Accurate, Low-Level Sulfur Analysis by ICP-MS Using MS/MS with Oxygen Reaction Cell Gas

Sulfur (S) is one of a small group of elements that have historically been difficult to analyze at low levels using conventional inductively coupled plasma–mass spectrometry (ICP-MS). Accurate low-level analysis of sulfur is required in applications in life science, clinical research, pharmaceutical development, food safety, environmental monitoring, geochemistry, and petrochemistry. ICP-MS analysis of sulfur is problematic because of the element’s high first ionization potential, which reduces sensitivity. Also, intense polyatomic interferences affect all the isotopes of sulfur. Background levels are also often high, due to the presence of S in many components used in laboratory equipment. ICP-MS fitted with a collision/reaction cell can use oxygen (O₂) cell gas to mass shift the ³²S⁺ ions to SO⁺ product ions at mass-to-charge (m/z) 48, avoiding the on-mass interferences at m/z 32. However, mass 48 can also be affected by interference from ⁴⁸Ca+, ⁴⁸Ti+, and ³⁶Ar¹²C⁺, as well as from overlapping product ions such as ³¹P¹⁷O⁺. Also, applications such as isotope ratio measurements and isotope dilution mass spectrometry (IDMS) require the accurate determination of a minor isotope, ³⁴S or ³³S, as well as ³²S. We describe how tandem mass spectrometry (MS/MS) controls the O₂ reaction gas chemistry to resolve multiple spectral overlaps on all the sulfur isotopes. Combined with appropriate selection of instrument components to reduce the sulfur background, MS/MS can provide accurate low-level analysis of sulfur and sulfur isotope ratios in aqueous and organic matrices.

Inductively coupled plasma–mass spectrometry (ICP-MS) is widely used for multielement analysis of trace analytes in a wide variety of sample types. A few elements, including sulfur (S), are difficult to measure using conventional (single)
quadrupole ICP-MS.

Applications that require low level sulfur analysis include petrochemical analysis (1,2), the study of biomolecules (such as proteins) in life science research (3,4), detection of sulfur compounds in foods (5), and development of sulfur-based pharmaceuticals (6,7). Accurate determination of sulfur at low concentrations is affected by several factors:

1.Sulfur has a high first ionization energy (or ionization potential) of 10.36 electron volts (eV), meaning it is only around 10% ionized in a normal ICP plasma. This translates into 10x lower sensitivity than for a fully ionized element.

2.The main isotope of sulfur (³²S, 95.04% abundance) suffers intense background overlap from polyatomic ions including ¹⁶O₂⁺ and ¹⁴N¹⁸O⁺, making trace analysis impossible by conventional quadrupole ICP-MS.

3.The next most abundant isotope, ³⁴S, accounts for only 4.2% of total sulfur, meaning a further reduction in sensitivity by a factor of more than 20. This isotope also suffers from intense background interferences, as does the final potentially useful isotope, ³³S (0.75% abundance). The fourth naturally occurring isotope, ³⁶S, is only 0.015% abundant, and is overlapped by a minor isotope of argon. The most significant background polyatomic ions that overlap the isotopes of sulfur are shown in Table I.

4.Sulfur is present in many laboratory reagents and components, causing background contamination.

The commercial development of collision–reaction cells (CRCs) for ICP-MS in 1997 (8) provided analysts with a novel approach to controlling interferences, solving many of the problematic spectral overlaps that had hampered ICP-MS analysis (9,10). However, a small number of analytes, including sulfur, remained difficult to determine at low concentrations.

Approaches to improve the detection of sulfur have included the use of xenon (Xe) cell gas to reduce the intensity of the ¹⁶O₂⁺ interference by collisional dissociation and charge transfer (11), and sector field high resolution ICP-MS to mass resolve the O^₂+ (4), However, the use of xenon cell gas does not give low enough detection limits for trace sulfur analysis, and sector field ICP-MS suffers from poor abundance sensitivity (peak tailing), which compromises the separation of the trace ³²S+ peak from the much more intense neighboring ¹⁶O₂⁺ peak.

