ANALYTICAL TECHNOLOGIES

Measuring trace ionic impurities in ultrapure acids and bases with ion chromatography

ManLi Wu and Jian-Ge Chen, Olin Research Center

Impurities in reagents used in semiconductor device production can adsorb on the surfaces of devices and thus affect their reliability.¹ The continuing increase in the sophistication and scale of integrated circuits has driven the need for monitoring trace levels of ionic contaminants. New parts-per-billion- or sub-parts-per-billion-level specifications for contaminants in concentrated acids have exceeded the detection limits of existing analytical technology.

Traditionally, determining trace ionic impurities in concentrated acids or bases has required labor-intensive, manual preconcentration or pretreatment methods prior to the taking of analytical measurements.² Most wet-chemistry analytical methods, including turbidity assays and colorimetric methods, are semiquantitative and not sufficiently sensitive. Since the invention of ion chromatography in 1975,³ its application to samples having highly concentrated ionic matrices had been limited until the introduction of moderate-capacity ion-exchange columns.⁴ These resins, packed in a smaller- diameter column (2 mm versus the conventional 4 mm), offer high background suppression capability when using a stronger eluent required for separating trace ionic contaminants from the sample matrix. As a result, these columns permit direct injection of highly concentrated acids without compromising trace ion detection limits that result from severalfold dilution of the concentrated sample. The application of such a system to the analysis of parts-per-billion levels of anionic impurities in 49% hydrofluoric acid (HF), 30% ammonium hydroxide (NH_⁴OH), and 25% trimethylammonium hydroxide (TMAOH) and ammonium (NH^₄+) in 98% sulfuric acid (H_²SO_⁴), and 37% hydrochloric acid (HCl) has been demonstrated.^5,6 The study on trace analysis of anionic impurities has been extended to cover a wider range of chemical reagents than previously analyzed, including concentrated H_²SO_⁴, HCl, phosphoric acid (H_³PO_⁴), and nitric acid (HNO_³). This article will summarize the matrix elimination approaches and problems encountered when attempting to achieve parts-per-billion-level quantitation of ionic impurities in samples containing interfering ions at high percent levels.

Ion Chromatography

Ion chromatography is a fast, efficient method for separating anions and cations. The separation is accomplished by two techniques: ion-exchange chromatography (IC) and ion-exclusion chromatography (IEC). IC uses a separator column that contains resins functionalized with a fixed charge and an associated counterion. After the sample introduction, a mixture of sample ions can travel through the column at different speeds as a result of their different affinities toward the charged site when the counterion is exchanged for the sample ion. IEC uses a column functionalized with a negatively charged membrane. This membrane, called a Donna membrane, is only permeable for neutral compounds. Totally dissociated anions cannot penetrate it because of their negative charge and therefore are excluded from the column. The difference in this permeability into the membrane results in varying travel times through the column and allows for the separation of weak acids from the inorganic anions. In this study, two-dimensional chromatography was applied by coupling different separation techniques, such as IC-IC or IEC-IC, to achieve the ion separation.

The basic components of a modern ion chromatograph are shown in Figure 1. In such a system, the analytical column provides the necessary separation of ions, while the suppressor can chemically reduce the background conductivity of the eluent. As a result of background suppression, the detection sensitivity is greatly improved, and the system can be used for analyzing parts-per-billion levels of anions and cations in water in conjunction with other on-line concentration techniques.

Figure 1: Diagram of a modern ion chromatograph.

Matrix Elimination

When analyzing trace ionic impurities in samples that contain large amounts of interfering anionic or cationic matrices, the analyte peaks can be obscured and the separation changed. In order to determine the level of trace anions or cations in concentrated acids or bases, the sample must be diluted or the sample matrix eliminated prior to analysis. Dilution reduces the sample amount introduced into the IC system and can result in less-sensitive detection of the ionic impurities.

Matrix elimination can be achieved by heat digestion of the acid sample to volatilize the matrix, followed by IC analysis. This method has the advantage of concentrating the sample, therefore improving the detection sensitivity. Nevertheless, heat digestion is more susceptible to contamination during sample treatment. An alternative method uses on-line matrix elimination while employing IC for the analysis. Several approaches—including autoneutralization, heart-cutting, and precipitation—have been explored and applied successfully to various acids and bases.

