Content uploaded by Daniel Hofstetter
Author content
All content in this area was uploaded by Daniel Hofstetter
Content may be subject to copyright.
Content uploaded by Daniel Hofstetter
Author content
All content in this area was uploaded by Daniel Hofstetter
Content may be subject to copyright.
Content uploaded by Daniel Hofstetter
Author content
All content in this area was uploaded by Daniel Hofstetter
Content may be subject to copyright.
Continuous wave operation of a 9.3
m quantum cascade laser
on a Peltier cooler
Daniel Hofstetter,
a)
Mattias Beck, Thierry Aellen, and Je
´
ro
ˆ
me Faist
University of Neucha
ˆ
tel, Institute of Physics, 1 Rue A.-L. Breguet, Neucha
ˆ
tel, CH 2000, Switzerland
Ursula Oesterle and Marc Ilegems
IMO, Physics Department, EPFL, Ecublens, 1015 Lausanne, Switzerland
Emilio Gini and Hans Melchior
Institute of Quantum Electronics, ETHZ, 8093 Zu
¨
rich, Switzerland
共Received 21 November 2000; accepted for publication 7 February 2001兲
High average power quantum cascade lasers at 9.3
m using InP top cladding layers and both
junction up and junction down mounting are presented.A3mmlong, junction up mounted device
emitted 54 mW average power at 30 °C and 11.5% duty cycle with a threshold current density of
3.72 kA/cm
2
. A similar, but only 1.5 mm long device with high reflection coating on both facets was
mounted junction down and tested at even higher duty cycles. At ⫺27°C, we achieved continuous
wave operation with a threshold current density of 3.3 kA/cm
2
.©2001 American Institute of
Physics. 关DOI: 10.1063/1.1360225兴
Midinfrared quantum cascade 共QC兲 lasers have reached
a high level of maturity and are now commercially available.
They are the ideal light source for environmental and medi-
cal sensors.
1–4
Many of those applications call upon high
average output powers; this is particularly true for photo-
acoustic trace gas spectroscopy in the parts per billion con-
centration range. Up to this point, the highest average output
powers were achieved by using superlattice active region QC
lasers with an InP lower cladding and an InAlAs-based top
cladding layer.
5,6
Additional performance improvements are
expected by lowering the thermal resistance of the devices.
This can be accomplished by the use of an InP top cladding,
by epitaxial side down mounting and the use of buried het-
erostructure 共BH兲 lasers.
7–10
In this letter, we present our
latest results on vertical transition QC lasers at 9.3
m which
were fabricated with InP top cladding layers. In addition, we
improved the thermal resistance of the devices by junction
down mounting.
Up to the active region, growth of this material was
based on molecular beam epitaxy 共MBE兲 of lattice matched
InGaAs/InAlAs layers on top of an n-doped InP
(Si, 2⫻10
17
cm
⫺3
) substrate. After the waveguide core with
the active region, an epitaxial overgrowth by metalorganic
vapor phase epitaxy 共MOVPE兲 of the InP top cladding and
contact layers completed the growth. The MBE growth pro-
cess started with the lower waveguide layer 共InGaAs, Si, 6
⫻ 10
16
cm
⫺3
, total thickness 0.225
m兲, proceeded with an
active region 共thickness 1.82
m兲 and was finished by an
upper waveguide layer 共InGaAs, Si, 6⫻ 10
16
cm
⫺3
, thickness
0.23
m兲. After thorough cleaning of the surface in H
2
SO
4
,
the samples were transferred to an MOVPE system, where
the top cladding layer 共InP, Si, 1⫻ 10
17
cm
⫺3
, thickness 2.5
m兲, the contact layer 共InP, Si, 7⫻ 10
18
cm
⫺3
, thickness 0.85
m兲, and the cap layer 共InP, Si, 1⫻ 10
20
cm
⫺3
, thickness 10
nm兲 were grown. The active region, which formed the cen-
tral part of the waveguide, consisted of 35 periods; those
were alternating n-doped funnel injector regions and un-
doped four quantum well 共QW兲 active regions. The laser
transition in the latter was vertical and the lower lasing level
utilized a double phonon resonance like the device outlined
in reference.
11
Like in a superlattice active region, this four
QW design makes use of the short lifetime of the lower laser
level. On the other hand, the thin QW right behind the injec-
tion barrier guarantees a good injection efficiency, similar as
in the diagonal, anticrossed three QW active region.
