GaAs - Dry Etching

GaAs - Dry Etching

Material Name: GaAs
Recipe No.: 10322
Primary Chemical Element in Material: Ga
Sample Type: Wafer
Uses: Etching
Etchant Name: None
Etching Method: Dry etching
Etchant (Electrolyte) Composition: All test wafers were taken as 3-inch S.I. GaAs wafers with thickness ~650 µm. These wafers were then coated within 30 µm thick positive photoresist AZ 4620 and exposed to define circular holes with 70 µm diameter. The patterned photoresist was post-baked at 120°C to improve plasma etch resistance and adhesion. The patterned wafers were then mounted on a carrier wafer with wax to make them loading compatible with ICP system. This is necessary in view of the fact that ICP etching requires cooling of the wafer during etching, which is very difficult to achieve without carrier wafer.

GaAs wafers were then etched in standard ICP system, one at a time. An oxygen plasma de-scum step prior to etching was utilized in order to remove any residual photoresist in the hole, which would have otherwise contributed to the roughness of the etched surface. Plasma of etcher is inductively coupled through a coil at 13.56 MHz, with independent energy control provided by 13.56 MHz RF biasing on the substrate. Helium gas was used to cool backside of the wafer. The substrate temperature was set at 20 °C for all test conditions. The etch chemistry was a mixture of Cl2/BCl3 through mass flow controlled process gas lines. The chamber was evacuated to a base pressure of ~ 9 x 10 exp(-6) Torr by a turbo-molecular pump backed by a dry mechanical pump before initiating the etch process. The etch gases mixture was introduced through an annular region at the top of chamber lid. ICP etching was carried out using positive photoresist mask to etch holes selectively. All the etching runs were carried out through a design of experiments by varying ICP power, process pressure, Cl2/BCl3 flow rate ratio and chuck bias power. The etch rate, etch depth, etch profile, mask selectivity and surface morphology of etched features were determined by Scanning Electron Microscopy (SEM). Deposition on the etched sidewall was analysed by high resolution Energy Dispersive X-ray Analysis (EDAX).

Effect of Pressure: Fig. (1) shows the variation of etch rate and the undercut with the process pressure. The process pressure is varied from 5 mTorr to 12 mTorr at a constant ICP power of 800W, RF bias power of 8 0W and Cl2:BCl3 ratio of 4:1. It is seen that the etch rate increases gradually from 4.6 µìm/min to 6 µm /min, when the pressure is raised from 5 mTorr to 12 mTorr. The increase in the etch rate with the pressure is attributed to an increase in the concentration of reactive chlorine species that enhances chemical component of the etching. This suggests reactant limited regime at lower pressures. In addition, it is evident from the figure that the undercut also increases from 19.6 µm to 21.5 µm when the pressure is increased from 5 mTorr to 12 mTorr. The reason for the enhanced undercut can be understood as follows. Actually the mean free path and plasma efficiency reduces as we raise the pressure. This results in a reduction in sputtering component on the etched surface. The decrease in sputtering component with increasing pressure results in poor surface morphology of etched surface with higher undercut. On the other hand, the etch profiles obtained at the pressure of 10 mTorr and 12mTorr under identical conditions are shown in Fig. (2). It is clear from the figure that the better surface morphology of etched surface is achieved at 10mTorr with slight narrowing at the bottom of etched profile.

