Mass production of GaAs-based advanced devices accelerates wafer scale-up and process optimization. Those devices include such as GaAs/AlGaAs or GaAs/InGaP metal semiconductor field effect transistors (MESFET), heterojuncton bipolar transistors (HBT) and high electron mobility transistors (HEMT). Recently, GaAs device industries try to increase its volume with processing of large size wafers such as 100 and 150 mm diameters. Especially, it is currently a transition time from 100 mm (4 in.) to 150 mm (6 in.) in mass-production of MESFET and HBT industries. It means that most of GaAs-based processes have to be ready for scale-up.
One of critical issues of dry etching of such a large GaAs wafer is to maintain a good uniformity. It is a key issue to make it possible to preserve the process uniformity in a large area GaAs wafer. Another issue for high productivity during GaAs etching is to minimize edge exclusion on a wafer. In advanced GaAs industries, the edge exclusion is getting tighter than ever. The limit for edge exclusion currently revolves from 8–10 to 4–6 mm. Minimization of edge exclusion can improve efficiency of throughput after dry etch processes.
Compactness of the reactor is also an important issue for business and maintenance point of view in industry. If the reactor is inadequately big for a certain size wafer processing without optimization, it wastes extra cost for expense. The chamber will require expensive vacuum pumps, which are bigger than it needs. It will also require more gas flow in order to maintain a given pressure than an optimized chamber will do.
The issue is now clear. It is how to improve the etch depth distribution on a wafer during plasma etching in a compact chamber. One of basic information that we want to know is to understand effect of gas flow distribution to that of etch depth in the chamber. In our previous experiment, it was found that the etch depth on GaAs wafer was generally higher at the edge and low at the center after plasma etching in both selective and non-selective ICP etching when the gas inserted from the edge of a reactor. Our curiosity goes to try to understand “Is there any geometrical effect in order to improve the dry etch uniformity?” and “How to optimize clamp design for uniform plasma etching if a reactor size is already fixed?”
In this paper, we will discuss about how to improve the gas flow uniformity in a chamber. Both simulation and experimental data showed that it was possible to have optimized chuck and electrode design for excellent gas flow in a reactor. It was also understood that size and height of clamp and focus ring significantly affected gas flow uniformity, etch depth topography and etch depth uniformity, finally.
A finite difference numerical method was used for gas flow simulation. A chamber was designed as 300 mm diameter and 300 mm height. A chuck was designed as 200 mm diameter and 150 mm height in the simulation. Several clamps and focus rings with different sizes and heights were introduced in the simulation. Plasma gases could be inserted through center, edge or a showerhead in the calculation. Simulation was done for both 100 mm (4 in.) and 150 mm (6 in.) diameter GaAs wafers.
A few simulated data was compared to real experiment results acquired with an ICP etching system. BCl
/N /SF /He gas was used
for GaAs etching. Full-size 100 mm GaAs wafers were used to collect the etch
depth data as a function of focus ring height for comparison of the data.
3. Results and discussion
shows simulated gas flow uniformity as a function of wafer diameter on the electrode. It is noticed that flow distribution by gas insertion from the edge has a similar trend to that through a shower head over the wafer. The results depicted that gas flow distribution was always high at the edge and low at the center if gas was introduced through either edge of the reactor or showerhead over the wafer. However, it was noticed that the gas insertion from the center could change the trend of gas flow distribution on the wafer. A reverse result of gas flux (i.e. high at the center and low at the edge) was obtained by changing the position of gas insertion in the chamber. Gas flow uniformity could be poor as high as ±25% in any condition with the model reactor. Width of the reactor was 300 mm and height of the electrode was 150 mm in the simulation model. No clamp on the electrode was assumed for the data on the
Fig. 1. Gas flow uniformity as a function of distance from an edge on 150 mm wafer (without a clamp).
shows simulation result on an effect of a 0.5 cm thick clamp for gas flow distribution in the reactor for 150 mm GaAs wafer processing. Introduction of a clamp in order to fix a wafer could affect the gas flow. Further simulation indicated that height of the clamp and size of the clamp were also important to control gas flow distribution on the wafer in the chamber. According to
Fig. 2. Gas flow uniformity as a function of distance from an edge on 150 mm wafer (with a clamp).
Simulated data for gas flow uniformity on a 100 mm diameter wafer was shown on
Fig. 3. Gas flow uniformity as a function of distance from an edge on 100 mm wafer (with a clamp).
The simulation results were compared with experimental data. Comparison of /N/SF/He) was inserted from the edge in the experiment. Different height of dielectric-based focus rings was introduced in sequential runs. The results showed that etch depth distribution exactly resembled those of gas flow, i.e. optimized height of focus ring existed in order to achieve minimum etch uniformity. A main role of the focus ring was to control the gas flow distribution in the chamber.
and indicated that the simulation of gas flow
distribution matched very well for etch depth uniformity for GaAs etching. The
etching gas of GaAs (BCl
Fig. 4. Etch depth distribution of GaAs as a function of distance from an edge on 100 mm GaAs wafer after inductively coupled BCl
/N /SF /He plasma
etching (with a clamp).
shows gas flow uniformity as a function of clamp size. A best configuration was achieved with a gas insertion through a showerhead in the simulation. According to the result, the gas flow uniformity varies as a function of clamp size as well as the height of the clamp even though the dimension of reactor and electrode were fixed. Notice that the lowest uniformity existed with a 1 cm distance from the wafer in the reactor. Clamp thickness was fixed as 0.5 cm in the simulation. Therefore, it meant that the height of clamp up to 1 cm from the wafer edge should be as low as wafer thickness in order to achieve excellent uniformity. The results showed that the flow uniformity could be reduced as low as <±1.5% up to the very edge (<2 mm) of the wafer.
Fig. 5. Gas flow uniformity as a function of distance from an edge on 100 mm wafer (with both a clamp and a showerhead).
Gas flow distribution on 150 mm (6 in.) diameter GaAs wafer was simulated at the same chamber and electrode with gas insertion through a showerhead (
Fig. 6. Gas flow uniformity as a function of distance from an edge on 150 mm wafer (with both a clamp and a showerhead).
4. Summary and conclusions
We compared simulated gas flow uniformity with experimental data of etch depth distribution for dry etching of large area GaAs wafer. The results showed that etch depth trends of GaAs in BCl
/N /SF /He followed the
trace of gas flow distribution simulated by finite difference numerical method.
It was confirmed that advanced design of gas flow distribution in a reactor was
very important to have excellent etch depth uniformity for a large area GaAs
wafer in plasma etching. Introduction of an optimized clamp and a focus ring
could help achieve great uniformity on the wafer for BCl /N /SF /He plasma etching of GaAs wafer.
Source: Solid-State Electronics