12
The
layer sequence of the structure, in nanometers, and starting
from the injection barrier, is as follows: 3.4/1.4/3.3/1.3/
3.2/
1.5/3.1/1.9/3.0/2.3/2.9/2.5/2.9/4.0/1.9/0.7/5.8/0.9/5.7/0.9/5.0/
2.2 nm. In
0.52
Al
0.48
As barrier layers are in bold,
In
0.53
Ga
0.47
As well layers are in roman, and n-doped layers
共Si 4⫻ 10
17
cm
⫺3
for S1840 and Si 2.5⫻ 10
17
cm
⫺3
for
S1850兲 are underlined. Laser fabrication proceeded then by
standard processing steps like outlined in Ref. 2. Device
S1840 was mounted junction up on a copper heatsink, while
S1850 was soldered junction down to improve thermal resis-
tance. In both cases, a 2
m thick thermally evaporated In
layer was used for soldering.
For testing, the devices were placed into a Peltier-cooled
aluminum box with a antireflection coated ZnSe window
共Alpes Lasers SA兲. Average output power and voltage vs
current (L–I–V) curves at temperatures of ⫺30, 0, 30, and
60°C were measured in this configuration. A pulse length of
40 ns was used with a variable pulse repetition frequency in
order to achieve a duty cycle between 1.5%(f
rep
⫽ 375kHz)
and 20%(f
rep
⫽ 5 MHz). For higher duty cycles, we left the
repetition frequency constant at 5 MHz and changed the
pulse length from 40 up to 200 ns. The average output power
was measured using a calibrated thermopile detector. For the
acquisition of emission spectra, we collected the light with a
Au-coated parabolic off-axis mirror (60°, f/1.33). After re-
flection on a second parabolic mirror (90°, f/3.75), the light
a兲
Electronic mail: daniel.hofstetter@unine.ch
APPLIED PHYSICS LETTERS VOLUME 78, NUMBER 14 2 APRIL 2001
19640003-6951/2001/78(14)/1964/3/$18.00 © 2001 American Institute of Physics
Downloaded 08 Apr 2005 to 130.125.4.20. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp
was focussed onto the input slit of a grating spectrometer
共Jobin–Yvon, d
focal
⫽ 0.3m兲.
In Fig. 1, we present a series of L–I–V curves of the
junction up mounted laser S1840 with InP top cladding. The
cavity length was 3 mm and a stripe width of 28
m was
used. At ⫺30 °C and for 15.5% duty cycle, a maximal aver-
age power of 115 mW was seen; the threshold current den-
sity under these conditions was on the order of 2.68 kA/cm
2
.
At 30 °C, the corresponding numbers are 47 mW and 3.72
kA/cm
2
; this was already slightly lower than the maximal
power of 54 mW at 11.5%. As shown in the inset of Fig. 1,
the emission wavelength of this laser was around 1070 cm
⫺1
,
or 9.3
m. Because of the vertical transition used, the wave-
length did not change at higher current injection. At low duty
cycle 共1.5%兲, we observed peak powers of 1.6 W (dP/dI
⫽ 387mW/A) for ⫺30 °C and 1.0 W (dP/dI⫽ 300 mW/A)
at 30 °C. This exceptional performance is due to specific im-
provements in the design of the active region. The slope
efficiency benefits from both the high injection efficiency
and the short lower state lifetime of the lasing transition.
From the threshold currents at the two extreme temperatures,
we deduced a T
0
value of 214 K.
Figure 2 shows thermal roll-over average power vs duty
cycle curves for normal pulsed operation at two typical tem-
peratures. The duty cycle with the highest average power
was 15.5% for ⫺30 °C and 11.5% for 30°C. The maximal
power values are 115 mW for ⫺30 °C and 54 mW for 30 °C.
A simple mathematical model allowed the numerical deter-
mination of the thermal resistance of this laser.
8
It starts with
known quantities such as input electrical power, threshold
current, operating voltage, duty cycle, and uses the thermal
resistance of the device as a fit parameter. With the curves
shown in Fig. 2, we found a value of R
th
⫽ 8.62K/W for the
total 共i.e., laser and holder兲 thermal resistance. The corre-
sponding theoretical curves for the maximal power are also
displayed in Fig. 2 共dashed lines兲; they agree well with the
experimental findings. The above R
th
value can be reduced
considerably when mounting the device junction down di-
rectly on a copper heatsink. In the case presented, the heat
needs to be transferred through the 150
m thick InP sub-
strate and a 2
m thick In solder layer to reach the heatsink.
