussek, U. Auer, F.J. Tegude, and
D. Jäger, Electro-optic probing of RF signals in
submicrometre MMIC devices, Electron. Lett.,
1995, Vol. 31, No. 25, pp. 2188-2189
[7] M.G. Li, E.A. Chauchard, C.H. Lee, and H.-L.A.
Hung, Two-dimensional field mapping of GaAs
microstrip circuit by electrooptic sensing, OSA
Proc. Picosecond Electronics and Optoelectron-
ics, March 13-15, 1991, Salt Lake City, USA, pp.
54-58
[8] G. David, R. Tempel, I. Wolff, and D. Jäger,
Analysis of microwave propagation effects us-
ing 2D electro-optic field mapping techniques,
Optical and Quantum Electronics, Special Issue
on Optical Probing of Ultrafast Devices and In-
tegrated Circuits, 1996, 919-931
[9] R. Majidi-Ahy, B.A. Auld, and D.M. Bloom, 100
GHz on-wafer s-parameter measurements by
electro-optic sampling, IEEE MTT-S, 1989, pp.
299-302
[10] Th. Braasch, G. David, R. Hülsewede, U. Auer,
F.-J. Tegude, and D. Jäger, Propagation of mi-
crowaves in MMICs studied by time- and fre-
quency-domain electro-optic field mapping, Proc.
Trends in Optics and Photonics Series (TOPS)
of OSA 1997 Spring Topical Meeting Ultrafast
Electronics and Optoelectronics, 1997, Lake
Tahoe, USA
[11] K.S. Giboney, S.T. Allen, M.J.W. Rodwell, and
J.E. Bowers, Picosecond measurements by free-
running electro-optic sampling, Phot. Tech. Lett.,
vol. 6, 11, 1994, pp. 1353-1355
[12] S. Loualiche and F. Clerot, Electro-optic micro-
wave measurements in the frequency domain,
Appl. Phys. Lett. 61, (18), 1992, pp. 2153-2155
3.4.5 Development of an experimen-tal setup for field probe measure-ments on nonlinear transmissionlines
D. KALINOWSKI AND R. HÜLSEWEDE
n experimental setup for field probe
measurements has been established.
The electrical fields on nonlinear transmis-
sion lines has been measured up to 60GHz.
Theoretical considerations have been done
up to 200GHz. Experimental results have
been compared with results using electro-
optic testing.
IntroductionIn recent years there has been a great
progress in the development on nonlinear trans-
3 RESEARCH74
A
RsRv jXAUUl
Fig. 2: Equivalent circuit for the electric field
probe
mission lines making them to key structures for
future microwave circuits [1]. To characterize
these structures measurements at external ports
are not sufficient. Different noncontacting probes
give a chance to take a look at the field distribu-
tion on the transmission lines. The probes are
the electro-optic probe [2], the magnetic field
probe [3] and the electric field probe [4]. Such
an electric field probe has been established. Its
potential has been demonstrated.
Experimental setupA sketch of the experimental set-up is shown
in Fig. 1. The nonlinear transmission lines
(NLTLs) are supplied by a microwave syn-
thesizer and a DC-voltage source. They can be
loaded with variable resistance. The probe de-
tects the electric field above the NLTL. Its sig-
nal is evaluated by a spectrum analyzer. A com-
puter controls the position of the probe and
stores the data measured by the spectrum ana-
lyzer. By that way two-dimensional field map-
pings can be done.
Theoretical resultsThe probe can be described by an equiva-
lent circuit which is shown in Fig.2 [5]. The volt-
age Ul depends on the electric field strength.
The probe impedance is given by RV and RS which
characterize the thermal and radiation losses and
XA describing the open line.
The relation between the indicated power at
the spectrum analyser and the square of the
measured field intensitity is shown in Fig. 3. It
demonstrates the steady increasing probe sen-
sitivity versus frequency. Thus the probe can
be used over the whole frequency range.
Measurements confirm this equivalent circuit.
Fig. 4 shows the relation between the indicated
signal on the spectrum analyser and the signal
frequency. The experimental results are given
by the dots (·), the theoretical results by the line
(-). A good correspondence is given between
these results. So the equivalent circuit can be
used to determine the electric field strength by
the spectrum analyzer signal.
0 40 80 120 160 2000
0,2
0,4
0,6
0,8
frequency (GHz)
sens
itiv
ity
(
)
pA Vm
Fig. 3: Sensitivity versus frequency.
3.4 Optical Sensor Systems 75
xy
z
NLTL
mixer synthesizer
dc - source
load
spectrum-analyser
bias tee
prober
Fig. 1: Sketch of the experimental Setup
Fig. 5:2-dim. field mapping of an NLTL (a)
signal with 7.4 GHz, (b) generated 3rd
harmonic at 22.2 GHz c) Sketch of the NLTL
Experimental resultsOne experimental result achieved with this
setup is presented in Fig. 5. A 4 µm NLTL has
been examined. The sketch of this line is shown
in Fig. 5c. Whereas one end has been connect-
ed with a synthesizer, the other end has been
unloaded. The synthesizer has supplied the line
with a 7.4 GHz microwave signal. The field dis-
tribution at this frequency is shown in Fig. 5a.
The 3rd harmonic generated on the line is shown
in Fig. 5b. The amplitude increases in the direc-
tion of propagation indicating the generation
along the transmission line. Furthermore an
asymmetrical transversal distribution is revealed.
ConclusionMeasurements up to 60 GHz have been done
successfully. Field distributions on transmission
lines with electrode widths down to 12 µm has
been shown. Comparisons with theoretical re-
sults and electo-optic probing are in good agree-
ment.
References[1] M.J.W. Rodwell, et al., Active and nonlinear
wave propagation devices in ultrafast electron-
ics and optoelectronics, IEEE Proc., Vol. 82, No.
7, pp. 1037-1059, 1994
[2] P.Bussek, G. David, Quantitative analysis of
two-dimensional electro-optically measured field
distributions in MMIC-structures, Annual Report,
Gerhard-Mercator-Universität - GH - Duisburg,
Fachgebiet Optoelektronik, 1995
[3] Y. Gao, I. Wolff, A new miniature magnetic field
probe for measuring three-dimensional fields in
planar high-frequency circuits, IEEE Transac-
tions on Microwave Theory and Techniques, Vol.
44, No. 6, June 1996, pp. 911-918
3 RESEARCH76
5 10 15 20 25 30 35 40-50
-46
-42
-38
-34
-30
-26
measurement
frequency (GHz)
theory
rel.
sign
al (
dB)
3.5 Technologies for Optoelectronic Components and Systems 77
3.5 Technologies for Opto-electronic Components andSystems
3.5.1 Development of a measurementsystem for the optical characteriza-tion of full-colour-LED-displays
M. WENNING, R. BUß, AND A. STÖHR
or the optical characterization of a full-
colour-LED-display, two measurement
systems have been developed. The first one
is to determine the spatial distribution of a
LED and the second one is to receive the
spectrum of a LED or a LED-pixel. The x, y, z
colour coordinates of the CIE chromaticity
diagram are evaluated from the spectrum. By
using the measurement system, an
optimization of a full-colour LED-display was
performed.
IntroductionAmong the various types of flat panel displays
(i.e., CRT, VFD, PDP, LCD, LED and EL), LED
displays are widely used as information boards
and as transportation terminal displays due to
their excellent reliability, service life and visibili-
ty.
Particularly as a result of the remarkable
progress made with high-brightness blue and
green LEDs, full-colour displays can now be
established for outdoors. Any colour can be pro-
duced using the three primary colours red, green
and blue. In this report, measurement systems
are developed to characterize a full-colour LED-
display. Furthermore, the LUMINO XTralux ML
4 C-Pixel and an optimized Pixel has been char-
acterized.