Autoneutralization

The difficulty of using IC to analyze anions in a concentrated base that contains large amounts of hydroxide ions is that these ions can override the eluent and decrease the separation efficiency of the anionic impurities. The AutoNeutralization pretreatment unit (ANPU; Dionex, Sunnyvale, CA) permits a sensitive, automated analysis of trace anions in concentrated bases and cations in concentrated acids.^7,8 The unit electrolyzes water to create the acid or base required for neutralization of the hydroxide or hydronium matrix, respectively. This can provide not only high-capacity neutralization but contamination-free matrix elimination as well.

The technique was applied to trace determination of Cl^—, NO^₃—, SO^₄2—, and PO^₄3— in 30% NH_⁴OH and 25% TMAOH.⁶ To achieve the detection limit at low parts-per-billion levels, column selection, separation conditions, system precleaning, sample-loop selection, fractionation volume, and quantitation technique were investigated and optimized. Both bases were analyzed directly without dilution. The recoveries of 10-ppb spikes of four ions into the samples were 103.8% ± 4.5% for Cl^—, 101.1% ± 5.6% for NO^₃— , 99.4% ± 4.6% for SO^₄2—, and 106.4% ± 6.1% for PO^₄3—. The detection limits were 5 ppb for Cl^— and SO^₄2— and 1 ppb for NO^₃— and PO^₄3—.

Using ANPU for cationic impurities such as NH^₄+ at < 50 ppb in 98% H^₂SO^₄ and 37% HCl was not feasible because of the low recovery in NH^₄+, which was a result of diffusion of the ion across the electrode membrane during neutralization. The unit is also not applicable to HCl because chloride can be oxidized on the electrode surface, leading to the destruction of the neutralizer. A novel approach was developed for achieving the analysis.⁶ A diagram of the analytical system is shown in Figure 2. In this system, the sample pretreatment column contains two neutralization columns placed in a series to provide neutralization of the acid matrix. Optimum separation conditions, injection volume, sample concentration, and quantitation technique were investigated. Both acids were diluted to half-strength and injected into the system for analysis. The recovery of 40-ppb NH^₄+ spikes in both acids was 102.0% ± 10.6% and the detection limit was 5 ppb.

Figure 2: Diagram of the instrumental setup for analysis of NH^₄+ in H^₂SO^₄ and HCl, and anions in HF, BOE, H^₃PO^₄, and HNO^₃.

Heart-Cutting

Hear-cutting is a technique which allows automatic reintroduction of a selected effluent fraction from the first IC into the second IC to achieve the desired separation efficiency. The high-concentration matrix anion is the major interfering ion for the IC analysis of trace anionic impurities in concentrated acids. Direct analysis of concentrated acid without sample pretreatment is difficult because peaks of analyte anions can be obscured by the broad peak of the matrix anion because of overloading. To reduce the overloading, concentrated acids are often greatly diluted to subpercent levels prior to their introduction to the IC for analysis.^9—12 The dilution largely reduces the detection limit from parts-per-billion to parts-per-million levels.

The purpose of the work was to develop procedures that can allow direct injection of concentrated acids in order to achieve parts-per-billion-level analysis of anionic impurities. One common technique involves two-dimensional chromatography, which uses two ICs placed in series so that acid can be preseparated in the first IC in order to partially isolate the anion matrix from anions of interest. A fraction of the effluent from the first IC is then selectively captured, concentrated, and introduced into the second IC for analytical separation and quantitation. Based on the chemical characteristics of the acids (HF, BOE, H^₃PO^₄, HNO^₃), two types of two-dimensional chromatography systems—IC-IC and IEC-IC—have been developed.

IC-IC. The IC-IC system has been used for analyzing Cl^—, SO^₄2—, and PO^₄3— in 70% HNO_³. Nitric acid is strong enough that NO^₃— is completely dissociated from the acid in water. In order to analyze anions at low-parts-per-billion levels, the acid sample cannot be diluted to less than percent level. However, with percent levels of HNO_³ injected onto an IC system, no existing IC column can be used for complete separation of the analyte anions from the NO_³ peak. It is, however, possible to choose a high-capacity column for the first IC to preseparate the majority of the matrix from anions. Some amounts of the matrix along with the majority of the anions can then be heart-cut into a second IC that contains a high-efficiency column for analysis. The setup of the two-dimensional IC system is shown in Figure 2. The sample pretreatment column contains an analytical separation column to provide the first-dimension separation of the matrix from the anions. Since the first IC dictates the feasibility of the analysis, parameters including column type, injection concentration, injection volume, separation conditions, and fractionation interval were carefully optimized. Figure 3 depicts the separation of Cl^— and SO^₄2— from NO^₃— matrix in 5.6% HNO^₃ using the optimized parameters.

