An evaluation of semiconductor processing fluids
07/01/2001
PROCESSING
by Carl Newberg
PTFE OR PFA PUMPS CAN REDUCE CONTAMINATION CAUSED BY METALLIC CONTAMINANTS IN SEMICONDUCTOR PROCESSING FLUIDS.
 |
The need for ultra-pure chemicals in the processing of semiconductors is well known. One factor of contamination is the level of trace dissolved metallic contaminants in semiconductor processing fluids. In particular, levels of iron, zinc, aluminum and sodium are considered contaminants for some critical semiconductor processes. It has been found that one source of this metallic contamination in those fluids is the pump and tubing used to move the fluid from the processing tanks, through filtration systems and into/out of storage tanks.
 |
One solution to this problem has been the development of pumps made out of only PTFE and/or PFA. The use of PTFE pumps has resulted in a significant decrease in extractable metal ionic contamination. Figure 1 is a graph obtained from one maker of PTFE pumps showing experimental results of an analysis of iron in recirculating hydrofluoric acid when a PTFE pump was installed. Figure 2 shows similar results for zinc, aluminum and sodium in an ammonia generator. In both cases metallic contamination levels dropped significantly when a PTFE pump was installed.
 |
One concern related to the use of a PTFE pump has been the issue of static electricity being induced into the fluids. That static charge could then cause device damage through ESD; or more catastrophically, a fire/explosion hazard when flammable fluids are used. Many flammable fluids are used in the semiconductor processing industry, such as isopropyl alcohol (IPA), methanol and various solutions used for photoresist strip operations, etc.
A literature search revealed very little regarding charge generation/ storage in fluids. Yatsuzuke1 performed experiments where pure water was rolled down Teflon sheets. The amount of charge developed in the water droplets was measured, and an electrostatic voltmeter was used to measure the resulting charge on the Teflon sheets. This experiment showed that charge could be introduced into the fluid as well as into the Teflon when fluid was transported across a Teflon surface.
Davies and Gerkin2 investigated the propensity for a flammable fluid to generate enough static charge to ignite when it contacted a grounded metal object.
Very little work has been done to investigate other processing fluids, as well as these fluids in a pumping operation, so a set of experiments was developed to measure the voltage present in various common semiconductor processing fluids when pumped through a PTFE pump and tubing.
This paper will document an investigation of the voltage generation in six different semiconductor processing fluids—air, deionized water, methanol, isopropanol, an aqueous ash residue remover and a solvent-based photoresist stripper/ash remover—as they are pumped through a PTFE pump and tubing.
Experimental procedure
Resistivity of test fluids. Because the ability of a material to conduct electrical current is paramount to understanding its performance in a pumping system, the electrical resistivity of the different fluids was measured. A cylindrical resistivity cell and a Keithley Model 156a electrometer/high resistance meter were used for these measurements. The resistance was measured, and the resistivity calculated via the geometry of the cell. The data from this testing is included as Table 1. Based upon this data, it was concluded that fluids (with the exception of air), were in the low static dissipative range, and will conduct electricity.
 |
Voltage generation measurements during the pumping operation. Conditions in the laboratory during the voltage measurements included an approximate 100 ft/min airflow over the test apparatus to simulate the airflow in a fume hood. The ambient temperature was 70 to 73 degrees F, and the relative humidity was 39 to 44 percent RH.
 |
A PTFE pump was supplied that had been modified slightly for this testing. Two holes were tapped into the dry side of the baffles to allow insertion of a metal "antenna" to measure the fields generated during operation of the PTFE pump (see Figure 3). Measurement of the voltage induced into the metal antenna by the field generated from the moving parts of the PTFE pump was measured with two different voltage measuring apparatus during a dry pumping operation.
 |
A Novx Model 5000 remote probe was used first to measure the voltage induced in the antenna. Figure 4 is a graph of the data obtained from one of these tests in one of the two chambers. The voltages measured using this technique were significantly lower than expected.
