Investigations Into Sulfur Speciation by Electrospray Mass Spectrometry* IAN I. STEWART DAVID A. BARNETT AND GARY HORLICK Department of Chemistry University of Alberta Edmonton Alberta Canada T6G 2G2 Sulfur is an element that manifests itself in a wide variety of chemical forms in the environment and it is generally not sufficient just to determine the elemental level of sulfur in environmental samples. In this study it is shown that electrospray mass spectrometry (ESMS) has the potential to provide a powerful direct probe for sulfur species in solution samples. The basic electrospray mass spectra to be expected for species such as SO:- S20,2- and SzO$- are presented and the important role of controlling and varying collision induced dissociation (0) conditions in order to validate the species identity is illustrated.It is also shown that ESMS can be used to monitor sulfur species such as S,O:- and S20:- which are products in certain redox reactions and also thiosulfate complexes such as Ag(S2O3);-. Finally quantitative results are presented for the determination of sulfate in waste water. Keywords Sulfur; electrospray mass spectrometry; collision induced dissociation; waste water; sulfur species The chemistry of sulfur is fairly complex as it can exist in several different forms. In aqueous inorganic solutions sulfur commonly exists in anionic form which can be as simple as sulfide or as complex as one of the numerous 0x0- or peroxo- acid forms. Sulfur species are present in many different aqueous systems such as ground and surface waters as well as biological fluids.In ground-water samples complex sulfur species prob- ably originate as a result of sulfide mineral oxidation at low to neutral pHs.’ Therefore the reaction products can give some indication as to the mechanisms involved in this process and thus the determination of such species is desirable for environmental reasons as well as from the point of view of understanding general inorganic sulfur chemistry. In biological samples such as urine sulfate determination is useful in the study of urolithiasis.2 Sulfate derivatives can also exist in many different organic molecules such as sulfonated azo dyes as well as steroidal c~mplexes.~ Over the years sulfur species and inorganic anions in general have been determined for the most part by chemical chromato- graphic and spectrochemical techniques.For example oxo- sulfur species have been determined by ion-exchange chroma- tography 4*5 ion-pair chroma tograp h y ‘-’ capillary elec trop h- oresis (CE)” and spectr~photornetry.’~-~~ For the most part these studies have focused on such species as sulfate sulfite and thiosulfate as well as polythionate species containing up to 13 sulfur atoms. In some cases however these types of techniques are inadequate as they are unable to provide the desired degree of selectivity or sensitivity. As a result research- ers have moved towards developing chromatographically coupled MS techniques such as HPLC-ICP-MS and CE-CP- MS in attempts to improve intra-element speciati~n.~~-’~ These techniques are good for differentiating between different oxi- dation state metal ion species and different polyatomic ion species such as arsenite monomethyl arsonic acid (MMAA) * Presented at the 1996 Winter Conference on Plasma Spectrochemistry Fort Lauderdale FL USA January 8-1 3 1996.Journal of Analytical Journal of Analytical Atomic Spectrometry and dimethylarsinic acid (DMAA) but are not entirely effective at differentiating sulfur species.” More recently a relatively new technique electrospray ioniz- ation mass spectrometry (ESMS) has been used successfully to determine inorganic cations and anions in solution ~ a r n p l e s . ’ ~ ~ ~ ~ In particular Agnes et a1.18 have shown that it can be used successfully to determine sulfur anion species in solution directly. A variation of electrospray (ES) ‘ion spray’ has also been coupled to CE for the MS determination of inorganic cations and anions,” which is desirable when exam- ining complex samples.The technique has been described extensively in the litera- t ~ r e ” - ~ ~ and so only a qualitative description will be given here. Simply ES is an electrostatic spraying process where a liquid surface becomes disrupted when an intense electric field is applied resulting in a spray of charged droplets. With ES a solution of optimal conductivity (typically a methanolic elec- trolyte solution of the order of 1 x 10-5-1 x moll-’) is pumped (x 1-10 p1 min-l) through a stainless-steel capillary (x 100 pm id 200 pm od) which is held at a high potential relative to a counter electrode (the mass spectrometer sampling orifice).This spraying phenomena at the capillary tip affords the transfer of solution ions to the gas phase where they can be sampled by a mass spectrometer. A negatively biased capillary results in negatively charged droplets and ultimately negative gas-phase ions; the opposite is true for a positively biased capillary. The formation of the spray the charging of the droplets and formation of gas-phase ions from such droplets have been described in detail in the l i t e r a t ~ r e . ~ ~ ? ~ ~ The focus of the present paper is on the effectiveness of ESMS in investigating inorganic sulfur species in solutions and is presented in part as a continuation of previous work.’