434
Anal. Chem. 1992, 64, 434-443
(7) TWChl, C. S.;OlllS, D. F. J . Cetel. 1000, 722, 178-192. (8) Fox, M. A. I n photocetalyss and Ennvkonment: Trmds and Appllca ?ions; Schlaveilo, M.. Ed.; Klewer Academic Publishers: Dordrecht, The Netherlands. 1988; pp 445-467. (9) Matthews, R. W. J. Phys. Chem. 1987, 97, 3328-3333. (lG) Ollls, D.; Peiluetti, E.; Serpone, N. I n Photoce?a&sls: Fundemen?als and AppryCetions; Pelizretti. E.. Serpone, N., Eds.; John Wiley 8 Sons: New York, 1989; pp 603-637. (11) Kawai, N.; Kawal. T.; Sekldo. S. Chem. Abstr. 1086, 705, 17609q. (12) Hafeman. D. G.; Parce, J. W.; McConnell, H. M. Sclence 1088, 240, 1182-1185. (13) Fox, M. A.; Tien, T. Anal. Cbem. 1088, 60, 2278-2282. (14) Furtak, T. E.; Canfleld, D. C.; Parkinson, B. A. J. Appl. Phys. 1080, 57, 8018-8021. (15) Butler, M. A. J. Electrod". Soc. 1084, 737, 2185-2190. (16) . . Vercruvsse. 0.:Rlaoie. W.: Gomes. W. P. Sol. Enerw Mater. 1985. 72, 157-167. (17) Selavka, C. M.; Jiao, K.; Krull. I . S.; Sheih, P.; Yu, W.; Wolf, M. Anal. Cbem. 1088, 60, 250-254. Cbem. (18) Lacourse, LaCourse, W. R.; Krull, I. S.; Bratin, K. Anal. Cbem. 1085, 5 7 , 1810-1814. . - . . .- . . . (19) LaCourse, W. R.; Krull, I. S. Anal. Cbem. 1087, 5 9 , 49-53. (20) Stulik, K.; Pacakova, V. Cri?.Rev. Anal. Cbem. 1084. 14, 297-351. (21) Yeung, E. S.; Synovec, R. E. Anal. Chem. 1086, 58, 1237A-1256A. (22) Rowien, K. L.; Birks, J. W.; Dueli, K. A.; Avery, J. P. Anal. Cbem. 1088. 60. 311-318. (23) Rowien, K. L.; Duell, K. A,; Avery, J. P.; Blrks, J. W. Anal. Cbem. 1080, 87, 2624-2630. (24) Gelderb, D. G.; Rowlen, K. L.; Birks, J. W.; Avery, J. P.; Enke, C. G. Anal. Chem. 1088, 58, 900-903. (25) Wrighton, M. S. J. Cbem. Educ. 1083, 60.877-881. (28) Gerlscher. H. I n Physlcal Cbemlstry: An Ahranced Treatise; Eying, H., Ed.; Academic Press: New York, 1970; pp 463-542. (27) Gerlscher, H. I n Soler Energy Conversion. SolM-State Physics Aspects; Seraphln, B. 0.. Ed.; Springer-Verlag: Berlin, 1979; pp 115-170. (28) Finklea, H. 0. J. Cbem. Educ. 1083, 6 0 , 325-327. (29) Finklea, H. 0. I n S " d u c ? o r Electrodes; Finklea, H. O., Ed.; Eisevier Science Publishers: Amsterdam, 1988; pp 1-42. '
-
'
'
(30) Heller, A.; Degani, Y.; Johnson. J. W.; Gallagher, P. K. J. Phys. Chem. 1087, 9 7 , 5987-599 1. (31) Hartig, K. J.; Getoff, N.; Nauer, G. In?. J. Hydrogen Energy 1083, 8 , 603-607. (32) Hartig, K. J.; LichtscheMi, J.; Getoff, N. 2. Naturforscb A 1081, 36, 51-58. (33) Dorschel, C. A.; Ekmanls, J. L.; Oberhottzer, J. E.; Warren, F. V.; Bidllngmeyer, 8. A. Anal. Chem. 1080, 67, 951A-968A. (34) Murphy, T. D. Cbem. Eng. (N.Y.)1077, 84, 168-182. (35) Morgan, S . L.; Deming, S. N. Anal. Cbem. 1074, 46, 1170-1181. (38) Deming, S. N.; Parker, L. R. Cri?. Rev. Anal. Cbem. 1078, 6 , 187-202. (37) Carey, J. H.; Oliver, B. G. Nature 1978, 259, 554-556. (38) Gerstner, M. E. J . Electrocbem. Soc. 1070, 726, 944-949. (39) Finklea, H. 0. I n Semiconductor Electrodss; Finklea, H. O., Ed.; Eisevier Science Publishers: Amsterdam. 1988: OD 43-148. (40) Finklea, H. 0.; Murray, R. W. J. Phys. C&m. 1070. 83. 353-358. (41) Maclel. 0. E.; Slndorf, D. W. J. Am. Chem. SOC. 1080, 702, 7606-7607. (42) Maciel, G. E.; Sindorf, D. W.; Bartuska, V. J. J . Cbromatcgr. 1081, 205. 438-443. (43) Klssinger, P. T. J. Cbem. Educ. 1083, 6 0 , 308-311. (44) Klssinger, P. T. Anal. Chem. 1077. 49, 447A-456A. (45) Secrlst, D. R.; Mackenzie, J. D. Prepr. Pap.-Am. Chem. Soc., Dlv. FuelCbem. 1067, 1 7 , 203-210. (46) Hardee, K. L.; Bard, A. J. J. Electrochem. Soc. 1077, 724, 215-224. (47) Hardee, K. L.; Bard, A. J. J. Electrod". Soc. 1075, 122, 739-742. (48) Vlachopoulos, N.; Liska, P.; Auguslynskl, J.; eatzel, M. J . Am. Cbem. Soc. 1088. 170, 1216-1220. (49) Morisaki, H.; Ono, H.; Yazawa, K. J. Electrochem. Soc. 1084, 737, 2081-2086. (50) Spitler, M. T.; Cary, J. D. J. Electrochem. SOC. 1080, 736, 2295-2299. (51) Kautek, W.; Gorbrecht, J.; Gerischer, H. Ber. Bunsen-Gss. Phys. Cbem. 1080, 8 4 , 1034-1040.
RECEIVED for review January 17, 1991. Revised manuscript received October 25, 1991. Accepted November 1, 1991.
Interfacing Ion Chromatography with Particle Beam Mass Spectrometry for the Determination of Organic Anionic Compounds John Hsu' Hazardous Materials Laboratory, California Department of Health Services, 2151 Berkeley Way, Berkeley, California 94704
I n this paper we demonstrate the successful lnterfaclng of bn chromatography wlth partlcle beam mass spectrometry through the use of a membrane suppressor for continuous desaltlng. The mkrobore verslon of the membrane suppressor does not Introduce slgnlflcant peak broadening. The detection limn for SIXaromatic sulfonic acids In the full scan mode is better than 0.4 pg Injected oncolumn. The utlllty of a mhted-phase column, havlng both reversed-phase and Ionexchange capabliltles, Is also demonstrated In the chromatography of a hazardous waste sample. For the determinatkn of anlonlc compounds In partlcular, the partlcle beam mass spectrometer Is found to exhlblt an analyte carryover phenomenon, except under the cleanest condltlons. The contamlnatlon problem Is found to be Independent of the choke of column or the use or absence of a membrane suppressor.
INTRODUCTION To respond to real-life emergencies like the aftermath of a hazardous chemical spill, we wish to obtain analytical results 'Mailing address: P.O. Box 1255, Berkeley, CA 94701. 0003-2700/92/0364-0434$03.00/0
in a very timely fashion. To isolate a specific chemical from a complex and unknown matrix, we wish to have the most versatile analytical tools at our disposal. In high-performance liquid chromatography (HPLC), it is useful to have a chromatographic column that could separate analytes by several distinct mechanisms. An example of such a column is one that has both ion-exchange and reversed-phase capabilities. The ability to interface such a multiphase column to a mass spectrometer to provide molecular weight or structural information is also attractive. Particle beam mass spectrometry (PBMS), in particular, offers the convenience of generating standard library-searchable 70-eV electron-impact (EI) mass spectra. This paper reports on the fiist successful interfacing of such a multiphase column with PBMS through the use of a membrane suppressor, for the analysis of organic anionic analytes. The cation-exchanger-based membrane suppressor, when connected to the effluent side of the column, continuously removes the metallic counterions (Na+) of the ionic eluants (e.g., OH-), to avoid their subsequent buildup in the mass spectrometer ion source. A membrane suppressor has been used in previous mass spectrometric applications for desalting between an anion-exchange column and a thermospray LC/MS interface ( I ) and between ion chromatography columns and an ion spray LC/MS interface (2). The 0 1992 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 64, NO. 4, FEBRUARY 15, 1992
latter study has utilized cation suppressor as well as anion suppressor in the analysis of both organic anions and cations. The versatility of the multiphase column we use here has been demonstrated previously (3) by the analysis of alcohols, heterocyclic amines, mono-, di-, tri-, and tetracarboxylic acids, and sulfonic acids as well as inorganic anions. The areas of applications range from water, wastewater, food and drinks, pharmaceuticals, and biological samples, to industrial discharge, hazardous wastes, ground water leachates, etc. The column can separate organic anions which structurally differ only slightly in the hydrophobic moiety (3). In this paper, we shall illustrate its use in the analysis of leachate samples from two hazardous waste dumpsites. At one of the sites (Stringfellow) where DDT and sulfuric acid wastes were dumped, p-chlorobenzenesulfonic acid (PCBSA) has been found to be a principal contaminant ( 4 , 5 ) . Aromatic sulfonic acids are therefore used as the model compounds in the current study. The sulfonic acids are relatively strong acids with low pK,L. Because of their high polarity, they are among the more difficult to analyze by HPLC, as the multiply charged compounds tend to produce broad tailing peaks. To match the optimal liquid flow rate of 0.4 mL/min or less for the PBMS, prototype, microbore versions of the analytical column (2.2 mm i.d.) and membrane suppressor are constructed. The gradient pump is also modified to deliver liquid a t the reduced flow rate. In the following, we shall demonstrate the analytical versatility of the column, report on the overall sensitivity of detection for aromatic sulfonic acid standards by the combined multiphase column-membrane suppressor-PBMS system, the quality of the mass spectra obtained, and the result of hazardous waste sample analysis, and discuss the experimental artifacts observed with PBMS and the precautions to keep in mind.
EXPERIMENTAL SECTION Reagents. The deionized water is obtained from a Milli-Q reagent water system (Millipore, Milford, MA). The HPLC solvent acetonitrile (ACN) is of glass-distilled OmniSolv grade from EM Science (Cherry Hill, NJ). The aqueous 50% (w/w) NaOH solution is Fisher Scientificcertified reagent. The aromatic sulfonic acid standards are of technical or higher grade. Benzenesulfonic acid, Na salt is purchased from Fluka (hnkonkoma, NY),p-chlorobenzene sulfonic acid, Na salt from Pfaltz & Bauer (Waterbury, CT), p-bromobenzenesulfonic acid, Na salt from Chem Service (West Chester, PA), and p-xylene-2-sulfonicacid from Sigma (St. Louis, MO). The remaining aromatic sulfonate standards are obtained from Aldrich Chemical Co. (Milwaukee, WI). Both the HPLC solvents and the standard solutions, prepared in water, are filtered before use. HPLC Systems and Chromatographic Conditions. Two HPLC pump systems have been used. The Hewlett-Packard Model 1090 HPLC system handles noncorrosive solvents. It is equipped with an autosampler and is used to make multiple flow injections of caffeine in a sensitivity check for the PBMS system. It is also used to locate the optimal PB nebulizer capillaryposition for each analytical standard; this is also performed by flow injections. The Dionex gradient pump can handle eluants over a wider pH range. After degassing, the organic and ionic eluants are stored in separate reservoirs pressurized under helium atmosphere in the ‘Eluant Degas Module” (of Dionex Corp., Sunnyvale, CA). The manual injector has a 5-pl sample loop. The analytical column used in this study is a prototype, 2.2 mm i.d., 25 cm long OmniPac PAX-500 column, supplied by the Dionex Corp. The mobile-phase flow rate is kept constant at 0.25 mL/min. A fixed-wavelength UV detector from Dionex Corp. utilizes a 254-nm band-pass filter to select the wavelength. Two principal seta of chromatographic conditions have been used. The first one uses a simultaneous acetonitrile and NaCl gradient. In the NaCl eluant, NaOH is added in small amount to maintain an alkaline pH. NaCl is effective in displacing or eluting a variety of analytes of different valency from the column and is therefore useful for general sample screening with a UV detector. A membrane suppressor is not used in this case. The
435
Table I. Typical Chromatographic Conditions for Use with the OmniPac PAX-500Column 1. SimultaneousACN and NaCl Gradient
time, min
% ACN
[NaCl],”M
18 0.03 5.0 18 0.03 35.0 45 0.30 2. SimultaneousACN and NaOH Gradient 0.0
time, min
ACN
[NaOH],M
0.0
10
8.0 13.0 18.0
20
0.009 or 0
%
27 40
0.05 0.08 0.12
“The 1M NaCl eluant used contains 4 mM of NaOH. second set of conditions uses a simultaneous acetonitrile and NaOH gradient. NaOH is most effective in displacing monovalent anions from the PAX column, is more effectivelyremoved by the membrane suppressor than NaCl, and is more suitable for use as an eluant in conjunctionwith PBMS applications. The second set of conditions is the one used here to obtain PBMS data. For gradient elution, typically there is an initial hold, followed by a 15-30-min ramp to the final condition (see Table I). Membrane Suppressor. Constructed by Dionex Corp., the prototype, microbore version (“AMsMS”) of the membrane suppressor (“anion micromembrane suppressor”) (6) comprises a central channel into which the HPLC effluent flows, separated by membranes from an outer channel in which an acidic regenerant flows in countercurrent direction. A ‘gCTeen” is placed within each channel to disrupt laminar flow, so as to ensure adequte contact of the liquid with the membrane. Both the Teflon-based membrane and the screen have cation-exchangefunctionalgroups affixed to them, to effect a more thorough cation removal from the central channel. The void volume of the inner channel is only 10 pL. The maximum back-pressure the membrane suppressor can withstand is about 100 psi. The exchange capacity of the suppressor is 88 Fequiv/min, or 350 mN Na+ (as NaOH) at 0.25 mL/min flow rate in the central channel. The exchange capacity is lower for larger, organic cations. Prior to incorporation into the suppressor, the functionalized membrane is extracted in a Soxhlet using organic solvents to remove foreign organic material which may reside on the membrane during manufacturing or processing. The regenerant used for the membrane suppressor is 0.02 N H$04, flowing at approximately13-20 mL/min. It is recirculaM continuously through a Dionex Corp. “Autoregen” unit which removes the cations exchanged out of the membrane suppressor and regenerates the H2S04regenerant in a closed loop. A conductivity detector from Dionex Corp. is used to monitor the effectiveness of the membrane suppressor in desalting and to measure the signal of the analytes following membrane suppression. When the ionic eluant is alkaline (e.g., NaOH), we can also check on the effectiveness of the ion suppression (in converting the alkaline NaOH to the neutral HzO) by pH paper. Mass Spectrometer and Particle Beam Interface. A schematic of the PB interface and the mass spectrometer ion source is shown in Figure 1. The features unique to our system and relevant to later discussion are described below. The mass spectrometeris a Hewlett-PackardModel 5988A, with a switchable EI/CI ion volume and two sample inlet porta at two opposite sidea of the ion source portion of the vacuum manifold. On switching from E1 to CI mode operation, a “plunger assembly” in the ion source is advanced t o w d the quadrupole m&98 analyzer, changing the effective ionization chamber from a more open to a more enclosed configuration. The ion volume is not of the type that can be removed or replaced independently of the rest of the ion source block. The particle beam interface is situated on the left side of the manifold, and a direct insertion probe (DIP) on the right. In this study, a solid dummy DIP tip with a flat surface toward the ion source is used. The DIP and ita removable tip can be withdrawn from the manifold for cleaning without breaking
436
ANALYTICAL CHEMISTRY, VOL. 64, NO. 4, FEBRUARY 15, 1992
Fused Silica
Nozzle
Two Skimmers
To Lenses and Quadrupoles
t
LC Effluent +
Desolvation Helium First Pump Stage
Second Pump Stage
U Plunger
Figure 1. Schematic of the particle beam interface and the mass spectrometer ion source. The drawing is not to scale.
the high vacuum. The DIP and the ion source block are heated (to 250 “C) independently by separate circuitry. The particle beam interface, Hewlett-Packmd Model 59980A, consists of a nebulizer, a desolvation chamber, a nozzle, followed by two skimmer cones. At the inlet to the nebulizer is a stainless steel frit. The position of the axial capillary in the nebulizer, i.e. the distance by which the capillary extends out of the nebulizer housing, can be adjusted by a micrometer. The desolvation chamber is kept at a moderately warm temperature (45 OC) by a heating jacket just outside the chamber, but the nebulizer is not independently heated. The compartmentsbetween the nozzle and the skimmer cones are evacuated continuously by two mechanical pumps. The PB nozzle and skimmer cones are made by Hewlett-Packard Co. with a proprietary liquid-crystalline material (plastic in appearance). They are replaced or cleaned by solvent at periodic intervals, depending on the degree of wear or the ability to hold vacuum in the proper pressure range. A transfer tube, about 1 ft. long, is located between the PB interface and the mass spectrometer ion source. In E1 mode of operation, the transfer tube is advanced to about 2.5 mm away from the sample cup at the inlet to the ion source, while in CI mode the tube is fully advanced. The PBMS system is tuned and mass-calibrated daily with PFTBA. The sensitivity check is done by multiple flow injections of a caffeine solution in methanol (50 ng of caffeine per injection). The electron multiplier voltage is adjusted to produce approximately equivalent signal from day to day. During the CI mode of operation, isobutane is used as the reagent gas. Hazardous Waste Samples and the Preparation of Sample Extracts. At the Stringfellow hazardouswaste dumpsite, aqueous leachate or groundwater samples were collected from near the dumpsite (sample “A”),further downstream (sample “B”),and after charcoal treatment of the water at the site (sample ”C”). The Casmalia site samples are more variable in composition since there are many storage drums containing many different types of hazardous wastes at that location. The procedurefor extraction of the principally aqueous samples by solvents has been described previously (5). Briefly, the procedure is as follows: 500-2000 mL of the sample is lyophilized to remove the water and the volatile organics. The residue is then dissolved in methanol. Acetone is added slowly to the methanol solution to precipitate out the inorganic salts. After filtration of the supernatant, the filtrate is concentrated down by evaporation under reduced vacuum. The inorganic salt precipitation step (adding acetone) may need to be repeated a few more times during the extract concentration process. After filtration, the concentrate in methanol is used directly for LC/MS analysis. PRINCIPLE Mixed-Phase OmniPac PAX-500 Column. The core of the PAX column resin particles, approximately 8 pm in diameter, is made of ethylvinylbenzene-based polymer (7). The polymer has been extensively cross-liked to minimize swelling by organic solvents. There are pores and winding paths extending deep into the core. The reversed-phase aspect of the chromatographic action takes place on these surfaces. The outside surface of the resin particles is coated with a uniform
layer of finer particles made of basically the same polymeric material which has, however, been functionalized with quaternary amine groups. It is at these surface groups that the anionic exchange aspect of the chromatographic action takes place. Membrane Suppressor. The countercurrent flows in the central and outer channels maintain the favorable concentration gradients which are the driving forces for cation exchange across the membranes. The gradients are typically a strong Na+ concentration differential and a pH gradient. While Na+ diffuses out, H+ diffuses into the central channel, combining with OH- to form the neutral H20. In the meantime, anionic analytes continue on in the central channel. The ionic eluants NaOH, Na2C03,and NaCl entering the suppressor are converted to HzO, HzC03,and HCI, respectively. NaCl is less effectively suppressed than NaOH or Na2C03, however, since the HC1 formed and the H2S04regenerant on the two sides of the membrane are both strongly acidic; thus a strong pH gradient across the membrane is no longer present. While the membranes offer some selective retention based on size, some neutral compounds may diffuse across the membranes driven by their own concentration gradients. Unlike their Na salts, the HzO, HzC03,and HCl formed in the membrane suppressor are volatile compounds which are pumped away in the particle beam interface and in the mass spectrometer and will not deposit in the ion source. Particle Beam Interface. The principle of operation of the PB interface has been described previously (8),but there are design differences in the hardware. In brief, effluent from the LC instrument or membrane suppressor is passed through a capillary and nebulized at the exit end by a fast surrounding flow of helium gas. The dispersion of helium, vapor, and liquid droplets formed is then passed at high speed through a nozzle into two successive stages of differentially pumped vacuum compartments. While the solvent further evaporates, the involatile solutes concentrate further down in the droplets and aggregate to form solid particles. The central beam of particles finally traverses the length of the transfer tube at near sonic speed to reach the mass spectrometer ion source which is housed in a high-vacuum manifold maintained a t about Torr. Solvent molecules may not have been completely removed from the particles a t the time the particles reach the mass spectrometer ion source. RESULTS AND DISCUSSION The ultimate objective of this study is to demonstrate the feasibility of interfacing the multiphase column with PBMS through the use of a membrane suppressor and then to check on the performance of the integrated system. We begin by testing the performance and the range of application of the prototype, microbore versions of the analytical column and the membrane suppressor. Then we wish to answer the following questions: Can the suppressor withstand the back-
ANALYTICAL CHEMISTRY, VOL. 64, NO. 4, FEBRUARY 15, 1992
497
CASMALIA I
10
0
30
20
TIME (min)
Figure 2. Separation of mono- and multivalent aromatic sulfonic acids
on an Omni-Pac PAX-500 column using a simultaneous ACN and NaCl gradient and UV detection at 254 nm.
r'
34.73'
I
0
7.61'
I
I
TIME
"A" 19.21'
"C"
h
0
. 20.
I
I
I
40
TIME (min) Flgure 3. Chromatography of the extracts of three Stringfellow aqueous leachate samples on a PAX-500 column using the simultaneous ACN and NaCl gradient of Table I and UV detection at 254 nm.
pressure introduced by the P B interface? Does the mass spectrometer detect any column or suppressor bleed material
I
20
I
I
40
(min)
Flgure 4. Chromatography of the extract of a Casmalia hazardous waste leachate sample on a PAX400 column uslng the simultaneous ACN and NaCi gradient of Table I and UV detection at 254 nm.
which is not detectable by the UV detector? How is the mars spectral quality and variability? What are the detection limits for the anionic organic analytes? Can we enhance the sensitivity of detection, say, by the addition of mobile-phase modifiers (9)? In the course of this study, the greatest difficulty we encountered, however, is seeing unexpected ions in the mass spectra collected. There is a contamination or analyte carryover problem in PBMS. Other investigators have alluded to similar difficulties ( 1 0 , I I ) . A lot of effort is then put into studying the nature of the contamination and in identifying the underlying processes or mechanism. This understanding has helped us to troubleshoot the instrument, to eliminate variables systematically, and to obtain mass spectra as free from contaminants as possible. The principal findings from this study are presented below. Multiphase Column Chromatography. Generally, we find the chromatography on the microbore and normal bore (3) versions of the PAX-500 column very similar. Na2C03 displaces a wider range of multivalent anions than NaOH, while NaCl is best for eluting the aromatic sulfonic acids of all valency. The range of separations achieved with a simultaneous acetonitrile and NaCl gradient (Table I) is shown in Figure 2, where the analytes are present at about 1pg each. First eluting are the monosulfonic acids, followed by the disulfonic acids, and finally a compound with three active functional groups. Following the solvent peak which shows a negative spike, the compounds in their elution order are the following: benzenesulfonic acid, p-chlorobenzenesulfonic acid, 1-naphthalenesulfonicacid and an impurity associated with it, 4hydroxybenzenesulfonicacid, 1,3-benzenedisulfonic acid, 1,5naphthalenedisulfonicacid, 2,6-naphthalenedisulfonicacid, and 2-hydroxynaphthalene-3,6-disulfonic acid. We fiid that the NaCl gradient cleanses the column well between runs. Applying the same gradient to the Stringfellow and Casmalia hazardous waste sample extracts, we obtain the chromatograms shown in Figures 3 and 4. The Stringfellow samples show excellent peak resolution. The three samples contain essentially the same constituents but in different proportions. We note that there is some dependence of the retention time on the concentrations of the analytes. The Casmalia sample is found to contain more earlier-eluting compounds than the Stringfellow samples. The sample is suitable for detailed study by the NaOH gradient.
438
ANALYTICAL CHEMISTRY, VOL. 64, NO. 4, FEBRUARY 15, 1992
FINAL % ACN/1 M NaCl 45/30
,
54/20
63/10
1
\
\
kl
66/7
67/6
i,
I
I
I
i0 TIME (min)
I
QO
Figure 5. Effect of increasing the reversed-phase component while decreasing the ion-exchange component of chromatographic separation on peak resolution for the Stringfellow sample A.
The advantage of having combined reversed-phase and ion-exchange capability in one column, a unique capability of the mixed-phase column, is demonstrated in the resolution of the late-eluting components in Stringfellow sample A. As shown in Figure 5, by utilizing a progressively stronger organic solvent gradient in conjunction with a progressively weaker
NaCl salt gradient, the labeluting bundle is resolved into four distinct peaks! For the chromatography in Figure 5, the initial condition is 18% ACN and 30 mN NaCl for all traces; the final conditions, for the top to bottom traces, are 45%, 54%, 63%, 66%, and 67% ACN, and 300,200,100,70, and 60 mN NaC1, respectively. The miscibility of ACN with the aqueous NaOH limits the highest eluant strengths we can go to. The final conditions used are at close to the miscibility limits. For the top trace, the initial hold is 0.4 min and the linear gradient is 20 min long. For the remaining chromatograms,the initial hold is 5 min and the gradient, 30 min long. Membrane Suppressor. The effectiveness of ion suppression by the microbore version of the membrane suppressor is verified by both pH and conductancemeasurements. When NaOH or Na2C03are the ionic eluants, the pH of the effluent remains constant at 5 throughout the gradient range utilized. The pH change accompanying the membrane suppression may affect or shift the solubility, absorption maxima as well as the absorptivity of the analyte. We have observed such change in the UV absorption spectrum with 4-nitrophenol. By adding a membrane suppressor before the UV detector, we find that the membrane suppressor introduces only very slight chromatographic peak broadening. When connected to PBMS, the membrane suppressor is able to withstand the back-pressure introduced by the PB interface without leakage. We do not detect any bleed from the membrane suppressor into the mass spectrometer. While all anionic compounds survive passage through the membrane suppressor, we find that caffeine does not. A possibility is that the nitrogens on the caffeine become protonated at the slightly acidic pH; the caffeine thus possesses positive charge and is treated as a cation. The loss of some neutral compounds (sugars) in the normal-bore version of the membrane suppressor has been previously reported (I). If only volatile eluants are used, we can use mass spectrometric detection with or without a membrane suppressor to study this phenomenon. However, membrane suppressors are used (2) primarily for the study of ionic compounds. We have not found any loss of the anionic organic compounds in this study. Particle Beam Interface. While higher organic mobilephase (ACN) composition is known to favor PBMS detection, a setting of 13.5 for the micrometer of the nebulizer is found to be generally adequate for all the aromatic sulfonate standards studied, across the organic solvent gradient range from 10% to 40% ACN. The optimal He pressure for the nebulizer is 50 psi. In Figure 6, we compare the chromatographic traces obtained by using three different types of detectors. We find no peak broadening attributable to the PB interface beyond that introduced by the membrane suppressor. For the measurements in Figure 6, a membrane suppressor is in-line between the column and the detector in all cases. For the top two traces, the UV detector is connected in series with the conductivity detector and preceding it. For the bottom trace, the PBMS system, operated in E1 mode and scanned from m / z 50 to 400, is the only detector. The compounds eluting after the solvent peak are benzenesulfonic acid (at 1.3 pg loading), p-toluenesulfonic acid (1.1pg) and an impurity associated with it that appeared as a shoulder peak in the PBMS trace and as a larger one in the conductivity trace, p-xylenesulfonic acid (1.4 kg), p-chlorobenzenesulfonicacid (1.5 pg), p-bromobenzenesulfonicacid (1.0 pg), and three very closely eluting compounds, each at approximately one-tenth of the above loading (0.1 pg): 4-hydroxybenzenesulfonic acid, 1naphthalenesulfonic acid, and 2-naphthalenesulfonic acid. The last three are much stronger chromophores at 254 nm compared to the earlier-eluting compounds.
ANALYTICAL CHEMISTRY, VOL. 64, NO. 4, FEBRUARY 15. 1992
II I I\
0.8 ug
L
:d
::I
499
L
n I\
I.
0.4 ug 1
0
,
1
1
1
1
10 20 TIME (min)
FIgm 6. Comparison of the chromatograms obtained by UV 254 nm, CondUctMty, 01 PBMS detection of a mixtwe of eigm a r m t i c sulfonic aclds, using a simuttaneous ACN and NaOH gradient, on a PAX-500 column. The membrane suppressor is in-line in all cases.
Sensitivity of Detection by the Combined Multiphase Column-Membrane Suppressor-PBMS System. Several dilutions of a standard mixture of six aromatic sulfonic acids are chromatographed with a simultaneous acetonitrile and NaOH gradient (Table I). The total ion chromatograms (TICS) obtained in full scan mode ( m / z 50-400) are shown in Figure 7. The loading levels for each analyte, from the top trace down, are 1.5, 0.8,0.4,0.2, and 0.1 pg, respectively. The anal* are benzenesulfonic acid, p-toluenesulfonic acid, p-xylenesulfonic acid, p-chlorobenzenesulfonic acid, 4hydroxybenzenesulfonic acid, and 2-naphthalenesulfonic acid. At 0.1 pg of each analyte, the TIC peaks are still distinctly recognizable. However, the molecular ions (which are not base peaks) are pronounced only in the 0.4% range and up-under the chosen data acquisition conditions. Contamination Problem in PBMS and the Effect on Mass Spectral Quality. The nature of the contamination has been studied at some length. The discussion below includes the following: the observed signs of contamination in the mass spectra; the types of contaminant ions observed; visual indications of PBMS hardware contamination; possible mechanisms of contamination, by examining the fate of the particles; and suggested PBMS user precautions. The suggested contamination processes are by no means exclusive or definitive, but they are consistent with our observations. At microgram levels of analyte loading, we find that the mass spectrometer is easily contaminated. Basically, the signs of contamination as seen in the mass spectra are the following. (A) Foreign m/z ions rise and fall in intensity along with the fragment ions of the analyte eluting, showing very similar elution profiles. At the baseline between the (well-resolved) chromatographic peaks, the background mass spectra are relatively free from the contaminant ions. Thus the contaminant ions appear only when there are incoming particles, ionizing in the mass spectrometer ion source along with them
....]
I
. .
0
m
IO 20 TIME (min)
Flgure 7. PBMS detection of a mixture of six aromatic sulfonic aclds at several bading levels, ctromatogaphed with the slmuttaneorcsACN and NaOH gradient of Table I.
at the same time. Likely, newly formed particles are physically dislodging the previously injected analytes deposited somewhere in the PBMS system. Less likely, the contamination is due to a continuous, gradual bleed, since such bleed would result in long tailing chromatographic peaks. Because the region between the peaks are rather clean, the mass spectra cannot be “corrected” by background subtraction. (B) The same contaminant ion is not associated with all analytes, suggesting that there is some dependence of the carryover phenomenon on the nature of the incoming or
440
ANALYTICAL CHEMISTRY, VOL. 64, NO. 4. FEBRUARY 15, 1992
resident analytes. The size, density, composition, and polarity of the newly formed particles can determine their ability to break loose or carry along formerly deposited particles or analytes. (C) When we look at the contamination of the spectrum of one particular analyte, we find the intensity of the contaminant ion observed to be roughly proportional to the amount of the analyte injected, suggesting that the contaminant is present in large excess and is “supplied upon demand”. The types of contaminant ions detected belong to the following categories: the molecular or fragment ions of the (aromatic sulfonic acid) standards previously injected, ions of compounds present in previously injeded samples, and ions of volatile buffer or ion-pairing agent previously used or presently being used in the mobile phase. The ions arising from the ammonium acetate buffer or ion-pairing agent (e.g., mlz 59,60,61) can be present as a very substantial fraction (e.g. 50%)of the TIC intensity. We note that the PB pumps cannot effectively remove the polar organics immediately and completely,as indicated by the acetic acid odor which could linger for days in the PB desolvation chamber. Since the pressure at the inlet to the PB pumps is generally from 0.1 to 0.5 Torr, the acetic acid vapor could be present at up to this pressure in the PB interface. Once the volatile organics are carried along by the particles into the mass spectrometer ion source which is maintained at 250 “C, they are expected to vaporize immediately, thus contributing instantaneously to the observed TIC intensity as a high percentage of it, giving the appearance of a “signal enhancement”. Such an increase in TIC intensity does enhance the detectability of the analytes in the sense of signaling us that some particles or analytes are reaching the mass spectrometer ion source. Since nanograms of many compounds are sufficient to give mass spectral signals, the fact that micrograms of samples are injected indicates that most of the analytes have not reached the electron multiplier detector of the mass spectrometer. If not entirely degraded, they must be residing somewhere. Anywhere along the path of the particles are places where the contaminants could possibly reside (see Figure 1). The visual signs of contamination of the PBMS hardware include the following. Most frequently, white deposits (the color of most of the aromatic sulfonic acids used) are found at the tip of the skimmer cones. The rim of the skimmer orifice is sometimes rough after some usage and appears damaged or discolored, or incoming particles appear to have fused with the skimmer cone material. Occasionally, white deposits are also found at the tip of the nozzle cone. We have not seen white deposits at the tip of the nebulizer capillary, but viscous material could linger there or at the nebulizer inlet frit. The result which we have seen in such an instance is a tailing contaminant ion profile superimposed on top of well-resolved, sharp specific analyte ion peaks to yield broad, tailing TIC peaks. In a narrow stretch at the end closer to the ion source, the inside bottom surface of the transfer tube occasionally has white deposits. These particles could be particles that did not travel far enough to reach the ion source or particles that bounced back after hitting the inlet cup of the source block. With use, the sample cup, the source block, and the channels and orifices within the block become coated with dark-colored charred or pyrolyzed organic material, which is not white or particulate in appearance. Not infrequently, a layer of very fine particles, approximately uniform in size to the eye, is found coating half of the flat surface of the dummy DIP tip, the half that is in the direct path of the incoming particles. This half is not blocked by the plunger assembly which in the E1 mode of operation
protrudes halfway into the left-to-right channel of the source block for the passage of the incoming particles. The color of the particles can be close to white or lightly brownish but generally not as dark as the pyrolyzed deposita on the source block. These particles on the DIP tip surface are also easier to clean off, suggesting that the heating of the DIP tip is less effective or less well regulated or that partial cooling of the DIP between runs allows the particles to survive. At times, the entire DIP tip surface facing the ion source is also found to be coated with a rather uniform layer of particles or deposit. Since half of the surface is not in the direct path of the incoming particles, multiple collisions probably occurred-if the particles were relatively dry. If wet, the incoming particles may deform and spread the analytes over a larger surface area. The passage way for the particles is quite uneven in the ion source with changing cross sections and with cavities of different sizes. In the CI mode, most incoming particles are blocked off by the fully protruded plunger assembly. Only a small fraction can enter the ionization chamber through a small hole in the plunger wall. At high analyte loading levels, the CI mass spectra obtained are thus even more seriously contaminated than the E1 spectra. To reduce the contamination problem, we have tried the following. Avoid overloading the mass spectrometer in the first place. Develop chromatography to produce very sharp peaks so that less analyte can be injected and, at the instant the analyte reaches the mass spectrometer, it is still present at high enough concentration to be detectable by the mass spectrometer. Flush the column and membrane suppressor with strong ionic and organic eluants before use to remove residues. Help to clean the PB interface by flushing the PB interface with solvent between runs. Make sure that the PB pumps are purged effectively. Watch out for early signs of contaminationduring a run by monitoring specific ion profiles. Visually check for contamination in the PB interface, especially on the skimmer cones, in the transfer tube, and on the DIP tip surface frequently, and clean or replace them as needed. Additional considerations include heating the DIP to a higher temperature to help volatilize the compounds. With or without a removable and replaceable ion volume, cleaning the ion source is still absolutely necessary. Having a spare ion source assembly and having isolation valves between the diffusion pumps and the high-vacuum manifold will decrease instrumental downtime. Experimentation with different manners of aerosol generation (8)and other types of PBMS designs (12) could also be helpful. The presence of the extraneous mass peaks has been found under all but the cleanest PBMS conditions during the analysis of the anionic analytes. We fiid it to be independent of the choice of the analytical column and the use or absence of the membrane suppressor but dependent on the history of usage of the PBMS system (in either the E1 or CI mode). The contamination appears to be most serious with the type of very polar compounds we are analyzing. Some of the cleanest mass spectra we ever obtained for the aromatic sulfonic acid standards are shown in Figure 8, where the ions are displayed from m / z 65 up. The four compounds are benzenesulfonic acid, p-toluenesulfonic acid, p-xylenesulfonic acid, and 2-naphthalenesulfonic acid. For p-xylenesulfonic acid, the molecular ion (mlz 186) has also been observed a t higher relative abundance than shown. The maw spectral variability is illustrated in Figure 9, which displays the mass spectra obtained for p-chlorobenzensulfonic acid on different days from either a sample (Stringfellow C, top spectrum) or standard (the remaining spectra) under relatively clean mass spectrometer conditions. Manual background subtraction has been performed in all cases. Known spurious ions in the mass spectra include mlz 104 and
ANALYTICAL CHEMISTRY. VOL. 64, NO. 4, FEBRUARY 15, 1992
441
Ill
0
z a n z 3 m a
mw 186
36,
70
W
L
3
4 w
U 17’
i’ a Y s 0 3 H
mw 172
Figure 8. E1 mass spectra obtained for four aromatic sulfonic acids under relatively interferencefree condltlons.
186 arising from p-xylenesulfonic acid, m / z 94 and 158 from benzenesulfonic acid, and m / z 133 from Cs+. The origin of the m/z 73 ion is unknown. The relative intensities of different mass peaks also vary. So as not to be misled in the interpretation of the mass spectra, it is important to check routinely for such variability. Comparison of the mass spectra in Figure 9 with that obtained previously for PCBSA (5)indicates that the m / z 79 peak in the previously published spectrum is a contaminant ion. The m / z 79 ion is one of two principal mass peaks of a compound that elutes before PCBSA in the Stringfellow sample (peak no. 2 in Table IV of ref 5). It is a recurring contaminant ion which we have observed from time to time. Analysis of Hazardous Waste Samples. By the same ACN and NaOH gradient used above for the standards, the extracts of the samples collected from the Stringfellow and Casmalia hazardous waste dumpsites are also analyzed with the combined multiphase column-membrane suppressorPBMS system. The chromatogramsobtained by W detection at 254 nm and by PBMS operated in the E1 mode are compared in Figures 10 and 11.
Stringfellow Samples. Comparison with Earlier Studies. In the earlier studies of the Stringfellow samples by anion-exchange chromatography (4, 5), a strong anionexchange “SAX”column from SGE (Ringwood, Australia) was utilized. While the acetonitrile gradient is ramped up, the concentrationof the volatile buffer, ammonium acetate, is held constant. On the basis of PBMS as well as UV detection at 230 and 265 nm, the presence of PCBSA, ita isomers, and some sulfonated DDT derivatives were confirmed or suggested. Compared to the other sulfonated compounds, PCBSA is
present a t considerably higher concentration. In the current study, we are using the same PBMS system, but a different batch of samples, and a different wavelength (254 nm) for W detection. When operated in principally the ion-exchange mode, the multiphase column is expected to produce chromatograms roughly equivalent to that produced by a purely anion-exchange column, where the elution order is roughly indicative of the strength and the number of active functional groups on the analytes. A comparison of the chromatograms in Figure 2 with those obtained by the SAX column (5) do show generally similar features with similar grouping of peaks. The difference in the relative intensity of the W absorption peaks can be accounted for by the sample batch difference as well as by the difference in the wavelength chosen for UV detection. As the reversed-phase component of the separation is increased (Figure 5), the order of elution shifts. The pattern of change in the elution order can be very complex and is very dependent on the types of constituents present in the sample (3). For Stringfellow samples, a comparison of the chromatographic separation by the multiphase column to those obtained by reversed-phase or anion-exchange columns will be the subject of discussion of a future publication. In agreement with earlier findings, the PBMS analysis performed in this study with NaOH as the ionic eluant (Figure 10) deteded three chromatographicpeaks (at around 10,16.5, and 23.5 min after injection) with very similar chromatographic profiles (broad tailing peak) and identical mass spectra. They correspond to PCBSA and its ortho and meta isomers, as tentatively assigned previously (5). Peak no. 2 in
442
ANALYTICAL CHEMISTRY, VOL. 64, NO. 4, FEBRUARY 15, 1992
c
75
100
A
I 0 * O 3 H
!
111
i 4'
m w 192
STRINGFELLOW b 11m11
192
d
twq 128
i
75
8
i
4
111
128
i
i
.
192
0
10
20
(min) Fkpe 10. Comparison of the chromatogaphic traces by UV or PBMS detection In the E1 mode for Stringfellow sample C using the simultaneous ACN and NaOH gradient of Table I. TIME
CASMALIA
M/z Flgure Q. Comparison of the mass spectra of pchlorobenzenesutfonic acid obtained on dlfferent days or at different concentrations. The top spectrum is obtained from the analysis of Stringfellow sample C, and the others are from the analytical standard.
I)
254 nm
Table 11. Electron Impact Mass Spectra of Compounds Detected in the Casmalia Sample peak no.
m/z of principal mass peaks (re1 intens)
1 2 3 4 5 6
60 (33), 61 (100) 70 (30), 74 (loo), 115 (97) 61 (26), 79 (loo), 96 (23) 59 (loo),80 (87), 82 (74) 59 (loo), 64 (31), 81 (29), 98 (26) 74 (31), 76 (loo), 104 (loo), 148 (23)
Table IV of ref 5 corresponds to the doublet eluting just before PCBSA in Figure 10. In an attempt to detect the group of minor components eluting after the PCBSA group, we have extended and doubled the NaOH gradient range. By UV detection, all the lateeluting peaks seen in Figure 5 are also detected, but the peaks are broader. The PBMS run yielded little additional information, however, since although these compounds are strong chromophores, they are present at below the limit of detection of the mass spectrometer. To detect these late-eluting compounds, the previous study utilizing the SAX column has resorted to overloading the column, with some loss of chromatographic peak resolution. By this process, however, we risk contaminating the mass spectrometer excessively while obtaining unreliable mass spectra. An alternative is to do preparatory scale sample fractionation prior to loading of the appropriate fraction to the LC/MS system. We intend to pursue this investigation at a later time.
0
10
20
TIME (min) Fkpe 11. Comparison of the chromatographictraces by UV or PBMS detection in the E1 rode for a Casmalla sample, using the simultaneous ACN and NaOH gradient of Table I. The last peak Is a background peak also s e e n in blank control. Casmalia Sample. For the Casmalia sample, the correlation between UV absorbance peaks and the TIC peaks by mas spectrometricdetection is not immediately obvious. The principal ions present in the electron-impactmass spectra are
ANALYTICAL CHEMISTRY, VOL. 64, NO. 4, FEBRUARY 15, 1992
*ooq
76
,
1000
1
104
I 3
1,2-8anz.ncdicai-boxyl
w
SAMPLE 148
105
\
i c acid
104
LIBRARY ENTRY
4W
K
600 400d
74
I
I
M/Z Figure 12. Comparison of the E1 mass spectrum of the last major peak in the Casmalia sample with the spectrum retrieved from the standard 7O-eV E1 mass spectral library by a library search.
listed in Table 11. We note that the mass spectrum for peak no. 4 is very similar to the spectrum for the doublet eluting just before PCBSA in the Stringfellow sample (Figure 10). A library search did not find a good match for most of these spectra However, as shown in Figure 12,the spectrum of peak no. 6,the last major peak eluting, matches very well with the spectrum of a phthalic acid retrieved from the EPA/NIH mass spectral library. The same Casmalia sample is also being investigated by another investigator (13)using several different types of columns. A fuller account of the findings will be published elsewhere. The reader may be interested in knowing that the original Casmalia leachate sample contains a high concentration of organic solvents, some of which are chlorinated. The volatile organics have been removed during the lyophilization step in our sample preparation.
ACKNOWLEDGMENT
443
Corp. and Robert D. Stephens of the California Department of Health Services for their roles in putting together the research team. Maria Rey, John R. Stillian, and Christopher A. Pohl of the Dionex Corp. have also been very supportive of our work. Registry No. Water, 7732-185;benzenesulfonic acid, 9811-3; p-toluenesulfonicacid, 104-15-4; p-xylenesulfonic acid, 609-64-1; p-chlorobenzenesulfonicacid, 9866-8;p-bromobenzenesulfonic acid, 138-36-3;4-hydroxybenzenesulfonic acid, 98-67-9;1naphthalenesulfonicacid, 85-47-2; 2-naphthalenesulfonic acid, 120-18-3.
REFERENCES (1) Simpson, R. C.; Fenselau, C. C.; Hardy, M. R.; Townsend, R. R.; Lee. Y. C.; Cotter, R. J. Anal. Chem. 1990, 62, 248-252. (2) Conboy, J. J.; Henion, J. D.; Martin, M. W.; Zwelgenbaum, J. A. Anal. Chem. 1990, 62, 800-807. (3) Slingsby, R. W.; Rey, M. OmnlPac Hendbook-klethods Devekpment 8nd TroubleshooNng; Dionex Corp.: Sunnyvale, CA, 1990. (4) Brown, M. A.; Kim, 1. S.; Roehl, R.; Sasinos, F. I.; Stephens, R. D. chemosphere 1989, 19, 1921-1927. (5) Klm, I . S.; Sasinos, F. I.; Stephens, R. D.; Brown, M. A. Envkon. Scl. Tedmd. 1990, 24, 1832-1836. (6) Stliilan, J. R. LCMeg. 1985, 3, 802-812. (7) Stiiilan, J. R.; Pohi, C. A. J . Chromtogr. 1990, 499, 249-266. (8) Wlibughby, R. C.; Browner. R. F. Anal. Chem. 1984, 56, 2626-2631. Wlnkler, P. C.; Perkins, D. D.; Williams, W. K.; Browner, R. F. Anal. Chem. 1988, 60,489-493. Browner, R. F.; Harrls, W. E.; Edman, K. Abstrects of Papers, Fourth Chemlcai Congress of the North A& can Continent, Dhrision of Analytical Chemistry, New York, NY, Aug 30, 1991; American Chemical Society: Washlngton, DC, 1991; Paper No. 122. (9) Kim, I . S.; Saslnos, F. I.; Stephens, R. D.; Brown, M. A. J . A m . Food Chem. 1990, 36, 1223-1226. (10) Bellar, T. A.; Behymer, T. D.; Budde, W. L. J . Am. Soc. Mess Spec@om. 1989, 1 , 92-98. Behymer, T. D.; Beliar, T. A.; Bud&, W. L. Anal. Chem. 1990, 62, 1686-1690. (11) Hsu. J. Interfacing Ion Chromatography with Partlck, Beam Mess Spectromeby; Callfornla Department of Health Services: Berkeley, CA, April 1, 1990 (available from the author upon request). See also, for example: Kim, I . S.; Sasinos, F. I.; Stephens, R. D.; Wang, J.; Brown. M. A. Anal. Chem. 1991, 63,819-823. (12) Ligon, W. V.; Dorn, S. B. Anal. Chem. 1990, 62, 2573-2580. Doerge, D. R.; Miles. C. I. Anal. Chem. 1991, 63, 1999-2001. (13) Kim, I. S.; Sasinos, F. I.; Stephens, R. D.; Brown, M. A. Prepdint Extended Abstmcts, 200th National Meeting of the American Chemical Society, Division of Environmental Chemistry, Washington, DC. Aug 26-31 1990; American Chemical Society: Washington, DC, 1990 pp 58-60.
I thank Rosanne Slingsby of Dionex Corp. for her extremely generous assistance, and In Suk Kim for providing the hazardous waste sample extracts which were prepared by Fassil I. Sasinos and D. K. Rishi for the Stringfellow and Casmalia samples, respectively. I would also like to thank I. S. Kim and Mark A. Brown for helpful discussions and for sharing their experience on the operation of the particle beam mass spectrometer. Ted Belsky kindly offered his interpretation of the mechanism for the observed analyte carryover phenomenon. Thanks are also due to Robert J. Joyce of Dionex
RECEIVEDfor review July 16,1991.Accepted October 31,1991. The paper was presented previously at the 200th National Meeting of the American Chemical Society in Washington, DC, before the Division of Analytical Chemistry, on Aug 30, 1990. A preliminary report was also given at the 1990 Pittsburgh Conference in New York, NY, on March 5,1990. The statements here do not represent the opinion of the California Department of Health Services or that of the Dionex Corp.