 |
The pump was connected to tubing in a configuration that allowed a continuous circulation of the fluid under test (see Figure 5). Voltage generation experiments were made by pumping the fluid through the system and measuring the voltage induced on the metal rods inserted into the fluid at three different points in the circuit (test points 1, 2 & 3 in Figure 5). The Novx 5000 remote probe was used for these measurements. A verification of the voltage using a Trek 541 was performed on several initial measurements. The data obtained with the Novx 5000 remote probe and the Trek 541 was nearly identical. The air actuated pump was run during these tests with a supplied air pressure of 30 to 40 psi. The system was first run with air only being pumped through the system. Figure 6 is a representative graph of the voltage measured at test point 1 while the air is pumped through the system.
 |
Deionized water was then measured. Figure 7 is a graph of the voltage measured in the fluid at test point 2 (see Figure 5) while the DI water was pumped in an ungrounded state. Figure 8 shows the voltage measured at test point 3 (see Figure 5) with the DI water grounded in the recirculation tank while being pumped. The spike seen at the beginning of the test was voltage induced during handling of the tubing during test setup. Note that the voltage quickly goes to near zero when the fluid is pumped in the grounded condition.
Methanol and IPA were tested next, followed by the solvent-based stripper, and then the aqueous ash remover.
 |
It was noted during the early tests that, particularly when the fluid was not grounded, a spike of voltage was seen upon initial contact of the probe to the test points. Through several measurements, it was found that this energy was the result of the tubing or pump being touched by the operator in-between tests (compare Figures 7, 8, 9 and 10). When the system was grounded and the fluid pumped briefly immediately prior to starting the measurements, no spike of energy was measured.
 |
To further investigate this phenomenon, the voltage was measured at test point 1 (closest to the output of the pump) and the pump housing was rubbed with a nitrile rubber glove. Figure 9 shows the voltage generated with the isopropanol ungrounded. Figure 10 is the same phenomenon with the IPA grounded. (Note the lower magnitude of voltage, and that the voltage in the ungrounded fluid does not bleed off). Figure 12 is a graph of the voltage generated in the ungrounded aqueous ash remover by rubbing the tubing prior to the start of the test. The solution is grounded part way through the test, and the voltage in the solution drops quickly to near zero volts.
 |
Figure 11 is a graph of the voltage generated at test point 1 (see Figure 5) in grounded isopropanol with the pump running initially, and then shut off part way through the test. Here it can be seen that the pumping does induce voltage into the IPA, and that grounding the solution helps some, but not completely. The higher resistivity of the IPA results in the generated voltage taking longer to discharge through to ground. As a result, some voltage is generated in the IPA (-30 to -50 volts), even when this solution is grounded (see Figure 13). Note that in Figure 14, the lower resistance fluid (solvent-based photoresist stripper), does not generate much voltage when running if it is grounded.
 |
To summarize the data obtained from this testing, See Tables 2 through 8. Tables 2 through 7 contain the data from the standard running tests. Table 8 is a summary of the "special" tests (rubbing pump, large spikes due to other conditions etc.) The maximum/ minimum columns are the absolute maximum and minimum measured during the test (including spikes). The "voltage while running" column is the typical voltage seen during the pumping operation and is taken from the graphs of each test.
 |
Conclusions
- While the pumping operation does generate static charge (Figure 11), the largest source of static charge in the fluid comes from the tubing/pump when it is touched by the operator (see Figures 9, 10, & 12).
- Grounding the static dissipative fluids assist in removing charge from the fluid. However, charge that is induced into the fluid when the fluid is stationary can build up to high levels (see Figures 8, 9, 10, & 12).
- The conductivity of the fluid effects the rate at which charge is bled off when the fluid is grounded. The more conductive fluids had lower voltage build-up when they were grounded (compare Figures 13 & 14).
 |
Carl Newberg is the president and owner of River's Edge Technical Service, an independent testing laboratory and consulting service to the ESD and contamination control industries. He has held positions as the ESD program manager for Western Digital Corp., and has been actively involved in the corporate ESD programs at Seagate Technology and IBM Corp. He is the co-chairman of the ESD subcommittee of IDEMA. He is also an active member of the ESD Association subcommittees participating in standards-writing and review.
References
- Yatsuzuke et.al. "Electrification of Polymer Surface Caused by Sliding Ultrapure Water," IEEE Transactions, Vol. 32, No. 4, July/August 1996.
- K. Davies & S. Gerkin, "Charging and Ignition of Sprayed Fluid," Proceedings of 1999 EOS/ESD Symposium, pp. 55 - 61.