* In particular the paper will discuss some background to the generation and sampling of gas-phase sulfur ions. A variety of solutions will be investigated ranging from standard laboratory solutions to reaction mixtures to real samples.Although the focus is on the qualitative aspects some preliminary data on the quantification of sulfur species will be presented. EXPERIMENTAL Instrumentation The ES source and mass spectrometer have been described previ~usly.’~ The high-pressure sampling region of the modified SCIEX-Perkin-Elmer ELAN Model 250 ICP-MS instrument is based on a sampling plate-skimmer configuration. There have however been some modifications inside the mass spec- trometer the shadow or first photon stop in the ion optics as well as the second stop located in the bessel box have been removed. This has led to an increase in sensitivity of one to one and a half orders of magnitude.As anions are being studied the mass spectrometer was operated in negative-ion mode. For these studies the ES tip was typically biased at - 3000 V the front plate at - 600 V the sampling plate was varied (typically between -2 and -60 V) to minimize or maximize collision induced dissociation (CID) in the high- pressure region and the skimmer was held at -2 V. As the Atomic Spectrometry September 1996 Vol. 1 1 (877-886) 877skimmer potential is held constant the potential drop and hence energy is proportional to the sampling-plate voltage (see ref. 24 for a discussion of this). The curtain gas used for these experiments was nitrogen at a flow rate of about 1.3 1 min-'. The ES needle tip had a 100 pm id and was operated at a flow rate of 1.0-2.0 p1 min-' via a syringe pump (Sage Instruments Model 341A).The tip position although optimized for each set of experiments was usually set 5 mm from the front plate and 1 mm off axis. It was found that careful selection of flow rate and applied capillary potential led to stable signals without the use of discharge suppressing gases. For quantitative investigations two different procedures were employed. In one a normal ES set-up was used where sample solutions were run in a sequential manner. The samples were prepared to contain the analyte in varying concentrations and a known constant concentration of internal standard. The flow rate was typically 2.00 pl min-'. In the second a Valco high- pressure injection valve with a 0.50 p1 internal sample loop was used.The samples were prepared to include the internal standard and were injected into a flowing stream of internal standard in methanol. The flow rate of the carrier-internal stan- dard solution was 2.50 pl min-'. Selected ions were monitored as a function of time. Reagents All solutions were prepared by dissolving the ACS-grade salts in distilled de-ionized water to form a stock solution. Aliquots of the stock solution were then diluted with HPLC-grade methanol to the desired concentration. This procedure was modified for the quantitative experiments where the methanol was distilled and the reagent salt was oven dried. These procedures result in the solution being primarily methanol with a water content in the range of 1-3% by volume.RESULTS AND DISCUSSION General Preliminary studies of inorganic sulfur species were carried out by Agnes et a1.I8 using ESMS. It was shown that a wide range of anionic sulfur species could be examined. For the most part anionic sulfur species are fairly amenable to ES and it is generally the case that if a species exists in an ionic form in solution and is stable in an ES solvent it can be observed by ESMS. These two stipulations can obviously pose some limi- tations. In addressing the first point take sulfide for example the reactive solution form is typically S2- which is difficult to observe by ES. This is due primarily to the fact that simple sulfides such as H2S are only partially ionized in aqueous solutions. The first- and second-dissociation constants are pK 6.8 and pK 14.15 respectively,2s which means that for an initial solution of 1 x moll-' only about 3 .6 ~ mol 1-' exists as HS- and there are virtually no S2- ions present. In addition the equilibrium can be shifted even further to the left when the sample is diluted in methanol which has a much lower relative permittivity than water z 33 as opposed to x78. The other concern is the stability of the species both in aqueous solution and the ES solvent. For the most part the commonly occurring sulfur anions especially the 0x0-sulfur species are fairly stable in solvents such as methanol however some species such as sulfite (SO,2-) are not; this example will be addressed later in the discussion. Although the generation of gas-phase ions from the droplets is still a controversial subject in the ES literature it is evident that ions tend to leave the droplet with a stabilizing solvation sphere of significant size.It is these well solvated ions that are sampled by the mass spectrometer and Fig. 1 will be used to illustrate the processes involved in sampling a solvated anion and stripping it down to a bare polyatomic ion via high- pressure CID at the interface. 2000000 1800000 1600000 14M1ooo 7 u) + 5 1200000 2 1OOOOOO .