Interaction of ClF3 with Metal Alloys and Polymer Gaskets
A.P. Taylor, B. Fruhberger, R. Hogle -- Semiconductor International, 7/1/1999
In recent years, the use of chlorine trifluoride (ClF3) as a chamber cleaning gas in semiconductor processing tools has become popular for a number of reasons. Plasma activation is not required as with more conventional cleaning gases such as tetrafluoromethane (CF4) or nitrogen trifluoride (NF3). Hence, thermally activated CVD single wafer and LPCVD batch wafer chambers are prime candidates for ClF3 cleaning.
For single wafer chambers, expensive plasma power supplies and matching networks are not needed for the chamber cleaning step. Examples include tungsten, tungsten silicide and titanium nitride CVD. Residues generated during deposition and those remaining in the chamber and vacuum lines are more aggressively attacked by the ClF3 during chamber clean compared with conventional plasma cleans. These applications generally require less than 0.5 slm ClF3 flow for cleaning. It is for these conditions that we designed the experiments.
For batch wafer systems such as LPCVD silicon nitride, in situ chamber cleans are made possible with ClF3. This leads to reduced downtime and possibly enhanced wafer yield due to improved cleaning efficiency in the chamber and also in the vacuum lines. The ClF3 flow rates are typically greater than 1 slm for these applications.
Another major advantage in using ClF3 for chamber cleans is that it is not a greenhouse gas like many other common etching gases. The use of PFC abatement equipment is not required.
 |
|
Fig.
1 XPS depth profile of a Ni 200 alloy sample exposed to
ClF3 at 200°C for 30 minutes. The calibrated sputter rate = 0.9
Å/sec. |
Semiconductor literature abounds with ClF3 materials compatibility studies. However, these studies have not been particularly designed to address concerns of the safe usage of ClF3 in the semiconductor industry. In this study, we looked at a number of materials used in ClF3 delivery systems, process chambers and vacuum lines. We exposed samples of these materials to dilute ClF3 in a controlled manner to determine their relative compatibilities. The two main classes of materials we investigated were: 1) metal alloys used in the construction of delivery systems and 2) polymers used as sealing materials in semiconductor processing chambers and vacuum lines.
|
Table 1 Metals and Metal Alloys Compatible with
ClF3 | |||
| Material | Oper. Temp.(°C) | Reference | Comments |
| Stainless Steels | - | 2 | |
| 302,321,347 | - | 2 | |
| 304,316 | RT-110 | 1 | |
| 304 | RT-120 | 3 | |
| 316 | RT-150 | 3 | |
| Carbon Steels | - | 2 | Limited Service |
| Carbon Steels | RT-150 | 3 | |
| Cr plated Steel | - | 1 | |
| Nickel | RT-500 | 1, 3 | |
| Nickel | - | 4 | Unattacked up to 400 °C |
| Ni Superalloy | - | 2 | |
| Copper | RT-100 | 1 | |
| Copper | RT-250 | 3 | |
| Aluminum | RT-400 | 1, 3 | |
| Al Alloy 356, | - | 2 | |
| 1100,2024,5052, | - | 2 | |
| 6061,6063,6066 | - | 2 | |
| Monel | RT-400 | 3 | |
| Lead | - | 3 | |
| Lead | - | 1 | Limited Service |
| Indium | - | 1 | |
Authors of the literature maintain that many metal alloys, including nickel 200 and stainless steels, are compatible with ClF3 due to formation of a passivating fluoride layer. Table 1 is a partial list of compatible and conditionally compatible metals and alloys taken from the literature. Table 2 is a list of metals and alloys that are incompatible with ClF3. It is our intent, with the metal alloy part of this study, to contribute to the understanding of this phenomenon.
We investigated details of the surface phenomena upon exposure of these materials to ClF3 with X-ray photoelectron spectroscopy (XPS) together with sputter depth profiling and scanning electron microscopy (SEM). We also investigated the interaction of ClF3 with the materials in the temperature range from room temperature up to 200°C. Our current study complements recent results published by Hattori et al.5 We found strong effects of temperature and alloy compositional variations on the observed surface phenomena.