Oxygen (O₂) cell gas can be used in the CRC of a single quadrupole ICP-MS instrument to promote the formation of SO⁺ product ions. The O₂⁺ interfering ions at m/z ³² do not react with the O₂ cell gas, so S+ can be mass-shifted away from the original O₂⁺ overlap (12). But the SO⁺ product ion masses are overlapped by existing ions, including ⁴⁸Ca⁺, ⁴⁸Ti⁺, and ³⁶Ar¹²C⁺ at m/z 48 for ³²S¹⁶O⁺. This approach is therefore not suitable for samples that might contain these other interfering elements, such as high Ca mineral waters and other environmental materials, or organic-based samples in petrochemical and life sciences applications (13).

Analysis of sulfur isotope abundance is of interest in several applications in environmental science, geochemistry, and hydrogeology. Such analysis requires interferences to be resolved on at least one of the minor isotopes of sulfur, ³⁴S or ³³S, as well as the major ³²S isotope. Minor isotopes must also be measured accurately to quantify sulfur using isotope dilution. CRC methods using O₂ cell gas with single quadrupole ICP-MS (with or without a bandpass cell) cannot control the formation of interisotope overlaps such as ³²S18O+ on ³⁴S¹⁶O⁺, so accurate isotope ratio and isotope dilution analysis that requires analysis of ³⁴S or ³³S is impossible. Potential matrix-based and interisotope overlaps on the SO⁺ product ion masses are shown in Table II.

The development of triple quadrupole ICP-MS (ICP-QQQ) in 2012 (14) offered a way to address the limitations of reaction mode methods on single quadrupole and bandpass ICP-MS. ICP-QQQ uses an additional quadrupole mass filter (Q1) before the CRC, allowing tandem mass spectrometry (MS/MS) operation. In MS/MS, Q1 functions as a 1 u mass filter to select the specific mass that is passed to the cell. This excludes any ions that might be present at the mass where analyte product ions would appear (15). As each analyte precursor mass is passed to the CRC while excluding all other masses, ICP-MS/MS also resolves interisotope overlaps so accurate isotope ratio and isotope dilution analysis is straightforward. We investigate the potential interferences on sulfur from several common matrix elements and demonstrate ICP-MS/MS conditions that resolve the interferences on the sulfur isotopes of interest. We compare these results with operating the ICP instrument without MS/MS (Q1 >1 u mass window), and show examples of applications where complete resolution of the interferences is needed.

Experimental

Instrumentation and Reagents

Measurements were performed using an Agilent 8900 ICP-QQQ (model #100, Advanced Applications configuration). The instrument was fitted with the standard glass concentric nebulizer, quartz spray chamber, quartz torch with 2.5 mm injector, and nickel sampling and skimmer cones. The instrument model used includes an inert, low contamination argon gas flow path, which reduces levels of sulfur (and silicon) in the gas supplied to the plasma torch. The effectiveness of this inert gas flow path is illustrated by the low ng/L (ppt) detection limit that can be achieved for sulfur in high purity semiconductor reagents (16). Samples were introduced via a peristaltic pump with 1.02 mm i.d. pump tubing. Operating conditions are shown in Table III.

Sulfur standards and matrix element solutions were prepared from single element stock standards (Agilent Technologies). Sulfur isotope standards were prepared from sulfur isotope CRMs IAEA-S-1, S-2, and S-3 (National Institute of Standards and Technology [NIST]). Ultrapure water was prepared in the laboratory using a Milli-Q Element system, and acids used for sample stabilization and matrix preparation were Ultra pure Normatom grade (VWR Chemicals).