- E 0 800000 Q) - E 600000 400000 200000 0 90 1 V 200 Fig. 1 the persulfate ion going from - 10 V potential difference to a - 30 V potential difference 3D CID spectra of ammonium persulfate ( 1 x moll-') in methanol. The plot illustrates the change in the solvated distribution of 878 Journal of Analytical Atomic Spectrometry September 1996 Vol.I 1A variable high-pressure CID plot of ammonium persulfate ( 1 x lo- mol I-’) in methanol is shown in Fig. 1 where different ‘CID energies’ are plotted as a function of intensity and m/z. It is important to note that these are not actual ‘collision energies’ but are just the potential differences between the sampling plate and skimmer which will be related to the collision energy. This can be simply understood by considering the fact that the greater the electric field in this region the greater the acceleration of an ion relative to the neutral nitrogen curtain gas (or other neutral gases in the expansion). This will in its simplest interpretation result in a greater collisional energy. The nature of the expansion will further complicate this as gas density and hence collisional frequency is a decreasing function of distance from the sampling orifice.The energy of collisions that occur in the later part of the expansion could conversely increase as a result of greater mean-free path and hence longer acceleration times between collisions. Whether the relationship between potential differ- ence and CID energy is a strictly linear relationship is unknown however the net result in all cases is that a greater potential difference affords a greater amount of stripping. Considering first the low CID potential (lOV) a low (rela- tively) intensity distribution at high m/z can be seen these are peroxodisulfate water cluster species with distributions ranging from four water ligands to 11 water ligands; the most intense peak in this distribution seems to be persulfate with seven water molecules at m/z 159.Since these species are 2- anions the distribution separation is 9 m/z units (18/2). When the potential difference is increased in increments of 5 V the distribution makes an obvious shift to lower m/z values and eventually at a CID value of 30 V yields primarily a desolvated bare persulfate anion at m/z 96. There is a minor peak at m/z 105 owing to the singly hydrated peroxodisulfate species. Past this point the energy supplied goes into molecular fragmen- tation and various decomposition products are formed. The actual process for the persulfate ion has been described in some detail by Agnes et ~ 1 . ’ ~ It is interesting to note the increase in signal intensity going from CID 10 V to CID 30 V.It would be expected that the total ion current should remain somewhat constant regardless of energy but this was not observed. One probable explanation for this is that ions leave the expansion region with greater velocities for higher potential differences and thus they have a much more efficient trans- mission through the ion optics and quadrupoles. Fig. 1 illustrates the different types of information that can be obtained and that it is important to be aware of sampling conditions while interpreting the mass spectrum. At potential differences greater than 30 V molecular fragmentation occurs and species such as HOSO,- HS04- SO3- and SO2- can be observed all of which are clearly not present in the original solution.These spectra obtained under harsher conditions have been discussed previously by Agnes et a1.I8 and are not shown here. Therefore great care must be taken when inter- preting spectra. Depending on the nature of the anion it might or might not be possible to ‘strip’ its solvation sphere down to a ‘bare’ molecular form. In the case of the persulfate ion above it is large enough to stabilize its bare gas-phase charge however other smaller anions might not be able to support multiple charges and therefore will seek to reduce their charge; this is usually achieved by fragmentation. Sulfate and thiosulfate are two such multiply charged ions that undergo charge reduction reactions in the gas phase. The sulfate anion which is perhaps the most commonly occurring of the sulfur-oxygen anion species can exist in solution in several different forms H2S04 HS04- and SO,2 - depending on the pH of the solution.For the most part it exists almost entirely as the ion in neutral solutions. Electrospray mass spectra of vanadyl sulfate in methanol are given in Fig. 2. The figure consists of two spectra a mild (relatively) CID condition spectrum in (a) and a harsher CID spectrum in (b). The distribution of solvated sulfate ions and also a certain amount of the protonated equilibrium species HS04- is shown in Fig. 2(a). The 2- water solvated species are separated by 9 m/z units corresponding to differences of one water ligand. It is observed that the sulfate anion requires at least four water ligands to maintain its two charges.Once this species is reached it will not simply lose another water ligand rather it will charge separate as shown in the scheme below in order to stabilize the charge S042-(H20)4-+HS04- +OH-(H20),+(3 -n)(HzO) Although it seems clear that when sampling with high-pressure CID the minimum water ligand