Fluorinated polymers are conditionally compatible with ClF3. The experiments were designed to compare the extent of surface degradation to samples exposed to ClF3 in the exhaust line of a vacuum pump used by the process chamber.
Metal alloy tests
A single wafer research reactor was used to control exposure of metal alloy samples in the chamber and polymer samples in the exhaust manifold of the vacuum pump to dilute ClF3. The Al chamber was configured with water-cooled walls, a 4 in. diameter monel heater block with temperature controller, two mass flow controllers with a common line to the chamber and a throttle valve for chamber pressure control. A BOC Edward's QDP80 dry pump with full gas ballast (47 slm N2) was used for vacuum service.
We chose five metal alloys for ClF3 exposure and detailed XPS analysis. The alloys included nickel 200, Hastelloy C22, Elgiloy and 304 and 316L stainless steels (SS). Samples were typically 1 in. in diameter and the finish on the before-exposed surfaces was 20 µ-in. RMS roughness or less. Samples were degreased with Freon. We then outgassed the heater block at 400°C under vacuum for about one hour. Afterwards, we set the heater block at 300°C and passivated the chamber with ClF3 for 30 minutes. For chamber passivation and during sample runs, we controlled the pressure set at 5 Torr and set the ClF3 and Ar flow rates at 200 sccm and 800 sccm, respectively. Identical conditions were used in the polymer study.
|
Table 2 Metals and Metal Alloys Incompatible with
ClF3 | ||
| Material | Reference | Comments |
| Tungsten | 1-3 | Forms volatile Fluoride |
| Titanium | 2,3 | Forms volatile Fluoride |
| Molybdenum | 1-3 | Forms volatile Fluoride |
| Brass | 3 | Dezincifies Material |
We simultaneously exposed the metal alloy samples to ClF3 for 30-min. intervals at sample holder temperatures of 24°C, 115°C and 216°C. Following exposure, we moved samples from the holders in atmosphere, placed in airtight containers and stored in a desiccator. We used this method of sample storage for the polymers as well.
ClF3 interaction with polymers
In the second phase of this study, we focused on polymers used as sealing materials. We chose five types of poly mer samples for ClF3 exposure and XPS analysis. The polymers included Kalrez (K#204, Compound 4079), O-ring Viton (Dowty Ref. #5560 and 4460), BOC Edward's dry pump Teflon shaft seal, Teflon O-ring and BOC gases Kel-F valve seat material. The polymer samples were held in place in the pipe between the exhaust manifold and silencer of the vacuum pump. The location for exposure was chosen because of the relatively high gas and surface temperatures (~120°C) at the exhaust manifold of the pump. Also, the presence of air in the exhaust line was desirable since it is in this environment that we would expect the most degradation of polymers.
 |
|
Fig.
2 XPS depth profile of a 316L stainless steel sample exposed
to a ClF3 at 200°C for 30 minutes. The calibrated sputter rate
= 0.9 Å/sec. |
In the process chamber, the surface temperature of the wafer chuck was held at 200°C to simulate process chamber conditions during ClF3 cleaning. The polymer samples were exposed to the ClF3 containing environment simultaneously two at a time for 30 minute intervals.
We performed surface analysis using a Surface Science Instruments SSX-100 small spot XPS system equipped with a monochromatic Al KaX-ray source, a low energy electron flood gun for charge compensation and a Leybold IQE 12/38 ion sputtering gun. Ar+ ions at an acceleration voltage of 3 keV were used for depth profiling of the metal alloy samples. We calibrated the sputter removal rate against a SiO2/Si standard. A Hitachi S520 SEM took micrographs of the metal alloy samples.