Figure 3: Separation of Cl^— and SO^₄2— fron NO^₃— matrix in 5.6 % HNO_³ under optimized conditions.

The analysis of trace phosphate in HNO^₃, however, is not possible. This is because the PO^₄3— ion, under the strongly acidic HNO^₃ matrix, forms an inorganic weak acid, H^₃PO^₄, with three pKa's (acid dissociation constants) of 7.52 x 10^—3, 6.23 x 10^—8, and 2.2 x 10^—13, respectively. This means the neutral H^₃PO^₄ is not retained on the column and will elute at the front end of the matrix peak. Other anions are retained on the column and elute at the fractionation where the anion matrix is significantly less than the front end of the matrix peak. The peak elution profile of HNO^₃ and the anions of interest from the first IC are illustrated in Figure 4.

Figure 4: Elution profile of NO^₃— matrix peak and the anions obtained from the first IC system.

After the analysis of five samples, it was found that the method gave high concentrations (hundreds of parts per billion) of Cl^— and SO^₄2—. The analysis of standard spiked samples also showed inconsistent recoveries. These inaccurate and scattered results as well as the discoloration of vials after contact with nitric acid suggested that the acid had oxidized the sample vials or the injector component and caused the leaching of contaminants from these materials. The use of a stainless steel—based injector in conjunction with glass sample vials should reduce contamination and provide more reliable results.

IEC-IC. The IEC-IC combination has been used to analyze anions at parts-per-million levels in fermentation broths that contain >=10% organic acids.¹³ In the study reported here, the IEC-IC system analyzed Cl^—, SO^₄2—, NO^₃—, and PO^₄3— in 49% HF, concentrated buffered etch oxide (BOE, a mixture of NH_⁴F and HF), and 85% H_³PO_⁴. These acids are classified as weak inorganic acids and are neutral at the pH of the matrix. IEC provides preseparation of the neutral inorganic acid from anions in the sample. The anions that elute before the inorganic acid peak are heart-cut and analyzed by the second IC system. To use the IC system in Figure 2 for IEC-IC, the sample pretreatment column is an ion-exclusion column.

In the HF analysis, sample concentration, fractionation interval, and standard calibration technique were optimized to provide detection limits of 2—10 ppb for the anions in 49% HF. The technique was applied to a BOE that contained a mixture of HF and NH^₄F. Unlike HF, which has a pH of <1, BOE has a pH that ranges from 4 to 6, depending on the relative concentrations of the two components in the mixture. The higher pH matrix gives rise to higher concentrations of F^— ion, causing much poorer resolution between it and other anions in the IEC system. As a result, the matrix F^— becomes a limiting factor when selecting a fraction from the effluent of the IEC system. The optimized fractionation interval, although containing minimum F^—, also contains fewer anions. This finding is shown in the chromatograms in Figure 5, in which Cl^—, NO^₃—, and PO^₄3— were not detectable. These anions' detection limits were at the 100-ppb level, with sulfate at about 10 ppb. The detection limits may be improved if another ion-exclusion column can be added to better resolve the anions and F^— matrix in the IEC system.

Figure 5: Overlaid chromatograms of the BOE analysis. The bottom chromatogram is BOE without any anion spike and the other two chromatograms are with 50- and 100-ppb anion mixure spikes, respectively.

