Three samples previous patterned by photolithography were etched by reactive ion etching process (RIE) using different etchant gases (Ar, CF4 and mixture of Ar and CF4 respectively). The thickness and dimensions of patterns were determined and the etch rates were calculated. It is found that the etch rate of Ar and CF4 is the highest among three etchant gases, but the exact etch rate is unknown because all photoresist has been removed. The etch rate of CF4 is higher than that of Ar because CF4 combines with both physical and chemical etching and has a higher relative molecular mass compared to Ar gas. The dimensions of patterns on the samples are 460 × 910 ?m2.
The objective of this laboratory is to operate SENTECH Reactive ion etching Etchlab 200 system and understand the difference between samples etched by different etchant gases.
Ion etching is utilizing the kinetic energy of ions to remove material deposited on thin films. The principle is based on the bombardment of ion particles that are accelerated to the surface of the sample inside a vacuum chamber by a strong electrical field. The generation of ion particles is by dissociation of the etchant gas (usually inert gas such as Ar). Before ion etching process, a selective patterned photoresist is applied to the surface of the sample. This photoresist will also be etched away during etching process, but the etch rate of photoresist is lower than that of the sample. Therefore, the material underlying the photoresist can still be protected for several hours if etch rate is not too high.
RIE is a type of dry etching technology that combines with chemical plasma etching and directional ion milling. The process begins with ionizing gas molecules and generating plasma in an electromagnetic field. Accelerated high-energy ions collide with the sample surface and chemically react with it, forming gaseous by-products that will later be removed by vacuum pumps. In addition, these ions can also transfer their kinetic energy and knock off some atoms out of the material. The etch rate is related to many process parameters, including etchant gas, pressure in the chamber, RF power and gas flow. A typical schematic diagram of the RIE system is showed in Figure 1, including etch gas inlets, vacuum pumps, a vacuum chamber, a wafer platter and RF power source.
Figure 1. Schematic diagram of RIE1
1. Sample Etching
All thin film samples used in this experiment were prepared by the demonstrator. Three samples (sample 1, sample 2 and sample 3) were etched using Ar, CF4, and a mixture of Ar and CF4 respectively with SENTECH Reactive ion etching Etchlab 200 system. The process began with pumping the system to high vacuum and waiting for several minutes to ensure the pressure was stable. The etchant gas was then injected into the chamber. When stable pressure was reached, the RF generator was opened and etch process started. After the process was finished, the sample was taken out of the chamber. The process parameters of each experiment are showed in Table 1.
Sample 1 2 3
Etchant gas Ar CF4 Ar & CF4
RF power (V) 300.0 300.0 100.0
Pressure (Pa) 60.00 60.00 10.00
Etching time (min) 8 4 2
Gas flow (sccm) 40.0 40.0 Ar: 5.0
Table 1. Process parameters
2. Photoresist Stripping
Acetone was used to strip the remaining photoresist, after which the sample was dried by N2. The process was repeated twice to ensure the photoresist was completely removed.
Thickness and dimensions of patterns were measured by Veeco Dektak 150 Surface Profiler. Sample 1 and sample 3 were measured only once and sample 2 was measured twice.
Considering the samples were not levelled when measured, the tilt of the raw profile was corrected and then the vertical difference between an etched area and an unetched area could be determined. The results of surface profiler for sample 1, sample 2 and sample 3 are showed in Figure 2, Figure 3 and Figure 4 respectively.
Figure 2. Results of surface profiler for sample 1 (using Ar)
Figure 3. Results of surface profiler for sample 2 (using CF4)
a) showing the width of patterns; b) showing the length of patterns
Figure 4. Results of surface profiler for sample 3 (using mixture of Ar and CF4)
The vertical difference between an etched and unetched area (thickness of removed materials) can be obtained from Figure 2, Figure 3 and Figure 4, and etch rate is calculated from thickness of removed materials and etch time. These results are tabulated in Table 2.