Results and Discussion

Comparing MS/MS and Single Quad Mode for Analysis of Sas SO+

The mass-shift approach to resolve interferences has been described extensively (17). To assess the ability of MS/MS to resolve matrix interferences on the sulfur isotopes, mass spectra were acquired covering the region where the S16O+ product ions would appear (m/z 48 for ³²S16O+, m/z 49 for ³³S¹⁶O⁺, and m/z 50 for ³⁴S¹⁶O⁺). The same acquisition was performed in MS/MS mode and “single quad” mode, where Q1 is operated as a simple ion guide (no mass filtering). A 10 μg/L (ppb) S standard and separate solutions containing potentially interfering matrix elements Ca, Ti, and C were measured. The overlaid spectra (sulfur standard and matrix solutions) are shown for both modes in Figure 1.

Figure 1: Overlaid spectra of 10 μg/L (ppb) S standard (gray), 100 μg/L Ca matrix (blue), 100 μg/L Ti matrix (green), and 5% isopropyl alcohol (IPA, pink). (a) ICP-triple quadrupole MS in “single quad” mode (Q1 operated as an ion guide). (b) ICP-MS/MS (Q1 operated as 1 u mass filter). Existing element overlaps on the SO+ product ion masses are eliminated with MS/MS, enabling accurate sulfur isotope abundance template match (red outlines).

The good isotope template fit and freedom from matrix overlaps for sulfur in MS/MS mode contrasts markedly with the severe overlaps apparent in “single quad” mode. The difference is due to the inability of the single quadrupole to reject the existing Ca⁺, Ti⁺, and ArC⁺ ions at m/z 48 to 50, where the SO⁺ product ions are measured. In MS/MS, these interfering ions are all rejected by Q1, which is set to pass only the mass of the respective precursor S⁺ ions, so m/z 32 for ³²S, m/z 33 for ³³S, and m/z 34 for ³⁴S.

Controlling Product Ion Formation with Q1 in MS/MS Mode

With O-atom addition reactions, the large (16 u) mass difference between the precursor and product ion masses means any existing ions at the product ion masses can be rejected, even if Q1 is operated in bandpass mode (mass window > 1 u). However, existing ion overlaps are not the only potential interferences that can affect the accurate measurement of S as SO+. If the sample contains phosphorus (P), ³¹P⁺ ions could also react with the O_² cell gas to form ³¹P¹⁷O⁺ product ions that would overlap the ³²S¹⁶O⁺ product ions at m/z 48. This is illustrated in Figure 2, where a scan of a sulfur standard (25 μg/L) is shown overlaid with scans of 0.5% H₃PO₄ and 0.5% HCl. The solutions were measured in MS/MS mode and with Q1 set to pass a 3 u mass window (simulating bandpass operation). In bandpass mode, the P matrix gives rise to an intense interference at m/z 32, while MS/MS ensures effective resolution of the ³¹P¹⁷O⁺ product ions. Rejecting ³¹P+ while allowing transmission of ³²S+ requires Q1 to operate as a 1 u mass filter (MS/MS mode).

Figure 2: Overlaid spectra of 25 μg/L (ppb) S standard (gray) and 0.5% H3PO4 matrix (pink); 0.5% HCl matrix (green) also shown. (a): ICP-triple quadrupole MS in “bandpass” mode (Q1 set to 3 u mass window). (b): ICP-MS/MS (Q1 operated as 1 u mass filter). PO+ product ion overlap on SO+ is eliminated with MS/MS, as confirmed by accurate S isotope abundance template match (red outlines).

Sulfur Isotope Analysis in Natural Waters

The relative abundance of the stable (non-radiogenic) sulfur isotopes varies a lot in natural materials due to fractionation during redox reactions. The variation in abundances is calculated as a deviation (or delta, δ) in the abundance of the minor isotope (such as δ³⁴S), expressed in per mil (parts per thousand, ‰). Natural materials such as groundwaters, soil, vegetation, and rocks can contain sulfur isotopes with δ³⁴S values ranging from -40‰ to +40‰, and occasionally more (18). This variation enables researchers to study processes such as sedimentary rock formation and biogeochemical cycles. High precision sulfur isotope analysis is usually performed using an isotope ratio mass spectrometer (IRMS), such as a thermal ionization mass spectrometer (TIMS), secondary ionization mass spectrometer (SIMS), or multi-collector ICP-MS. However, the large variation in natural sulfur isotope abundances means that an ICP-MS/MS method with O₂ cell gas can measure the differences in S ratios, albeit less precisely.

Determining accurate S isotope abundances by conventional ICP-MS can be affected by errors from interisotope overlaps. Where the mass that enters the CRC is not tightly controlled, multiple sulfur isotopes can react to form product ion overlaps. For example, ³²S¹⁸O⁺ appears at the same mass as ³⁴S¹⁶O⁺ (m/z 50), so accurate determination of ³⁴S abundance is only possible if ³²S⁺ is excluded from the CRC. As with adjacent mass overlaps (such as ³¹P¹⁷O⁺ on ³²S¹⁶O⁺), resolving inter-isotope overlaps requires Q1 to operate as a 1 u mass filter (MS/MS).

Sulfur isotopes were measured in a range of mineral water, river water, spring water, and seawater samples, using ICP-MS/MS in mass-shift mode with O₂ cell gas. Sulfur isotope calibration was based on IAEA-S-1, and bracketing standards were used for mass bias correction. The results are summarized in Figure 3. Excellent δ³⁴S precision of less than 0.9‰ SD (n = 10) was obtained. The accuracy of the method was confirmed by the measured δ³⁴S value for NASS 5 seawater (21.5‰), which agreed well with the global average value for oceanic seawater of 21‰ (18).

Figure 3: Sulfur isotope variation (δ34S, ‰) in water samples measured using ICP-MS/MS. Error bars show ±1 SD (n = 10). Precise S isotope ratios were obtained with SD less than 0.9‰, and clear differences in 34S isotope abundances could be seen. The measured δ34S value of 21.5‰ for NASS 5 seawater CRM agreed well with the global average δ34S value for oceanic seawater of 21‰ (18). Note: the symbol ‰ indicates parts per thousand (also called per mil) used to report isotope ratio precisions.

Sulfur isotope analysis in geochemistry applications may require the measurement of the ³³S:³²S ratio as well as ³⁴S:³²S. The ³³S isotope can be measured as ³³S¹⁶O⁺ using ICP-MS, but the large difference in abundance between ³²S and ³³S isotopes can cause errors in the analysis of the minor isotope. In ICP-MS/MS analysis using O₂ mass shift, ³²S must be excluded from the CRC when ³³S is measured. This separation can be degraded either by poor Q1 resolution (mass window >1 u), or by poor abundance sensitivity (peak tail of ³²S overlaps ³³S). In both cases, the ³²S peak would contribute to the signal at the adjacent mass 33, affecting the ³³S:³²S ratio.

This potential error was assessed by measuring digested sulfide minerals and calculating the mass-independent deviation in the measured abundance of the ³³S isotope (δ³³S). Mass independent fractionation of sulfur isotopes rarely occurs naturally, so the defect in ³³S abundance calculated from ∆³³S = δ³³/³²S - δ³⁴/³²S x 0.515 should equal zero, assuming equilibrium fractionation (19). A measured value of zero (within experimental error) therefore indicates no contribution from ³²S on ³³S. The results obtained with ICP-MS/MS are shown in Table IV, with a δ³³S of -0.1 confirming no contribution from ³²S on ³³S measured as the ³³S¹⁶O⁺ product ion.

Conclusions

ICP-MS/MS using O₂ cell gas to mass shift the S⁺ ions to their respective SO⁺ product ion masses provides the opportunity to determine sulfur accurately at lower levels than was previously possible by ICP-MS. By allowing only the target S⁺ precursor ion into the CRC, MS/MS eliminates potential interferences from other reactive ions (such as ³¹P⁺) and from existing ions at the product ion masses (such as ⁴⁸Ca⁺, ⁴⁸Ti⁺, and ³⁶Ar¹²C⁺).

MS/MS also excludes the possibility of inter-isotope overlaps, enabling accurate sulfur isotope analysis in complex sample matrices. These measurements cannot be performed successfully using conventional single quadrupole or bandpass ICP-MS.

References

1. B. Bouyssiere, P. Leonhard, D. Profrock, F. Baco, C. L. Garcia, S. Wilbur, and A. Prange, J. Anal. At. Spectrom.19, 700–702 (2004).
2. J. Heilmann, S.F. Boulyga, and K.G. Heumann, Anal. Bioanal. Chem. 380, 190–197 (2004).
3. J.G. Martinez-Sierra, F.M. Sanz, P.H. Espilez, R. Santamaria-Fernandez, J.M.M. Gayon, and J.I.G. Alonso, J. Anal. At. Spectrom. 25, 989–997 (2010).
4. C. Rappel and D. Schaumlöffel, Anal Bioanal Chem. 390, 605–615 (2008).
5. B. Lajin, and W. Goessler, Anal. Chim. Acta 1092, 1–8 (2019).
6. E.H. Evans, J.C. Wolff, and C. Eckers, Anal. Chem. 73, 4722–4728 (2001).
7. C.J. Smith, I.D. Wilson, L. Weidolf, F. Abou-Shakra, and M. Thomsen, Chromatographia Suppl. 59, S165–S170 (2004).
8. P.J. Turner, T. Merren, J. Speakman, and C. Haines, G. Holland, and S.D. Tanner (eds), Plasma Source Mass Spectrometry: Development and Applications. (The Royal Society of Chemistry, Cambridge, United Kingdom, 28–34,1997).
9. N. Yamada, J. Takahashi, and K. Sakata, J. Anal. At. Spectrom.17, 1213–1222 (2002).
10. E. McCurdy and G. Woods, J. Anal. At. Spectrom.19, 607–615 (2004).
11. D. Pröfrock, P. Leonhard, and A. Prange, Anal. Bioanal. Chem. 377, 132–139 (2003).
12. D.R. Bandura, V.I. Baranov, and S.D. Tanner, Anal. Chem. 74, 1497–1502 (2002).
13. K. De Wolf, L. Balcaen, E. Van De Walle, F. Cuyckens, and F. Vanhaecke, J. Anal. At. Spectrom. 25, 419–425 (2010).
14. S.D. Fernandez, N. Sugishama, J.R. Encinar, and A. Sanz-Medel. Anal. Chem. 84, 5851–5857 (2012).
15. L. Balcaen, G. Woods, M. Resano, and F. Vanhaecke, J. Anal. At. Spectrom. 28, 33–39 (2013).
16. K. Nakano, Ultra-low-level determination of phosphorus, sulfur, silicon and chlorine using the Agilent 8900 ICP-QQQ (Agilent application note 5991-6852EN, 2016).
17. L. Balcaen, E. Bolea-Fernandez, M. Resano, and F. Vanhaecke, Anal. Chim. Acta 894, 7–19 (2015).
18. T.B. Coplen, J.K. Böhlke, P. De Bièvre, T. Ding, N.E. Holden, J.A. Hopple, H. R. Krouse, A. Lamberty, H. S. Peiser, K. Revesz, S.E. Rieder, K.J.R. Rosman, E. Roth, P.D.P. Taylor, R.D. Vocke Jr., and Y.K. Xiao, Pure Appl. Chem. 74(10), 1987–2017 (2002).
19. E.D. Young, A. Galy, and H. Nagahara, Geochim. Cosmochim. Acta 66(6), 1095–1104 (2002).

Ed McCurdy and Glenn Woods are withAgilent Technologies LDA (UK) Ltd, in the United Kingdom. Bastian Georg is with Agilent Technologies in Canada. Naoki Sugiyama is with Agilent Technologies International in Japan. Direct correspondence to: ed_mccurdy@agilent.com.