The behavior of the H^₃PO^₄ matrix was very similar to that of the HF matrix, except that H^₃PO^₄ had three pKa's. Depending on the pH in the mobile phase of the IC system, phosphate can be found in any of the four forms: neutral, monobasic, dibasic, and tribasic. The pH can therefore affect the dominant H^₃PO^₄ form. At higher pH or mobile-phase concentration, the tribasic form is favored and therefore is retained on the column for a longer time. The anions, on the other hand, elute faster at high pH because of the higher mobile-phase strength. Because of this trading effect, eluent strength plays an important role on the ability to separate anions from phosphate. Figure 6 compares chromatograms obtained by separating the same fractionation under different pHs or mobile-phase concentrations in the IC system. The lowest eluent strength gave the optimum separation. Similar parameters of the HF analysis were investigated, and 85% H^₃PO^₄ was injected directly onto the system for analysis. The spike recoveries at the 10-ppb level for Cl^—, 40-ppb level for NO^₃—, and 200-ppb level for SO^₄2— were 112% ± 6%, 86.2% ± 1.4%, and 119% ± 2.7%, respectively. The detection limits of all three anions in 85% H^₃PO^₄ were about 10 ppb.

Figure 6: Chromatograms of anions in H^₃PO^₄ obtained under three levels of mobile-phase strength.

Precipitation

An anion matrix in concentrated acid can also be removed by selective precipitation of the main anion with a cation fixed onto an ion-exchange resin. The method has been applied to the determination of anions in H_³PO^₄ and H_²SO_⁴.^14,15 In this study, the approach has been used to analyze parts-per-billion levels of anions in 37% HCl and 98% H_²SO_⁴.

Figure 7 shows a diagram of the precipitation/IC system. The acid sample containing percent-level HCl is introduced into the precipitation column where the Cl^— matrix is retained. Following the precipitation column is a cation retention column that catches the residual cations bleeding from the precipitation column. Since the anionic impurities are diluted after the precipitation, the effluent is preconcentrated on a concentrator column before entering the separation column. While the anionic impurities are being analyzed by the IC system, the precipitation column is regenerated for the next injection so that the analysis can be fully automated. With optimized conditions, 19% HCl was injected onto the system; the chromatogram of the analysis is shown in Figure 8. The spike recoveries of 20-ppb anions in a 37% HCl sample were found to be 103% ± 2.0% for SO^₄2—, 95.5% ± 2.4% for PO^₄3—, and 104% ± 1.9% for NO^₃—. The detection limits of all three anions were 10 ppb.

Figure 7: Diagram of the instrumental setup for the analysis of anions in HCl and H^₂SO^₄.

Figure 8: Chromatogram of the anionic analysis in 37% HCl.

The application of the precipitation/IC system to the analysis of trace anions in H^₂SO^₄ was an even more difficult challenge. Most of the precipitation resins tested came only with smaller pore sizes. When precipitation occurred, the pores were easily plugged by the precipitate, preventing the precipitation sites from being accessible to additional SO^₄2— ions. When macroporous resins were used, the capacity increased sufficiently for effective removal of the SO^₄2— matrix. The regeneration of the column was also a challenge for automating the H^₂SO^₄ analysis using an on-line precipitation column. A novel approach has been developed that regenerates the precipitation column for continuous use without interrupting the analysis. In this approach, a chemical is pumped through the precipitation column to solubilize the precipitate. After rinsing the column with water, a solution containing the precipitating cations is used to regenerate the column. While the precipitation column is being regenerated, the second IC performs the analytical separation of the previous fraction taken from the first IC. As Figure 9 shows, this allows for an automated analysis of concentrated H^₂SO^₄.

Figure 9: Chromatogram of the anionic analysis in 98% H^₂SO^₄.

There was also a problem associated with the contamination caused by anions in the resins, which required thorough cleaning and conditioning prior to the analysis. In addition, a concentrator column was needed to preconcentrate the effluent from the precipitation column since the analyte anions were greatly diluted. The use of a concentrator column, however, has resulted in low recoveries of the anionic impurities, which is caused by the overloading of the concentrator column with the SO^₄2— matrix. Further work will investigate the preparation of precipitation columns with higher load resins and the selection of an effective concentrator column to achieve accurate analysis of the trace anions in concentrated H^₂SO^₄.

Conclusion

The work reported here has demonstrated that the analysis of trace ionic impurities can be achieved for most semiconductor-grade reagents using ion chromatography. The use of on-line matrix elimination and preconcentration allows for fast, reliable analysis of parts-per-billion levels of analyte ions in concentrated acids and bases. Further studies in selecting effective sample pretreatment columns or adequate instrumental components should eventually solve the remaining problems being encountered for BOE, HNO₃, and H₂SO₄.