Sample 1 2 3
Etchant gas Ar CF4 Ar & CF4
removed materials (nm) 407.77 335.73 5.02
Etch rate (nm/min) 50.97 85.47 2.51
Table 2. Thickness of removed materials and etch rate when using different etchant gases
Considering three samples have the same patterns, only the dimensions of patterns on sample 3 are recorded. According Figure 3, the width and length are obtained and listed in Table 3.
Width (?m) 460
Length (?m) 910
Table 3. The dimensions of patterns
Discussion and analysis
Sample 1 and sample 2 were made with the same pressure, gas flow and RF power but using different etchant gases. The etch rate of sample 2 (using CF4) was faster than that of the sample 1 (using Ar). There are two possible reasons for this result. One is that etching by Ar is ion milling, which is only based on physical bombardment of ion beam on surface film, while using CF4 is RIE technique, which combines with ion bombardment and chemical reactions with surface film. Free radical (F) is very active and easy to react with the sample, forming gaseous by-products that can be removed by vacuum pumps. For example, if the sample is made of Si, then free radical F can react with Si and form SiF4, which is gaseous and easy to be pumped out. This process is illustrated in Figure 5. Another reason is that the relative molecular mass of CF4 (88) is higher than that of Ar (40). CF4 can be dissociated to CF3, CF2, CF, F, C and their ions (as shown in Figure 6). Those heavier positive charged ions have higher kinetic energy and transfer more energy to the sample so that more atoms will be knocked out of the surface.
Figure 5. Chemical reactions between free radicals and the surface of the sample1
Figure 6. Various reactions and species in a plasma1
The etch rate for sample 3 is not acceptable because it seems that the etch rate was too high and all photoresist was removed by RIE process. From the microscopic image of sample 3 (not shown in this report), almost no clear patterned area can be found on the sample surface. In addition, there is no clear platform in figure 4, which indicates that all photoresist is removed and there is no protection for the material underlying the photoresist. Therefore, the calculated etch rate cannot be regarded as the real etch rate.
The sample 3 was etched with the lowest pressure, the lowest RF power, the lowest gas flow and the shortest etch time, but this sample exhibited the highest etch rate. There are several reasons for this phenomenon. The most possible reason is that the addition of Ar gas contributes more ion bombardment, which may physically damage the chemical bond of the materials and create more activated sites for chemical reactions between free radicals and the sample surface. This bombardment may also help remove by-products absorbed on the sample surface. Without this inhibitor layer, free radicals are easier to react with the sample. Another possible reason could be that the pressure in the chamber and gas flow were too high when sample 1 and sample 2 were etched. Mean-free path of an atom or a molecule in a gas ambient is inversely proportional to pressure. If pressure is too high, mean-free path will be small, which means that the probability of collisions between gas molecules increases and that less gas molecules can bombard the sample surface. The third reason may be too high RF power used in sample 1 and sample 2. High energy ions accelerated by high RF power may go too deep inside the sample instead of knocking atoms out of the surface.
In our experiment, all the measurements are only performed once due to the limited time. If possible, these measurements should be repeated to obtain more reliable data. Another mistake is that the photoresist of sample 1 was not completely stripped before dimensional measurements because some dark patterned areas can be found in the microscopic image. These patterned areas should be bright yellow if all photoresist is removed. This problem can be solved by repeating photoresist stripping process for more times.
Three samples were etched using Ar, CF4 and mixture of Ar and CF4 respectively. The etch rate of mixture of Ar and CF4 is the highest in three samples, but the exact etch rate is unknown because the etch rate is too high and all photoresist is removed. The etch rate of Ar gas (50.97 nm/min) is higher than that of CF4 (85.47 nm/min). This is because CF4 can both bombard and chemically react with the sample surface. In addition, its relative molecular mass is higher than that of Ar so that more kinetic energy can be transferred to the sample. The length of patterns is 910 ?m and the width is 460 ?m.
1 P. K. Petrov, MSE 410 Lecture 8.5(patterning-etching).