InAs/(GaIn)Sb short-period superlattices for focal plane arrays

Figure 9a shows a SEM image of mesas with 40 µm pitch in a 256x256 FPA arrangement after epitaxial overgrowth
with the 150 nm thick Al0.5Ga0.5As0.04Sb0.96 layer. In Figure 9b the corner of such an FPA pixel is shown at higher
magnification. The excellent surface morphology without polycrystalline phases is evident. In order to prevent
oxidation of the Al-containing passivation layer, a 200 nm thick silicon nitride layer was deposited after removing the
sample from the MBE system. Since the Fermi level at the p-n junction on the diode sidewall is defined by the
Al0.5Ga0.5As0.04Sb0.96 layer, the electrical properties of the photodiode are not affected by this dielectric layer. The
silicon nitride was used as an etch mask for wet chemical etching of the Al0.5Ga0.5As0.04Sb0.96 passivation and the GaSbcap
layer. The wet chemical etchant chosen for this process step offers a sufficiently high selectivity to the Incontaining
SL. Finally, ohmic n- and p-contact metalization layers were deposited.
Various test diodes were wire-bonded after processing and characterized by dark I-V measurements. Since surface
leakage currents occur at the mesa sidewalls, dark current measurements in dependence of the perimeter-to-area (P/A)
ratio were carried out to separate the surface leakage currents from the bulk contributions. A convenient measure for
this investigation is the R0A product, since it is inversely proportional to the square of the limiting Johnson noise
contribution [10].
(R 0 A) -1 (Ω -1 cm -2 )
10 3
10 2
10 1
10 0
10 -1
10 -2
EDGE LENGTH OF
QUADRATIC MESA (µm)
800 700 600 500 400 300200 200 100
Dielectric passivation
Passivation by overgrowth
λ c = 10 µm
(a) 77 K (b)
77 K
0.02 0.04 0.06
PERIMETER-TO-AREA (µm -1 )
70 µm edge length
P/A = 0.057 µm -1
10 3
10 2
10 1
10 0
10 -1
0 2 4 6 8 10 12 10-2
DIODE #
Figure 10. (a) Inverse R0A product vs. P/A ratio of InAs/GaInSb superlattice photodiodes with 10 µm cut-off at
77 K. (b) Histogram of the (R0A) -1 product of the diodes from (a) with 70 µm edge length (P/A = 0.057 µm -1 ).
Figure 10a shows the (R0A) -1 in dependence of the P/A ratio on a logarithmic scale at 77 K. Since the P/A ratio depends
on the size and geometry of the mesa, the edge length of a quadratic mesa is depicted on the upper scale for
convenience. The (R0A) -1 data of the sample passivated by overgrowth is compared with a dielectrically passivated
sample with 10 µm cut-off wavelength. R0A values obtained with a dielectric passivation are low and the inverse R0A
increases linearly with increasing P/A ratio. Obviously the employed dielectric layer is not well suited for the
passivation of FPA pixels with 40 µm diameter and less. In contrast, the R0A product obtained with the passivation by
AlGaAsSb overgrowth is more than three orders of magnitude larger compared to the dielectric passivation.
Furthermore, the best diodes align along the dashed line corresponding to R0A ≈ 10 Ωcm 2 which is the known bulk
value of this material. The independency of (R0A) -1 on the (P/A) ratio indicates suppression of surface leakage currents
(R 0 A) -1 (Ω -1 cm -2 )