Bonding-induced mechanical stress in GaAs-based laser diodes is studied by numerical and experimental techniques.
This stress, induced by the soldering processes, appears when cooling down the assembly because of Coefficient of
Thermal Expansion (CTE) mismatch, and dimensional disparities. Detailed mechanisms taking place are not fully
understood. Residual stress is also known to influence device reliability. Composite submounts studied are composed of
a CuW heat spreader on an AlN bottom plate in standard and optimized designs to lower mechanical stress levels. CuW
shows a high thermal conductivity and a matched CTE with GaAs. Plain AlN submounts are studied as a reference. The
numerical technique is a Finite Element Method calculation to compute the stress tensor induced in GaAs-based laser
diodes during the soldering process on submounts with 80-20 AuSn eutectic solder pads. Starting with 31 MPa on the
plain AlN submount, the standard composite submount gives 23.5 MPa while the optimized version is as low as 12 MPa.
The experimental technique consists of Degree of Polarization (DoP) measurements of the photoluminescence emitted
by a planarized diode bonded to a submount. From the DoP, relative stress variations induced by the submount are
estimated. Starting with DoP referenced at 100% on plain AlN submounts, the standard composite submount gives 46%
DoP reduction while the optimized version is expected to exhibit a reduction larger than 65%. Composite submounts
with reduced mechanical stress and preserved thermal properties were studied experimentally and theoretically. An
optimized design allows reducing the mechanical stress by a factor 2.5 at least.
For novel devices such as quantum dot lasers, the usual thermal characterization using temperature induced wavelength
shift is ineffective due to weak thermal shift of the inhomogeneously broadened gain-peak. This calls for new thermal
characterization techniques for such devices. To this end we have analyzed bulk thermal properties of broad area
quantum dot lasers theoretically, and have experimentally verified these calculations using the novel technique of microthermography.
InGaAs/GaAs 950 nm emitting, 50 μm wide and 1.5 mm long, large optical cavity quantum dot lasers
were used for the study. Our two-dimensional steady-state model self-consistently includes current spreading and
distributed heat sources in the device and using finite element method reproduces high resolution temperature maps in
the transverse cross section of the diode laser. A HgCdTe based thermocamera with detection spectral range 3.5-6.0 μm
was employed for micro-thermography measurements. Its microscope with 6x magnification has a nominal spatial
resolution of 4 μm/pixel for full frame images of 384×288 pixels. A ray tracing technique was used to model the
propagation of thermal radiation inside the transparent laser die which in turn links calculated and experimentally
derived temperature distributions. Excellent agreement was achieved which verifies the model-calculation and the
thermal radiation propagation scheme inherent in the experimental approach. This result provides a novel means for
determining reliable bulk temperature data from quantum dot lasers.