Effect of ICP Power: In Fig. (3), we show the variation of etch rate with ICP power for two different pressures of 10 mTorr and 12 mTorr. It is obtained here that the etch rate gets increased with the ICP power up to 600 W at 10 mTorr pressure and 80 W RF bias power, suggesting that the etching is limited by reaction rate mechanism. This is attributed to higher concentration of reactive chlorine radicals, which increase the chemical etching component and hence ion flux that enhances the bond breaking and sputter de-sorption efficiency of the etch process. Further increase in the ICP power (above 600 W) however leads to a relatively constant etch rate, which is mainly due to competition between the sputtering and the etch reaction as a consequence of increased ions that are sputtering the adsorbed species (neutrals or ions) out of the surface prior to etch reaction. This sputter de-sorption at higher ICP powers results in etched surface with better surface morphology at the same RF chuck bias power.
Fig. (4) shows the etch profile obtained at higher ICP power of 1000 W, when the pressure is kept as 10 mTorr. Under this situation, the etch profile with smooth etched sidewalls is obtained. Fig. (3) also shows that the reaction limited regime is enhanced up to 800 W with an increase in the process pressure to 12mTorr at 60 W RF bias power. Further, as the concentration of reactive species is increased with pressure, their sputter de-sorption at 600 W is not sufficient to remove active species. Hence, the etch rate is found to saturate after 700 W up to 1000 W at 12 mTorr and 60W RF bias power.

Effect of RF Bias Power: The dependence of etch rate and mask selectivity on the RF bias power is shown in Fig. (5). It is clear from the figure that the etch rate depends strongly on the bias power and it gets enhanced rapidly from 4.73 µm/min to 5.36 µm /min, when the bias power is increased from 60W to 90W. These observations were made at 800 W ICP power, 10 mTorr pressure and 4:1 as the Cl2:BCl3 ratio for 15 min as the etch time. Enhancement in the etch rate is mainly attributed to higher physical etching component at higher bias power due to enhanced sputtering as a result of increased ion energy, which improves removal of etch by-products and assists in bond breaking. In addition, the etch an-isotropy is found to enhance as RF bias power is increased. This is expected as the ion energy gets increased due to the higher DC bias, which leads to the straight sidewall profile of the etched feature.
The GaAs: photoresist etch selectivity was determined by the ratio of GaAs etch rates to photoresist etch rates. This is portrayed in Fig. (5), where photoresist selectivity shows a strong dependence on the RF bias and ICP power. Mask selectivity is found to deteriorate from 15.53 to 13.85 when the bias power is increased from 60 W to 90 W. This is mainly caused by enhanced physical etching component at higher RF biases. It would be worth mentioning that this is one of the best reported selectivity using photoresist mask for GaAs etching applications at low pressures. The increase of resist etch rate with RF bias power and ICP power is due to dominant physical etch mechanism. Actually larger RF bias power increases the energy of ions bombarding the wafer surface and the higher ICP power increases the density or flux of the ions bombarding the wafer surface. Both of these activities finally enhance the physical etch mechanism. On the other hand, the reduced resist etch rate with increasing pressure is mainly due to less physically and higher chemically driven processes as a result of increased reactant concentration at higher pressure.
The desired etch profile with controlled side wall angles and surface smoothness can be obtained with a proper balance between enhanced sputter de-sorption of etch byproducts and surface protection by by-products. The sputter de-sorption of etch by-products is a function of ion energy that depends upon pressure and ICP/bias power. Based on above experimentation we arrived at a optimum recipe with ICP power of 800 W, bias power of 60 W and pressure of 12 mTorr that resulted in etch profile with smooth surface and controlled sidewalls with angle of ~84º, without narrowing at the bottom, as shown in Fig. (6).
The straighter sidewall at the bottom is perhaps due to the balanced chemical and physical etching components at 800 W ICP and 60 W RF bias. The EDAX spectrum of optimum etched surface, as shown in Fig. (7), also indicates negligible deposition of CClx polymer, most probably due to a better balance between enhanced sputter de-sorption of etch by-products and surface protection by by-products. Moreover, the etched GaAs surface roughening is strain induced, which tends to be more at defect sites due to higher strain that leads to non-uniform etching.

