Two effects that contribute to the wafer bow were carefully separated: the lattice mismatch between layer and substrate as well as the vertical temperature gradient across the wafer resulting from temperature difference between wafer pocket and shower-head. In result, the lattice match of InGaAs to InP could be tuned in-situ with a ±50 ppm resolution – an accuracy that formerly could be achieved only by ex-situ X-ray diffraction (XRD). Download the PDF of the talk here
Fig. 1a shows the 633 nm refractive index of InGaAsP and InGaAlAs in the full composition range at three relevant growth temperatures T1<T2<T3. With n(x,T) and k(x,T) available for the full range of lattice matched quaternary compositions, precise and quantitative process control becomes straight forward.
Fig. 1: Control of device related InGaAsP and InGaAlAs film growth on InP: When lattice matched growth is validated by in-situ wafer bow sensing, both quaternary material systems can be treated as an effective quasi-ternary mixture: (InGaAs)x(InP)1-x and (InGaAs)x(AlGaAs)1-x , respectively.
Fig. 1a) The respective composition range (x=0…1) covers the reference (PL) wavelength range 0.92–1.65 µm for InGaAsP and 0.85–1.65 µm for InGaAlAs. The three lines for each material system give the 633 nm refractive index at three wafer temperatures.
Fig. 1b) The quaternary/ternary layers (steps 17–29) are lattice matched to InP (step 15) as can be seen from the unchanged wafer curvature (green line). Hence, the measured 633 nm reflectance data (blue line) of an InP/InGaAsP device tructure can be exactly fitted (red line) yielding all compositions x and all growth rates. The –3 K reduction in wafer temperature due to the changed As/P ratio is a real effect.
For this highly precise nk database, growth rates and accurate lattice match were carefully gauged to ex-situ XRD. Wafer temperatures were measured by EpiTT that had been calibrated by AbsoluT.
For more information please feel free to contact firstname.lastname@example.org or call +49(0)30 89 00 55-0.
LayTec's new release of the control and analysis software EpiNet 2016 offers completely new analysis features for our customers interested in high-accuracy statistical process control (SPC) of related device growth processes. Fig. 1 gives an example:
Fig. 1: Screenshot of the EpiNet 2016: data analysis of an InGaAsP/InP device structure on InP(001): the thickness of the three very thin InGaAsP layers in steps 2, 6, 10 is: 28.5 nm, 48.7 nm and 100.3 nm respectively. The table in the lower part of the figure gives the sequence of analysis functions for routine and automated SPC of this device growth process.
The thickness of very thin InGaAsP layers in a device stack grown in an AIXTRON Planetary Reactor® on InP(001) is determined by a well selected set of automated analysis operations. First, several InP layers are utilized for permanent in-situ high-accuracy re-calibration of all reflectance channels (yellow lines) in long lasting epi runs. Second, the lattice matching of the quaternary layers is verified by wafer bow analysis (not shown). Third, the composition of the quaternary material is determined at the thick InGaAsP layer in step #14. And finally, based on this information, the thickness of the thin InGaAsP layers in steps #2, #6 and #10 is accurately measured by double-wavelength thickness analysis.
For better understanding of growth processes, LayTec offers related training courses for process engineers and quality managers. To learn more, please contact email@example.com or call our sales engineer +49 30 89 00 55-0.
Products for SEMI materials