Results and discussion
This section outlines the salient features of the metal alloy and polymer experiments. The ClF3 -induced corrosion on five materials (Hastelloy C-22, Nickel-200, Elgiloy, 304 SS, 316L SS) is compared using XPS sputter depth profiling and SEM. The materials had been exposed to ClF3 at room temperature, ~100°C and ~200°C, as described in the previous section. At room temperature, the amount of metal halogenide formation on all the alloy surfaces was comparable after exposure. At elevated temperatures, we saw dramatic differences in the thickness of the passivation layer from one material to the other.
|
Table 3 Thickness of Corrosion Layers After ClF3 Exposure
at the Indicated Temperatures | |||||
| Hastelloy C-22 |
Nickel-200 | Elgiloy | 316LSS | 304SS | |
| No exposure | 0 Å | 0 Å | 0 Å | 0 Å | 0 Å |
| Room temp. | 50 Å | < 50 Å | 15 Å | 30 Å | 25 Å |
| 100°C | 60 Å | < 50 Å | 140 Å | 100 Å | 600 Å |
| 200°C | 80 Å | < 50 Å | 900 Å | 900 Å | > 2800 Å |
Figure 1 is a representative XPS depth profile obtained from the Ni-200 sample after ClF3 exposure at ~200°C. Ni-200 alloy contains nominally 99.2 weight % of Ni, with Fe, Cu, Si, C and Ti (in decreasing order of concentration) present at <0.4 weight % each. The concentrations of Ni (Ni2p3/2), F (F1s), Cl (Cl2p), O (O1s) and C (C1s) were determined as a function of sputter time. Elemental concentrations in the depth profiles are expressed in calculated atom percent. The element specific signal intensities, corrected by the relevant sensitivity factors, are expressed relative to the sensitivity factor corrected sum of all signal intensities, set to 100 atom percent. The interface between the corroded layer and the underlying sample was not well defined, as seen by the slow decrease in halogen concentration as a function of sputter time (i.e., depth). This makes the quantitative determination of the thickness of the corroded layers somewhat arbitrary.
The Ni-200 samples were found to be the least reactive to ClF3 of all materials investigated. A slight increase in the fluorinated layer thickness was observed with increasing temperature of exposure. The layer's thickness remained below ~50Å, even after exposure to ClF3 at ~200°C. Small concentrations of chloride were also detected. The binding energy of the Ni2p3/2 XPS signal in the fluorinated layer was consistent with the formation of a NiF3 layer.
Figure 2 is an XPS depth profile obtained from the 316L SS sample after ClF3 exposure at ~200°C. The nominal bulk composition of 316L SS is as follows (weight %): 16.0-18.0 Cr, 10.0-14.0 Ni, 2.0-3.0 Mo, 2.0 Mn, 1.0 Si, <0.05 P, C, S with the balance of Fe. The concentrations of Fe (Fe2p3/2), Cr (Cr2p3/2), Ni (Ni2p3/2), Mo (Mo3d), F (F1s), Cl (Cl2p) and O (O1s) were determined as a function of sputter time. After exposure to ClF3 at ~200°C, a fluorinated layer with thickness ~900 Å had formed. The binding energies for all metals in the layer on 316L SS were consistent with the formation of the metal fluorides. The depth profiles suggest that the Fe and possibly the Mo concentrations in the fluorinated layer are reduced relative to those of the other metals in the alloy. Fe and Mo are known to form volatile halogenides (FeCl3, MoF6). Loss of iron chloride could explain the apparent relative decrease of Fe within the layer. It could also have led to the increase in layer thickness with increasing temperature, by affecting the structural integrity of the alloy.
The chlorides formed on 316L SS clearly exhibited a temperature dependence. While the concentration of detected fluorine went up with temperature of exposure, the concentration of chlorine went down. This behavior was evident for all the metal alloys.
Table 3 contains the estimated corrosion layer thicknesses for all the samples. The Ni-200 and Hastelloy C22 alloys were least affected and the 304 SS samples were most affected by exposure to ClF3 at elevated temperatures.
The polymer samples were exposed to dilute ClF3 in the exhaust of a QDP80 dry pump as described in the previous section. The calculated elemental surface composition and high resolution spectra of the C 1s and F 1s peaks were determined using XPS for each sample. The F/C ratio was calculated for the exposed and unexposed samples for comparison purposes.
