Chapter 4 Results and Discussion
4.2 The device fabrication
After GaN bulk etching test, we applied PEC etching on fabrication process of AlGaN/GaN HEMTs. Due to the absence of sample of AlGaN/GaN HEMTs with n+-GaN capping layer, we only can finish the part of mesa isolation. The meaning of mesa isolation, the first step for HEMTs process, is to define the active region of device and to isolate the devices.By the database of PEC etching, we chose the standard etching recipe (KOH concentration 0.015M, light intensity 30~50mW/cm2, stirred solution and Ti etching mask) to fabricate mesa isolation. Due to the specific quality of device sample, we faced a lot of obstacles for mesa isolation. The first one was that PEC etching can’t etch the second layer (GaN bulk) of device, because the barrier of 2DEG was too high for electron to transmit from GaN buffer into AlGaN. Figure 4.15 was the band diagram.
Figure 4.15 band diagram for AlGaN/GaN HEMTs
Figure 4.16 showed the etched surface after 50min, 90min, and 130min PEC etching respectively. From the SEM figures, there was no etching depth for GaN even over 130min PEC etching.
In order to solve the transmission problem, we used the ohmic etching mask to avoid the failure of transmission mechanism of buffer layer. We solved successfully the unetched phenomenon, but there was the second obstacle after using ohmic etching mask. From the Figure 4.17, there was nonuniform etching at the region between devices. We thought it was due to the electron accumulations on the etching mask, resulting in the nonuniform etching near the mask. We tried to minimize the effect of nonuniform etching by using 0.0075M KOH solution, but the method can’t solve this problem thoroughly.
Finally we overcame the obstacles by using double Ti etching mask and finished the device mesa isolation successfully. The flow of mesa process was shown in Figure 4.18. We used the first Ti mask to etch the AlGaN, then deposited the second Ti mask to etch the GaN buffer.
After mesa isolation, we finished the ohmic contact, optical gate, and nitride passivation consequently. The performances of DC characteristics were shown in Figure 4.19. The device size was a 100um gatewidth, 2um gatelength, 5um source-drain spacing, and 2um G-D spacing. The IDss was 510mA/mm and Gm was 120mS/mm for PEC etching device.
Besides of wet etching, we fabricated the second device by dry etching.
Figure 4.20 showed the DC characteristics. The device size was the same with that fabricated by PEC etching. The IDss was 460mA/mm and Gm is 110ms/mm for ICP dry etching. According to the DC performance in Figure 4.19 and Figure 4.20, we can identify that the double Ti etching
mask solved successfully the process step of device mesa isolation, and the device fabricated by PEC etching had the better characteristics than that by dry etching. We believed the difference of the DC performances between PEC and dry etching resulted from the etching damage existing on etching surface.
Chapter 5 Conclusions
This work studies the PEC etch for GaN materials. The PEC etching
was successfully applied to n+-GaN and n--GaN and a new etching mechanism was proposed.The etching rates of n+-GaN with Ti mask at 0.0075M, 0.015M, and 0.03M KOH concentration were 700Å/min, 1000Å/min, and 1400Å/min respectively. From the data, we know that the higher the KOH concentration is, the faster the etching rate is. The etching rate of n+-GaN using ohmic etching mask was 2000 Å/min (0.015M KOH concentration).
Apparently the etching rate for ohmic mask was higher than that for Ti mask. The roughness was 16.0nm and 15.2nm for Ti and ohmic mask at 0.015M KOH concentration respectively. The etched roughness of the samples were in the same order for the n+-GaN etched at different etching conditions.
The etching rates of n--GaN were 350Å/min and 800Å/min for using Ti etching mask and ohmic etching mask respectively and the surface roughnesses were 68.327nm and 60.531nm for Ti and ohmic etching mask respectively. From the comparison for n+-GaN and n--GaN, we know that the higher dopant concentration would minimize the selective etching effect, and result in the smooth etching surface.
In this work, we found the transmission mechanism for etching mask in PEC etching, and explained why the etching rate fast of ohmic mask
was. Besides of this, we used double Ti mask to solve successfully the mesa isolation, and the PEC etching was applied to the AlGaN/GaN HEMT process for mesa etch. The results were compared to those AlGaN/GaN processed with conventional ICP dry etch for isolation.
From the DC measurement data, the device fabricated by PEC etching had the better characteristics with Gm 115mA/mm and IDss 510mS/mm;
meanwhile the device using dry etching for isolation had Gm 110mA/mm and IDss 460mS/mm
References
1. D.A. stocker, E.F. Schubert, J.M. Redwing, Appl. Phys. Lett. 73 (18), 2 November1998, “Crystallographic wet chemical etching of GaN.”