Etch Rate Variation with Cl2: BCl3 Flow Rate Ratio: Fig. (8) shows the etch rate variation with Cl2/BCl3 flow rate ratio in ICP plasma for an etch time of 60 min at the pressure of 15 mTorr. Here we observe that the etch rate is increased from 2.3 µm/min to 2.7 µm/min as the flow ratio is enhanced from 3 to 6.5. This is mainly due to an increase in chlorine radicals with increasing percentage of Cl2 in gas mixture that is responsible for chemical etching. Etch byproducts staying on the surface block the fresh enchant species from reacting. This saturates the etch process and eventually leads to a reduction in etch rate at the above ratio of Cl2:BCl3 as 6.5. In other words, above 6.5 ratio, reactant limited regime dominates that reduces the etch rate. This behavior suggests that the balance between physical etching component and chemical etching component gets deteriorated with the flow rate ratio, which degrades the surface smoothness. On the other hand, the de-sorption of chemical species limits the etching process at higher Cl2 flow rates, leaving some of the residue on the surface itself. This is shown in Fig. (9).
The EDAX spectrum of etched surface clearly indicates the deposition of CClx polymer at high flow rates and pressures (Fig. 10), which would have otherwise sputtered away at low pressures due to higher ion energy.

Effect of Etch Time: Fig. (11) shows the dependence of average etch rate and etch depth on the etch time for ICP power of 800 W, 12 mTorr pressure and 60 W bias power. The average etch rate is found to decrease with etch time, from 5.26 µm/min at an etch depth ~ 79..m to 1.71µm/min at an etch depth ~ 162..m. This is mainly due to the reduction in the effectiveness of supplying reactive species and removal of etch by-products with increased depth. The etch rate reduction can be compensated by increasing the gas flow rates and pressure. However, this in turn results into deeper etch profiles with severe by-product deposition on the sidewalls. This makes it necessary to incorporate a polymer cleaning process for deeper profiles after completion of etching step.
Procedure (Condition): No data
Note: GaAs etch characteristics like etch rate, etch profile sidewall angle, etch surface morphology and selectivity are studied as a function of Inductively Coupled Plasma (ICP) power and Cl2/BCl3 flow rate ratio in ICP at low pressure (<15mTorr) and low RF bias power (<100W) regime to achieve moderate GaAs etch rate with an-isotropic profiles and smooth surface morphology. The low pressure regime etching at Cl2/BCl3 flow rate ratio of 4:1 has resulted in vertical etch profiles with controlled sidewall angle ~84 deg., smooth surface morphology and good mask selectivity ~15 without significant deposition of CClx polymer on the etched sidewalls but with limited etch depth ~ 100 µm using photoresist mask. The mask selectivity is found to be a strong function of RF bias power and ICP power and a weaker function of process pressure. The resultant etch depth increases with an increase in pressure and flow rate ratio at the expense of etch surface morphology, as the desorption of chemical species limits the etching process at higher Cl2 flow rates and leaves some of the residue on the surface.
Reference: D.S. Rawal, et al., A Highly Selective Low Pressure Inductively Coupled Plasma Etching Process for GaAs Using Photoresist Mask, The Open Plasma Physics Journal, 2011, 4, 34-39.

Figure 1: Variation of etch rate and undercut with the pressure.

Figure 2: Etch profiles obtained under identical conditions at the pressure of (a) 12 mTorr and (b) 10 mTorr.

Figure 3: Variation of etch rate with ICP power.

Figure 4: Etch profile obtained at 10 mTorr pressure.

Figure 5: Dependence of etch rate and mask selectivity on RF bias power.

Figure 6: Optimum etch profile with smooth surface morphology and controlled sidewall angle of ~84º.

Figure 7: EDAX spectrum of etched surface at 12mTorr pressure, 800 W ICP and 60 W RF bias.

Figure 8: Etch rate variation with Cl2/BCl3 flow rate ratio at 15 mTorr/1000 W ICP/60 W RF bias.

Figure 9: Etch profile obtained at 15m Torr with Cl2= 125 sccm and BCl3= 20 sccm flow rates and 1000 W ICP/60 W RF bias.

Figure 10: EDAX spectrum of etched surface at high gas flow rates.

Figure 11: Dependence of average etch rate and etch depth variation on etch time.