The XPS spectra of the Kalrez O-ring showed very little alteration of the polymer surface as a result of the ClF3 exposure. The C 1s spectra showed the characteristic lineshape expected for Kalrez with three resolved spectral features, attribu-table in order of decreasing binding energy to carbon in -OCF3, -CF2 and -CH2 functional groups6.
|
Table 4 Results for the Polymer
Studies | ||||||||||
| Material | C 1s | O 1s | F 1s | Cl 2p | Si 2p | S 2p | Mo 3p3/2 | Mg 2s | Ca 2p | F/C |
| Kalrez | 30.89 | 4.74 | 64.37 | - | - | - | - | - | - | 2.08 |
| Kalrez | 32.67 | 4.57 | 62.76 | - | - | - | - | - | - | 1.92 |
| Kel-F | 34.71 | 10.51 | 49.38 | 2.36 | 3.04 | - | - | - | - | 1.42 |
| Kel-F | 28.58 | 9.98 | 60.04 | 1.40 | - | - | - | - | - | 2.10 |
| Teflon1 | 31.53 | 1.01 | 66.55 | - | - | 0.58 | 0.33 | - | - | 2.11 |
| Teflon1 | 29.51 | 8.43 | 62.06 | - | - | - | - | - | - | 2.10 |
| Teflon2 | 31.67 | - | 68.33 | - | - | - | - | - | - | 2.16 |
| Teflon2 | 29.70 | 7.19 | 63.11 | - | - | - | - | - | - | 2.12 |
| Viton | 41.26 | 6.31 | 47.27 | - | 2.38 | - | - | 1.22 | 1.55 | 1.14 |
| Viton | 29.94 | 8.37 | 60.59 | 1.10 | - | - | - | - | - | 2.02 |
For all other polymers, differences in the surface chemistry were found after exposure to ClF3. Table 4 lists the results of the polymer experiments. The Kel-F sample showed an increase in the F/C atomic ratio after exposure. The most prominent differences in the C 1s spectra were found for the feature at lower binding energy, which is located in the spectral region consistent with -CH2 carbon or filler carbon. This spectral feature was found to be strongly reduced for the exposed sample.
The Teflon shaft seal sample exhibited low intensity spectral contributions from Mo, S, and O from a MoS2 filler and light surface oxidation. The exposed sample showed an increased level of surface oxygen and absence of signal contributions from Mo and S. The F/C atomic ratio did not change significantly upon exposure. The filler, however, appeared to be removed from the surface after exposure. In addition, the C 1s from the exposed sample showed a second feature on the high binding energy side of the single -CF2 carbon feature. This suggests ClF3-induced surface modification of the polymer. Oxygen appeared to either participate in the reaction or the material surface became more prone to oxidation, which may have taken place upon exposure to air.
The Teflon O-ring sample showed similar surface modification as the Teflon shaft seal. The O-ring did not contain any filler materials. The surface composition of the exposed sample showed spectral contribution from oxygen, and the C 1s was modified after exposure in accord with the shaft seal sample.
 |
|
Fig.
3 XPS spectra of an unexposed Viton O-ring
sample. |
The Viton O-ring sample exhibited the most dramatic differences after exposure. Figures 3 and 4 are the C 1s and F 1s spectra for the Viton O-ring sample before and after exposure. The unexposed C 1s spectrum showed the characteristic lineshape for Viton with three resolved spectral features. These features are attributable in order of decreasing binding energy to C in -CF3, -CF2 and -CH2 functional groups of the polymer.6 The filler materials in the Viton were magnesium silicate and calcium silicate. The XPS spectral contributions from Mg, Ca, Si and O were detected before exposure but not afterwards. As with the other polymers with fillers, surface removal is evident. The exposed sample showed an increase in the F/C ratio. The C 1s spectrum showed a decrease in signal from the -CH2 functional group and an increase in signal from the -CF3 group after ClF3 exposure. This suggests preferential fluorination of the -CH2 functional groups of the polymer.