number for sulfate is four the charge separation reactions could also occur to a lesser degree in the more solvated species (i.e. five water ligands). The charge separation itself is not straight forward as it could occur with some variation as indicated below S042-(H20)4+HS04-(HzO) + OH-(H20)2 +etc.. . The above reactions probably account for the presence of the HS04- species observed in Fig. 2(a). At extremely mild sam- pling conditions the HSO,- species contribution is greatly reduced.Sulfate with less than four water ligands is not observed using the high-pressure CID interface. The above studies have also been observed by low-pressure CID.26 A spectrum acquired under harsher CID conditions is shown in Fig. 2(b) where all the sulfur species have essentially been converted into one species (HS04-) prior to mass analysis. Note also the presence of various solvated hydroxide species characteristic of the charge separation step. The mass spectrum has distributions owing to basic ions such as OH- under gentler CID conditions and so the origin of these species should be viewed with caution. However as a consequence of CID if they did exist from other sources such as free OH- they would exist primarily as the bare OH- ion at m/z 17.Therefore the distribution observed in Fig. 2(b) would most probably be there as a result of the fragmentation of the solvated sulfate ion. Also note the peaks at m/z 129.5 and 179 that have not been definitely assigned but could be assigned with caution vanadyl [VO(S04)22-] and vanadate species [ V 0 2 ( S 0 ) - ] respectively. Thiosulfate although not as stable and somewhat more reactive in solution than sulfate also exists primarily as the 2 - ionic species in solution. Three sodium thiosulfate spectra are given in Fig. 3 each acquired under different CID region conditions. Under fairly gentle conditions [Fig. 3 (a)] where AV=14V a distribution of solvated thiosulfate ions with a solvation distribution ranging from three to nine water mol- ecules is observed.A distribution maximum is noted at m/z 101 corresponding to the S2032- (H20)5 species. Under gentler conditions the distribution maximum shifts to higher values. Similar to the sulfate anion a minimal solvation sphere must be maintained by the thiosulfate anion in order to maintain its charge. For thiosulfate it was found that at least three water ligands are required corresponding to an m/z of 83 any number less than this leads to some interesting results. Intuitively one might expect a charge separation similar to that of sulfate to occur as in the scheme below S2032-(H20)3 -+HS203- + OH-(H20)2 This is not strictly observed. The spectrum shown in Fig. 3(b) was acquired under slightly harsher conditions (AV = 19 V) here the solvated distribution can be seen to shift to a maximum Journal of Analytical Atomic Spectrometry September 1996 Vol.11 879200 - 150- 100- 7 'W 50- u) c 9 0 + 0 0 n = 7 n n = 8 0 7 5 8 0 8 5 90 95 100 105 110 115 120 125 130 135 140 145 150 155 160 165 7 s .- 10 2 0 30 4 0 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 fli2 Fig. 2 potential differences (a) - 12.5; and (b) -22.5 V Mass spectrum of vanadyl sulfate (1 x moll-l) in methanol. The spectra were acquired at two different sampling plate to skimmer at m/z 83 and some decomposition products resulting from the continued stripping of this species are also observed. Although there is evidence for the expected HS203- anion at (m/z 113) it is not the dominant ion formed. The dominant species formed is based on the S203- (m/z 112) ion and its solvated precursors S203-( H20) and S203-(H20)2 at m/z 130 and 148 respectively.It is unclear how these products form however an electron must be lost in coincidence with the loss of a water ligand. Examination of the third spectrum in the series [Fig. 3(c)] illustrates that the solvated species can be stripped down to the bare polyatomic ion S203- at m/z 112. The exact nature of the process going from the solvated 2- polyatomic ion to the 1- bare ion is currently under investi- gation. The sodium adduct (m/z 135) is most likely formed as a result of droplet preconcentration. The formation of the S203- species is a consistent product when sampling with the high-pressure CID source however it is unknown whether this species forms in the same dominant manner under the conditions of low-pressure CID.From the above discussion it is apparent that some ions cannot be stripped down to their 'true' polyatomic ions however they can be stripped down to characteristic poly- atomic ions as indicators. In the above case the characteristic ions are HSO,- (m/z 97) for sulfate and S203- (m/z 112) for thiosulfate. It is clear that with some background knowledge of the sampling behaviour of these species by high-pressure CID it should be possible to differentiate between the various anions in solution. A test sample was prepared of equimolar concen- trations of ammonium persulfate sodium sulfate and sodium thiosulfate where each was 1 x lo- moll-' in methanol. Two spectra are shown in Fig. 4 where spectrum (a) was acquired under fairly harsh CID conditions (AV = 30 V) and spectrum (b) was acquired under gentler conditions (AV= 10 V).Based 880 Journal of Analytical Atomic Spectrometry September on the knowledge of the potential difference and characteristic peak assignments spectrum (a) indicates that the persulfate anion the sulfate anion and the thiosulfate anion could be present however the assignment is ambiguous at best. It is not until the ions are sampled under gentler conditions [spec- trum (b)] that this is confirmed. Examination of Fig.4(b) illustrates the presence of the three distinct distributions of persulfate sulfate and thiosulfate each highlighted in its own pattern. Not all species are stable in ES solvents let alone in aqueous solutions. One such species is the sulfite ion which is only finitely stable in aqueous solutions and is reactive in methanol solutions. Upon dilution of freshly prepared aqueous sulfite solution in methanol the species CH30S02- is readily observed by ES.This is illustrated in Fig. 5 where in addition to the CH30S02- species (m/z 95) being observed the presence of the minor species HS04- is also detected. Although the origins of the HS04- species is unclear the ester could be formed via the scheme HS03- + CH30H-+H20 + CH30S02- When an aqueous sulfite solution is allowed to age ( z 1 month) the only observable species when diluted by methanol is sulfate. This indicates that sulfite slowly decomposes in aqueous solution to yield sulfate. Other species such as dithionite (S2042-) are also fairly unstable in aqueous solutions and can decompose rapidly depending on the pH.25 Its decomposition products if stable in solution can of course be detected by ESMS.From the above discussion it is obvious that the chemical nature of the species is critical when obtaining information about them by ESMS. It was shown that large multiply charged anions are able to stabilize the charge internally whereas smaller multiply charged ions need a supporting 1996 Vol. 11x = 5 I . . " " . " " ' ' " " ' " ' ' I " ' I " ' " ' " ' I " " ' " " I""""' 0 I ....,.... I ....,....,....,.... 70 80 90 100 110 120 130 140 150 160 170 70 80 90 100 110 120 130 140 150 160 170 7 0 80 90 100 110 120 130 140 150 160 170 m/z Fig. 3 Mass spectra of sodium thiosulfate (1 x potential differences (a) - 14 V; (b) - 19 V; and (c) -24 V in methanol.The spectra were acquired at different sampling plate to skimmer solvation sphere to be stable. If the ion gets to the point where it no longer has the necessary solvation then charge-separation reactions occur. It is important to be able to understand these processes in order to interpret the resulting mass spectra properly. Interpreting Reaction Mixtures Sulfur ions in solution especially various 0x0-anions can be fairly reactive. They can act both as reducing agents and oxidizing agents of various strengths in solution. For instance the thiosulfate and sulfite ions are known to have a moderate reducing character whereas the peroxodisulfate ion is known to be a powerful oxidizing agent. This characteristic can be fairly useful analytically for example Koh and co-worker~l'-~~ have determined many sulfur species spectrophotometrically based on their redox chemistry.Perhaps one of the most well known reactions is the use of thiosulfate in the volumetric determination of iodine which proceeds according to the reaction 2S2032- +I2 +2I- + S4062- where the products are iodide and tetrathionate ions. This simple reaction system was examined by ES. The mixture consisted of 1.02 x mol I-' thiosulfate in aqueous solution. Upon mixing and dilution the dark purple iodine solution immediately became clear which is characteristic of its reduction to iodide. The solution after standing for 2 h was then diluted with methanol (100-fold) and the resulting solution was electrosprayed.Two spectra acquired under different sampling conditions are given in Fig. 6 (a) at a potential difference of 15 V and (b) at 27 V. Examination of Fig. 6(u) reveals two distributions the tetra- thionate distribution stripped down to the bare tetrathionate ion at m/z 112 and accompanying hydrates as well as a distribution of thiosulfate species which were present in excess. The spectrum just highlights the hydrated species however there are species such as S4062-(CH30H) S40,2-(CH30H)2 and S4062-(CH30H)(H20) present at m/z 128 144 and 137 respectively. The iodide ion is also present at m/z 127. If the spectrum is acquired at a greater potential difference [Fig. 6(b)] the CID products shown below moll-' iodine and 2.23 x S4062- -+S203 - + S203 - S4062- +S303- + SO3- are formed which would not normally be found in solution. The tetrathionate ion because of its size can be stripped down Journal of Analytical Atomic Spectrometry September 1996 Vol.1 1 881400 300 200 100 T ln ln K O c 2 V 80 90 100 m 140 c .- ln 2 120 - 1 00 80 60 40 20 0 (b ) Sulfate Thiosulfate Persulfate I 110 120 130 140 70 80 90 100 110 120 130 140 150 160 170 nVz Fig. 4 Mixture of 0x0-sulfur species ammonium persulfate sodium thiosulfate and sodium sulfate each 1 x lop4 moll-' in methanol. Spectrum (a) was acquired with a sampling plate to skimmer potential difference of - 30 V and spectrum (b) with a potential difference of - 10 V CH,OSO,- / to its bare form at m/z 112. However increasing the CID energy leads to the decomposition products as described above and as observed in Fig.6(b). A second redox mixture system that was studied by ESMS is given in the scheme below MnO +2SO,,- +4H+ +Mn2+ +S2062- +2H,O where sulfite is oxidized by manganese(1v) oxide yielding the dithionate ion ( S 2 0 6 2 - ) . The reaction mixture consists of MnO (0.087 g in 100 ml total volume) 2.05 x mol 1-1 SO,2- and 4 x lo- moll-' HN03. The reaction mixture was allowed to stand for 1 d to ensure complete reaction upon which time it was diluted with methanol (100-fold) and then examined by ES. The results are given in Fig. 7 the spectrum was acquired under fairly harsh stripping conditions in order to illustrate that the dithionate ion can be stripped down to lJ+ . ,,- , .~ it" L, ,1.,t, :- TI.:- :- -L,,..-.,.I ^ C - - - / - o n - - A :- confirmed by the isotopic distribution and in addition there is some residual hydrate at m/z 89.If the CID energy is greatly increased the dithionate ion decomposes according to the m/Z Fig- Mass spectrum of sodium sulfite ( 1 mol 1-1) in anol. The spectrum was acquired with a sampling plate to skimmer potential difference of - 20 V scheme S206,- - SO - + SO - 882 Journal of Analytical Atomic Spectrometry September 1996 Vol. 1180 9 0 100 110 120 130 140 150 160 170 180 60 80 100 120 140 160 180 200 220 240 260 m/z Fig. 6 Mass spectra of the reaction products from a solution of iodine (1 x lo-’ moll-’) and sodium thiosulfate (2 x lo-’ moll-’) diluted in methanol 100-fold. The spectra were acquired at sampling plate to skimmer potential differences of (a) - 15; and (b) -27 V d Z Fig.7 Mass spectrum of the reaction products from a solution of MnO (1 x lo-’ moll-l) sodium sulfite (2 x lo-’ moll-’) and nitric acid (4 x lop2 moll-’) diluted in methanol 100-fold.The spectrum was acquired with a sampling plate to skimmer potential difference of -22v In addition to the 2- species the protonated and sodium adducts are also observed at higher m/z. The spectrum also shows two other ions of interest those at m/z 97 (HS04-) and m/z 111 (CH,OSO,-) for which there are two possible expla- nations. The first is the oxidation of sulfite by manganese(1v) both in aqueous and methanolic solutions to yield the two products the second is the reaction of the dithionate ion with methanol uia the following scheme S 2 0 6 2 - + CH30H+HS04- +(CH,O)SO,- It is unlikely that the above scheme occurs on its own in solution and could be acid or manganese(1v) catalysed.The above scheme could also be occurring as a result of some gas- phase chemistry. Sulfur species are also useful complexing agents and for example thiosulfate is used in photographic processes. Thiosulfate solutions are typically used to complex unphoto- lysed silver bromide from a photographic emulsion. In fact thiosulfate is able to complex with most silver ions in solution provided it is added in excess. When silver is the dominant ion an Ag2(S,0,) precipitate will form which then gradually decomposes to Ag2S. However in the case of thiosulfate being used in the dissolution and complexation of silver halides it is typically added far in excess to ensure the reaction AgX+2S2032- +Ag(S203),,- + X - The above reaction holds for any of the insoluble silver halides as the net stability constant for the above complex is of the order of 5 x lo1,.To examine this system a solution was prepared where AgCl (0.1433 g in a total volume of 100 ml) was dissolved in excess of S2032- (2.5 x lo-’ moll-l). The solution was allowed to sit for 1 d and then diluted 100-fold in methanol and examined by ESMS. The results are shown in Fig. 8. The spectrum in Fig. 8(a) was acquired under fairly gentle conditions and shows the distribution of solvated thios- ulfate as expected being added in excess and also a second distribution of solvated Ag( S,03)2 - species. This distribution ranges from five water ligands (m/z 140.3) to 11 water ligands (m/z 176.3) and the species are separated by 6 m/z units (18/3 = 6).The Ag(S20,),3- species is only stable down to five hydrates. The spectrum in Fig. 8(b) is a higher resolution spectrum and was acquired under slightly harsher sampling conditions. The isotopic fine structure due primarily to the silver and sulfur isotopes becomes compressed three-fold as it Journal of Analytical Atomic Spectrometry September 1996 Vol. 11 8830 i7j 4000; 3000 - 2000 - 1000 - x = 6 x = 7 l 5 I 10 135 140 145 150 155 160 160 170 180 190 200 210 220 230 ,m/z Fig. 8 Mass spectra of the reaction products from a solution of silver chloride ( 5 x mol l-’) dissolved by sodium thiosulfate (2.5 x lo-’ moll-’) in aqueous solution. The sample was then diluted 100-fold in methanol.The spectra were acquired at different sampling plate to skimmer potential differences (a) -9; (b) - 14 (high resolution); and (c) -24 V is a 3 - species. The spectrum in Fig. 8(c) was acquired under fairly harsh conditions and shows- what happens when the Ag( S203);- is stripped lower than the minimum stabilizing ligand number (ie. five water ligands). Consistent with the previous discussion on thiosulfate one of the thiosulfate ligands loses an electron in some unknown manner to form an electron deficient species as shown at m/z 165.5 [Ag(S203)22-]. The complex further decomposes to the Ag(S,O,)- species as observed at m/z 219 with increased CID energy. The isotopic abundances of the above species are consistent with the calculated values as shown by the two examples given in Fig.9. The above discussion illustrates the strength of ES as an investigative probe into solution chemistry where reaction mixtures or complex mixtures of species can be investigated and meaningful results obtained. In some cases however the 5 130 134 138 142 146 150 niques must be used in order to evaluate a sample properly. The use of ES or ion spray coupled techniques has focused for the most part on biological c o m p o ~ n d s ~ ~ ’ ~ ~ however some groups have investigated inorganic systems.20 It is clear that ESMS can be a powerful probe for inorganic solution species. Quantitative Studies Preliminary Data To date there has been a decided lack of information regarding the quantitative aspects of ESMS.This is probably because of the subtle complexity of the technique. The concentration response is inherently non-linear and has been described by Agnes and Horli~k.~’ There is also a lack of CRMs for many anionic compounds such as sulfur species. As a result there is a real challenge to find ways to quantitate samples accurately n 20 $ 0 .I 3 100 I? 80 60 40 20 0 160 164 168 172 176 180 m/Z c sample could be too complex and therefore separation tech- .- (b) by ESMS. such work is Currently being invef&ated in this laboratory and an in-depth study will be presented in a future Fig. 9 Relative isotopic abundances of silver thiosulfate complexes (a) Ag(S,0,)2(H20),3-; and (b) Ag(S,O,),- 884 Journal of Analytical Atomic Spectrometry September 1996 Vol. 112.001 y = 7.4800e-2 + 3.0630e+4x 0 RA2 = 0.995 f C a (*1 c C .- j~u~fatel/mol I-' Fig.10 Log-log calibration curve for sodium sulfate in methanol by ESMS. The log of the sulfate intensity as ratioed against a 1 x lop4 moll-' KI internal standard response and plotted versus the log of the sulfate concentration paper. Some preliminary results of that study which are encouraging but by no means definitive are presented here. To achieve linearity in the calibration curve the analyte signal is typically ratioed against an internal standard. In addition to account for matrix affects which can be fairly severe in some cases the preferred method of quantification is standard additions. A calibration curve for sulfate is presented in Fig. 10. Harsh CID conditions were utilized thus the signal measured was that for HS04- at m/z 97 [see Fig.2(b)].Iodide (I- at m/z 127) was used as the internal standard and was present in all solutions at a level of 1 x lo- moll-'. The response was linear over the two orders of magnitude shown by the slope of the log-log plot being 1.01 & 0.02. The correlation coefficient was 0.998. The DL for sulfate based on three times the standard deviation of the background at m/z 122 was 1 pg 1-'. The background position at m/z 122 was selected as there is no ion present at this point under the conditions sampled and thus this value should reflect the detection limit of a sample not contaminated by sulfate. Sulfate is a fairly common anion and could be present as an impurity in reagents or solvents.Therefore great care must be taken in preparation of samples and standards. As an example a sample was run consisting of just 1 x moll-' iodide in methanol. The background sulfate concentration was found to be 3.60k0.26 pg 1-' based on the above plot. In either case the above results indicate the types of sensitivities obtainable by ES and detection limits could be blank limited and thus higher than the 1 pg 1-' indicated above. The applicability of this technique to a real-life sample was investigated. A waste-water sample was examined using ESMS. The sample was originally collected from a landfill site to examine a series of cations and anions in the waste-water. Sulfate was determined to be present at 6.10f0.61 x loW3 moll-' in the sample by an independent laboratory using ion chromatography.A mass spectrum of the sample was obtained (see Fig. 11) by diluting the sample 100-fold in methanol and with a sampling plate to skimmer potential difference of 35 V. Note the presence of CH,O- (m/z 31) and CH30C02- (m/z 75) which are typical background ions indicative of basic solutions. The presence of sulfate in the sample is indicated by the signals for HS04- and NaS0,- at m/z 97 and 119 respectively. Other anions observed include F - C1- NO2- and NO3-. A major problem associated with quantification by ES is that the response of the analyte and internal standard do not necessarily behave similarly upon addition to or variation of the sample matrix. Variation in the intensity of an ion with concentration has been discussed in the 1iteratu1-e.~~ It is therefore desirable to match the matrix as closely as possible 20 40 60 CH,OCO,- HSO NaSO,(H,O)- /I NalSo4- I 80 100 120 140 160 m/z Fig.11 Mass spectrum of a waste-water sample. The sample was diluted 100-fold in methanol and acquired with a sampling plate to skimmer potential difference of -40 V to minimize these effects. The standard additions technique should allow for this. Sample preparation involved diluting 0.33 ml of the sample with KI internal standard solution (1 x lo- moll-'). Prior to dilution aliquots of a standard Na,SO solution were added to the sample to afford the standard additions. The samples were then injected from a 0.50 pl sample loop into a flowing stream of internal standard (1 x mol 1-' KI) solution. The standard additions calibration curve is shown in Fig.12. The signal level used for sulfate was the sum of the intensities for HSO,- (at m/z 97) and NaS0,- (at m/z 119) and five replicates were used for each point. These values were then normalized to give the total sulfur concentration. Using least- squares calculations the sulfate concentration was determined to be 4.87 & 0.33 x lop3 moll-'. If this value is compared with the reported value of 6.10k0.61 x lo- mol I-' (determined earlier by an independent laboratory) it was found to be 20.2% lower then expected. The difference is fairly significant and could be due to several factors such as the original sample analysis could have been inaccurate the sample that was run 1 month after collection could have had its speciation altered during this time or the response or activity of the sulfate ion or adduct ions in methanol need not be the same as in aqueous solutions.The fact that the isotope ratio was assumed to be constant and therefore only the major isotope needed to be monitored could also have contributed to error in no small amount. Although the results are not accurate enough to seem 0.30 0.15 0.10 8 0.05 u o .- .- -20 iy = 5.98056-2 + 41 11 .lx iR"2 = 0.997 [Added sulfate]/prnol I-' Fig. 12 Standard additions calibration curve for sulfate ion in a waste-water sample by ESMS. The total sulfate intensity is ratioed against a 1 x lop4 moll-' KI internal standard response and plotted versus added sodium sulfate concentration Journal of Analytical Atomic Spectrometry September 1996 Vol.11 885analytically useful they are promising especially as these are only preliminary results. CONCLUSIONS The ability of ES to generate intact gas-phase ions representa- tive of solution ions opens up a whole new avenue for speciation work. In many ways it is an ideal source to couple with MS as it retains key chemical information about the nature of the elements in a sample that MS is capable of determining (i.e. molecular form and charge state) information that for example is generally destroyed when using ICP-MS. The sampling of these ions is for the most part understood however there are areas open for further fundamental studies which could concentrate on energy deposition in the desolv- ation process (e.g. ion kinetic energies as a function of CID energy and state of desolvation and fragmentation on the resultant energies).However it is also clear from the present study that for every determination considerable work must be carried out to understand fully the qualitative results of ESMS and in particular more work is required to define the concen- tration response of ions by ES in order to quantitate them accurately and properly. Financial support by the Natural Sciences and Engineering Research Council of Canada (NSERC) and the University of Alberta are gratefully acknowledged. REFERENCES Goldhaber M. B. Am. J. Sci. 1983 283 193. Singh R. P. and Nancollas G. H. J. Chromatogr. 1988,433 373. Bruins A. P. Covey T. R. and Henion J. D. Anal. Chem. 1987 59 2642. Takano B. McKibben M. A. and Barnes H.L. Anal. Chem. 1984,56 1594. Miura Y. Tsubamoto M. and Koh T. Anal. Sci. 1994,10 595. Steudal R. and Holdt G. Chrornatogruphia 1986 651 379. 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 Rabin S. B. and Stanvury D. M. Anal. Chem. 1985 57 1130. De Backer B. L. Nagels L. J. Alderweireldt F. C. and Van Bogaert P. P. Anal. Chim. Acta 1993 273 449. Geissler M. and Van Eldik R. Anal. Chem. 1992 64 3004. Wildman B. J. Jackson P. E. Jones W. R. and Alden P. G. J. Chromatogr. 1991 546 459. Koh T. and Taniguchi K. Anal. Chem. 1974 46 1679. Koh T. Miura Y. and Suzuki M. Anal. Sci. 1988 4 267. Koh T. Miura Y. Ishimori M. and Yamamuro N. Anal. Sci. 1989 5 79. Koh T. Takahashi N. and Yokoyama K. Anal. Sci. 1994 10 765. LaFreniere K. E. Fassel V. A. and Eckels D. E. Anal. Chem. 1987 59 879. Gjerde D. T. Weiderin D. R. Smith F. G. and Mattson B. M. J. Chromatogr. 1993,640 73. Olesik J. W. Kinzer J. A. and Olesik S. V. Anal. Chem. 1995 67 1. Agnes G. R. Stewart I. I. and Horlick G. Appl. Spectrosc. 1994 48 1347. Agnes G. R. and Horlick G. Appl. Spectrosc 1994 48 655. Huggins T. G. and Henion J. D. Electrophoresis 1993 14 531. Yamashita M. and Fenn J. B. J. Phys. Chem. 1984 88 4451. Hayati I. Bailey A. I. and Tadros T. H. F. J. Colloid Interface Sci. 1987 117 205. Kebarle P. and Tang L. Anal. Chem. 1993 65 972A. Stewart I. I. and Horlick G. Anal. Chem. 1994 66 3983. Cotton F. A. and Wilkinson G. Advanced Inorganic Chemistry. A Comprehensive Text Interscience New York 5th edn. 1988 p. 500. Blades A. T. and Kebarle P. J. Am. Chem. SOC. 1994,116 10761. Smith R. D. Wahl J. H. Goodlett D. R. and Hoftstadler S. A. Anal. Chem. 1993 65 574A. Smith R. D. Barinaga C. J. and Udseth H. R. Anal. Chem. 1988 60 1948. Agnes G. R. and Horlick G. Appl. Spectrosc. 1994 48 649. Paper 5/075068 Received November 16 1995 Accepted March 15 1996 886 Journal of Analytical Atomic Spectrometry September 1996 Vol. 1 1