References

1. Shafer H, and Budde KJ, "Study of Contamination Mechanisms at Silicon Surfaces during Wet Chemical Processes," Siemens AG Research Laboratories.

2. Book of SEMI Standards (BOSS)1995—Process Chemical Volume, Mountain View, CA, Semiconductor Equipment and Materials International, 1995.

3. Small H, Stevens TS, and Bauman WC, "Novel Ion Exchange Chromatographic Method Using Conductimetric Detection," Analytical Chemistry, 47(11):1801, 1975.

4. Siriraks A, Pohl CA, and Toofan M, "Determination of Trace Anions in Concentrated Acids by Means of a Moderate-Capacity Anion-Exchange Column," Journal of Chromatography, 602:89—95, 1992.

5. Chen JG, and Wu ML, "Using Two-Dimensional Ion Chromatography to Measure Contaminants in Ultrapure Chemicals," MICRO, 15(1):31—37, 1997.

6. Chen JG, and Wu ML, "Measuring Anions and Ammonium in Bases and Acids with Ion Chromatography," MICRO, 15(6):51—62, 1997.

7. Siriraks A, and Stillian J, "Determination of Anions and Cations in Concentrated Bases and Acids by Ion Chromatography-Electrolytic Sample Pretreatment," Journal of Chromatography, 640:151—160, 1993.

8. "Determination of Trace Anions in Concentrated Bases Using Autoneutralization Pretreatment/Ion Chromatography," Application Note 93, Sunnyvale, CA, Dionex, August 1994.

9. Murayama M, Suzuki M, and Takitani S, "Determination of Trace Anionic Impurities in Concentrated Inorganic Acids by Recycle Ion Chromatography," Journal of Chromatography, 466:355—363, 1989.

10. Vanderford G, "Determination of Trace Anions in Hydrofluoric Acid by Ion Chromatography," Journal of Chromatography, 602: 75—78, 1992.

11. Stahl R, "Ion Chromatographic Determination of Chloride,
Chlorate, and Perchlorate in Sulfuric Acid Solutions," Chromatographia, 37(5/6):300—302, 1993.

12. Buldini PL, Sharma JL, and Sharma S, "Determination of Trace Amounts of Anionic Impurities in Hydrochloric Acid by Ion Chromatography," Analyst, 119(1):121—124, 1994.

13. Loconto PR, and Hussain N, "Automated Coupled Ion Exclusion Chromatography for the Determination of Trace Anions in Fermentation Broth," Journal of Chromatography, 33(2):75—81, 1995.

14. Tada Y, Ohgi K, and Okada T, "Determination of Trace Nitrate Ions in High Purity Phosphoric Acid," Toso Kenkyu Hokoku (in Japanese), 38:65—70, 1994.

15. Fernandez J, "Determination of Anionic Impurities in Concentrated Inorganic Acids by Ion Chromatography," Analysis (in French), 22:430—439, 1994.

ManLi Wu, PhD, is a consulting scientist and supervisor for the chromatography group in the central analytical department of Olin Corp in Cheshire, CT. Her field of specialization is analysis of industrial chemicals and trace analysis of impurities using high-performance liquid chromatography, gas chromatography, ion chromatography, and other advanced techniques. Recently, she has been involved with GLP studies of biocides in the product chemistry and environmental fate areas. Wu received her BS from Tamkang University and MS from Tsing Hwa University, both in Taiwan, and her PhD in analytical chemistry from the University of Connecticut (1984). (Wu can be reached at 203/271-4298.)

Jian-Ge Chen, PhD, is a research scientist in the pharmaceutical R&D department of DuPont Merck (Wilmington, DE), with responsibility for developing analytical methods for evaluating novel pharmaceutics and supervising staff scientists. Prior to taking this position, he was a senior research chemist in the central analytical department at Olin Corp. (Cheshire, CT) specializing in trace analysis using microseparations, including multidimensional chromatography. He received his BS in chemistry (1990) from the State University of New York at Stony Brook and his PhD in analytical chemistry (1995) from the University of Pittsburgh. His publications include a review article, a book chapter, and more than 10 research papers.

MicroHome | Search | Current Issue | MicroArchives
Buyers Guide | Media Kit