It is difficult to apply the polymer sample results to real world applications since the extent of surface degradation to the materials is unknown. Most O-rings in vacuum systems are only partially exposed to the gases in the system and only the exposed surfaces would degrade. Clearly, Kalrez was least altered by the ClF3 and would be the best candidate for service. We currently offer and recommend Kalrez O-ring kits for the exhaust portion of our dry pumps used for ClF3 service. We recommend the standard Viton O-rings supplied in the exhaust portion of our pumps be replaced after six months for ClF3 service.
 |
|
Fig.
4 XPS spectra of an unexposed Viton O-ring
sample. |
Summary and conclusions
We exposed samples of five different metal alloys and five different polymers to dilute ClF3 in a controlled manner to determine their relative compatibilities. We used X-ray photoelectron spectroscopy (XPS) and scanning electron microscopy (SEM) to analyze the exposed and unexposed samples.
We investigated the interaction of ClF3 with the metal alloys in the temperature range from room temperature up to 200°C. We used a single wafer research reactor to control ClF3 exposure of metal alloy samples in the chamber and polymer samples in the exhaust manifold of the vacuum pump.
We compared the ClF3-induced corrosion on five materials (Hastelloy C-22, Nickel-200, Elgiloy, 304 SS, 316L Stainless Steel) using XPS sputter depth profiling and SEM. At room temperature, the amount of metal halogenide formation on all the alloy surfaces was comparable after exposure. At elevated temperatures, we saw dramatic differences in the thickness of the passivation layer from one material to the other.
The Ni-200 and Hastelloy C22 alloys were least affected and the 304 SS most affected by exposure to ClF3 at elevated temperatures. For the Ni-200, the thickness of the halogenide layer remained below ~50 Å, even after exposure to ClF3 at ~200°C for 30 min. For the 316L sample after exposure under the same conditions, a layer with thickness of ~900 Å had formed.
Fluorinated polymers are conditionally compatible with ClF3. We
studied Kalrez (K#204, Compound 4079), O-ring Viton (Dowty Ref. #5560 and 4460),
BOC Edward's dry pump Teflon shaft seal, Teflon O-ring and BOC gases Kel-F valve
seat material. We designed the experiments to compare the extent of surface
degradation to samples exposed to ClF3 in the exhaust line of a
vacuum pump used by the process chamber. The XPS spectra of the Kalrez O-ring
showed very little alteration of the polymer surface as a result of the
exposure. For all other polymers, we found differences in the surface chemistry
after exposure. The Viton O-ring sample exhibited the most dramatic differences
after exposure. 
References
1. ClF3 Technical Data, Iwatani International Corp., April 12, 1994.
2. ClF3 Handling Manual, Rocketdyne Corp. (TRW), Air Force Cont. No. AF33 (6116)-6939, Proj. No. 3148, Task No. 30196, Sept. 1961.
3. Chlorine Trifluoride Brochure, Rev. 6, C. Gugliemini, Air Products Corp., Feb. 18, 1997.
4. Matheson Gas Handbook, Matheson Corp., 164.
5. T. Hattori, A.N. Liyanage, H. Sakuma and E. Ozawa, ISSM Proc. 1996, Tokyo, 329.
6. I.V. Bletsos et al., Polymer Reprints 1991, 32(2), 256-257.
|
Anthony Taylor has been an applications engineer for BOC Edwards specializing in dry pumping in the semiconductor industry for the past five years. He received a Ph.D. in experimental condensed matter physics of surfaces from Rensselaer Polytechnic Institute in 1993. |
|
Richard Hogle is development manager at the San Marcos, Calif., New Products Center for BOC Edwards, where his work has concentrated on purification and packaging of HBr, WF6, HF, Cl2, BCl3 and ClF3 |
|
Bernd Fruhberger, Ph.D. is a senior staff scientist at Sensor Research and Technology Corporation (SRD), Orono, Maine, where his responsibilities include the development of semiconducting metal oxide based gas sensor elements. He received a Ph.D. from the University of Heidelberg, Germany, in 1994. |