[4] D. Kalinowski Entwicklung eines Feldsonden-
meßplatzes zur zweidimensionalen Analyse
elektromagnetischer Feldverteilungen auf
nichtlinearen Leitungen, Diploma thesis,
Gerhard-Mercator-Universität Duisburg, 1996
[5] R. Geißler, et al., Taschenbuch der Hochfre-
quenztechnik Band 2: Komponenten, Springer-
Verlag, Berlin-Heidelberg, 1992
F
Measurement systemsThe spatial distribution of a LED is measured
with the setup shown in Fig. 1. The LED is fixed
on a LED-holder and is driven by a constant
current. By rotating the
swivel-arm in 1°- steps,
the data of the spatial
distribution is received.
The Fig. 2 shows
the setup to determine
the spectral distribu-
tion of a LED or LED-
pixel. After the spec-
tum is measured, the
x, y, z colour coordi-
nates are determined
[1,2].
The CIE (Commis-
sion Internationale de
lEclairage) diagram is
the standard colouri-
metric system. The x, y, z axis of this diagram
are based on three colour-matching functions,
each of which is related to the spectrum of red,
green and blue. A sequence of single-wavelength
computer
IEEE
Lock-in-amplifierIn Ref.
chopper
single-LED orLED-Pixel
light stop with aperture
lens hole
detector =lens + photodiode
monochromatorwith
stepper motor
Fig. 2: Measurement setup - spectrum
3 RESEARCH78
0,0 0,2 0,4 0,6 0,80,0
0,2
0,4
0,6
0,8
yellowish greengreen
red
blue
single-wavelenght colours
D65
y
x
Fig. 3: CIE diagram of all LEDs
0,0 0,2 0,4 0,6 0,80,0
0,2
0,4
0,6
0,8single-wavelenght colours
D65
red
green
blue
y
x
0,0 0,2 0,4 0,6 0,80,0
0,2
0,4
0,6
0,8single-wavelenght colours
D65
red
green
blue
y
x
Fig. 4: (a) CIE diagram of LUMINO Xtralux-pixel and (b) of the optimized pixel.
colours can be expressed as a curve in the x, y,
z space of the CIE diagram and the projection of
the curve on the x, y plane is a horseshoe-shaped
pattern (Fig. 3). Any colour can be expressed as
a point inside of this horseshoe-shaped curve.
Experimental resultsThe colour coordinates of all measured LEDs
are shown in Fig. 3 which are determined from
the spectrum of the LEDs. In Fig. 4 (a) every
colour in the triangle region, of which the verti-
ces indicate the three primary colours of the
LUMINO -Xtralux LEDs, can be radiated by ad-
justing the luminous intensity of each LED. As
0,0 0,2 0,4 0,6 0,80,0
0,2
0,4
0,6
0,8
80°
60°
40°20°
0°
single-wavelenght colours
D65
y
x
0,0 0,2 0,4 0,6 0,80,0
0,2
0,4
0,6
0,8
90°
80°60°
40°20°0°
single-wavelenght colours
D65
y
x
Fig. 5: (a) Colour-shift of LUMINO-Xtralux-pixel and (b) of the optimized pixel.
3.5 Technologies for Optoelectronic Components and Systems 79
(a) (b)
(a) (b)
0 20 40 60 800,0
0,2
0,4
0,6
0,8
1,0
green
blue
red
viewing angle /°
rela
tive
Int
ensi
ty
0 20 40 60 800,0
0,2
0,4
0,6
0,8
1,0
green
red
blue
viewing angle /°
rela
tive
inte
nsi
tyFig. 6: (a) Spatial distribution of the LUMINO-Xtralux-pixel and (b) of the optimized pixel.
seen in the diagram, it is not possible to obtain
the standard white D65, which is outside the tri-
angle. The colour coordinates of an optimised
Pixel are shown in Fig. 4 (b). The LEDs are well-
chosen to obtain a greater triangle region. In
Fig. 5 the colour coordinate variation versus the
viewing angle of these two Pixels is shown. The
XTralux ML 4 C-Pixel (Fig. 5 (a)) has a large co-
lour shift to the red primary colour, which can
explained with a wider spatial distribution of the
red LED (cf. Fig. 6 (a)).
The small shift of the optimized Pixel was
performed by using a red, green and blue LED
having nearly the same spatial distribution. This
was obtained by modification of the lense-form
(encapsulation) of the LED. Furthermore, the
LED-surface is roughend.
ConclusionsWithin the scope of this thesis, two measure-
ment setups have been developed. Further-
more, LEDs and the LUMINO XTralux ML 4 C-
Pixel have been characterized. With the
knowledge of colour metrics and the colour co-
ordinates of the LEDs, an optimized Pixel has
been assembled. A greater colour range and a
nearly constant colour coordinate versus view-
ing angle of the pixel has been achieved.
References[1] Heinweg Lang, Farbmetrik und Fernsehen,
R.Oldenburg München Wien, ISBN 3-486-
20661-3, 1977
[2] DIN 5033, Farbmessung
3.5.2 Opinion poll on the evaluationof the legibility of LED-based dis-plays
R. HEDTKE AND R. BUß
n this report the legibility of LED-based dis-
plays is evaluated on the base of
interviews with passengers of the public
local traffic. To reach this aim a model ex-
plaining the causal connection between
several external influences and the legibility
is made up according to DIN 1450. Based on
this model a questionnaire for the interviews
3 RESEARCH80
(a) (b)
I
is developed. The gained data of the inter-
views is evaluated using statistic methods.
IntroductionToday, the permanent availability of informa-
tion has become of increasing importance,
where the transmission of visual information cer-
tainly becomes to play a more and more impor-
tant role. Due to the permanent development in
the area of LED-technology it has become pos-
sible to produce so-called super-luminescence
light emitting diodes (SLED), having a very high
brightness. In the public local traffic LED-based
displays find an increasing application. In co-
operation with the company LUMINO/Krefeld, a
producer of such displays, methods to improve
the legibility of those display-sys-
tems (cf. Fig. 1) are acquired.
The causal connectionThe causal connection based
on DIN 1450 is modified referring
to the requirements of the ques-
tionnaire. The result is shown in
Fig. 2 where the arrows describe
the influences. It can be seen that
there are eight major points hav-
ing an influence on the legibility:
The type face and size, brightness and colour,
the distance between each letter, the distance
of view, the light conditions, and personal influ-
ences.
The used questionnaireBased on the causal connection, the ques-
tionnaire shown in Fig. 3 has been developed
to proof and determine the level of influence of
each point. The first three questions are so called
icebreaker-questions to start the conversation
with the person to be interviewed. The following
questions are related to the legibility of the dis-
played text. Legibility, type size, distance be-
tween each letter, type face, brightness, and
colour are assessed using marks between one
and five. This type of assessment has been cho-
sen because everyone
knows this marks from
school. This causes
that comprehension
problems are avoided.
Furthermore, tenden-
cies like small, right,
and large where re-
corded if possible. Fi-
nally, the other points
Fig. 1: LED-based display used in the public local traffic.
TYPE FACE BRIGHTNESS COLOUR
TYPE SIZE LEGIBILITYDISTANCE BETWEEN LETTERS
DISTANCE OF VIEW LIGHT PERSONALINFLUENCES
Fig. 2: Model of the causal connection.
3.5 Technologies for Optoelectronic Components and Systems 81
Fig. 3: The used questionaire
3 RESEARCH82
of influence are recorded after the main part of
the interview.
InterviewFor the interviews, locations in the following
cities are selected according to the practicabili-
ty of the interviews and considering the local
conditions: Duisburg, Essen, Düsseldorf; Ober-
hausen, Leipzig, Stuttgart.
At each of these locations interviews with 50
passengers have been carried out. To reach a
high comparability of the gathered data, the in-
terviews only took place on platforms for the pub-
lic local traffic.
ConclusionThe Fachgebiet Optoelektronik at the Ger-
hard-Mercator-Universität Duisburg has inves-
tigated the judgement of the legibility of
LED-based displays in co-operation with the
company LUMINO/Krefeld, a producer of such
displays. The aim of this study was the record-
ing of subjective opinions of the users of the
public local traffic.
The legibility of the analysed LED-based dis-
plays has been valued by up to 90% of the in-
terviewed passengers with well or very well.
Especially the displays based on yellow LEDs
in the city of Leipzig have been rated very pos-
itively. The majority of the passengers have felt
the brightness to be right. As well in Leipzig, the
valuation has been the same, even if the dis-
play was exposed to direct sunlight. As a rule,
type size, letter distance, and type were also
valued as right. Up to 80% of the passengers
have felt as being informed very well by those
displays.
In summary, the users of the dynamic pas-
senger-information-system are contented with
the quality and information provided by these
systems. It should be pointed out that the LED-
technology serves the high requirements refer-
ring to the demands in the public local traffic.
Even under unfavourable conditions like direct
sunshine the judgement is good or very good.
References[1] DIN 1450, Beuth, Berlin, Juli 1993
[2] V. Dreier, Datenanalyse für Sozialwissen-
schaftler, Oldenbourg, München-Wien, 1994
[3] K. Holm (Hrsg.), Die Befragung 1, Franke, Mün-
chen, 1975
[4] E. Noelle, Umfragen in der Massengesellschaft,
Rowohlt, Reinbek bei Hamburg, 1963
3.5.3 Evaluation of possible improve-ments to enhance the UV-power effi-ciency of a xenon flashlamp system
B. NEUHAUS AND A. STÖHR
his paper will present experimental
results of the characterisation of a
xenon flashlamp system with a flat alumini-
um rear reflector fabricated by Bläsing Elek-
tronik GmbH. The spatial distribution of the
UV-radiation is determined. Studies about
typical gas components and some materials
of the discharge tube result in possibilities
to optimize the arc lamp. Furthermore the
shape of a single flash is recorded and the
efficiency is measured in dependence of the
pulse repetition rate. To optimize this sys-
tem several rear reflectors with different geo-
metrics are tested and the reflection index
of five different materials is measured in the
UV and NIR wavelength range. For all these
examinations different measurment setups
are worked out.
3.5 Technologies for Optoelectronic Components and Systems 83
T
IntroductionThe UV-drying process is of great significance
to the industry working with printing procedures
e.g. the silk-screen printing. New UV-paints are
free from solvents and they harden only by irra-
diation with UV-light [1]. Therefore powerful
lamps with great emission in the UV-range are
very important. Pulsed UV-lamps offer several
advantages compared with conventional UV-
burners. With flashlamps the intense heat emis-
sion can be reduced; they radiate only for a short
pulse duration but the hardening process is more
effective because the radiation output of
flashlamps is greater than of conventional lamps
[2]. The UV-flash drying process will become
an economical and nonpolluting alternative pro-
cess in the future.
The intention of the following investigations
is to characterize such a flashlamp system and
to optimize it based on the experimental results.
Arc lamp construction and priniciple of aflashlamp system
Fig.1 shows the construction of a typical arc
lamp. The electrodes are set in a clear quartz
tube. The type of quartz depends on the desired
output spectrum. The electrodes are made of
tungsten to enhance
electron emission. Arc
lamps are filled with inert
gas under several atmo-
spheric pressure or a
mixture of gas and a def-
inite amount of mercury.
The internal pressure in
the tube increases during
operation to 15-75 bar,
depending on the lamp
type.
The flashlamp system operates by sending
an electric charge from a pulse generator to the
gas filled lamp. The gas absorbs the energy by
storing it in its atoms and subsequently it re-
leases the energy by emitting photons, which
results in a high intensity flash of light [3]. Light
emssions in all directions can be guided by in-
stalling a reflector behind the lamp which col-
lects the light and reflects it onto a surface which
should be treated. Caution: these lamps produce
high intensity UV radiation and ozon. Precau-
tions are necessary during operation mode.
Experimental setupsFig. 2 and Fig. 3 show the two basic experi-
mental setups used for the measurements. In
order to determine the spatial distribution of
emitted radiation and to determine the UV-effi-
ciency of the flashlamp in the UV-range the set-
up in Fig.2 is used. The pulse generator sup-
plies the flashlamp with electrical impulses. The
pulse repetion rate (1,56Hz, 3,12Hz, 6,25Hz,
12,5 Hz, 25 Hz) can be adjusted by a rotary
switch. The light pulses, radiated by the
3 RESEARCH84
cathode
anode
tube
filling gas
+
-
Fig. 1: Con-
struction of arc
lamps [2].
Fig. 2: Experimental setup to measure the
relative optical output of an arc flashlamp.
flashlamp, are detected by a special
silicon carbide (SiC; spectral range: 210-
380nm) photodiode.
Fig. 3 shows the setup to determine
the UV-reflectivity of different reflector
materials. The light beam of a mercury
lamp is deflected by a beam splitter in
definite directions. The reflector reflects
the incoming light beam which is then
focused on a SiC photodiode. By us-
ing the second detector it is possible
to compare the input power with the re-
flected power in order to determine the
reflection index.
For all investigations in the NIR
wavelength range a monochromator
and a silicon photodiode (BPW20; 375-1100nm)
are used to receive spectral values; the light
source is a halogene lamp. To record a single
pulse an oscilloscope is used.
Experimental resultsa) Fig. 4 shows the relative optical output pow-
er with respect to time of the flashlamp type
UVL-1500/TP1 fabricated by Bläsing Elektron-
ik GmbH. The pulse shape has a fast rise time
t r( % %)10 90− of 31,5ms and slower decay. The
pulsewidth t FWHM is 520ms. These are typical
values for flashlamps [2]. Further experiments
have shown, that the flash energy as well as
the pulswidth stay constant for all frequencies
in the range from 1,56Hz until 25Hz.
b) In order to determine the spatial distribution
of the radiation from the flashlamp type UVL-
1500/TP1 the SiC-photodiode is led along a
semicircle around the lamp as shown in Fig.
5. One can record the radiant intensity (rela-
tive) for a number of angles. The radius is r =
100cm. The values are recorded in steps of
one degree and they are related to the 90°-
direction. Fig. 6 shows the results (normal
curve) given in a polar coordinate system. It
can be seen that the distribition of the radia-
tion is symmetrical to the axes. Besides, it
can be noticed that over an angle of a= 168°
fifty percent of the radiant power compared
with the power in direction of 90° is still radiat-
3.5 Technologies for Optoelectronic Components and Systems 85
Fig. 3: Experimental setup to measure the
reflection index.
t sFWHM = 520µ
t sr( ,10% 90%) 315− = µ
− 2 5. ms 2 5. ms0 0. s
500µs div/
op
tica
l po
wer
(r
el.)
0
t ime t
Fig. 4: Relative optical output power of the flashlamp
UVL-1500/TP1.
ed by the flashlamp. So the angle of radiation
is very wide. Mostly such a wide angle is un-
desirable because one looses most of the
radiant in the borderlands when no object is
placed very closed to the radiant source. For
plenty of applications a distribution like a club
(dashed curve) is desired. It is possible to
achieve such a distribution with special rear
reflectors.
c) Fig. 7 represents the UV-efficiency in addic-
tion to the pulse repetition rate from the
flashlamp type UVL-1500/TP1. The values
are calculated for a wavelength range from
210-380nm as well as for the range from 270-
280nm. The detector was about two meters
away from the light source and the measure-
ments were repeated several times for all rep-
etition rates which are adjustable. From these
values a mean value is formed for each fre-
quency. This mean value is used for the cal-
culation of the efficiency for flashlamps. The
illustration shows that the efficiency stays
nearly constant for each frequency. The con-
3 RESEARCH86
flashlampUVL 1500-TP/1
SiC-detectorJEC-1210-380nm
r = 100cm
Fig. 5: Principle to record the spatial distribu-
tion of the radiant.
Fig. 6: Principle to record the spatial distri-
bution of the radiant.
effic
ienc
y
ηUV270 280−
ηUV210 380−
η ⋅ −( )10 9
pulse repetition rate f (Hz)Pu ls
Fig. 7: UV-efficiency in addiction to the pulse
repetition rate.
780 800 820 840 860 880 900 920 940 960 980 1000-5
0
5
10
15
20
25
30
35
40
45
aluminium UV-mirror reflector III reflector II reflector I
IR-r
efle
ctiv
ity
wavelength (nm)
Fig. 8: .Spectral IR-reflection index of differ-
ent materials.
clusion is, that the pulse repetition rate has no
essential influence on the efficiency at least at
small frequencies below f PULS = 25Hz.
d) The printing industry is interested in rear re-
flectors with a high reflection index in the UV-
range for great UV-efficiency and a low re-
flection index in the IR-range because of the
undesirable heat, which is produced by infra-
red radiation. Therefore five different materi-
als are tested for the reflectivity in the UV
(210-380nm)- and IR-(780-1000) wavelength
range. These five materials and their UV-re-
flection index are:
- highly polished aluminium, used so far by
Bläsing Elektronik GmbH; the UV-reflec-
tion index is ρUV = 90,24%.
- a special UV-glass-mirror that transmits the
IR-radiation and reflects the UV-radiation;
the UV-index results in ρUV = 97,17%.
- three metallic reflectors I, II and III with dif-
ferent coatings; The compositions of these
three coatings and their designation are
unkown.
During the measurements it was apparent,
that the three reflectors I, II and III are in all prob-
ability diffused and diffused/directed reflecting
materials with the consequence
that the absolute UV-reflection
index could not be determined.
But a second measurement meth-
od could prove, that all five mate-
rials could increase the radiant
power better than the aluminium
reflector.
Fig. 8 shows the spectral IR-
reflectivity of these five materials
over a wavelength range from
780-1000nm. The aluminium re-
flector is worse than the other four
materials. In comparison with all
3.5 Technologies for Optoelectronic Components and Systems 87
groups and experimental results the reflector
coatings I and III are most suitable to optimize
the xenon flashlamp system.
Proposals for improvementsThe proposals to optimize the flashlamp sys-
tem can be divided in four groups:
1. gas filling
2. tube material
3. reflector material
4. reflector form
Gas fillingFour typical fillings for arc lamps are xenon
(Xe), mercury (Hg), the mixture mercury-xenon
(Hg(Xe)) and deuterium (D 2 ). To compare the
effectiveness of these fillings in the UV-area the
efficiency as a function of the input power of dif-
ferent arc lamps with these gases is calculated
over a range from 210-380nm. In Fig. 9 this com-
parison is shown. To achieve these results some
spectral irradiance curves from the L.O.T.-Oriel
company [2] catalog are evaluated. The xenon
gas lamps are plainly worse than the other
lamps. Increasing the lamp power does little in-
fluence to the UV-efficiency of the xenon burn-
D2-deuterium
Xe-xenon
Hg-mercury
Hg(Xe)-mercury/xenonη *10 9−
ηUV210 380−
lamp power (W)
Fig. 9: UV-efficiency ηUV210 380− of different discharge lamps.
er. The mixture Hg(Xe) provides the best optical
radiation power but to get the optimal power a
warm up time of about 15 minutes is necessary.
Tube materialQuartz is undoubtedly the best material for
the discharge tube and it is normally used by
the industry. Quartz guaranties the mechanical
and thermal durability. The type of quartz de-
pends on the desired UV-output. There exist
several special quartz types: a) UV grade quartz
that transmits the output to below 200nm, and
b) ozone free quartz which absorbs short wave-
lengths to prevent ozone generation. Above
280nm the special types do not offer advantag-
es compared with standard quartz variants.
Reflector materialLiterary investigations give some new infor-
mation about reflector materials to optimize the
radiation power. a) Labsphere Ltd. company
developed a diffuse reflecting material, spectral-
on, with reflectivity over the range from 250-
2500nm shown in Fig. 10. Spectralon is a ther-
moplastic resin and it is thermally
stable to > 350°C. The reflectance
is > 95% over this range. Surface
contamination only decreases the
reflectance at the lower ends of
the spectral range. B) metallic re-
flectors are very sensitive to sur-
face contamination and to over-
heating. These facts can decrease
the reflectance greatly as well as
the permanent irradiation with UV-
light.
Reflector formFig.11 shows how the spatial
distribution of radiation could be
changed when two aluminium reflectors with dif-
ferent forms are used. These forms are described
in Fig.12 and Fig.13. The influence on the distri-
bution is tested when the distance between the
tube and the reflector varies. The values are
related to the 90°-direction and to the values of a
refl
ec
tan
ce
wavelength (nm)
Fig. 10: Spectral reflection grade of spectralon [4]
0
20
40
60
80100
120
140
160
1801,51,5 1,251,0 1,01,25
angle (°)
without reflector R1 1,5 cm R1 4,0 cm R1 6,5 cm R2 1,5 cm R2 4,0 cm R2 6,5 cm
Fig. 11: Spatial distribution of the radiation
with the reflectors I/II in dependence on the
distance between tube and reflector.
3 RESEARCH88
measurement without a reflector. So you are in
position to determine the increase of the radiant
intensity as well. The reflector R2 could increase
the radiant power best, with nearly 40% com-
pared with the
values without a
reflector. The dis-
tance between
the tube and the
reflector is impor-
tant as well. The
reflector R1 pro-
duces the best
results at a dis-
tance of d=4,0cm
whereas the
second reflector
was best at a dis-
tance of
d=1,5cm.
ConclusionThe analysis of
the flashlamp system of Bläsing Elektronik GmbH
provided the following results: The flashlamp ra-
diates only for the duration of a pulse. The pulse
repetition rate up to 25Hz has no signifcant
influence on the radiation intensity. The spatial
distribution of the radiation is extremly wide. Over
an angle of 168° fifty percent of the radiant power
compared to the power in direction of 90° is still
radiated by the flashlamp.
The investigations to optimize this system
provided the following possibilties: The mixture
mercury-xenon has the best UV-efficiency and
it is suitable to use for fillings in arc lamps. The
change of the reflector form and material could
increase the radiant power as well. First exper-
imental measurments point to the assumption that
e.g. the combination of the reflector material III
with the form R2 which could both increase the
intensity best, would provide a better efficiency.
References[1] Erhardt D. Stiebner, Bruckmann´s Handbuch
der Drucktechnik, Bruckmann, München 1992
[2] L.O.T. ORIEL catalog Vol.II, Light Sources,
Monochraomators & Spectrographs, Detectors
& Detection Systems, Fiber Optics, Oriel Cor-
poration, USA, 1994
[3] Polygon flashlamps http://www.polygon1.com/
technology.html
[4] Labsphere catalog; Diffuse reflectance Coatings
And Materials, Labsphere, North Sutton, 1996
3.5.4 Construction of a flip chip de-vice for bonding integrated circuits
J. ERVENS AND R. BUß
o bond integrated circuits in flip chip tech-
nology a heating device is required, enabling
precise adjustment and soldering of the sol-
der bumps. This device was designed,
built and put into operation. Further-
more, an alternative process to electroplat-
ing [1] was tested to put solder bumps onto
microelectrodes. Therefore, testing chips
were constructed on which solder bumps
were evaporated. In the completed device test
soldering points were carried out and ana-
lyzed.
IntroductionIn the age of space-saving integration of semi-
conductor circuits the use of flip chip technolo-
gy is getting more and more interesting. A spe-
cial advantage is the possibility to connect silicon
technique to III-V-compound semiconductors,
Fig. 12: Cross-section
of the reflector R1.
Fig. 13: Cross-section
of the reflector R2.
3.5 Technologies for Optoelectronic Components and Systems 89
T
which are of high importance to optoelectronic
applications.
Optoelectronics have many components with
vertical radiation. Therefore a direct connection
of the silicon substrate to the light emitting chip
is very useful for the third dimension. Another
point of interest is the self-adjustment of the
chips by the surface tension of the melted sol-
der bumps. As shown in Fig. 1 an alignment er-
ror of several micrometer in the adjustment can
be balanced out.
Description of the mode of operation of theheating device
The complete equipment consists of
- heating unit
- adjustment unit
- temperature control
- temperature measuring instrument
The heating unit is shown at the top of Fig. 2.
It comprises the halogen radiators and the mir-
ror reflectors and serves for fastening the sam-
ple holding device. In the device, which consists
of 10 millimeter thick aluminium plates, the chips
get soldered, lying on top of each other. A mem-
brane pump sucks the top chip to a heat-resis-
tant pane of glass as shown in Fig. 3. The bot-
tom chip is put down on an appliance fastened
to the adjustment unit which consists of a rotary
table, that can be moved by hand, and an x-y-z-
manipulator, see Fig. 3. The adjustment of the
chips can be observed with an ocular from the
top.
After the adjustment is complet-
ed in x- and in y-direction the
bottom chip is brought into con-
tact with the top chip by the z- ma-
nipulator. Then the nitrogen valve
is opened and a low pressure is
adjusted by a manometer, to
heat the chips in a nitrogen atmosphere. This
disables oxidation of the surfaces of the solder
bumps. The halogen radiators are started and
controlled by the temperature control unit, which
consists of a phase-angle control, that influenc-
es the electric power. By means of a rotary po-
tentiometer the temperature of the radiators is
also influenced. A fast-response thermocouple
juts as a measuring sensor into the heating unit.
This digital measuring instrument shows the pre-
dominant temperature inside. After about ten min-
utes the chips are soldered and after cooling
down thay can be taken out off the heating unit.
Fig. 2: Front view of flip chip device.
Fig 1: Self-adjustment of solder bumps [2].
3 RESEARCH90
Construction of test chipsAt first the substrate is covered with a steel
resist pattern mask and fastened in the deposi-
tion apparatus. Then a gold layer is evaporat-
ed, because the adhesive bond strength of gold
on the substrate is significantly higher than that
of soft solder. The soft solder layer is evaporat-
ed in several steps of the operation.
Results of the test seriesThe temperature in the heating unit is suffi-
cient in order to melt the solder bumps and to
connect the chips. In case of stronger mechan-
ical demand the gold layer dissolves off the sub-
strate, which means the soldering of the solder
bumps was successful. Self-adjustment can not
be recognized, because the layer thickness
achieved in the evaporation process is far too
thin. If a higher layer thickness can be achieved,
evaporation will be applicable, but regarding to-
days knowledge electroplating is preferred.
References[1] G. Sadowski, D. Zeidler: Mikrogalvanik für die
Herstellung lötfähiger Bumpsysteme, me Bd 6,
1992, pp 358-361
[2] M. Wale, M. Goodwin: Flip-Chip Bonding Opti-
mizes Opto-ICs, Circuits and Devices, pp 25-
31, Nov. 1992chip 1chip 2
xy
z
sealing ring
diminished pressure panes of glass
rotatable and inx-y-z-directionshiftable appliance
Fig. 3: Principle of chip arrangement.
3.5 Technologies for Optoelectronic Components and Systems 91
4 TEACHING ACTIVITIES92
4 Teaching activities
4.1 Lectures, Excercises, andpractical studies
Technical Electronics 3: Optoelec-tronics
D. Jäger and A. Stöhr
The course Technical Electronics 3: Opto-
electronics covers the basic theory and tech-
nology of modern semiconductor photonic de-
vices as well as applications of these devices in
optoelectronic integrated circuits (OEICs). The
course starts with the fundamental physical phe-
nomenon of light-material interaction in semi-
conductors, such as fundamental absorption,
spontaneous and stimulated emission. Subse-
quent lectures deal with the theory and technol-
ogy of photoconductive devices, photodiodes,
modulators, light emitting diodes (LEDs), and
laser diodes. Special attention is given to mod-
ern quantum well waveguide laserdiodes and
their applications in optical communication sys-
tems, medicine, and material processing.
Ultra High Frequency TransmissionTechniques: Optical Signal Trans-mission
D. JÄGER AND R. BUß
The course Ultra High Frequency Transmis-
sion Techniques: Optical Signal Transmission
starts with the propagation of electromagnetic
waves considering the features of optical waves
at surface boundaries, such as reflection and
refraction. Proceeding with the description of
such fundamental physical effects like scatter-
ing, absorption and dispersion, optical wave
propagation in various types of dielectric
waveguides is discussed. Based on this funda-
mentals the design, properties and technologi-
cal realization of waveguides based on III/V com-
pound semiconductors are discussed. Another
main part of this course deals with fiber optic
waveguides: Wave propagation in graded index
fibers as well as in stepped index fibers is de-
rived where both advantages and disadvantag-
es of each type are elucidated. Problems such
as signal distortion in fibre optic waveguides are
analyzed and solutions to avoid them are giv-
en. Following the topic of wave propagation, the
most important devices for optical and optoelec-
tronic integrated circuits (OEIC) are presented.
The properties and technological realization of
waveguide laser diodes, vertical cavity surface
emitting laser diodes (VCSEL), modulators, and
detectors are discussed. Finally, economical
aspects of optical communication techniques
and future prospects like fiber to the home are
touched
Special Areas of Optoelectronics:Lasers
D. JÄGER AND A. STÖHR
The first lectures within the course Lasers
cover the basic principles and the mathemati-
cal description of electromagnetic waves. The
course proceeds with the quantum mechanical
interactions between electromagnetic waves
and atomic materials resulting in the two most
important requirements for light amplification by
stimulated emission of radiation (laser). Special
attention is then given explaining the basic con-
cepts, the functionality, and the characteristic
4.1 Lectures, excercises, and practical studies 93
specifications of different laser sources of im-
portance, such as the Helium-Neon laser, the
Ar-ion laser, Excimer lasers, the Ti:Sapphire la-
ser, semiconductor laser diodes etc.. Finally,
examples of laser applications in various indus-
trial areas (medicine, communication, material
processing etc.) are discussed together with fu-
ture trends.
Optical Signal Processing
D. JÄGER AND R. BUß
The course Optical Signal Processing starts
with the basic theory of non-linear optical effects
both in dielectric materials and in semiconduc-
tors. The causes for optical bistability are de-
scribed and principles like optical switching are
applied to the realization of optical memories
and logic elements. Within the next section of
this course, the phenomenon of opto-electronic
bistability is introduced. It is shown that the inte-
gration of a light modulator and a photodetector
is leading to so-called self-electro-optic effect
devices (SEED), showing various forms of
switching behaviour which can be controlled
both optically and electrically. Finally, the main
advantages of optical signal processing are
pointed out while discussing applications such
as optical switching networks, image process-
ing systems, optical neural networks, optical
phased array antennas, optical computing, and
optical interconnects.
Multimedia-Techniques
D. JÄGER AND R. BUß
This course elucidates Multimedia from
three different points of view: The optoelectron-
ic area, the informatic area and the area of data
processing. Starting with optoelectronic devic-
es and interfaces for fiber-optic networks (LAN,
WAN, FDDI), multiplexing (TDM, WDM) and
routing techniques in the optical domain are in-
troduced. Problems of high capacity data stor-
age using optical techniques and mobile con-
nections to the internet are discussed. The
second part deals with modern techniques for
data compression, coding and security problems
together with the discussion of pattern recogni-
tion using neural networks. Large electronic
databases, techniques for data retrieval, video
indexing methods and electronic data inter-
change are presented. The last part of this
course elucidates todays computer hard- and
software such as Pentium MMX technology,
multimedia PCs , WWW, Internet phone, elec-
tronic mail and more. Next, various types of net-
work protocols (ATM, FDDI, Ethernet, TCP/
IP, ...) suitable for multimedia applications are
discussed. Finally applications such as
teleteaching, teleworking, edutainment (Educa-
tion and Entertainment), video on demand, world
wide web and video conferencing are treated.
