March Etcher O2 Plasma Ash Performance

Roger Robbins10/10/2016

Purpose

This paper investigates the actual etch rate for removing S1813 photoresist as a de-scum process or removal step to verify the performance.

The de-scum step is intended to remove any residue in the bottom of developed photoresist patterns in order to enable uniform etching or prevention of “lift-off” adhesion issues. The strip process is intended to remove all the resist after pattern etch so the next device layer can be formed.

Experiment Description

Since the tool is used for de-scum operation via a low powered short exposure plasma etch, as well as longer timed photoresist removal, we undertook to examine the short time etch rate as well as a brief look at longer runs to see the actual etch rate characteristics.

Test 1 – Short Term Etch (De-Scum)

In this test, we exposed a 4” wafer coated with about 1.8 microns of Shipley S1813 photoresist, patterned and developed, to a timed plasma de-scum process consisting of a 50 Watt plasma of O2 at 200 mTorr pressure for 5 sec to 120 sec . Data was collected on the Nano Spec tool to achieve better statistical repeatability than the profilometer could produce. The etch rate data is shown in the graph of Figure 1. Note that there is anetch rate anomaly during the first few seconds. This is a repeatable phenomenon and coupled with visual observation of a bright flash as the plasma starts up, appears to be related to a high energy plasma startup surge and perhaps a tuning undershoot before the etch rate settles out at a constant level. Our de-scum process has always been set at 15 seconds and so it “accidently” fell at the standard etch rate level. However, this graph illuminates the region around the standard etch time and shows that you cannot make small adjustments to the process time to slightly increase or decrease the de-scum rate.

Figure 2shows a more useful plot in which you can estimate the amount of film removed versus etch time.

Test 2 – Long Term Etch (Resist Strip)

In this test, we used a 200 Watt O2 Plasma at 200 mTorr to determine the S1813 stripping (removal) rate as a function of time. This was also conducted on a 4 inch dia. Si wafer and the data measured on the Nano Spec tool. Figure 3 shows the etch rate as a function of time and Figure 4 shows the depth of material removed as a function of time for ease of estimating the time required to strip all of the resist. Figure 5shows the uniformity pattern for round large area wafers.

Figure 1. Etch Rate of O2Plasma 50 Watt De-Scum process.

Figure 2. Extended depth of S1813 photoresist removed as a function of time using the 50 WattDe-Scum process.

Figure 3. Etch Rate Calibration chart for 200 Watt Strip process for Shipley 1813 photoresist.

Figure 4. Depth of S1813 Photoresist removed by a 200 Watt plasma as a function of stripping time. This enables an estimate of how long it takes to completely strip an S1813 film of various thicknesses.

Figure 5. This etch pattern is different from the Technics in that the incoming gas originates from two gas dispensing pipes at the side of the front door and exits in the center of the rear side wall. This causes a more rapid etching at the front door and side edges of the wafer (wafer flat side in this instance) than at the center and rear. (Color rings result from thinning film thicknessesselectingthin film interference wavelengths that produce colors.) The data in this document was taken in the center of the wafer, so it will predict longer etch times than actually required for clearing small samples.