When using the junction down mounting technique, the ac-
tive region is only 2.5
m away from the wafer surface,
which is soldered onto the heatsink. An even further im-
provement of the thermal behavior at high duty cycle opera-
tion is expected when using BH lasers. In this case, the ac-
tive region is entirely embedded in InP, which will lead to a
very efficient heatsinking.
Similar experiments as shown for S1840 were made with
the slightly lower doped, junction down mounted sample
S1850. Due to the lower doping in the active region, it
showed lower output powers and slope efficiencies; how-
ever, the threshold current density was somewhat better than
the one of S1840. At low duty cycle,a3mmlong and 28
m wide laser exhibited 147 mW/A at ⫺30°C (j
th
⫽ 2.44kA/cm
2
) and 120 mW/A at 30 °C (j
th
⫽ 3.0kA/cm
2
).
Because of the copper heatsink being somewhat longer than
the laser cavity, the lower slope efficiency might be partly
due to a shadowing effect. A T
0
value of 226 K was calcu-
lated for this device.
FIG. 3. Thermal roll-over average power vs duty cycle curves of the device
S1850 at 0 °C and at 30 °C for burst mode 共top兲 and for normal pulsed
operation 共bottom兲. A burst duty cycle of 20% and a repetition frequency of
10 kHz were used for this experiment. The dashed lines are theoretical fits
using a thermal resistance value of 2.9 K/W 共burst兲 and 5.67 K/W 共normal兲.
FIG. 1. Light vs current and voltage vs current curves ofa3mmlongand
28
m wide device S1840 at four representative temperatures and a duty
cycle of 15.5%. The inset shows a typical emission spectrum of this device.
FIG. 2. Thermal roll-over average power vs duty cycle curves of the device
S1840 at ⫺30 °C and 30 °C. The dashed lines correspond to the theoretical
fit using a thermal resistance value of 8.62 K/W.
1965Appl. Phys. Lett., Vol. 78, No. 14, 2 April 2001 Hofstetter
et al.
Downloaded 08 Apr 2005 to 130.125.4.20. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp
If such a device is operated at high duty cycle, the Peltier
cooler, which can dissipate 20 W of thermal power, can
reach the limit of its capabilities. This is the reason why we
tested our lasers also in burst mode. In Fig. 3, we compare
the device performance of S1850 at high duty cycle in nor-
mal pulsed operation and in burst mode operation. In 20%
burst mode, the thermal load on the Peltier cooler is reduced
considerably, because the laser is in ‘‘normal’’ pulsed mode
only during 20
s; this short period is followed by a 80
s
long time during which the device remains unbiased. As
shown in the upper part of Fig. 3, the duty cycle within the
burst could be cranked up to a value of 100% 共which corre-
sponds to 20% ‘‘real’’ pulsed operation兲 for both 0 °C and
30°C. The maximal average power during the burst was 112
mW for 0 °C and 60 mW for 30 °C. At 100% duty cycle
within the burst, average powers of 36 and 10 mW were
obtained for 0°C and 30 °C, respectively. As with device
S1840, we calculated the thermal conductance of the laser
and found a value of 410 W/cm
2
K(R
th
⫽ 2.9K/W). The
lower half of Fig. 3 shows the measurements at real pulsed
operation. Here, the maximal thermal roll-over output power
for 0°C 共66.5 mW兲 was reached already at 27% duty cycle.
The dashed line corresponds to the simulated curve using a
total thermal resistance value of R
th
⫽ 5.67K/W. The thermal
resistance difference between regular pulsed operation and
burst mode corresponds to the thermal resistance of the laser
submount. It is obvious that the latter contributes still a con-
siderable fraction, namely almost 50%, to the total thermal
resistance. In addition, the deviation of the experimental
points from the theoretical curve around 40% duty cycle is
probably an indication of an early device aging process.
In Fig. 4, we show L–I–V curves of a different S1850
device. This particular laser had a length of only 1.5 mm.