2. C. Youtesey, I. Adesida, G. Bulman, Elec. Lett. Vol.33 No.3 (1997) “Broad-area photoelectrochemical etching of GaN”
3. C. Youtesey, I. Adesida, G. Bulman, Appl. Phys. Lett. 71 (15), 13 October 1997, “Highly anisotropic photoenhanced wet etching of n-type GaN.”
4. J. Skriniarova, A. Fox, P. Bochem, P Kordos, IEEE 2000,
“photoenhanced wet etching of gallium nitride for gate recessing.”
5. M. S. Minsky, M. While, E. L. Hu, Appl. Phys. Lett. 68 (11), 11 March 1996, “Room-temperature photoenhanced wet etching of GaN.”
6. C. Youtesey, I. Adesida, L.T. Romano, G. Bulman, Appl. Phys. Lett. 72 (5), 2 February 1998, “Smooth n-type GaN surfaces by photoenhaced wet etching.”
7. T. Rotter, D. Mistele, J. Stemmer, F. Fedler, J. Aderhold, J. Graul et al.
Appl. Phys. Lett. 76 (26) 26 June 2000, “Photoinduced oxide film formation on n-type GaN surfaces using alkaline solutions.”
8. J.L. Weyher, F.D. Tichelaar, H.W. Zandbergen, L. Macht, P.R.
Hageman, Appl. Phys. Lett. 90 (12) 15 December 2001. “Selective photoetching and transmission electron microscopy studies of defects in heteroepitaxial GaN.”
9. J. Skriniarova, A. Van Ver Hart, H.P. Bochem, A. Fox, P. Kordos, Materials Science and Engineering B91-92 (2002)298~302,
“Photoenhanced wet chemical etching of n+-doped GaN.”
10.C. Youtesey, L.T. Rommano, I. Adesida. Appl. Phys. Lett. 73 (1998) 6, (Gallium nitride whiskers formed by selective photoenhanced wet etching of dislocation.)
0 1 2 3 4 5 6 0
1000 2000 3000 4000 5000 6000 7000 8000
Etching depth (A)
Time (min)
0.03M 0.015M 0.0075M
Figure 4.1 the etching rate of Ti etching mask for n+ GaN at 0.0075M, 0.015M, and 0.03M respectively.
(a)
(b)
Figure 4.2 the SEM image for n+ GaN with 0.0075M KOH concentration (a) after PEC etching, (b) after KOH treatment
(a)
(b)
Figure 4.3 the SEM image for n+ GaN with 0.015M KOH concentration (a) after PEC etching, (b) after hot KOH treatment
(a)
(b)
Figure 4.4 the SEM image for n+ GaN with 0.03M KOH concentration (a) after PEC etching, (b) after hot KOH treatment
(a)
(b)
(c)
(d)
Figure 4.5 the AFM image for n+ GaN (a)unetched (b) 0.0075M (c) 0.015M (d)0.03M
0 2 4 6 8 0
2000 4000 6000 8000 10000 12000 14000
Thickness (A)
Time (min)
Ti mask ohmic mask
Figure 4.6 the etching rate of Ti and ohmic etching mask for n+ GaN at 0.015M solution concentration.
Etching depth (Å)
(a)
(b)
Figure 4.9 the AFM image for n+ GaN after hot KOH treatment (a)Ti etching mask (b)ohmic etching mask
(a)
(b)
Figure 4.10 the SEM image for n+ GaN with ohmic mask using 0.015M KOH concentration (a)after PEC etching, (b) after hot KOH treatment
Figure 4.11 the etching rate of Ti and ohmic etching mask for low doping GaN at 0.015M KOH concentration.
0 2 4 6 8 10 12 14 16 18 20 22
(a)
(b)
Figure 4.12 the SEM images for low doping GaN with Ti mask using 0.015M KOH concentration (a)after PEC etching, (b) after hot KOH treatment
(a)
(b)
Figure 4.13 the SEM images for low doping GaN with ohmic mask using 0.015M KOH concentration (a)after PEC etching, (b) after hot KOH
treatment
(a)
(b)
(c)
Figure 4.14 the AFM image for low doping GaN after hot KOH treatment (a)unetched surface (b)Ti etching mask (c)ohmic etching mask
(a)
(b)
(c)
Figure 4.16 the SEM images for HEMTs device using the standard PEC etching (a) 50min (b) 90min (c) 130min
(a)
(b)
Figure 4.17 the optical microscope images for HEMTs device using ohmic etching mask (a) 0.015M (b) 0.0075M
Ti mask AlGaN
GaN
Figure 4.18 the process flow of device mesa isolation
(a)
550 Vg=0 to -5 volts, step=-1 volt
Ids(mA/mm)
Vds(volt)
(a)
(b)
Figure 4.20 the DC performances of the device fabricated by dry etching (a) IV curve (b) Gm
500 Vg=0 to -5 volts, step=-1 volt
Ids(mA/mm)