Information Technology 1 + 2
D. JÄGER AND CO-WORKERS
Practical studies for students with emphasis on
Information Technology (E3 I/IT)
Exp. 1: Optical Transmission
Exp. 2: Optical Signal Processing
Exp. 3: Optoelectronic Sensors
Exp. 4: Optical Neural Signal Processing
4 TEACHING ACTIVITIES94
4.2 Seminars and Colloquia
Seminar on Optoelectronics
D. JÄGER AND CO-WORKERS
M. Groß, Dimensionierung und Entwicklung
eines thermooptischen Schalters im polynmer-
en Materialsystem, Apr. 1996
M. Alles, Tagungsberichte: IPRM 96,
22.04.-25.04., Schwäbisch Gmünd und IPR 96,
29.04.-02.05., Boston, May 1996
H. Slomka, Reinstwassererzeugung für die
Optoelektronik, May 1996
R. Hülsewede, Nichtlineare Leitungsstruk-
turen zur Frequenzerzeugung und Frequenz-
vervielfachung, May 1996
R. Buß, ZEMAX: Ein Softwarepaket zur Sim-
ulation optischer Systeme, May. 1996
M. Engel, 2D-Simulation von INT-HEMT für
OEIC, Jun. 1996
M. Wenning, Entwicklung einer Meßtechnik
zur Bestimmung der Physikalischen und fotome-
trischen Eigenschaften von LED-basierten Full-
Color-Displays, Jun. 1996
S. Redlich, Nichtlineare Vielschichthetero-
strukturen für die Microwellenphotonik, Jun.
1996
R. Hedtke, Demoskopische Untersuchung
und Beurteilung der Leserlichkeit von LED-basi-
erten Anzeigesystemen, Jun. 1996
B. Neuhaus, Aufbau einer Meßtechnik zur
Charakterisierung und Optimierung der UV-Aus-
beute von Blitzlampen für den Einsatz in der
Druckindustrie, Jul. 1996
D. Kalinowski, Entwicklung eines Feldson-
denmeßplatzes zur 2-dimensionalen Analyse
elektromagnetischer Feldverteilungen auf nich-
tlinearen Leitungen, Jul. 1996
V. Wendrix, Herstellung und Charak-
terisierung eines Wanderwellen-Photodetek-
tors, Jul. 1996
A. Kreuder, Ankopplung von Sende- und
Emfpangsmodulen an eine faseroptische Über-
tragungsstrecke, Oct. 1996
S. Redlich, Ladungsträgertransport über
Heterobarrieren - Simulationsmethoden, Oct.
1996
M. Groß, Stand des Projekts EPI-RET, Oct.
1996
A. Lüddecke, Simulation der Millimeter-
wellengeneration eines Wanderwellenphotode-
tektors, Nov. 1996
J. Evens, Aufbau einer Flip-Chip Apparatur
zur Verbindung integrierter Schaltungen, Nov.
1996
P. Karioja, Overview on activities at VTT,
Nov. 1996
4.2 Seminars and Colloquia 95
V. Mezentsev, Nonlinear Problems Related
To The Modern Optical Communication, Nov.
1996
M. Meininger, Entwicklung photovoltaischer
Zellen zur Energieversorgung einer künstlichen
Sehprothese, Dec. 1996
I. Ryjenkova, Millimeterwave propagation in
nonlinear transmission lines, Dec. 1996
O. Berger, Bestimmung des HF-Er-
satzschaltbildes von Photodetektoren mit Hilfe
der Netzwerkanalyse, Jan. 1997
T. Baumeister, Systemanalyse des
optoelektronischen Energie- und Signalübertra-
gungssystems im Projekt EPI-RET, Jan. 1997
A. Kreuder, Untersuchung der dynamischen
Eigenschaften von nichtlinearen Vielschicht-
heterostrukturen, Jan. 1997
M. Groß, Stand des Projekts EPI-RET, Feb.
1997
O. Lotz, Entwicklung einer Seelaterne in
LED-Technik in den Farben Rot und Grün, Apr.
1997
R.S. Johnson, Silicon Motherboards for Fi-
bre-Chip Coupling, Apr. 1997
M. Schmidt, Elektronische Eigenschaften
von Bor, May 1997
I. Ryjenkova, Nichtlineare Leitungen für das
Millimeterwellengebiet, Jun. 1997
T. Braasch, Bericht über die Messe Laser
1997, Jul. 1997
J. Ervens, Experimentelle Untersuchungen
zum Ladungsträgertransport über eine GaAs/
AlGaAs-Barriere, Oct. 1997
R. Heinzelmann, Bericht über die OFS97
in Williamsburg, Nov. 1997
R. Hedtke,Entwicklung einer optischen En-
ergie- und Signalübertragungsstrecke, Nov.
1997
C. Kampermann, Implementierung eines
analytischen Modells zur Simulation der optis-
chen Eigenschaften nichtlinearer Halbleiter-Het-
erostrukturen, Nov. 1997
B. Ponellis, Simulation des optischen Kon-
versionswirkungsgrades von Wanderwellen-
Photodetektoren, Dec. 1997
4 TEACHING ACTIVITIES96
Colloquium on Optoelectronics
D. JÄGER AND LECTURERS WITH EMPHASIS ON OPTO-
ELECTRONICS
Prof. Dr. H.G. Schuster, Universität Kiel,
Komplexe Adative Systeme, Jan. 1996
Dipl.-Phys. G. David, Universität Duisburg,
Elektrooptische Feldverteilungsmessungen zur
Höchstfequenz-Charakterisierung von mono-
lithisch integrierten Mikrowellenschaltungen,
Feb. 1996
Dr. A.L. Ivanov, Universität Frankfurt,
Switching Kinetics of a Low-Intensity Electro-
Optical Element due to Intrinsic Photoconduc-
tivity, May 1996
Dr. J.-Uwe Meyer, Fraunhofer-Institut St. In-
gberg, Mikrotechnologien zur Kontaktierung von
biologischen Zellen und Geweben, May 1996
Dipl.-Ing. R. Heinzelmann, Universität Du-
isburg, Elektrooptische Wellenleitermodulator-
en für optische Übertragungssysteme, May
1996
Dipl.-Ing. S. van Waasen, Universität Duis-
burg, 20 Gb/s Wellenleiter-pin/Wanderwellen-
verstärker OEIC: Jüngste Ergebnisse, Jun.
1996
Ass. Prof. Dr. A. Driessen, Univ. of Ensche-
de, Netherlands, Advanced Micro-Systems for
Optical Networks (AMON), Jun. 1996
Dr. N. Vodjdani, THOMSON CSF, Orsay Ce-
dex, France, Integrated Optoelectronics for
Optical Microwave Links and Optical Communi-
cations, Oct. 1996
Prof. Dr. M. Dragoman, Time-frequency
characterization of optical pulses, Oct. 1996
Dipl.-Ing. R. Buß, Fachgebiet Optoelektro-
nik, Duisburg, Optoelektronik in der Neurotech-
nologie, Oct. 1996
Dr.-Ing. M. Martin, Hahn-Meitner-Institut,
Berlin, Entwicklung von GHz Komponenten am
Hahn-Meitner-Institut, Nov. 1996
Dipl.-Ing. M. Alles,Fachgebiet Optoelektron-
ik, Duisburg, 60 GHz Wanderwellen-Photode-
tektoren für optische Millimeterwellenverbindun-
gen, Dec. 1996
Dipl.-Phys. T. Braasch, Fachgebiet Optoele-
ktronik, Duisburg, Elektrooptisches Testen zur
on-wafer-Charakterisierung von MMICs , Jan.
1997
Dipl.-Ing. A. Brennemann, Fachgebiet Halb-
leitertechnik/-technologie, Duisburg, Neuartige
Photoreceiver auf Basis einer Kombination von
pin-Diode und Permeable Junction Base Tran-
sistor (PJBT), Jun. 1997
Prof. Dr. W. Sohler, Universität Paderborn,
Integrierte Optik in LiNbO3: neue Entwicklun-
gen, Jun. 1997
Prof. Dr.-Ing. R. Schwarte, Universität
Siegen, Neuartiges optisches 3D-Meßsystem
für die schnelle Formerfassung, Jul. 1997
4.2 Seminars and Colloquia 97
4 TEACHING ACTIVITIES98
4.2 Seminars and Colloquia 99
4.3 Doctoral, Diploma, andGraduate theses
Doctoral theses
Gerhard David, Höchstfrequenz-Charak-
terisierung von monolithisch integrierten Mikro-
wellenbauelementen und -schaltungen durch
zweidimensionale elektrooptische Feldvertei-
lungsmessungen
Steffen Knigge, Nichtlineare Optische
Eigenschaften von Vielschichtheterostrukturen
Andreas Stöhr, Entwicklung und Real-
isierung elektrooptischer Wellenleiter-Schalter
für photonische Systeme im Wellenlängenbe-
reich um 1 µm
Ralf Kremer, Optisch gesteuerte Koplanar-
leitungen als III-V-Halbleiter-Bauelemente für die
Mikrowellen-Signalverarbeitung
Stefan Zumkley, Vertikale elektrooptische
Modulatoren für optische Verbindungstechnik im
Gbit/s-Bereich
Diploma theses
Ludger Brings, Implementierung eines
rechnergestützten Syntheseverfahrens zur Re-
alisierung monolithisch integerierter periodischer
Hochfrequenzleitungen
Peter Bussek, Quantitative Auswertung von
zweidimensionalen elektrooptischen Feldvertei-
lungsmessungen zur Charakterisierung von
monolithisch integrierten Mikrowellenschaltun-
gen
Thomas Alder, Herstellung und Charak-
terisierung von Wellenleitermodulatoren für den
Wellenlängenbereich um 1,3 µm
Michael Wenning , Entwicklung einer
Meßtechnik zur Bestimmung der physikalischen
und fotometrischen Eigenschaften von LED-
basierten Full-Color-Displays
Thomas Engel, 2D-Simulation von InP-
HEMTs für Verstärker in Empfänger-OEICs
Dirk Kalinowski, Entwicklung eines Feld-
sondenmeßplatzes zur zweidimensionalen Ana-
lyse elektromagnetischer Feldverteilungen auf
nichtlinearen Leitungen
Thomas Baumeister, Systemanalyse des
optischen Energie- und Signalübertragungs-
moduls für eine künstliche Sehprothese
Andreas Kreuder, Untersuchung der
dynamischen Eigenschaften nichtlinearer
Vielschichtheterostrukturen
4 TEACHING ACTIVITIES100
Mark Meininger, Entwicklung photovoltais-
cher Zellen zur Energieversorgung einer kün-
stlichen Sehprothese (Retina Implant)
Oliver Lotz, Entwicklung einer Seelaterne
in LED-Technik in den Farben Rot und Grün
Jutta Ervens, Experimentelle Untersuchun-
gen zum Ladungsträgertransport über eine
GaAs/AlxGa1-xAs-Heterobarriere
Uwe Weimann, Entwicklung einer Infrarot-
Datenübertragungsstrecke auf rotierenden Bild-
Text-Systemen
Claus Kampermann, Implementierung
eines analytischen Modells zur Simulation der
optoelektronischen Eigenschaften nichtlinearer
Halbleiter-Heterostrukturen
4.3 Doctoral, Diploma, and Gratuate theses 101
Graduate theses
Michael Heinsdorf, Herstellung und Char-
akterisierung von Wanderwellen-Photodetek-
toren auf InP-Substrat
Ralph Hedtke, Demoskopische Untersu-
chungen zur Beurteilung der Leserlichkeit von
LED-basierten Anzeigesystemen
Birgit Neuhaus, Aufbau einer Meßtechnik
zur Charaktierisierung und Optimierung der UV-
Ausbeute von Blitzlampen für den Einsatz in der
Druckindustrie
Jutta Ervens, Aufbau einer Flip-Chip-Appa-
ratur zur Verbindung integrierter Schaltungen
André Lüdecke, Simulation der Millimeter-
wellengeneration eines Wanderwellen-Photode-
tektors
Oliver Berger, Bestimmung des HF-Er-
satzschaltbildes von Photodetektoren mit Hilfe
der Netzwerkanalyse
Bernd Ponellis, Simulation des optoelektro-
nischen Konversionswirkungsgrades von Wan-
derwellen-Photodetektoren
5 PUBLICATIONS AND PRESENTATIONS102
5 Publications and presenta-tions
[1] G. David, R. Tempel, I. Wolff, and D. Jäger,
Analysis of microwave propagation effects
using 2D electro-optic field mapping tech-
niques, Optical and Quantum Electronics,
Special Issue on Optical Probing of
Ultrafast Devices and Integrated Circuits,
1996, pp. 919 - 931
[2] G. David, P. Bussek, and D. Jäger, High
resolution electro-optic measurements of
2D field distributions inside MMIC devices,
Proceedings of CLEO 96, Anaheim, USA,
1996, pp. 450-451
[3] G. David, R. Tempel, I. Wolff, and D. Jäger,
In-circuit electro-optic field mapping for
function test and characterization of
MMICs, 1996 IEEE MTT-S Int. Microwave
Symp., June 17-24, San Francisco, USA,
1996, pp. 1533-1536
[4] R. Kremer, S. Redlich, L. Brings, and D.
Jäger, A novel type of constant impedance
travelling wave phase shifter for InP- based
MMICs, 1996 IEEE MTT-S Int. Microwave
Symp., June 17-24, San Francisco, USA,
1996
[5] M. Alles, Th. Braasch, and D. Jäger, High-
speed coplanar Schottky travelling-wave
photodetectors, Int. Conf. on Integrated
Photonics Research, Conference Proceed-
ings pp. 380-383, Boston, USA 1996
[6] G. David and D. Jäger, Analysis of in-cir-
cuit electro-optic measurements of MMICs,
XXVth General Assembly of the URSI,
August 28 -September 5, 1996, Lille,
France
[7] M. Alles, Th. Braasch, R. Heinzelmann, A.
Stöhr, and D. Jäger, Optoelectronic de-
vices for microwave and millimeterwave
optical links, 11th Int. MIKON 96, Confer-
ence Proceedings, Workshop Optoelec-
tronics in Microwave Technology, pp. 68
- 76, Warsaw, 27-30 May 1996 (invited
paper)
[8] M. Alles, R. Heinzelmann, R. Hülsewede,
R. Kremer, S. Redlich, A. Stöhr, and D.
Jäger, Wave propagation in planar struc-
tures for travelling wave semiconductor
devices, Progress In Elektromagnetic Re-
search Symposium PIERS 96, Confer-
ence Proceedings, p. 136, July 1996,
Innsbruck, Austria (invited paper)
[9] P. Berini, A. Stöhr, K. Wu, D. Jäger, Nor-
mal Mode Analysis and Characterization
of an InGaAs/GaAs MQW Field-Induced
Optical Waveguide Including Electrode Ef-
fects, IEEE/OSA J. Lightwave Technol.,
Vol. 14, No. 10, pp. 2422 - 2435, October
1996
103
[10] D. Jäger, M. Alles, T. Braasch, R.
Heinzelmann, and A. Stöhr, Integration
Technology for Microwave Photonic De-
vices, Interaction technology of Micro-
waves and Light-Waves-Systems and
Devices, XXVth General Assembly of the
URSI, August 28 -September 5, 1996, Lille,
France (invited paper)
[11] D. Jäger, R. Hülsewede, I.V. Ryjenkova,
V.K. Metzentsev, S. L. Musher, Microwave
Propagation on Nonlinear Transmission
Lines, XXVth General Assembly of the
URSI, August 28 -September 5, 1996, Lille,
France
[12] D. Jäger, V.K. Metzentsev, I.V. Ryjenkova,
S. K. Turitsyn and R. Hülsewede, Micro-
wave Propagation on Nonlinear Transmis-
sion Lines, Proceedings of PIERS96,
Hong Kong
[13] M. Alles, T. Braasch, and D. Jäger, Trav-
elling Wave Photodetector for Optical Gen-
eration of Microwave Signals, Indium
Phosphide and Related Materials IPRM
96, Proc. Part II, pp. 30 - 31, Schwäbisch
Gmünd, 1996 (post deadline paper)
[14] R. Hülsewede, V.K. Mezentsev, S.L.
Musher, I.V. Ryjenkova, S.K. Turitsyn, and
D. Jäger, Travelling wave generation of
millimeter waves in bi-modal NLTLs, 26th
European Microwave Conference EuMC
96, Prague
[15] R. Heinzelmann, A. Stöhr, Th. Alder, R.
Buß, and D. Jäger, EMC measurements
using electrooptic waveguide modulators,
International Topical Meeting on Micro-
wave Photonics MWP 96, Conference
Proceedings, Technical Digest, December
3-5, 1996, Kyoto, Japan
[16] R. Hülsewede, V.K. Mezentsev, S.L.
Musher, V. Ryjenkova, S.K. Turitsyn, and
D. Jäger, Millimeter Wave Generation on
Nonlinear Transmission Lines, 1996 Int.
Workshop on Millimeter Waves Digest,
1996, Orvieto, Italy
[17] D. Jäger, Optically Controlled Microwave
Devices, International Topical Meeting on
Microwave Photonics MWP 96 technical
digest, December 3-4, 1996, Kyoto, Japan
[18] D. Jäger, V.K. Mezentsev, S.L. Musher and
I.V. Ryjenkova, Millimeter wave power gen-
eration on nonlinear transmission lines,
Asia Pacific Microwave Conf. APMC 96,
New Delhi, India
[19] I. Ryjenkova, V.K. Mezentsev, S.L.
Musher, S.K. Turitsyn, R. Hülsewede and
D. Jäger, Nonlinear Transmission Lines for
Millimeter Wave Applications, INMMC 96,
Duisburg
[20] Th. Braasch, G. David, R. Hülsewede, U.
Auer, F.-J. Tegude and D. Jäger, Propa-
gation of Microwaves in MMICs Studied by
Time- and Frequency-Domain Electro-Op-
tic Field Mapping, Spring Topical Meeting
Ultrafast Electronics and Optoelectronics,
March 17-19, 1997, Lake Tahoe, USA
5 PUBLICATIONS AND PRESENTATIONS104
[21] Th. Braasch, G. David, R. Hülsewede, U.
Auer, F.-J. Tegude and D. Jäger, Fre-
quency and time domain characterization
of nonlinear transmission lines using
electro-optic probing techniques, MIOP
97, April 22-24, 1997, Sindelfingen, Ger-
many
[22] M. Alles, U. Auer, F.-J. Tegude and D.
Jäger, Millimeterwave Photodetectors, Mi-
crowaves and Optronics, MIOP 97, April
22-24, 1997, Sindelfingen, Germany
[23] M. Alles, U. Auer, F.-J. Tegude and D.
Jäger, High-Speed Travelling-Wave Pho-
todetectors for optical Millimeterwave
transmission operating at 1.55 µm, Work-
shop Mobile Millimeter Communications
MMMCom, Dresden, 12-13. Mai 1997
[24] S. Redlich, A. Kreuder, and D. Jäger, Dy-
namics of nonlinear electro-optical GaAs/
AlAs multilayer-heterostructures, Interna-
tional Conference on Low Dimensional
Structures (LSDS) 97, May 19-21, 1997,
Lissabon, Portugal
[25] A. Stöhr, Heterostructure Semiconductor
Photonic Devices and Systems,
Euroconference on Advanced Heterostruc-
tures, July 1997, Grenoble, France
[26] I. Ryjenkova, M. Alles and D. Jäger, Non-
linear travelling wave photodetector for mil-
limeter wave harmonic frequency genera-
tion, Journal of Communications Special
Issue, Microwave Photonics, Vol. 48, Aug.
1997, pp. 14-17
[27] I. Ryjenkova, V.K. Mezentsev, S.L.
Musher, S.K. Turitsyn, R. Hülsewede, and
D. Jäger, Millimeter Wave Generation on
Nonlinear Transmission Lines, Publication
in annales des télécommunications (spe-
cial issue), Vol. 52, No. 3-4, 1997, pp. 134-
139
[28] M. Alles, U. Auer, F.-J. Tegude, and D.
Jäger, High-speed Travelling-Wave Pho-
todetectors for Wireless Optical Millimeter
Wave Transmission,MWP 97, Sep. 3-5,
1997, Duisburg/Essen, Germany
[29] Th. Braasch, G. David, R. Hülsewede, and
D. Jäger, 1D- and 2D-elektro-optic field
mapping to study nonlinear effects in
NLTLs, MWP 97, Sep. 3-5, 1997,
Duisburg/Essen, Germany
[30] S. Redlich, and D. Jäger, Nichtlineare
Vielschichtheterostrukturen für die Mikro-
wellen-Photonik, Photonik Symposium,
Oct. 7-9, 1997, Schwäbisch Hall, Germany
[31] S. Redlich, C. Kampermann, and D. Jäger,
Vielschichtheterostrukturen: Neue
Materialien für die Mikrowellen-Photonik,
Photonik-Symposium, Oct. 8-10, 1997,
Würzburg, Germany
[32] S. Redlich, C. Kampermann, and D. Jäger,
Modeling and simulation of nonlinear hy-
brid AlGaAs/GaAs Bragg reflectors, 10th
III-V Semiconductor Device Simulation
Workshop, Oct. 16-17, 1997, Torino, Italy
105
[33] A. Stöhr, R. Heinzelmann, T. Alder, W.
Heinrich, T. Becks, D. Kalinowski, M.
Schmidt, M. Groß, and D. Jäger, Optically
Powered Integrated Optical E-Field Sen-
sor, 12th International Conference on Op-
tical Fiber Sensors, Conference Proceed-
ings, Oct. 1997, Williamsburg, Virginia,
USA
[34] M. Groß, T. Alder, R. Buß, R. Heinzelmann,
M. Meininger, and D. Jäger, Micro Photo-
voltaic Cell Array for Energy Transmission
into the Human Eye, EPVSEC14, 1997,
Barcelona, Spain, Vol. 1, pp. 1165 - 1167
[35] M. Alles, U. Auer, F.-J. Tegude, and D.
Jäger, High-Speed Travelling-Wave Pho-
todetectors for Optical Generation of
Millimeterwaves, APMC 97, Dec. 2-5,
1997, Hongkong, China
[36] I. Ryjenkova, D. Jäger, Nonlinear RTD Cir-
cuits for High-Speed A/D Conversion,
APMC 97, Dec. 2-5, 1997, Hongkong,
China
5 PUBLICATIONS AND PRESENTATIONS106
107
6 Guide to the Department of Optoelectronics
Travel by car - The Department of Optoelectronics, now located in the Center for Solid-State Elec-
tronics and Optoelectronics (ZHO), can easily reached by car via various highways: A3 from the
north and south, A40 from the east and west. Exit at Duisburg-Kaiserberg or Duisburg-Wedau, see
map for details.
Travel by plane - From Düsseldorf International Airport take the city-train (S-Bahn) S1 to Duis-
burg main station (Hauptbahnhof, Hbf).
Travel by train - From Duisburg main station (Hauptbahnhof, Hbf) it is a 20 min. walk to the Depart-
ment of Optoelectronics and the ZHO. You can either go by Taxi or take the bus 933 or 936 to
Universität or take the tram 901 to station Zoo/Uni.
108
Notes: