Volume 59, Number 12, 2005APPLIED SPECTROSCOPY
? 2005 Society for Applied Spectroscopy
Tunable Laser Diode System for Noninvasive Blood Glucose
JONATHON T. OLESBERG,* MARK A. ARNOLD, CARMEN MERMELSTEIN,
JOHANNES SCHMITZ, and JOACHIM WAGNER
Optical Science and Technology Center and the Department of Chemistry, 100 IATL, University of Iowa, Iowa City, Iowa 52242
(J.T.O., M.A.A.); and Fraunhofer-Institut fuer Angewandte Festkoerperphysik (IAF), Tullastrasse 72, D-79108, Freiburg, Germany
(C.M., J.S., J.W.)
Optical sensing of glucose would allow more frequent monitoring
and tighter glucose control for people with diabetes. The key to a
successful optical noninvasive measurement of glucose is the collec-
tion of an optical spectrum with a very high signal-to-noise ratio in
a spectral region with significant glucose absorption. Unfortunately,
the optical throughput of skin is low due to absorption and scatter-
ing. To overcome these difficulties, we have developed a high-
brightness tunable laser system for measurements in the 2.0–2.5 ?m
wavelength range. The system is based on a 2.3 ?m wavelength,
strained quantum-well laser diode incorporating GaInAsSb wells
and AlGaAsSb barrier and cladding layers. Wavelength control is
provided by coupling the laser diode to an external cavity that in-
cludes an acousto-optic tunable filter. Tuning ranges of greater than
110 nm have been obtained. Because the tunable filter has no mov-
ing parts, scans can be completed very quickly, typically in less than
10 ms. We describe the performance of the present laser system
and avenues for extending the tuning range beyond 400 nm.
Index Headings: Glucose; Tunable laser; Near-infrared spectrosco-
py; Noninvasive sensing; Acousto-optic tunable filter; AOTF; Ex-
ternal cavity; GaInAsSb; Diabetes.
The benefits of tight glycemic control in people with
diabetes are well-documented.1–3Hyperglycemia over ex-
tended periods is the primary cause of the severe com-
plications associated with diabetes, including premature
death, blindness, kidney failure, amputations, heart dis-
ease, and stroke. Effective glycemic control requires fre-
quent blood glucose monitoring to provide the informa-
tion needed to administer the proper amount of insulin
while avoiding hypoglycemia.
Frequent blood glucose monitoring would be more
widely practiced with the availability of an analytical sys-
tem that operates in a manner that is accurate, painless,
sample-free, and easily implemented by the diabetic pa-
tient during his/her normal daily routine. State-of-the-art
glucose monitoring technology falls considerably short of
these requirements. Current test-strip technology requires
a blood sample for each measurement. The pain associ-
ated with such measurements can inhibit frequent moni-
toring, especially in children. Frequent monitoring is also
discouraged by the need to handle and dispose of the
blood sample and by the difficulty of implementing the
test in social settings. Current industry-wide efforts to
reduce the size of the required blood sample and to short-
en the analysis time are beneficial, but they do not ad-
Received 8 March 2005; accepted 12 October 2005.
* Author to whom correspondence should be sent. E-mail:
dress the fundamental limitations of an invasive proce-
Noninvasive optical sensing of glucose has been pro-
posed by many research groups for the frequent and pain-
less measurement of glucose in people with diabetes.4
The concept is to pass a selected band of near-infrared
radiation through a vascular region of the body and ex-
tract the glucose concentration from the resulting spectral
information. Near-infrared spectroscopy is a promising
approach for noninvasive sensing because of the unique
near-infrared absorption spectrum of glucose and the sig-
nificant penetration of near-infrared light into human tis-
One of the primary difficulties with performing optical
absorption measurements in skin is the low optical
throughput. In addition to the strong water absorption
(0.9 AU/mm at 2.2 ?m wavelength), skin is highly scat-
tering. In separate noninvasive measurements using con-
ventional Fourier transform infrared (FT-IR) instrumen-
tation,5the peak transmission through a sample with an
effective aqueous path length of 0.6 mm is typically 0.1–
1.0%. In order to maximize the signal-to-noise ratio of
the measurement, it is helpful to have the brightest source
possible. We presently use 50 W tungsten filament bulbs
that operate at 3050 K. Although higher power bulbs can
be obtained, their brightness (optical power per unit ra-
diating area) does not increase with power.
Laser diodes would be a very useful tool for tissue
spectroscopy because of their brightness, which enables
large optical powers to be collected onto a small, low
noise detector. Laser diodes, being solid-state devices,
could lead to more compact and rugged spectrometers
compared to a system that requires a sensitive interfer-
ometer. In order to be useful for noninvasive spectros-
copy, however, a laser diode system must be capable of
acquiring a spectrum consisting of measurements at a
number of wavelengths. The need for a spectrum rather
than measurements at one or two discrete wavelengths is
a consequence of the broad and highly overlapped nature
of near-infrared absorption bands. The need for measure-
ments over a range of wavelengths is illustrated in Fig.
1, which shows the absorptivity spectrum of glucose
compared with that of several other biomolecules. The
glucose absorption spectrum is unique, but it is highly
overlapped with the other spectra. Disentangling the glu-
cose signal from a composite spectrum can be accom-
plished with multivariate techniques, such as partial least-
squares regression. Multivariate techniques inherently re-
quire measurements at a number of wavelengths.
in the 2.0–2.5 ?m wavelength range. The glucose absorptivity spectrum
is unique, but exhibits significant overlap with the absorptivity spectra
of other biomolecules.
Absorptivity of glucose and a sampling of other biomolecules
ode operating just above threshold. (b) Reflectivity of the anti-reflec-
tion-coated facet. The vertical line shows the free lasing wavelength of
(a) Free lasing spectrum of the anti-reflection-coated laser di-
There are at least two ways to collect a spectrum using
a laser diode. The first is to use a set of diodes, each
operating at a fixed wavelength. To be useful for trans-
cutaneous spectroscopy, however, this would require at
least 12–18 devices, each of which is locked at a partic-
ular wavelength. This arrangement represents a straight-
forward but cumbersome solution. The second way to
obtain a spectrum is to use a single emitter whose wave-
length is tunable.
With regards to spectroscopy, tunable laser diodes have
been used across both near-infrared6–10and mid-infra-
red11–13wavelengths. The vast majority of spectroscopic
work performed using tunable laser diodes has been di-
rected at gas sensing. The requirements for noninvasive
aqueous sensing, however, are very different from those
of gas sensing because of the difference between gas-
phase and condensed-phase absorption spectra. Gas-
phase absorption spectra comprise several sharp features
that correspond to the rotational modes of the gas mol-
ecule. By contrast, the aqueous glucose spectrum is com-
posed of features with widths of 25–100 nm (50–200
cm?1). A tunable laser diode system designed to measure
gas will typically scan across a single rotation line to
quantify the species (0.1 nm of tuning). For aqueous
spectra, however, much broader tunability is required
(250–400 nm, or 500–800 cm?1). While gas sensing re-
quires very narrow laser linewidths to resolve the sharp
features, narrow linewidths are not necessary for aqueous
sensing. The typical spectral resolution utilized in our
transcutaneous measurements is 8 nm (16 cm?1).5
Several tuning strategies have been employed with tun-
able laser diode systems. The most convenient strategies
involve only electronic wavelength control. For example,
current can be injected into regions of the device in order
to modify the index of refraction of the material, which
leads to a shift in wavelength. Alternatively, the current
used to drive the device can be ramped in order to rapidly
vary the device temperature, which causes a shift in
wavelength. Both of these approaches provide tunability
over narrow wavelength bands, which is sufficient for gas
sensing but not for aqueous spectroscopy. Tuning can
also be achieved by directly modifying the temperature
of the device heat-sink, but this provides limited tuning
range and is slow. Recently, wide tuning ranges have
been obtained by subjecting the laser device to high-pres-
sures, but it is unclear whether this can be accomplished
in a practical manner without damaging the devices.14–16
An additional strategy is to employ an external cavity,
which can provide a wider tuning range than electrical
or temperature tuning. Tuning ranges of up to 242 nm
(1170 cm?1) have been reported in the literature.17,18Ex-
ternal cavity systems at longer wavelengths have dem-
onstrated tuning ranges from 3.40–3.68 ?m (224
The work described here was performed with coated
Fabry–Perot laser diodes fabricated at the Fraunhofer In-
stitute for Applied Solid-State Physics. These devices
are based on an active region with three strained
Ga0.70In0.30As0.06Sb0.94quantum wells, Al0.28Ga0.72As0.02Sb0.98
barriersand separate confinement
Al0.85Ga0.15As0.07Sb0.93waveguide cladding layers.21–24The
devices were 1 mm long ridge waveguide lasers with a
ridge width of 16 ?m. The devices were mounted epi-
layer-side down on a copper C-mount heat-sink. The op-
timal device for wavelength tuning had a 95% high-re-
flective coating on one end and a 0.3% anti-reflection
coating on the other. The device operates with a natural
wavelength of 2.29 ?m with a threshold current of 150
mA at a heat-sink temperature of 20 ?C. Figure 2a shows
the natural lasing spectrum of the device at a current of
170 mA. Shown in Fig. 2b is a measurement of the re-
flectivity of the anti-reflection coated facet, which is de-
signed to reach a minimum reflectivity at the natural las-
The laser diode was mounted on a thermoelectric tem-
perature-controlled mount and coupled into an external
cavity arrangement. A photograph and simplified sche-
matic of the system are shown in Fig. 3. The laser output
from the facet with the anti-reflection coating was col-
lected with a high numeric aperture asphere (NA ? 0.55
and f ? 4.51 mm) and imaged onto a flat end-mirror
(which also serves as an output coupler) 20 cm away. A
beam-splitter could optionally be inserted in the cavity in
order to pick off a small portion of the beam during align-
ment. An acousto-optic tunable filter (AOTF) was placed
Volume 59, Number 12, 2005
a simplified schematic.
(a) Photograph of the present external cavity system, and (b)
tem. The fringes are due to feedback from the internal laser diode cav-
Tuning spectra obtained with the present external cavity sys-
between the collimating lens and end-mirror. The tunable
filter (Brimrose Corp.) is based on a temperature-stabi-
lized TeO2crystal and optimized for the 2.0–2.5 ?m
wavelength range with a pass-band width of 12 nm and
a diffraction efficiency greater than 50%. The tunable fil-
ter causes a 6? horizontal deflection of the selected wave-
length band, which requires that the end-mirror be rotated
by 6? off-normal with respect to the output from the laser
diode. The tunable filter is driven by a custom-designed
radio-frequency (RF) driver based on an Analog Devices
AD9854 direct-digital-synthesis signal generator chip.
The synthesizer chip is controlled by a microcontroller,
which programs the synthesizer to generate a chirped sine
wave with frequencies running from 40–45 MHz. The
chirp is amplified by a 4 watt RF power amplifier and
delivered to the tunable filter. A small portion of the sys-
tem output is picked off using a pellicle beam splitter and
delivered to an extended-wavelength InGaAs detector to
provide a reference channel. The system as a whole is
mounted on a 12 in. optical breadboard so that the system
can be moved and positioned as a unit.
RESULTS AND DISCUSSION
The present system is capable of tuning over a wave-
length range of more than 110 nm (220 cm?1), as is il-
lustrated by the spectra in Fig. 4, which were collected
at a series of fixed RF drive frequencies using an FT-IR
spectrometer. The external cavity system begins to lase
at a threshold current of 105 mA and obtains optimal
tunability at a drive current of 200 mA. At higher cur-
rents, the output power continues to increase although the
tuning range narrows slightly due to carrier loss to inter-
nal cavity lasing. Note, however, from the schematic in
Fig. 3b, that the internal lasing signal is not collinear with
the external cavity output and can be eliminated with a
beam-stop. An external cavity utilizing a grating in the
Littrow configuration has also been employed. The grat-
ing-based external cavity had the same threshold current
and wavelength tuning range as the AOTF-based system.
With a 90% reflective external cavity output coupler,
the optical power at 200 mA is 0.5 mW at the center of
the tuning range. Powers of several milliwatts should be
obtainable with improved anti-reflection coatings and
waveguide geometries. Optical power levels as high as 1
watt have been obtained with shorter wavelength near-
infrared external cavity systems.25The optical power
presently available for our noninvasive glucose work
originates from a tungsten filament. The integrated radi-
ant power from this broad-band source collected onto the
detector over the 2.0–2.5 ?m band is approximately 0.05
mW. Thus, the present 0.5 mW from the tunable laser
system represents a factor of 10 improvement in available
signal with respect to the broadband source.
In Fig. 4, fringes are visible in each of the lasing spec-
tra. These fringes are due to significant feedback from
the anti-reflection-coated facet of the device. The pres-
ence of the fringes indicates that the laser is operating in
a coupled-cavity regime, which is undesirable.26–28The
coupling of the internal and external cavities can be re-
duced by decreasing the feedback from the anti-reflec-
Spectra of four chemical components were measured
using the tunable laser system and an FT-IR spectrometer
to test the tunable laser system’s ability to resolve spectral
features of glucose and other biomolecules. Solutions of
glucose, urea, and acetate were measured in a 1 mm path
length cuvette and referenced to water. Absorbance spec-
tra for the solutions are shown in Fig. 5 for measurements
made using the tunable laser system and a Nicolet Nexus
670 FT-IR spectrometer. Spectra collected using the FT-
IR spectrometer were obtained from 128 coadded inter-
ferograms (60 s collection time) with 16 cm?1resolution
using a thermoelectrically cooled, extended wavelength
InGaAs detector. A band pass filter was placed before the
detector to limit the spectral range to the 2.0–2.5 ?m
wavelength band. The spectra collected using the tunable
laser are the average of 2000 individual scans collected
over 60 s. Thermoelectrically cooled InGaAs detectors
were used for signal and reference channels.
The spectra recorded using the two systems are very
similar in shape and magnitude. Vertical shifts in absor-
bance spectra are common in the 2.0–2.5 ?m wavelength
range due to the strong temperature dependence of water
absorption and are not analytically significant. Solution
present tunable laser diode spectrometer and (b) an FT-IR spectrometer.
(c) The absorbance spectra measured with an FT-IR spectrometer over
a wider spectral range. The highlighted region is the portion of the
spectra shown above. Note that vertical shifts are common in this wave-
length range and are not analytically significant.
Absorbance spectra of four analytes measured with (a) the
using 14-band K·p theory for a series of carrier densities. ?externalis the
gain required for lasing with the external cavity. ?R?0.3%and ?R?0.01%are
the gains required for lasing without the external cavity with front facet
reflectivities of 0.3% and 0.01%, respectively.
Gain spectra for the triple-quantum-well laser diode calculated
temperatures were not controlled during these measure-
The present tunability of the system is sufficient for in
vitro quantification of biological analytes such as glucose
in moderately complex samples (e.g., for quantification
of glucose in interstitial fluid or clear effluent from a
bioreactor). However, our experience with in vivo mea-
surements indicates that a wider spectral range is required
for transcutaneous glucose measurements. Based on our
preliminary work with animal models,5we estimate that
a minimum useful tuning range of 250 nm is required for
in vivo tissue measurements, whereas 400 nm would be
The tuning range that can be obtained with an external
cavity system depends primarily on three factors: the
width of the gain spectrum of the laser diode, the effec-
tive reflectivity provided by the external cavity, and the
reflectivity of the laser diode facet. The gain spectrum of
the laser diode broadens with increasing drive current un-
til the device reaches lasing threshold based on feedback
from the front facet. In order to maximize the tunability
of the system, the reflectivity from the front facet of the
diode needs to be minimized while the amount of light
coupled back into the device from the external cavity
needs to be maximized.
We have performed calculations of the gain spectrum
of the triple-quantum-well material using 14-band K·p
theory. This formalism has been used extensively in the
past to model the optical and electrical characteristics of
near- and mid-infrared quantum wells and superlatti-
ces.29–35The calculated modal gain spectrum of the de-
vice is shown in Fig. 6 for quantum-well-carrier densities
of 3.5–5.5 ? 1017cm?3. Also shown in the figure are the
threshold gain required for operation with the existing
external cavity (?external) and the threshold gain for lasing
based on the 0.3% reflectivity of the front device facet
(?R?0.3%). The threshold gain for lasing off the internal
cavity (?R?0.3%) was calculated based on the cavity length,
facet reflectivities, and previously measured values for
internal loss.21Threshold gain for the external cavity
(?external) was calculated from ?R?0.3%and the difference in
threshold current for operation with and without the ex-
The diode will begin to lase based on feedback from
the front facet at a carrier density at which the modal
gain exceeds ?R?0.3%, which in the present device occurs
at a carrier density of 4.2 ? 1017cm?3. The tuning range
that can be obtained with the external cavity can then be
calculated by examining the spectral range over which
the gain curve for this carrier density exceeds ?external. The
estimated tuning range for the present devices, which is
shown in the figure with a gray bar, matches the experi-
Two types of modifications to the laser device will ex-
tend the tuning range that can be obtained. The first type
of modification involves improvements that allow for
more efficient coupling of the laser emission back into
the device. Wagner et al. have recently reported improved
clad designs that reduce the spatial divergence of the laser
emission, which allows for higher-efficiency coupling of
the emission back into the device and reduction of
The other type of modification that can be used to ex-
tend the tuning range involves suppressing feedback from
the front facet of the device. This can be done by im-
proving the efficiency of the anti-reflection coating ap-
plied to the facet, by shortening the laser diode, or by
utilizing a curved or angled stripe geometry.38–42In prin-
ciple, reflectivities as low as 10?6can be obtained by
combining a curved stripe geometry and an anti-reflec-
tion coating.38Reducing the reflectivity of the front facet
to 10?4would increase the threshold gain for internal las-
ing to 55 cm?1, which is shown by the line labeled
?R? 0.01%in Fig. 6. This would allow operation at a carrier
density greater than 5 ? 1017cm?3and provide tuning
from 2.12–2.34 ?m (220 nm, or 440 cm?1). By combin-
1484 Download full-text
Volume 59, Number 12, 2005
ing a reflectivity of 10?4with a shortened laser diode (0.5
mm), the tuning range can be extended well beyond 400
nm, which will be ample for noninvasive sensing.
We report the initial development of a tunable laser
diode spectroscopy system designed for the measurement
of glucose in in vivo transcutaneous spectra. The high-
brightness source provides 0.5 mW of power tunable over
the 2.21–2.33 ?m (4300–4520 cm?1) range. Further im-
provements designed to reduce the optical feedback from
the anti-reflection-coated facet will extend the tuning
range. Tuning ranges as wide as 400 nm are expected for
optimized laser diodes. These capabilities will make the
system a highly attractive platform for near-infrared spec-
troscopy in high optical density samples.
This research was supported by grants from the National Institute of
Diabetes and Digestive and Kidney Diseases of the National Institutes
of Health (DK-60657 and DK-02925).
1. Diabetes Control and Complications Trial Research Group, ‘‘The
effect of intensive treatment of diabetes on the development and
progression of long-term complications in insulin-dependent dia-
betes mellitus’’, N. Engl. J. Med. 329, 977 (1993).
2. Diabetes in America, NIH Publication No. 95-1468 (National In-
stitutes of Health, National Institute of Diabetes and Digestive and
Kidney Diseases, 1995), 2nd ed.
3. N. F. Ray, M. Thamer, E. Gardner, J. K. Chan, and R. Kahn, Dia-
betes Care 21, 296 (1998).
4. O. S. Khalil, Clin. Chem. 45, 165 (1999).
5. J. T. Olesberg, L. Liu, V. Van Zee, and M. A. Arnold, Proc. SPIE-
Int. Soc. Opt. Eng. 5325, 11 (2004).
6. J. Wang, M. Maiorov, J. B. Jeffries, D. Z. Garbuzov, J. C. Connolly,
and R. K. Hanson, Meas. Sci. Technol. 11, 1576 (2000).
7. J. Wang, M. Maiorov, D. S. Baer, D. Z. Garbuzov, J. C. Connolly,
and R. K. Hanson, Appl. Opt. 39, 5579 (2000).
8. J. C. Nicolas, A. N. Baranov, Y. Cuminal, Y. Rouillard, and C.
Alibert, Appl. Opt. 37, 7906 (1998).
9. L. H. Xu, Z. F. Liu, I. Yakovlev, M. Y. Tretyakov, and R. M. Lees,
Infrared Phys. Technol. 45, 31 (2004).
10. R. D. May, J. Geophys. Res. 103, 19161 (1998).
11. M. D. Brookes and A. R. W. McKellar, J. Chem. Phys. 109, 5823
12. C. R. Webster, G. J. Flesch, D. C. Scott, J. E. Swanson, R. D. May,
W. S. Woodward, C. Gmachl, F. Capasso, D. L. Sivco, J. N. Bail-
largeon, A. L. Hutchinson, and A. Y. Cho, Appl. Opt. 40, 321
13. K. Namjou, S. Cai, E. A. Whittaker, J. Faist, C. Gmachl, F. Capasso,
D. L. Sivco, and A. Y. Cho, Opt. Lett. 23, 219 (1998).
14. P. Adamiec, A. Salhi, R. Bohdan, A. Bercha, F. Dybala, W. Trze-
ciakowski, Y. Rouillard, and A. Joullie, Appl. Phys. Lett. 85, 4292
15. P. Adamiec, F. Dybala, A. Bercha, R. Bohdan, W. Trzeciakowski,
and M. Osinski, Proc. SPIE-Int. Soc. Opt. Eng. 4973, 158 (2003).
16. F. Dybala, P. Adamiec, A. Bercha, R. Bohdan, and W. Trzecia-
kowski, Proc. SPIE-Int. Soc. Opt. Eng. 4989, 181 (2003).
17. M. Bagley, R. Wyatt, D. J. Elton, H. J. Wickes, P. C. Spurdens, C.
P. Seltzer, D. M. Cooper, and W. J. Devlin, Electron. Lett. 26, 267
18. H. Tabuchi and H. Ishikawa, Electron. Lett. 26, 742 (1990).
19. H. Q. Le, G. W. Turner, J. R. Ochoa, M. J. Manfra, C. C. Cook,
and Y.-H. Zhang, Appl. Phys. Lett. 69, 2804 (1996).
20. H. Q. Le, G. W. Turner, J. R. Ochoa, M. J. Manfra, C. C. Cook,
and Y.-H. Zhang, Proc. SPIE-Int. Soc. Opt. Eng. 3001, 298 (1997).
21. C. Mermelstein, S. Simanowski, M. Mayer, R. Kiefer, J. Schmitz,
M. Walther, and J. Wagner, Appl. Phys. Lett. 77, 1581 (2000).
22. S. Simanowski, N. Herres, C. Mermelstein, R. Kiefer, J. Schmitz,
M. Walther, J. Wagner, and G. Weimann, J. Cryst. Growth 209, 15
23. S. Simanowski, C. Mermelstein, M. Walther, N. Herres, R. Kiefer,
M. Rattunde, J. Schmitz, J. Wagner, and G. Weimann, J. Cryst.
Growth 227, 595 (2001).
24. S. Simanowski, M. Walther, J. Schmitz, R. Kiefer, N. Herres, F.
Fuchs, M. Maier, C. Mermelstein, J. Wagner, and G. Weimann, J.
Cryst. Growth 202, 849 (1999).
25. D. Mehuys, D. Welch, and D. Scifres, Electron. Lett. 29, 1254
26. M. Ahmed and M. Yamada, J. Appl. Phys. 95, 7573 (2004).
27. S. G. Abdulrhmann, M. Ahmed, T. Okamoto, W. Ishimori, and M.
Yamada, IEEE J. Sel. Top. Quant. Electron. 9, 1265 (2003).
28. K. Petermann, IEEE J. Sel. Top. Quant. Electron. 1, 480 (1995).
29. C. H. Grein, M. E. Flatte ´, J. T. Olesberg, S. A. Anson, L. Zhang,
and T. F. Boggess, J. Appl. Phys. 92, 7311 (2002).
30. W. H. Lau, J. T. Olesberg, and M. E. Flatte ´, Phys. Rev. B 64,
31. J. T. Olesberg, W. H. Lau, M. E. Flatte ´, C. Yu, E. Altunkaya, E.
M. Shaw, T. C. Hasenberg, and T. F. Boggess, Phys. Rev. B 64,
32. S. A. Anson, J. T. Olesberg, M. E. Flatte ´, T. C. Hasenberg, and T.
F. Boggess, J. Appl. Phys. 86, 713 (1999).
33. M. E. Flatte ´, C. H. Grein, T. C. Hasenberg, S. A. Anson, D.-J. Jang,
J. T. Olesberg, and T. F. Boggess, Phys. Rev. B 59, 5745 (1999).
34. D.-J. Jang, M. E. Flatte ´, C. H. Grein, J. T. Olesberg, T. C. Hasen-
berg, and T. F. Boggess, Phys. Rev. B 58, 13047 (1998).
35. J. T. Olesberg, S. A. Anson, S. W. McCahon, M. E. Flatte ´, T. F.
Boggess, D. H. Chow, and T. C. Hasenberg, Appl. Phys. Lett. 72,
36. M. Rattunde, A. Huelsmann, J. Schmitz, G. Kaufel, J. Weber, M.
Mikulla, and J. Wagner, Proc. SPIE-Int. Soc. Opt. Eng., paper in
37. J. Wagner, Proc. SPIE-Int. Soc. Opt. Eng., paper in press (2005).
38. T. Swanson, Photon. Spectra 36, 78 (2002).
39. I. F. Lealman, A. E. Kelly, L. J. Rivers, S. D. Perrin, and R. Moore,
Electron. Lett. 34, 2247 (1998).
40. B. Mason, J. Barton, G. A. Fish, L. A. Coldren, and S. P. DenBaars,
IEEE Photon. Technol. Lett. 12, 762 (2000).
41. C. A. Williamson, M. J. Adams, A. D. Ellis, and A. Borghesani,
Appl. Phys. Lett. 82, 322 (2003).
42. C.-F. Lin, Y.-S. Su, and B.-R. Wu, IEEE Photon. Technol. Lett. 14,