Surface-finishing techniques for stainless steel passivation
05/01/2006
Sunniva R. Collins, Swagelok Co., Solon, Ohio
There is considerable value in identifying optimum surface-finishing techniques for 316L stainless steel used in ultrahigh purity gas applications in the semiconductor industry. Purity and corrosion resistance are extremely important, with new aggressive chemistries being introduced and device feature sizes becoming ever smaller. Nevertheless, comparative data on the many types of surface-finishing techniques have not been readily available to the industry.
Analysis technique
A research study examining how surface-finishing techniques affect corrosion resistance has recently been completed. Variables in the evaluations included 316L composition; material melt method (argon-oxygen decarburation [AOD], AOD/vacuum arc remelting [AOD/VAR], or vacuum induction melting/vacuum arc remelting [VIM/VAR]); and surface treatment (mechanically finished, electro-polished, mechanically finished and passivated, or electro-polished and passivated). For each combination, the passive oxide layer was quantified using glow-discharge optical emission spectroscopy (GD-OES).
GD-OES combines sputtering and atomic emission to provide an extremely rapid and sensitive technique for element depth profiling. The GD-OES consists of a glow discharge source and one or more optical spectrometers. During analysis, a plasma is generated in the analysis chamber by the applied voltage between the anode and the cathode (the sample surface) in the presence of argon under low pressure. Ionized argon atoms cause sputtering of the sample area. Sputtered atoms excited in the plasma rapidly de-excite by emitting photons with characteristic wavelengths. After calibration, the technique can provide a quantitative depth profile (QDP) of materials.
The parameters measured from a typical QDP include oxide thickness in angstroms (Å), the chromium/iron (Cr/Fe) ratio at 10Å, maximum Cr/Fe ratio, carbon (C) layer thickness in Å, and maximum phosphorus (P) in atomic percent (at.%). To evaluate the performance of the passive layer, critical pitting temperature tests (CPT) were performed per ASTM G150, “Standard Test Method for Electrochemical Critical Pitting Temperature Testing of Stainless Steels.”
As a result of these experiments, a considerable body of data was generated. In all, more than 70 different combinations of conditions were examined, with a minimum of six replicates of each condition. When examined as a whole, the data represent characteristics of the oxide layer that forms on 316L stainless steel, regardless of composition variations, melt variations, or differences in surface treatments.
Surface-analysis figures of merit
SEMI F19-0304, “Specification for the Surface Condition of the Wetted Surfaces of Stainless Steel Components,” defines the acceptable ranges of values for surface-analysis figures of merit for wetted surfaces intended for use in semiconductor applications. In these experiments, GD-OES was used to measure all relevant parameters.
Oxide thickness (Å). The thickness of the oxide layer was determined from the full width at half maximum (FWHM) of the oxygen peak. For high purity and ultrahigh purity surfaces, SEMI F19-0304 requires a minimum oxide thickness of 15Å, and also indicates that an oxide thickness >60Å may describe a mechanically polished surface with high carbon contamination.
Cr/Fe ratio at 10Å (Cr/Fe10Å). The ratio of Cr to Fe at 10Å was a measure of both chromium enrichment in the oxide, and the cleanliness of the surface (as seen by a minimization of contaminants such as C or P).
Maximum Cr/Fe ratio (Cr/FeMax). This measure was determined by finding the location in the oxide with the largest difference between Cr and Fe, and taking the ratio of the two values. It also measures chromium enrichment in the oxide and in general is at least equal to (and more often greater than) the Cr/ Fe10Å.
Carbon layer thickness (Å). The thickness of the carbon layer was determined by the FWHM of the carbon peak. SEMI F19-0304 states (paragraph 6.6.1): “adsorbed carbon contamination shall be <30at.%, declining to base levels within 15Å of the initial surface.” This parameter measures surface cleanliness, with lower values indicating a cleaner surface.
Maximum phosphorus (at.%). The height of the P peak measures the relative cleanliness of the surface after mechanical finishing or electropolishing. This measure is not typically reported, but was tracked as part of these evaluations to determine the effect of phosphorus contamination on the resultant performance characteristics of the passive layer. Phosphorus can be introduced to the surface during mechanical polishing and during electropolishing, which is typically performed with a sulfuric/phosphoric solution.
Results
Perhaps the most interesting finding was that phosphorus contamination in the oxide has a very strong depressing effect on CPT, as shown in Fig. 1. This result can be useful in setting limits on maximum P in the oxide. If P is >2%, CPT values reach a ceiling of 25°C. Therefore, passivation methods that remove phosphorus, such as those based on nitric acid and sodium dichromate, are preferable to methods such as citric acid-based passivation that do not.
Dichromate passivation is associated with higher levels of carbon contamination. Though carbon contamination on the surface does not degrade corrosion resistance, SEMI F19-0304 does not permit high levels of carbon contamination. If carbon can be shown to be an acceptable contaminant, mechanically polished surfaces with dichromate passivation may be used to provide a surface with enhanced corrosion resistance.
Chromium enrichment in the oxide can be measured, yielding both Cr/Fe10Å and Cr/ FeMax parameters; in general, Cr/ FeMax is ≥Cr/Fe10Å. Cr/ FeMax, however, can be very high in oxide layers that are heavily contaminated with P and C, as indicated in Fig. 2. For this reason, the measure of Cr/FeMax can be confounded by the effects of contamination. As a measure of Cr enrichment and surface cleanliness, Cr/Fe10Å is therefore more accurate in differentiating among surface treatments. Figure 3 shows the effect of P contamination on Cr/Fe10Å, and indicates that Cr/Fe10Å is a better metric for evaluating surface cleanliness than Cr/ FeMax (Fig. 2).
In general, a thinner oxide is less contaminated with C and P, and is more enriched in Cr. At oxide thicknesses <70Å, CPT increases with decreasing oxide thickness, indicating that a thinner, more compact oxide is preferable for enhanced corrosion resistance. Most oxide thicknesses >70Å are derived from surface-finishing techniques involving mechanical polishing, thermally grown oxides, or unacceptable levels of surface contamination.
Nitric acid-based passivation minimizes contamination, both maximum P in the oxide layer and the carbon layer thickness. It also maximizes Cr/Fe10Å. Since citric passivation is associated with higher levels of phosphorus contamination, CPT results for this method are lower.
Dichromate passivation returns the highest average CPT values, but with increases in the carbon layer thickness as well. Using dichromate passivation, it is possible to grow relatively thick oxides (>60Å), and since carbon contamination does not negatively affect corrosion resistance, oxide thicknesses >70Å show an increasing trend in CPT values.
Sunniva R. Collins, Ph.D., is a manager, standards and product regulatory compliance, at Swagelok Co., 31400 Aurora Rd., Solon, Ohio; ph 440/349.5934, ext. 4128; [email protected].