High reflection facet coatings 共55% reflectivity兲 were used to
maintain a low threshold current density. In order to protect
the device from catastrophic failure, the measurement was
always stopped after having reached threshold. At a tempera-
ture of ⫺27 °C 共245 K兲, this laser could be operated under
continuous wave 共CW兲 conditions with a threshold current of
1.38 A (j
th
⫽ 3.3kA/cm
2
). The threshold current increased
considerably when going from 1.5% (I
th
⫽ 0.95A) via 65%
(I
th
⫽ 1.1A) up to 100% duty cycle (I
th
⫽ 1.38A). When tak-
ing into account the T
0
value just discussed, this threshold
current increase allows us to calculate the approximate tem-
perature increase ⌬T in the active region. The relatively high
value of ⌬T⫽ 85K illustrates that the active region suffers
from a thermal stress which might be play an important role
for device failure.
In conclusion, we have presented QC lasers at an emis-
sion wavelength of 9.3
m with InP top cladding layers and
both junction up and junction down mounting. Especially the
junction down device showed a reasonable thermal resis-
tance value and could be operated CW at ⫺27°C. Maximal
average output powers of 115 mW at ⫺30 °C and 54 mW at
30°C were seen for the junction up laser.
The authors gratefully acknowledge Michel Rochat and
Ste
´
phane Blaser for technical assistance. This work was fi-
nancially supported by the Swiss National Science Founda-
tion and the Science Foundation of the European community
under BRITE/EURAM projects UNISEL 共No. CT97-0557兲
and SUPERSMILE.
1
J. Faist, F. Capasso, D. L. Sivco, A. L. Hutchinson, and A. Y. Cho,
Science 264, 553 共1994兲.
2
D. Hofstetter, J. Faist, M. Beck, A. Mu
¨
ller, and U. Oesterle, Appl. Phys.
Lett. 75, 665 共1999兲.
3
C. Gmachl, J. Faist, J. N. Baillargeon, F. Capasso, C. Sirtori, D. L. Sivco,
S.-N. G. Chu, and A. Y. Cho, IEEE Photonics Technol. Lett. 9, 1090
共1997兲.
4
B. Mizaikoff, C. S. Murthy, M. Kraft, V. Pustogow, A. Mu
¨
ller, D. Hof-
stetter, J. Faist, and N. Croitoru, Proceedings of the PITTCON 2000 Con-
ference, New Orleans, LA, Session 185, 1418.
5
A. Tredicucci, F. Capasso, C. Gmachl, D. L. Sivco, A. L. Hutchinson, and
A. Y. Cho, Appl. Phys. Lett. 73, 2101 共1998兲.
6
A. Tredicucci, F. Capasso, C. Gmachl, D. L. Sivco, A. L. Hutchinson, A.
Y. Cho, J. Faist, and G. Scamarcio, Appl. Phys. Lett. 72,2388共1998兲.
7
C. Gmachl, A. M. Sergent, A. Tredicucci, F. Capasso, A. L. Hutchinson,
D. L. Sivco, J. N. Baillargeon, S. N. G. Chu, and A. Y. Cho, IEEE
Photonics Technol. Lett. 11, 1369 共1999兲.
8
M. Beck, J. Faist, U. Oesterle, M. Ilegems, E. Gini, and H. Melchior,
IEEE Photonics Technol. Lett. 12, 1610 共2000兲.
9
X. P. Jiang, H. Temkin, M. MacDonald, R. A. Logan, and D. Coblenz,
Electron. Lett. 30, 1680 共1994兲.
10
D. Hofstetter, D. Sun, C. Dunnrowicz, M. Kneissl, and D. W. Treat, IEEE
Photonics Technol. Lett. 10, 1371 共1998兲.
11
D. Hofstetter, M. Beck, T. Aellen, and J. Faist, Appl. Phys. Lett. 78,396
共2001兲.
12
J. Faist, C. Sirtori, F. Capasso, D. L. Sivco, J. N. Baillargeon, A. L.
Hutchinson, and A. Y. Cho, IEEE Photonics Technol. Lett. 10, 1100
共1998兲.
FIG. 4. Light vs current and voltage vs current curves for a 1.5 mm long and
28
m wide device S1850 with double-sided facet coating at five different
duty cycles between 64% and 100% and at ⫺27 °C.
1966 Appl. Phys. Lett., Vol. 78, No. 14, 2 April 2001 Hofstetter
et al.
Downloaded 08 Apr 2005 to 130.125.4.20. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp