摘要:

H i g h l i g h t Recent developments in the analysis of light isotopes by continuous flow isotope ratio mass spectrometry Andrew J. Midwood*a and Brian A. McGawb a Macaulay Land Use Research Institute, Craigiebuckler, Aberdeen, UK AB15 8HQ. E-mail a.midwood@mluri.sari.ac.uk; Fax: +44 0 1224 311556; Tel: +44 0 1224 318611 b School of Applied Sciences, The Robert Gordon University, St Andrew Street, Aberdeen, UK AB25 IHG Received 21st June 1999, Accepted 19th July 1999 Introduction Over the last 30 years there has been an exponential increase in the use and application of stable isotopes in many areas of research, including agriculture, environmental science, marine and clinical sciences.Until about the 1960s however, the measurement and use of stable isotopes was largely restricted to isotope geochemistry with commercially available mass spectrometers tailored to meet the needs of this specialisation. These instruments were designed to measure relatively small differences in isotope ratios ( < 1 part in a 1000 or 1‰ with respect to a standard) with a high level of precision and accuracy (±1 to 0.01‰).Although automated, sample throughput was, and in some cases still is, limited to 10s of samples per day, due largely to complex off-line sample preparation procedures. In many biological applications, where natural variability may be a significant factor, the requirement for high precision instrumentation may be overshadowed by the desire to analyse large numbers of samples (100s).This is the case for example, when assessing trophic levels in an ecosystem, where it is more important to ensure a representative sample of the populations, rather than generating very accurate isotope analyses for a restricted data set. In the early 1980s the desire to undertake this type of work provided the impetus for the development of a simple rapid means of analysing light isotopes such as 15N in marine and agricultural environments. Up until this point 15N analysis was extremely laborious.Firstly, Kjeldahl digestion was required to convert the sample N to ammonium for quantitative analysis, followed by oxidation to nitrogen gas by the Rittenburg technique before mass spectrometric analysis using a dual inlet gas isotope ratio mass spectrometer (GIRMS). In such instruments, repeated alternate measurements of the sample and a reference gas are obtained, individual analyses typically taking 15 min.As a first step to streamlining this process Barsdate and Dugdale1 linked an automated commercially available Dumas combustion system for elemental N analyses to a mass spectrometer (MS). The N2 generated from the combustion of a solid sample was trapped, purified and then introduced into the dual inlet of the MS for isotopic analysis. Although this was a marked improvement in terms of speed, it was still not a fully automated system. The first continuous flow-isotope ratio mass spectrometer (CF-IRMS) was built by Preston and Owens and involved the coupling of a stand alone elemental N analyser directly to the source of a GIRMS.The term coined, ‘Continuous Flow’, related to the use of a carrier gas, which was used to carry the sample from the preparation system into the MS. Helium tends to be the carrier of choice since it is chemically unreactive and is readily available in high purity grades. Preston and Owens were the first to demonstrate that precise and rapid 15N and total N analysis could be performed in this highly automated fashion; obtaining a precision of ±0.7‰ at natural abundance levels.Nitrogen isotope analysis was soon followed by CO2 analysis to determine the 13C/12C ratio of solid samples. Modern CF-IRMS are now compact, bench top instruments capable of analysing both isotopes sequentially in the same sample, in a matter of minutes. A schematic of a typical CF-IRMS system capable of this type of analysis is shown in Fig. 1. In this paper we report on some of the latest developments using this form of mass spectrometry which relate to the analyses of 2H in H2O, volatile organics and solids, and 18O and 34S in solid samples. Analysis of 2H in H2O, volatile organic compounds and solids by CF-IRMS The classical method for analysis of 2H in aqueous samples involves the off-line batch conversion of sample H to H2 followed by analysis using a dual inlet GIRMS. Several schemes have been used to produce H2 and involve the chemical reduction of H2O in sealed glass tubes at high temperature.The chemicals used have included Zn at 450–500 °C,3 Cr at 700 °C4 and Mn at 520 °C.5 As an alternative to these batch methods, dynamic methods have also been used and involve repeatedly passing H2O vapour held in a vacuum line over U held at 600 °C or Zn turnings at 450 °C.3 The use of U is complicated by the persistence of a severe sample memory effect so that when analysing highly enriched samples, there is a marked carry over of 2H from one sample to the next.3 Without exception, all of these methods are time consuming and prone to isotope fractionation if the utmost care in preparation is not exercised. These techniques also tend to be sensitive to dissolved salts in the H2O samples, the salts poisoning the metals completely or partially preventing reaction with the H2O.Over the last few Fig. 1 Typical schematic of a CF-IRMS capable of sequential analysis of 15N and 13C in a solid sample.The Conflo interface depicted in this particular system has a He dilution facility which is used to deal with the fact most biological samples have more C than N. This interface equalises the N2 and CO2 peaks as they elute from the elemental analyser by diluting the CO2 with He prior to entering the MS. Anal. Commun., 1999, 36, 291–294 291years, there has been a drive, therefore, to develop an analysis procedure based on a CF-IRMS system.These efforts have attempted to reduce the analysis time, eliminate isotope fractionation and marked memory effects, and minimise the influence of dissolved salts. Before analysing H2 using a CFIRMS two fundamental problems had to be addressed. The first problem is related to limited resolution which can be achieved using a bench top MS and the second to a suitable means of converting H2O to H prior to analysis. Analysis of H2 by CF-IRMS In a continuous flow mode as the He carrier gas ionises within the source of the MS it generates a 4He+ signal 104 to 105 larger than that produced by 2H1H+ of the sample.The result is significant interference of the relatively weak 2H1H+ signal (note that at low 2H concentrations all 2H is found as 2H1H). To overcome this, Prosser and Scrimgeour6 used a modified N2– CO2 MS incorporating a ‘flared flight tube’ to resolve the 4He+ and 2H1H+ beams. By altering the geometry of the MS, the collectors for 1H2 + and 2H1H+ could be placed much further apart, allowing a greater dispersion of the ion beams and complete resolution of the 4He+ and 2H1H+.In fact the 2H1H+ was detected using a totally separate collector spur whilst the 1H2 + was collected in the standard triple collector assembly of the instrument. This approach has also been used by Fourel et al.7 who have recently produced a MS incorporating a flared flight tube for H2 analyses. Tobias et al.8 took a completely different approach to this problem and used a selective H2 filter, composed of Pd foil. When this filter is heated to 550 °C it becomes permeable to H2, but not to the He carrier gas.The advantage of this system is that it is relatively simple and can be linked directly to a conventional MS capable of analysing H2 and hence the need to alter the MS geometry is eliminated. However, since the filter only allows the H2 into the ion source, the source becomes pressure sensitive and peak intensities must be carefully matched to allow accurate calibration. Also, transmission rates through the filter are low (0.1% of the total H2). Tobias et al.8 suggested that ~1 mmol of H2 injected into the carrier gas stream was sufficient to produced adequate signal intensity (measured peak height of 1 to 2 V ± 100 mV) in a conventional high precision GIRMS.The analysis of H2 is further complicated by the ion/molecule reaction which occurs within the ion source of the mass spectrometer: 1H2 + 1H2 + ? 1H3 + + 1H The production of 1H3 + must be corrected for, otherwise an overestimation of the 2H1H+/1H2 + ratio occurs.The formation of 1H3 + is a second order reaction, so that the concentration of 1H3 + is proportional to the square of the partial pressure of H2 in the ion source.9 Therefore a linear relationship exists, between the measured 2H1H+ and 1H2 + ion beam intensity; in a conventional dual inlet GIRMS, correction for the 1H3 + is achieved by measuring the m/z 3+m/z 2 ratio over a range of m/z 2 ion beam intensities and generating a linear correction term. During sample analysis the gas enters the ion source at a constant pressure, resulting in the proportion of 1H3 + being fairly constant.It is relatively simple therefore to correct for 1H3 + species. However, in a continuous flow system the pressure of H2 entering the ion source is constantly changing over the width of the peak, so that correction is not straight forward.A mathematical algorithm has been determined by Prosser and Scrimgeour,6 who established a 1H3 + correction by introducing the same working reference 3 times such that a progressively increasing 1H3 + signal was generated. However, due to an enrichment effect this procedure must be repeated using a second reference of a different 2H content to the first. Ideally, the enrichment of these ‘working references’ should bracket those of the samples to be subsequently analysed.In a similar manner Fourel et al.7 used alternate gas injections of two calibrated H2 gas samples into the He stream to generate an appropriate correction algorithm for their instrument. Correction for 1H3 + when using the Pd filter is more complex due to the pressure sensitivity of the source to sample size. Tobias et al.8 found that after repeated introduction of the same sample to produce a steadily increasing 1H2 + ion beam intensity the d2H ranged over 300‰.The relationship to 1H2 + intensity was also non-linear and best described by a third-order polynomial; however, a mathematical correction can be made. These authors did not discuss the possibility of an enrichment effect so it is not known if this is a problem with a Pd filter system. Preparation of sample H A number of different approaches have been used to convert sample H2O to H2 prior to analysis in continuous flow mode. These have included equilibration of H2O with H2 or reduction of H2O to H2 using Ni, a technique which is also applicable to volatile organics.Equilibration of H2O samples with H2 is carried out in the presence of a catalyst, usually Pt. The activity of the Pt catalyst is strongly suppressed when covered with H2O; to overcome this, Pt has in the past been incorporated into a hydrophobic resin. For example, Hokko Beads™ (Shoko Corp., Tokyo, Japan) which consist of a styrene-divinylbenzene (SDB) copolymer porous resin doped with 3% w/w Pt (density 0.2 g cm23, grain size 125–250 mm). This resin ensures that the Pt stays at the H2O/H2 interface and equilibration is achieved in about 1 h.An alternative to Hokko Beads™ is to use a Pt-onalumina catalyst which is more readily available and can be physically separated from the H2O sample by simply using a small sample tube placed inside a Vacutainer™ (Becton Dickinson, Rutherford, New Jersey, USA) containing the H2O sample. Prosser and Scrimgeour6 used this approach and reported that, since equilibration times are considerably longer with this catalyst compared to Hokko Beads™ (being up to 3 days) temperature induced fluctuations in the 2H content of the H2 are greatly reduced.This is an important consideration since the equilibrium is temperature sensitive, changing by about 6‰ °C21. The need to tightly control temperature can however, be avoided altogether by simply equilibrating calibrated working standards with the samples.Equilibration procedures such as the two just described involve relatively simple sample processing and are ideally suited to automated analysis through the use of standard autosamplers. Several CF-IRMS systems are already commercially available which use autosamplers for handling gas samples held in 10 to 25 ml Vacutainers™. Typically these systems are designed for the analysis of 13C labelled CO2 in breath or 18O in H2O which is measured indirectly following equilibration of CO2 with H2O. No modification of these sample handling systems is required to permit H2 analysis. Probably the biggest drawback of equilibration methodology however, is the depleted 2H content of the H2 after equilibration.For example, H2 equilibrated with Vienna Standard Mean Ocean Water (V-SMOW) at 20 °C has a d2H = 2748.0‰. This restricts the accuracy and precision of this technique. Prosser and Scrimgeour obtained a precision of ±1.5‰ using a Pt-onalumina catalyst and an analyser incorporating a flared flight tube.In comparison the Zn reduction batch method yields a typical precision of ±0.7‰. Another disadvantage is the large volume of sample H2O required, typically 0.1–1 ml, whereas reduction methods use an order of magnitude less sample. On-line reduction of H2O has been investigated as a potential sample handling system for use with CF-IRMS. Typically, a gas chromatograph (GC) is used to vaporise the H2O sample which is then reduced to H2 as it flows through a Ni furnace.Tobias et al.8 used a Ni furnace heated to 850 °C in conjunction with a Pd foil filter system to analyse the 2H content of 0.4 ml aliquots of H2O. This system yielded a precision of < 10‰. Begley and Scrimgeour10 used carbonised Ni held in an alumina reactor tube at 1050 °C to reduce H2O to H2 and CO, the latter allowing 292 Anal. Commun., 1999, 36, 291–29418O as well as 2H analysis. Using a MS fitted with a flared flight tube, a precision of ±2.0‰ was achieved for 0.5 ml H2O samples. Both these systems have been successfully used to analyse volatile organics. Tobias and Brenna11 obtained precisions of < 6‰ for a range of organics including ethylbenzene and cyclohexanone, and benzene of variable 2H enrichments (248 to 372‰).Begley and Scrimgeour10 analysed 2H enriched samples of ethanol, ethyl benzene and acetone and reported precisions better than ±12‰ for highly enriched samples (d2HV-SMOW Å 5200‰).Recently, Kelly et al.12 reported a method which allowed the analysis of 2H in solid organic samples. This procedure was based on an elemental analyser fitted with a pyrolysis column held at 1080 °C and packed with glassy carbon grit. Pyrolysis gases produced from the decomposition of the sample were passed through a GC column packed with molecular sieve and heated to 60 °C. The H2, N2 and CO generated were separated and the H2 analysed using a MS fitted with a flared flight tube.High temperature pyrolysis has come to the fore recently as probably the most amenable method for the analysis of 2H in a wide range of samples. It is this technology which is set to develop in the future and make H analysis by CF-IRMS commonplace. Analysis of 18O in solid organic materials using pyrolysis and CF-IRMS A recent development in CF-IRMS has been the analysis of 18O in solid materials such as plant tissues and vegetable oils.In the past, this analysis has relied on a number of complex reaction procedures with the emphasis on CO2 as the gas of choice for analysis, the reasons being that CO2 is readily formed, easily trapped/purified, and amenable to IRMS analysis. For solid materials, reaction schemes such as the Shuetze–Unterzaucher method have been used,3 where the sample is decomposed by pyrolysis over platinized carbon at 950 °C; CO produced is then oxidised to CO2 by reaction with I2O5 at 100 °C. 5CO + I2O5 ? 5CO2 + I2 This method suffers from memory and blank problems associated with the quartz reaction vessel used and there is a dilution of the sample oxygen by the I2O5 which contributes oxygen to the CO2. A variety of other similarly complex reaction schemes have also be documented for 18O analysis and have been reviewed by Wong and Klein.3 Recently, Farquhar et al.13 presented a method for analysing 18O in plant tissues using CF-IRMS system. This system developed earlier work of Brand et al.,14 who used the Scheutze–Unterzaucher reaction to produce CO from ml quantities of H2O.A key feature of this work was the direct analysis of the CO generated from reaction of the H2O with a carbon source without any conversion to CO2. Farquhar et al.13 developed this approach to allow the analysis of solid organic materials for 18O again using CO. Samples were pyrolysed using an elemental analyser fitted with a reaction column packed with nickelized carbon and held at a temperature of between 1080 and 1100 °C.A similar system has also recently being described by Bréas et al.15 for the analysis of 18O in vegetable oils. A problem to be overcome when analysing organic compounds, many of which contain nitrogen, is the isobaric interference of N2 + (m/z 28) with 12C16O+. This has been resolved by using a GC column to separate the two gases as they elute from the pyrolysis system prior to entry into the MS. Also, contributions of 13C17O+ to the m/z 30 signal and the unrecorded 13C18O+ have been calculated to be insignificant and for a typical sample the correction produces a change of just 0.01‰.13 At the high temperatures used, the nickelized carbon may react with the quartz glass of the reactor tube and generate CO. To avoid this Farquhar et al.13 lined the reactor with nickel foil over the section holding the nickelized carbon.Also, to ensure the pyrolysis and CO formation, a temperature sensitive reaction occurring in the hottest part of the reactor column, the lower section of the reactor was drawn out into a 120 cm capillary. This modification avoided the need to pack the lower half of the reactor with an inert material such as quartz chips.Errors in d18O analysis may occur due to a blank contribution from the Sn capsules, reactor packing material or the reactor itself. Also, absorption/release characteristics of O2 and N2 within the pyrolysis system, and CO with the nickel may result in sample memory.13,15 Memory effects become more problematical as the pyrolysis column ages, and may be evident in columns after the analysis of as few as 50 samples.13 It has been reported that doping the column with a chlorinated hydrocarbon may reduce this effect and also improve the efficiency of the pyrolysis,16,17 however, this improvement may only be short lived.The quality of the nickelized carbon and absolute carbon content may be important.13 Despite these factors Farquhar et al.13 reported a reproducibility (s = 0.2‰, n = 10) comparable to that of any existing methods with the major advantages of rapid analysis (typically just 7 minutes per sample) and the relatively simple reaction scheme.Analysis of 18O in water using this approach is particularly suited to small sample volumes in the ml range. There are however, certain difficulties when working with such small sample volumes. Care must be exercised to ensure the samples are completely sealed in the Sn cups used since any evaporative loss would have a marked influence on the d18O measured.Farquhar et al.13 found losses of only 1 to 2% from prepared samples immediately before analysis, and reported an analysis precision close to those obtained for solid samples, however, they observed that up to 20% of the oxygen in the H2O sample was not recovered. In place of nickelized carbon, glassy carbon has also been used to convert the oxygen content of mainly cellulose samples to CO for 18O analysis by CF-IRMS. Recently, Saurer et al.18 illustrated that continuous flow methodology is a fast and reliable technique for cellulose analysis yielding results comparable to established off-line methods.When using glassy carbon however, water analysis is not possible. The analysis of 18O in water and more importantly solid samples using this technology is clearly a significant step forward in terms of speed and reduced complexity.Prolonging the longevity of the pyrolysis column is clearly going to be a target for this technique in the future. This may involve using different column packing materials or alteration of the operating parameters or a combination of both. Furthermore, there is a lack of biological reference materials which cover a range of certified 18O contents, hindering comparative evaluations of accuracy and the factors which effect this. Analysis of 34S in solid samples The common gas of choice for 34S analysis in materials other than geological specimens is SO2 generated from the S content of the sample using a series of chemical reactions.Sample S is converted to BaSO4 which is then reduced to H2S and converted to Ag2S, finally this is oxidised to SO2. A major disadvantage of this method is the relatively large amount of S required (typically 3–7 mg) to ensure sufficient material for analysis. As with any complex chemical reaction scheme, sample throughput is slow (estimated to be ~ 4.5 hours per sample) with the ever present danger of inadvertently altering the sample d34S through isotope fractionation, particularly by incomplete conversion to SO2.The analysis of the total S content of soils and biological materials using an elemental analyser has been routine for many years. A natural progression of CF-IRMS to the analysis of 34S would at first sight seem straight forward. However, there have been relatively few reported d34S analyses made using a CFAnal.Commun., 1999, 36, 291–294 293IRMS.19,20 The gas SO2 has a reputation as being difficult to work with tending to ‘stick’ to the metal components of the MS leading to serious memory problems. In the past, GIRMS have overcome this problem by heating the mass spectrometer inlet and gas handling components to ~ 110 °C. In a CF-IRMS however, the flow of He through the system and the use of materials such as PTFE for tubing and fittings has reduced these problems.Giesemann et al.19 adapted a CF-IRMS to allow the analysis of 34S in biological materials, they obtained a reproducibility of ±0.2‰ for d34S with samples containing as little as 20 mg of S. With this system S was extracted from the samples and precipitated as BaSO4, which was then wrapped in an Sn cup with 0.1 mg of V2O5 and combusted at 1100 °C with a 5 ml pulse of O2 using a Carlo Erba NA1500 elemental analyser. The gases produced were carried by a He stream through a single oxidation-reactor column filled with tungstic anhydride, CuO and Cu, the CuO and reduced Cu being used to reduce traces of SO3 to SO2 and trap any surplus O2.Magnesium perchlorate was then used to absorb any H2O formed before separating N2, CO2 and SO2 by passage through a GC and analysing the SO2 in the MS. The S isotopes were observed to separate slightly as SO2 eluted from the GC,19 with enrichment of 34S being observed at the start of the SO2 peak.However, the measured ratio is not sensitive to the amount of SO2 entering the MS (changing < 0.01‰ per 1 V change in signal intensity).19 However, this does not appear to be the case with all systems. Eriksen20 observed a pronounced and variable pressure sensitivity when analysing 34S in soil samples with a variable S content. To overcome this, isotope ratio analyses were undertaken using two concentrations of S in both a standard and the sample. A line was then drawn between each pair of analyses.If the pressure dependence of the sample and standard was similar, the vertical distance between the two parallel lines was then used to calculate the d34S of the sample. Sample memory in a CF-IRMS system has received no attention as yet, which is surprising since this has been identified as a problem in dual inlet instruments. The absence of publications relating to 34S analysis by CFIRMS reflects the fact that considerably fewer studies are conducted which involve this isotope when, for example, compared to 2H or 18O.The reasons for this are that work involving 34S often requires the measurement of relatively small changes in natural isotope levels, and as such favours high precision and accuracy analysis by GIRMS. Also, for tracer work there is a limited availability of 34S labelled compounds, those which are available tend to be prohibitively expensive and although this may change in the future at present there is no application driven need to develop CF-IRMS analysis of 34S.The future Clearly CF-IRMS technology has developed considerably since Preston and Owens first devised a system for 15N analysis. In terms of the mass spectrometers used; the design and efficiency of pumping systems, ion optics, collectors systems and flared flight tubes have all contributed to improve precision and response linearity. In recent years however, undoubtedly the biggest area of development has been in relation to the sample introduction systems.This has allowed the use of CF-IRMS technology with an increasingly diverse range of sample types. At present, limiting factors to precision and accuracy tend to be associated with these sample introduction systems and problems such as memory and blank offsets. Scope exists through design and operating parameters to minimise these undesirable effects. The compact, robust and relatively simple nature of these instruments has already enabled a greater spread of application and use of 13C and 15N in the scientific community.Future developments are set to expand these application still further and encompass 2H, 18O and possibly 34S in a variety of sample types. Acknowledgement Manuscript preparation was supported by SOAEFD. References 1 R. T. Barsdate and R. C. Dugdale, Anal. Biochem., 1965, 13, 1. 2 T. Preston and N. J. P. Owens, Analyst, 1983, 108, 971. 3 W. W. Wong and K. D. Klein, Mass Spectrom. Rev., 1986, 5, 313. 4 M. Gehre, P. Hoefling, P. Kowski and G. Strauch, Anal. Chem., 1996, 68, 4414. 5 A. Tanweer and Han Liang-Feng, Isot. Environ. Health Stud., 1996, 32, 97. 6 S. J. Prosser and C. M. Scrimgeour, Anal. Chem., 1995, 67, 1992. 7 F. Fourel, T. Merren, J. Morrison, L. Wassenar and K. Hobson, Mircomass Application Note AN300/LA Version 1, 1998. 8 H. J. Tobias, K. J. Goodman, C. E. Blacken and J. T. Brenna, Anal. Chem., 1995, 67, 2486. 9 R. Gonfiantini, in Stable Isotope Hydrology. Deuterium and Oxygen- 18 in the water cycle, Technical Report Series No. 210, ed. J. R. Gat and R. Gonfiantini, Vienna, 1981. 10 I. S. Begley and C. M. Scrimgeour, Anal. Chem., 1997, 69, 1530. 11 H. J. Tobias and J. T. Brenna, Anal. Chem., 1996, 68, 3002. 12 S. D. Kelly, L. G. Parker, M. Sharman and M. J. Dennis, J. Mass Spectrom., 1998, 33, 735. 13 G. D. Farquhar, B. K. Henry and J. M. Styles, Rapid Commun. Mass Spectrom., 1997, 11, 1554. 14 W. A. Brand, A. R. Tegtmeyer and A. Hilkert, Org. Geochem., 1994, 12(6/7), 585. 15 O. Bréas, C. Guillou, E. Sada and F. Angerosa, Rapid Commun. Mass Spectrom., 1998, 12, 188. 16 W. J. Kirsten, Anal. Chim. Acta, 1978, 100, 2799. 17 W. J. Kirsten, Microchem. J., 1977, 22, 60. 18 M. Saurer, I. Robertson, R. Seigwolf and M. Leuenberger, Anal. Chem., 1998, 70, 2074. 19 A. Giesemann, H.-J. Jager, A. L. Norman, H. R. Krouse and W. A. Brand, Anal. Chem., 1994, 66, 2816. 20 J. Eriksen, Commun. Soil Sci. Plant Anal., 1996, 27(5–8), 1251. Paper 9/04908H 294 Anal. Commun., 1999, 36, 291–294

ISSN:1359-7337    

DOI:10.1039/a904908h

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Communication Investigation of problems associated with the determination of iodine in glacial acetic acid samples using flow injection analysis-inductively coupled plasma-mass spectrometry† Kathryn L. Ackley, Jason A. Day, Karen L. Sutton and Joseph A. Caruso* Department of Chemistry, University of Cincinnati, PO Box 0172, Cincinnati, OH 45221-0172, USA Received 28th May 1999, Accepted 7th July 1999 Determination of iodine in glacial acetic acid is a major concern of acetic acid manufacturers and consumers.The use of ICP-MS for iodine determinations in acetic acid is hindered by memory effects that produce an elevated background signal necessitating long rinse times between samples. In this work, different analysis methods are employed in an attempt to minimize memory and matrix effects allowing for the accurate determination of iodine in glacial acetic acid using ICP-MS. Ammonium hydroxide solutions (3.7 and 7.4 M) were better at reducing the elevated iodine signal present after the introduction of an acetic acid sample than water or 0.3 M nitric acid.Memory effects were decreased when the sample was introduced by flow injection rather than constant sample aspiration. Peak areas generated by flow injection decreased significantly with increasing ammonium hydroxide concentration in the carrier solution. Iodine determinations made with 1.7 M ammonium hydroxide as the carrier solution were higher than determinations made with 3.7 M ammonium hydroxide as the carrier solution for the same samples, however, the percentage difference between the two determinations varied widely from sample to sample.All samples were analyzed by the method of standard additions in an attempt to compensate for matrix effects. This work illustrates the importance of the carrier solution in the determination of iodine in glacial acetic acid samples. Introduction Acetic acid in the United States is almost exclusively produced by the Monsanto process which involves the carbonylation of methanol with an organometallic catalyst.Iodine, in the form of methyl iodide, is also necessary at molar concentrations in the reactor to facilitate the insertion of CO because methanol alone is not sufficiently reactive.1 Also, to improve the economics of the Monsanto process, proprietary technology has been introduced by all major US manufacturers that increases the likelihood of iodine contamination of the product acetic acid.2 This contamination poses a significant problem since a major portion of the acetic acid produced in the US is used in the production of vinyl acetate, and low levels of iodide have been known to poison the palladium catalyst used in the synthesis of vinyl acetate.Therefore, acetic acid must be monitored for iodine concentration. Acetic acid synthesis methods generate a number of alkyl iodides (C1–C8 and higher). Quantification of several species has been achieved by gas chromatography with electron capture detection (GC-ECD). However, non-volatile organic and inorganic iodides cannot be detected using this method.A variety of techniques have been employed for the determination of total iodine. Neutron activation analysis3 has been used successfully, but the technique is expensive and requires access to a nuclear reactor. Inductively coupled plasma atomic emission spectroscopy (ICP-AES),4–8 microwave induced plasma-atomic emission spectroscopy,9 and cathodic stripping voltammetry10 may be used as well.Isotope dilution has also been employed for the determination of iodine.11–13 Inductively coupled plasma-mass spectrometry (ICP-MS) is a popular technique for trace element determinations because of its excellent sensitivity and element specificity.14 However, the determination of iodine by ICP-MS poses some significant challenges. The ICP-MS detection limit for iodine is higher than for most metals because of iodine’s higher ionization potential. In addition to this decreased sensitivity, a signal memory effect is a major consideration for iodine analysis by ICP-MS.The selective evaporation of iodine as HI or I2 from droplets present in the spray chamber is a possible cause of the memory effect.15 Regardless of the source, this memory effect produces an elevated background signal and necessitates long rinse times between samples to return the background counts to acceptable levels.Several strategies have been utilized to minimize the memory effect associated with iodine analysis by ICP-MS. Larsen and Ludwigsen16 prepared plant and animal materials for analysis by wet ashing to mineralize organic matter and convert volatile iodine to nonvolatile species. A second strategy is to prepare the sample in alkaline media to prevent the reduction of I2 to I2 or the formation of HI. Water,17 milk and milk powder,16,18,19 serum,19 urine,20 and soil samples21 have been analyzed by ICPMS in alkaline solution.This method is useful in cases where samples are easily made alkaline. However, titrating acetic acid samples until they are basic is not feasible since the high concentration of the resulting salt would accumulate on the sampler cone orifice causing signal degradation and instrument drift. Also, the heat generated during this titration would likely drive off volatile iodine species that could be present in a sample.When samples such as acetic acid are not amenable to titration, alternative strategies for controlling the memory effect associated with iodine by ICP-MS must be employed. Allain and coworkers22 observed cross-over contamination from iodine when analyzing urine samples. To compensate for the observed memory effect, multiple analyses were made of samples with low iodine concentrations which followed samples with high concentrations of iodine. Rinsing the sample introduction system for 3 min with a 0.5% v/v ammonia solution has also been used to overcome memory effects.15 Flow injection analysis (FIA) has been used previously for the analysis of iodine using ICP-MS detection.12–19 Sturup and Buchert19 analyzed alkaline solutions of milk and milk powders using FIA, as did Kerl et al.12 who analyzed plant and tissue materials using an acid–hydrogen peroxide digestion process.The principle advantage of FIA is that the amount of analyte entering the spray chamber is reduced, which helps to minimize † Presented at the 1999 European Winter Conference on Plasma Spectrochemistry, Pau, France, January 10–15, 1999.Anal. Commun., 1999, 36, 295–298 295memory effects. In addition, high sample throughput is possible because long rinse times to reduce background signal are avoided.9 The goal of this work was to investigate strategies for the determination of total iodine concentrations in acetic acid samples using ICP-MS. Experimental Instrumentation ICP-MS conditions are listed in Table 1.Flow injection analyses were performed using a Gilson Minipuls 3 peristaltic pump (Middleton, WI, USA). The pump was connected to the nebulizer via a 30 cm length of 0.002 in id PEEK (polyetheretherketone) tubing. A six port Rheodyne 4396 injector (Cotati, CA, USA) with a 20 mL PEEK injection loop was used for introduction of analyte into the carrier solution. Samples were loaded in 2.5 mL disposable polypropylene/polyethylene syringes (Fortuna, W.Graf Gmbh & Co., Wertheim, Germany). All samples were loaded into syringes immediately prior to analysis. Four replicate injections were made for each sample, each 2 min apart. The carrier solution flow rate to the nebulizer was 1.2 mL min21. Two commercially available glacial acetic acid samples were purchased from their manufacturers, and two samples of glacial acetic acid not commercially available were obtained from a third manufacturer.None of the samples analyzed were certified for iodine concentrations. Samples were diluted with 18 MW cm21 deionized water (Barnstead, Boston MA, USA) (1 + 3) based on preliminary investigations for method sensitivity. Four aliquots of each sample were analyzed, and each aliquot was spiked with indium standard to achieve a final concentration of 5 ng mL21 indium. The aqueous indium solution was prepared from a 10 mg mL21 stock solution (Claritas PPT, SPEX CertiPrep, Inc., Metuchen, NJ, USA).One aliquot of each sample was diluted with no iodide spike. The remaining three samples were spiked with differing amounts of a 10 mg mL21 aqueous iodide standard (High-Purity, Charleston, SC, USA) to achieve final concentrations of 20 ng mL21, 100 ng mL21, and 200 ng mL21. One of the samples, not commercially available (sample D), contained a high concentration of iodine. For this sample, the aqueous iodide standard was used to spike samples by 120, 200, and 300 ng mL21.The iodine memory effect was studied with five different carrier solutions, 18 MWcm21 deionized water, 1.7 M NH4OH, 3.7 M NH4OH, 7.4 M NH4OH, and 0.3 M HNO3 (Fisher Certified ACS Plus, Fisher Scientific, Fair Lawn, NJ, USA). The molarities of the NH4OH solutions correspond to 3.5, 7.0, and 15% v/v ammonia solutions prepared in water. The ammonium hydroxide solutions were prepared from a 14.8 M ammonium hydroxide solution (Fisher Scientific). Plots of the washout profiles (signal intensity vs.time) for each rinsing solution were used to determine which solution to use for the analysis. Fig. 1 Washout profiles obtained when different rinsing solutions are used following the aspiration of an acetic acid sample with a high iodine concentration. Table 1 ICP-MS Parameters ICP-MS instrument Perkin Elmer Sciex Elan 6000 (Ontario, Canada) Sample introduction system Cross flow nebulizer and a quartz double pass spray chamber, water cooled to 5 °C Dwell time 100 ms Isotopes monitored 127I and 115In Data acquisition Sweeps/reading = 1 Replicates = 1 Readings/replicate varied to yield analysis times of the desired length Nebulizer gas flow 1.0 L min21 Lens voltage 7.5 V Radiofrequency power 1000 W 296 Anal.Commun., 1999, 36, 295–298Discussion Washout studies Significantly elevated iodine baselines were observed during preliminary work involving the analysis of iodine in acetic acid by ICP-MS without flow injection.The background decreased with time, but the elevated levels persisted for over 5 min after each sample was analyzed. A study was performed to see if the elevated background signal for iodine could be reduced using different rinsing solutions following the introduction of an acetic acid sample containing iodine. The iodine signal was monitored after an acetic acid sample containing a high concentration of iodine (approximately 250 ng mL21 in the diluted form as estimated by FIA-ICP-MS using standard additions) was introduced to the ICP-MS for 1 min.Water and 0.3 M HNO3 were selected as rinse solutions since these are commonly used in ICP-MS analyses. Both solutions caused the iodine signal to decrease slowly. A 3.7 M NH4OH solution was investigated since reports in the literature indicate memory effects may be reduced by preparing samples in alkaline solutions.15,17 Fig. 1 shows the iodine washout profiles obtained with the different rinsing solutions. The water and 0.3 M HNO3 solutions produced iodine signals that remained elevated longer than when 3.7 M NH4OH was the rinse solution.The appearance of the 3.7 M NH4OH washout profile was better than the nitric acid solution or the deionized water, so a 7.4 M NH4OH solution was used as a rinse solution to determine if a higher base concentration would be more effective. Fig. 2 shows the washout profiles obtained when 3.7 and 7.4 M ammonium hydroxide solutions were used as rinsing solutions.The iodine background signals were significantly lower for the 3.7 and 7.4 M solutions than water alone or the nitric acid solution, and the iodine background signal quickly returned to baseline for both base solutions. The same acetic acid sample was introduced to the instrument using flow injection, and the iodine signal was monitored for tailing due to memory effects. Flow injection was investigated in an attempt to minimize memory effects, transient acid effects and perturbations to the plasma due to the organic and acid content of the samples.The memory effects were dramatically decreased as one would expect since a significantly smaller amount of sample (20 mL) was introduced to the system for each flow injection peak. From the flow injection data, 2 min between sample injections was determined to be sufficient to allow the iodine signal to return to the baseline. Peak areas during FIA decreased significantly with increasing NH4OH concentration in the carrier solution.The peak area obtained when 3.7 M NH4OH was used as the carrier solution was roughly 2.5 times as large as the peak area obtained when the same sample was injected in a 7.4 M NH4OH carrier solution. A 1.7 M NH4OH solution was also used as a carrier solution to see if the same discrepancy between peak areas would exist between the more dilute 1.7 M and the 3.7 M carrier solutions. Fig. 3 shows the peaks obtained when the same acetic acid sample was injected into 1.7, 3.7, and 7.4 M NH4OH carrier solutions.The decreased signal may result from differences in solvent transport efficiencies with differing concentrations of base. Also, the ionization properties of the Fig. 3 Peaks obtained when the same acetic acid sample is injected into carrier solutions containing differing ammonium hydroxide concentrations. Fig. 2 Washout profiles obtained when a 3.7 and 7.4 M ammonium hydroxide solution was used as the rinsing solution following the aspiration of an acetic acid sample with a high iodine concentration.Anal. Commun., 1999, 36, 295–298 297plasma may be affected by the larger concentrations of base.23 Standard additions using flow injection analysis Two NH4OH solutions (3.7 and 1.7 M) were used as carrier solutions for the determination of iodine in the acetic acid samples by FIA-ICP-MS. These carrier solutions were selected to determine if the difference in peak areas observed with different concentrations of NH4OH would have an effect on the iodine determinations of the acetic acid samples.Table 2 shows the iodine concentration determined with each carrier solution. In all cases, the determined iodine concentration was lower with the 3.7 M carrier solution, but the percentage difference between the determinations made with the two mobile phases varied widely. These differences may be due to acid transient effects that have been described elsewhere in the literature,23 or to changes in the physical properties of the plasma caused by the increased concentration of NH4OH entering the plasma.Interestingly, the percentage difference between the iodine determinations was almost the same for the samples containing the lowest iodine concentration (sample A) and the highest concentration (sample D), but a strong correlation can not be made between the concentration of iodine in the sample and the difference observed between the two determinations.Ascertaining which carrier solution provides the most accurate iodine determinations was difficult. No acetic acid sample certified for total iodine is available, and spiking a sample to determine the percent recovery is hindered by the fact that nearly, if not all, commercially available acetic acid is expected to have trace levels of iodine. A comparison should be made between the data obtained by FIA-ICP-MS and another technique capable of determining total iodine concentration such as neutron activation analysis.Currently this study shows that more work must be done to develop a method for routine determination of iodine in acetic acid. The results do show that a basic carrier solution (for FIA) or a basic rinsing solution (for constant sample aspiration) returns the iodine signal to baseline more quickly than water or dilute nitric acid, and the selection of the carrier phase for FIA affects the peak area and the determined iodine concentration. With further method development, FIA-ICP-MS may be a useful technique for the monitoring of iodine in acetic acid production.Acknowledgements The authors gratefully acknowledge Dr. Mikhail Belkin for his assistance and helpful discussions. References 1 F. E. Police and J. F. Roth, Chem. Commun., 1968, 1578. 2 Personal communication with acetic acid producer, 1998. 3 R. R. Rao and A. Chatt, Analyst, 1993, 118, 1247. 4 S.P. Dolan, S. A. Sinex, S. G. Capar, A. Montaser and R. H. Clifford, Anal. Chem., 1991, 63, 2539. 5 B. S. Sheppard, J. A. Caruso, K. A. Wolnik and F. L. Fricke, Appl. Spectrosc., 1990, 44, 712. 6 J. A. Nobrega, Y. Gelinas, A. Krushevska and R. M. Barnes, J. Anal. At. Spectrom., 1997, 12, 1243. 7 T. Nakahara and T. Mori, J. Anal. At. Spectrom., 1994, 9, 159. 8 W. E. Braselton, K. J. Stuart and J. M. Kruger, Clin. Chem., 1997, 43, 1429. 9 F. Camuna, J. E. Sanchez-Uria and A. S.Medel, Spectrochim. Acta, 1993, 48B, 1115. 10 W. Holak, Anal. Chem., 1987, 59, 2218. 11 G. Radlinger and K. G. Heumann, Anal. Chem., 1998, 79, 2221. 12 W. Kerl, J. S. Becker, H. J. Dietz and W. Dannecker, J. Anal. At. Spectrom., 1996, 11, 723. 13 M. Haldimann, B. Zimmerli, C. Als and H. Gerber, Clin. Chem., 1998, 44, 819. 14 Inductively Coupled Plasma Mass Spectrometry, ed. A. Montaser, Wiley-VCH, New York, 1998. 15 H. Vanhoe, F. V. Allemeersch, J. Versieck and R. Dams, Analyst, 1993, 118, 1015. 16 E. H. Larsen and M. B. Ludwigsen, J. Anal. At. Spectrom., 1997, 12, 435. 17 Y. Takaku, T. Shimamura, K. Masuda and Y. Igarashi, Anal. Sci., 1995, 11, 823. 18 H. Baumann, Fresenius’ J. Anal. Chem., 1990, 338, 809. 19 S. Sturup and A. Buchert, Fresenius’ J. Anal. Chem., 1996, 354, 323. 20 P. Schramel and S. Hasse, Mikrochim. Acta, 1994, 116, 205. 21 H. Yamada, T. Kiriyama and K. Yonebayashi, Soil Sci. Plant Nutr., 1996, 42, 859. 22 P. Allain, C. D. Mauras, L. Jaunault, T. Delaporte and C. Beaugrand, Analyst, 1990, 115, 813. 23 I. I. Stewart and J. W. Olesik, J. Anal. At. Spectrom., 1998, 13, 843. Paper 9/04296B Table 2 Concentrations of iodine in acetic acid samples determined by FIA-ICP-MS 1. 7 M NH4OH carrier solution 3.7 M NH4OH carrier solution Correlation Iodine RSD for 4 Correlation Iodine RSD for 4 coefficient for concentration/ injections of coefficient for concentration/ injections of Sample spiked samples ng mL21 unspiked sample spiked samples ng mL21 unspiked sample A (commercially available) 0.982 235 13.8 0.982 132 7.63 B (commercially available) 0.974 467 12.4 0.977 436 3.47 C (not commercially available) 0.986 627 7.75 0.991 180 11.8 D (not commercially available) 0.978 1762 5.22 0.951 1048 5.85 298 Anal. Commun., 1999, 36, 295–298

ISSN:1359-7337    

DOI:10.1039/a904296b

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Communication Mixed-mode capillary electrochromatographic separation of anionic analytes Emily F. Hilder, Miroslav Macka and Paul R. Haddad* School of Chemistry, University of Tasmania, GPO Box 252-75, Hobart, Tasmania 7001, Australia. E-mail: Paul.Haddad@utas.edu.au; Fax: +61 3 6226 2858 Received 7th June 1999, Accepted 21st July 1999 In this work, mixed-mode capillary electrochromatography is introduced as a method for selectivity manipulation in the separation of charged analytes and is investigated for a number of analytes.This concept involves utilising a component of the eluent to permit the chromatographic and capillary electrophoresis (CE) separation mechanisms to contribute in varying proportions to the separation. This approach was first investigated using a combination of CE with reversed-phase liquid chromatography (RP-LC) for hydrophobic, charged analytes (aliphatic sulfonates), and using the concentration of organic modifier in the eluent to control the contributions of CE and RP-LC. However, the use of reversed-phase columns was found to be problematic for mobile phases with less then 50% organic modifier due to the hydrophobicity of the stationary phase causing the column bed to overheat and dry, and low electroosmotic flow (EOF) values ( m @ 17.8 3 1029 m2 V21 s21) caused additional restrictions. In a second case, ion-exchange stationary phases were used, with the type and concentration of a competing anion in the eluent being used to control the contributions of ion chromatography (IC) and CE to the separation.Nine common inorganic anions were separated using a silica based anion-exchange column and phosphate (pH 7.20) or sulfate (pH 8.2) as eluent with direct UV detection at 214 nm and 17 inorganic and small organic anions were separated using a nitrate eluent (pH 6.80) with indirect UV detection at 214 nm. The separation selectivity was shown to be a combination of IC and CE.Introduction Capillary electrochromatography (CEC) has developed rapidly as a technique that can offer separation efficiencies far superior to traditional reversed-phase liquid chromatography (RPLC). 1–3 In particular it has shown promise for separations of neutral analytes using an octadecylsilica (C18) stationary phase, with solvophobic effects being the dominant retention mechanism. Such systems are usually very similar to RP-LC with regard to the separation selectivity, but offer increased separation efficiencies due to the flat flow profile of the mobile phase generated by the electroosmotic flow (EOF).If charged analytes are considered, in addition to increased efficiencies, CEC has the potential to offer new separation selectivities due to the superimposed capillary electrophoresis (CE) separation mechanism. In this case the migration of the analytes will be governed both by their individual electrophoretic mobilities and also by their interaction with the stationary phase.This approach has been demonstrated previously by the separation of several charged, hydrophobic analytes including strong, moderate and weak acids and bases using reversed-phase capillary columns. 4,5 There have been few reports of CEC separations using stationary phases other than reversed-phase materials despite the fact that ion-exchange stationary phases offer the unique opportunity to introduce new separation selectivities for ionic species.Smith and Evans6 were the first to demonstrate this with the separation of some highly polar tricyclic antidepressants using a strong acid cation-exchange (SCX) stationary phase. This separation is expected to be based on a combination of cation-exchange and reversed-phase behaviour, with the analyte movement also being influenced by electrophoretic mobility. Following this, separations of other basic drugs have also been investigated using SCX stationary phases.7 Recently, Wei et al.8 reported the separation of some basic drugs using bare silica as stationary phase.Results indicated that the separation mechanism was a combination of cation-exchange, reversed-phase and normal-phase chromatographic interactions. In the above cases involving organic analytes of relatively large molecular weight and hence small electrophoretic mobilities, the contribution to the separation due to electrophoresis is likely to be minor. This is in contrast to small inorganic ions that have large electrophoretic mobilities, where the separation selectivity should be influenced strongly by the electrophoretic separation mechanism.Comparison of the separation selectivities of CE and ion chromatography (IC) for inorganic ions shows that there is a strong sense of complementarity between the two techniques,9 hence combining IC and CE separation mechanisms in the form of ion-exchange capillary electrochromatography (IE-CEC) offers the possibility of new selectivities for these analytes.This approach has not yet been investigated fully and only a few reports of separations of inorganic anions10,11 and some cations by IE-CEC11 have appeared. Specifically, Kitagawa et al.11 showed the separation of several anions (SO4 22, SO3 22, S2O3 22), cations (lanthanides; Li+, Na+, NH4 +, K+) and anions and cations simultaneously (Li+, Na+, K+, Cl2, NO22, NO32, I2, ClO42, SO4 22) with indirect detection. The system used was pressure driven, with low voltages (+4 kV, 21 kV) being used, so consequently the selectivity changes introduced by the CE mechanism were relatively small.Similarly, Li et al.10 separated I2, IO32 and ReO42 and found that the separation selectivity in the IE-CEC system followed the ion-exchange elution order and there was a relatively small contribution by the CE mechanism. Further separations of inorganic ions have not yet been reported. This paper further develops the concept of mixed-mode CEC as a means of manipulating separation selectivity for anionic analytes.The technique has been examined for the separation of aliphatic sulfonates by mixed CE and RP-LC mechanisms, and some common inorganic anions by mixed CE and IC mechanisms. Experimental Instrumentation All experiments were performed using a HP3D CE system (Hewlett-Packard, Waldbronn, Germany), equipped with a diode array detector operated at 214 nm. Helium gas was connected to the system and was used as the external pressure source.A pressure of 12 bar was applied to both ends of the column and the column was thermostatted at Anal. Commun., 1999, 36, 299–303 29920 °C during all experiments. All samples were injected eletrokinetically at 25 kV for 3 s unless stated otherwise. A non-metallic alignment interface was used instead of the standard metallic interface to avoid problems associated with using a negative separation potential (see Results and discussion).Materials and reagents Fused silica capillaries (75 mm id 3360 mm od) were purchased from Polymicro Technologies Inc. (Phoenix, AZ, USA). Water was purified using a Milli-Q water (Millipore, Bedford, MA, USA) system. The columns were packed with 3 mm C18 Hypersil (Hewlett-Packard, Melbourne, Australia) or 3 mm SAX (capacity Å 200 meq g21, 2.5% carbon; XTec Consultants Ltd., Clwyd, UK) stationary phase. All chemicals used were of analytical reagent grade. Phosphate buffer was prepared from sodium dihydrogen phosphate and titrated to pH 7.2 with sodium hydroxide.Sulfate buffer was prepared from sulfuric acid and titrated to pH 8.2 with tris(hydroxymethyl)aminomethane (Tris). Nitrate buffer was prepared from nitric acid and titrated to pH 6.8 with 1,3-bis[tris(hydroxymethyl)methylamino]propane (Bis-Tris propane). All buffer solutions were filtered through a 0.45 mm membrane filter of Type HA (Millipore, Bedford, MA, USA) and degassed before use.Stock solutions (10 mM) of the anions were prepared from the sodium or potassium salts. The electroosmotic flow was determined by injections of thiourea (0.2 mM in water). CE mobilities were measured in an open capillary (flushed with 1 M NaOH then background electrolyte before use) under the same conditions as the CEC experiments. Column preparation Untreated fused silica capillaries were packed using a slurry packing technique similar to that described by van den Bosch et al.12 A 50 cm length of capillary was connected using a piece of 0.4 mm id polyetheretherketone (PEEK) tubing to a HPLC in-line filter containing a metal frit. The other end of the capillary was connected to a stainless steel slurry reservoir (100 mm 3 2 mm id).Approximately 5 mm of the capillary protruded into the slurry reservoir. The slurry was prepared by adding 30 mg of the packing material to 1 mL of acetonitrile (in the case of C18 material) or water (SAX material) and inserting the mixture into an ultrasonic bath for 5 min.The slurry was then transferred rapidly into the reservoir and pumped into the column using a Haskel 40102 air driven fluid pump (Haskel, Brisbane, Qld, Australia) at a pressure of 9000 psi† with water used as the driving liquid. When the packing was completed, the pump was switched off and the column was allowed to decompress for at least 1 h. The column was then flushed with water at 3000 psi. While flushing with water a frit was sintered near the middle of the column using a homemade heating device.The frit making device was based on that used by van den Bosch et al.12 and consisted of a piece of nichrome ribbon in which a 0.5 mm hole had been drilled. The capillary was threaded through this hole. The temperature of the ribbon was adjusted by varying the voltage across the ribbon and was approximately 500 °C. The heating time varied depending on the packing material (e.g. 2 cycles of 15 s for SAX material).Following fabrication of the outlet frit the in-line filter was removed and the excess packing flushed out. An inlet frit was then sintered in the same way, 25 cm before the outlet frit. The excess capillary before the inlet frit was then removed and a detection window was made by burning off the polyimide coating 15 mm past the outlet frit. The final column (34.5 cm total length, 25 cm packed bed) was then mounted in the HP cartridge and conditioned with mobile phase before use. Results and discussion Approaches to mixed-mode IE-CEC For charged analytes, different separation selectivities are usually obtained depending on whether an electrophoretic (CE) or chromatographic (LC or IC) separation method is used. This is illustrated by the example shown in Fig. 1 for some common inorganic ions. The lines shown connect the same ion in the various techniques hence any crossover of these lines indicates differences in selectivity.This is perhaps best illustrated in the case of fluoride which is eluted very quickly in an IC system, but migrates relatively slowly in a co-EOF CE system. Therefore, if the IC and CE techniques could be combined, the selectivity for fluoride could be controlled by varying the relative contributions of the two separation mechanisms. The term ‘mixed-mode’ is then used to describe such separations in which electrophoresis and chromatography are combined as a means of controlling selectivity in IE-CEC.This approach is in contrast to previous IE-CEC separations in which ion-exchange has been the primary separation mechanism, and the mobile phase was electrokinetically driven.4–6,10 For selectivity manipulation to be possible by the combination of electrophoresis and chromatography, then a method of controlling the influences of the two methods on the separation is necessary. Some possible approaches to this control are outlined below. In the case of hydrophobic, charged analytes, and using a reversed-phase capillary column, the concentration of organic modifier in the mobile phase can be potentially varied as the control parameter.Such a scheme is demonstrated in Fig. 2a, which shows typical separation selectivity for aliphatic sulfonates in RP-LC and in CE. In a mixed-mode CEC system, an organic modifier such as methanol in the mobile phase can be used to control the relative contributions of RP-LC and CE. If no methanol is added then the separation will be dominated by interactions with the stationary phase, but if sufficient methanol is added (such that in the RP-LC system the analytes would be co-eluted) then the separation would be entirely due to differences in analyte mobilities.In this way the relative contributions of the two separation mechanisms can be controlled and the system may be ‘tuned’ for a particular selectivity. A different situation arises for analytes exhibiting little solvophobic interaction, such as inorganic ions.Using an ionexchange stationary phase, the migration of the analytes depends both on the ion-exchange interaction with the stationary phase (i.e. an IC component) and also on the electrophoretic mobility of the analyte (i.e. a CE component). The relative contributions of ion-exchange and electromigration can potentially be controlled by varying the concentration of a competing co-ion in the eluent. The separation selectivities for some common inorganic anions are illustrated in Fig. 2b. A low † 1 psi Å 6.894 757 3 103 Pa. Fig. 1 Separation selectivities for ion chromatography (IC) and capillary electrophoresis (CE) of inorganic anions. The lines join the same anion in the different techniques and where these lines cross is an indication of potential selectivity changes when the techniques are combined. CE counter-EOF indicates that in this mode the analytes migrate in the opposite direction to the electroosmotic flow (EOF), whereas co-EOF describes the analytes migrating in the same direction as the EOF. 300 Anal. Commun., 1999, 36, 299–303concentration of competing co-ion would result in a small degree of ion-exchange displacement, so that ion-exchange interactions between the analyte and the stationary phase would dominate the retention of the analyte. Conversely, a high concentration of competing co-ion would cause effective displacement of the analyte ions from the exchange sites on the stationary phase, so that migration of the analyte ions would be influenced predominantly by electromigration effects.In this way the relative contribution of IC to the overall IE-CEC mechanism and the separation selectivity can be varied. Reversed-phase-CEC system (RP-CEC) The use of reversed-phase capillary columns for the separation of hydrophobic, charged analytes was found to be problematic for several reasons. First, in order to change the relative contributions of RP-LC and CE (i.e.for selectivity control to be possible), some flexibility was required in the composition of the mobile phase, especially in the content of organic modifier. Using the C18 stationary phases available (Hypersil ODS, Waters NovaPak, Lichrosorb RP-18), the system became unstable when < 50% methanol or acetonitrile was added to the mobile phase due to drying and overheating of portions of the column bed. Drying of the packing bed has been observed by others13–15 and has been attributed to differences in EOF in the packing and frits (as sintering the frits results in removal of the ODS coating).With large amounts of methanol being required for stable operation, the potential for control of analyte interaction with the stationary phase was therefore quite limited. Another problem encountered using this approach was the low EOF values observed. Specifically, the highest EOF achieved was 17.8 3 1029 m2 s21 V21 (pH 8.0, 80% acetonitrile), compared with 71.0 3 1029 m2 s21 V21 for an open capillary.For the homologous series of aliphatic (C1 to C8) sulfonates, their electrophoretic mobilities range from 231.1 3 1029 to 253.7 3 1029 m2 s21 V21, so these analytes would not migrate towards the detector in a counter-EOF mode. When separated in the co-EOF mode, the separation selectivity is the same as the ion-exchange selectivity, which excludes the possibility of selectivity manipulation. Considering these practical limitations, selectivity control for hydrophobic, charged analytes using RP-CEC is not feasible using the packing materials available for this study.Ion-exchange-CEC system (IE-CEC) When a silica-based ion-exchange packing material (strong anion-exchanger) was used, problems with adequate wetting on the stationary phase and low EOF did not occur and completely aqueous eluents of varying concentrations (within some constraints due to current considerations) could be used.Additionally, the separation could be performed in the co- EOF mode hence the low magnitude of the EOF ( Å 210 3 1029 m2 s21 V21) was not a problem. However, the use of an anodic separation voltage and a standard HP metallic alignment interface caused arcing to occur and the column to break at the detection window. Whilst the reason for this behaviour is not known, it has been observed previously for negative voltages16 and was remedied by use of a standard HP non-metallic alignment interface.Separations obtained by IE-CEC are shown in Figs. 3–5. For the inorganic anions used as analytes the ion-exchange selectivity coefficients follow the order:17 ClO42 > SCN2 > I2 > S2O3 22 > MoO4 22 > CrO4 22 > SO4 22 > CO3 22 > NO32 > Br2 > NO22 > Cl2 > OCN2 > HCOO2 > BrO32 > ClO32 > F2. On the other hand, the expected CE migration order (in the co-EOF mode) based on the measured mobilities is: BrO32 < HCOO2 < CrO4 22 < F2 < OCN2 < ClO32 < NO32 < NO22 < ClO42 < SO4 22 < S2O3 22 < Cl2, CO3 22, I2 < SCN2 < Br2 < MoO4 22.From this it can be seen that the selectivity observed in the IE-CEC system (Figs. 4–6) was neither solely due to ion-exchange interactions nor was a result of the electrophoretic mobilities of the analytes, but was a combination of the two. Considering the mobilities of the anions in a CE system (mCE) and the IE-CEC system (mCEC), then the difference between these values (Dm) describes the degree to which ion-exchange effects influence the migration of a particular analyte.Mobility data for the phosphate eluent are illustrated in Fig. 6a, from which it can be seen that the Dm values generally followed the expected IC elution order. For example, BrO32 is eluted earlier in an IC system than CrO4 22, which in turn is eluted more quickly than S2O3 22 and the observed mobility data supported this, i.e.Dm(BrO32) < Dm(CrO4 22) < Dm(S2O3 22). As there are relatively few UV absorbing inorganic anions, indirect detection is commonly used in IC17 but has not been utilised in IE-CEC except for a few anions and some cations.11 Therefore, to increase the range of anions that could be detected, indirect detection was investigated. Nitrate, which allows indirect detection at 214 nm, was chosen because it has a moderate ion-exchange selectivity coefficient on strong base anion-exchangers. Separation of 17 anions is shown in Fig. 4 and the mobility data for these analytes are presented in Fig. 6b. These analytes showed the same general trend as observed earlier with the retardation of the analyte increasing as the ionexchange interaction increased. For example, an analyte such as F2 would be eluted close to the void volume in an IC system, and showed a much smaller ion-exchange interaction with the stationary phase in the IE-CEC system than Cl2 which would be eluted later in an IC system.Further, SO4 22, S2O3 22 and SCN2 are known to be strongly retained in an IC system and Fig. 2 a, Separation selectivity for aliphatic sulfonates by RP-LC and counter-EOF CE. Selectivity may be controlled in RP-CEC by varying % methanol in the mobile phase. b, Separation selectivity for some common inorganic anions by IC and co-EOF CE. Selectivity may be controlled in the IE-CEC system by varying [eluent]. Anal. Commun., 1999, 36, 299–303 301subsequently the mobilities of these analytes showed a large relative change in the mixed-mode IC-CEC system.These results showed that the observed separation selectivity was due to both IC and CE influences. However, in order to manipulate the separation selectivity, either the ion-exchange capacity of the stationary phase or the composition and concentration of the eluent must be varied. Fig. 5 illustrates the effect of changing both the eluent type and/or concentration.Comparison of Fig. 3 (5 mM phosphate) and Fig. 5b (2.5 mM sulfate) shows that the separation selectivity was altered by changing the nature of the eluent and hence its eluotropic strength. For example, BrO32 has a small ion-exchange selectivity coefficient and hence its mobility was not affected noticeably by the increase in eluent strength (from phosphate to sulfate). However, Br2 shows a stronger ion-exchange interaction and therefore its mobility was affected more by the increase in eluent strength. This factor, coupled with a higher electrophoretic mobility than BrO32, caused Br2 to be eluted more quickly in the sulfate system, leading to separation of BrO32 and Br2.Fig. 5 also shows that the separation selectivity may be manipulated by varying the eluent concentration. In this case [SO4 22] was varied from 1.25 to 10 mM (at sulfate concentrations > 10 mM the current was too high for stable Fig. 3 Separation of inorganic anions by IE-CEC with direct UV detection at 214 nm.Eluent was 5 mM phosphate, pH 7.20. Separation voltage was 225 kV. Peak identities 1 = Br2, 2 = BrO22, 3 = NO22, 4 = NO32, 5 = I2, 6 = SCN2, 7 = S2O3 22, 8 = MoO4 22 and 9 = CrO4 22 (all 0.2 mM). Fig. 4 Separation of inorganic anions and aliphatic sulfonates by IE-CEC with indirect UV detection at 214 nm. Eluent was 2.5 mM nitrate, pH 6.80. Separation voltage was 225 kV. Peak identities 1 = Cl2, 2 = Br2, 3 = ClO32, 4 = OCN2, 5 = F2, 6 = HCOO2, 7 = C1-SO32, 8 = ClO42, 9 = CO3 22, 10 = SCN2, 11 = S2O3 22, 12 = C2-SO32, 13 = C3-SO32, 14 = C4-SO32, 15 = C5-SO32, 16 = SO4 22 and 17 = C6-SO32 (all 0.2 mM).Fig. 5 Varying eluent concentration (sulfate eluent, pH 8); a, 1.25 mM; b, 2.5 mM; c, 5 mM; and d, 10 mM. Separation voltage was 225 kV. Peak identities 1 = Br2, 2 = BrO32, 3 = NO22, 4 = NO32, 5 = I2, 6 = SCN2, 7 = S2O3 22, 8 = MoO4 22 and 9 = CrO4 22 (all 0.2 mM). Fig. 6 Mobilities (1029 m2 s21 V21) of inorganic anions in CE and IECEC for a, phosphate eluent and b, nitrate eluent. is CE mobility (mCE), is IE-CEC mobility (mCEC) and 8 is the difference between CE and IECEC mobilities (Dm). 302 Anal. Commun., 1999, 36, 299–303operation) and within this range it can be seen that varying the eluent concentration allowed the selectivity to be manipulated. If a higher eluent concentration or a stronger eluent was to be used, greater selectivity control should be possible.Conclusions The concept of mixed-mode IE-CEC has been introduced as a method for manipulating separation selectivity for inorganic ions by combining IC and CE separation mechanisms. IE-CEC has been used for the separation of nine inorganic anions (with direct UV detection) and 17 inorganic and small organic anions with (indirect UV detection). The number of separated anions is greater than previous IE-CEC separations for inorganic anions and it has been shown that the separation selectivity is a result of the combination of IC and CE.Further it has been shown that by varying the eluent type (phosphate or sulfate) or concentration (1.25 to 10 mM sulfate) the separation selectivity can be manipulated by controlling the ion-exchange contribution to the separation mechanism. Acknowledgements The authors would like to thank Professor Peter Myers (Xtec Consultants Ltd.) for providing the ion-exchange packing material used in this study. References 1 A. L. Crego, A. Gonzalez and M. L. Marina, Crit. Rev. Anal. Chem., 1996, 26, 261. 2 M. M. Robson, M. G. Cikalo, P. Myers, M. R. Euerby and K. D. Bartle, J. Microcolumn Sep., 1997, 9, 357. 3 M. G. Cikalo, K. D. Bartle, M. M. Robson, P. Myers and M. R. Euerby, Analyst, 1998, 123, 87R. 4 I. S. Lurie, T. S. Conver and V. L. Ford, Anal. Chem., 1998, 70, 4563. 5 I. S. Lurie, R. P. Meyers and T. S. Conver, Anal. Chem., 1998, 70, 3255. 6 N. W. Smith and M. B. Evans, Chromatographia, 1995, 41, 197. 7 W. Wei, G. A. Luo and C. Yan, Am. Lab., 1998, 30, C. 8 W. Wei, G. A. Luo, G. Y. Hua and C. Yan, J. Chromatogr. A, 1998, 817, 65. 9 P. R. Haddad, J. Chromatogr. A, 1997, 770, 281. 10 D. M. Li, H. H. Knobel and V. T. Remcho, J. Chromatogr. B, 1997, 695, 169. 11 S. Kitagawa, A. Tsuji, H. Watanabe, M. Nakashima and T. Tsuda, J. Microcolumn Sep., 1997, 9, 347. 12 S. E. van den Bosch, J. C. Heemstra Kraak and H. Poppe, J. Chromatogr., 1996, 755, 165. 13 H. Rebscher and U. Pyell, Chromatographia, 1994, 38, 737. 14 J. H. Knox and I. H. Grant, Chromatographia, 1991, 32, 317. 15 R. M. Seifar, W. T. Kok, J. C. Kraak and H. Poppe, Chromatographia, 1997, 46, 131. 16 G. Rozing, personal communication, 1998. 17 P. R. Haddad and P. E. Jackson, Ion Chromatography, Principles and Applications, Elsevier, Amsterdam, 1990. Paper 9/04506F Anal. Commun., 1999, 36, 299–303 303

ISSN:1359-7337    

DOI:10.1039/a904506f

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Communication Ceramic microchips for capillary electrophoresis–electrochemistry Charles S. Henry,a Min Zhong,a Susan M. Lunte,a Moon Kim,b Haim Baub and Jorge J. Santiagoc a Department of Pharmaceutical Chemistry, University of Kansas, Lawrence, KS 66047, USA b Department of Mechanical Engineering and Applied Mechanics, University of Pennsylvania, Philadelphia, PA 19104-6315, USA c Department of Electrical Engineering, University of Pennsylvania, Philadelphia, PA 19104-6315, USA Received 16th June 1999, Accepted 1st July 1999 A capillary electrophoresis–electrochemistry chip constructed from low-temperature co-fired ceramic (LTCC) tape is presented.This is the first report of such a chip constructed in this manner using these materials. Electroosmotic flow at pH 7 is demonstrated by the migration of a neutral marker, catechol. The separation and detection of two phenolic compounds are presented. Introduction Capillary electrophoresis (CE) in the microchip format is a powerful separation technique1 of considerable research interest as evidenced by the number of related publications.2 The advantages of microchip CE include smaller size, reduced reagent volume and cost, the ability to design highly parallel systems and to perform high speed, efficient separations.3,4 The increase in speed without loss of efficiency is a result of decreased injection and detection volumes, shorter delay times between sample loading and separation, and the use of higher field strengths. This results in less band broadening than is seen in standard CE systems.5 One major concern in the construction of microchip CE systems is the choice of substrate material.Because most detection systems are based on optical detection, transparent substrates such as glass, plastics, and elastomers have frequently been used for microchip applications.4 Glass provides a rigid substrate with electroosmotic flow similar to that of fused silica.Unfortunately, high quality glass is costly, fragile, and difficult to bond.6 Plastic and elastomer substrates are less expensive than glass and are amenable to mass production.7,8 A limitation of polymer substrates is the lack of information on their electroosmotic properties. It is desirable to develop microchip CE systems from materials that have properties similar to those of glass but are easier and less expensive to fabricate. The ceramic materials used in this report fit these criteria.The material is an aluminium borosilicate ceramic and should, therefore, have electroosmotic properties similar to those of fused silica. The material may be purchased in large, relatively inexpensive sheets. Fabrication can be accomplished using either milling or laser ablation. Electrical conductors and electrodes can be printed on individual layers, and interlayer hydraulic and electrical connections can be provided by vertical vias. Bonding of multiple layers (up to 80 total) is simple, requiring a lamination and firing step.Ceramic tape technology is, thus, the convergence of layered manufacturing and rapid prototyping. In contrast to standard microfabrication techniques, ceramic tape technology does not require clean rooms, which reduces the cost of fabrication.9 The use of electrochemical detection (EC) with microchip CE has many potential advantages.10–12 Electrochemical instrumentation is inexpensive in comparison to the laser systems typically used for fluorescence detection. The mass sensitivity and selectivity for EC are comparable to fluorescence detection.The microchip need not be transparent. Additionally, both the detection (microelectrodes) and control (potentiostat) components of the system can easily be miniaturized onto the microchip with the CE. These benefits, combined with the range of naturally electrochemically active analytes available,12 make the further development of microchip CE–EC attractive.In this report, we describe the first use of ceramic substrates for the construction of a microchip CE system. Electrochemical detection was used for the characterization of electroosmotic flow. Although electrode placement limited the separation efficiency, two model analytes, catechol and dopamine, were separated in less than 1 min. Experimental Catechol, dopamine, sodium hydroxide, sodium chloride and Ntris-[ hydroxymethyl]-methyl-2-aminoethanesulfonic acid (TES) were obtained from Sigma (St.Louis, MO) and used as received without further purification. All solutions were prepared from deionized water (LabConco). Platinum wire was used for the anode, cathode, working electrode and auxiliary electrode (Goodfellow, Cambridge, UK). The diameter of the working electrode was 25 mm, all other electrodes were 0.5 mm in diameter. A 1 mm Ag wire was used for the reference electrode (Goodfellow). Green Tape 951AX was obtained from DuPont (Wilmington, DE) and processed according to their specifications.Ceramic microchips were produced using conventional milling equipment and ceramic processing procedures. This began with the patterning of two pieces of ceramic tape (Fig. 1). In one piece, a shallow capillary trench was cut with a CNC milling machine (Fadal VMC 15XT). In the second piece, which serves as the cover plate, the openings for the reservoirs were cut using the same CNC machine. These two pieces were stacked, with a third piece of unmodified tape on the bottom for extra strength, and aligned using alignment pins.The stack of materials was laminated for 10 min at 80 °C and 3000 psi† using a Labpress (Carver Inc. Wabash, Indiana) fitted with heated plates. After lamination, the three pieces formed a single unit. Final firing of the ceramic tapes was accomplished in a temperature-controlled oven with a maximum temperature of 875 °C. A standard cross-configuration (Fig. 1) was used for this work.1 The distance between the anode and cathode was 3.0 cm and the effective distance between injection point and detection was 2.5 cm. The capillary channel was 25 mm deep and 100 mm wide. † 1 psi Å 6.894 757 3 103 Pa. Anal. Commun., 1999, 36, 305–307 305Electrochemical detection was accomplished using a standard three-electrode setup. End-column detection was employed as first reported by Gavin and Ewing10 for conventional CE and later modified for the microchip format by Mathies and co-workers.11 A Pt wire was manually mounted at the exit of the capillary channel into the waste reservoir and held in place using epoxy (Miller–Stephenson 907).Ag wire and Pt auxiliary electrodes were mounted in the waste reservoir. The Ag/AgCl reference electrode was generated by oxidizing the Ag wire in the presence of Cl– ions for 5 min. NaCl was added to the waste reservoir to a final concentration of 0.1 M to ensure stability of the reference potential.The oxidation of catechol and dopamine was monitored using a computer-controlled potentiostat (LC- 4C, BioAnalytical Systems, West Lafayette, IN) and data collection system (DA-5, BioAnalytical Systems). Results and discussion To date, two classes of substrate materials have been evaluated for microchip CE systems—glass and polymers. These substrate materials were selected based on the need for optical transparency due to the use of fluorescence detection.4 When electrochemical detection is employed, the need for transparency is eliminated.This led us to evaluate ceramic tape, which has several advantages for microchip CE. One benefit is reduced cost of production, as the base material is less expensive than glass and the same price as many plastics. Additionally, the chips need not be manufactured in a clean room setting. The system described in this communication was constructed totally outside of a clean room. Finally, multiple layers may be integrated into a single device, permitting an increased functionality per unit surface area.The first parameter considered in testing the ceramic microchips was the wall roughness. The use of a milling tool for formation of the channels should result in walls that are more uneven than those generated using photolithography. The increased roughness may result in band broadening and a decrease in peak efficiency. A micrograph of a channel cut with the milling tool is shown in Fig. 2. The observed roughness is approximately 0.2 mm, which is similar to the roughness measured in channels fabricated by laser ablation of polymers. 13 The basic material in the ceramic tape is aluminium borosilicate, which should generate an electroosmotic flow similar to that of fused silica capillaries. The migration of catechol, an electrochemically active molecule that is neutral at pH 7.0, was investigated. The capillary channel was treated initially with 0.1 M NaOH for 1 h.Catechol (25 mM) was injected by applying 600 V between the sample and sample waste reservoirs (Fig. 1) for 5 s. During this time, the intersection of the separation and sample channels became filled with sample solution. The shape of the plug is not well characterized at this time because the top plate is opaque; however, it should resemble that seen by others using a similar geometry and injection technique.14 The electrophoretic mobility of catechol was measured at pH 7.0 with a field strength of 200 V cm21.Fig. 3 shows a representative electropherogram. The electroosmotic flow was calculated to be 3.00 ± 0.05 3 1024 cm2 V21 s21 (n = 4), which is very similar to that of fused silica (4 31024 cm2 V21 s21) at the same pH.15,16 As expected, the linear velocity increased with increasing voltage, going from 0.053 cm s21 at 167 V cm21 to 0.066 cm s21 at 225 V cm21. The use of higher voltages with ceramic chips was not tested during this initial phase; however, no problems are foreseen as the breakdown voltage and thermal conductivity of this material are similar to those of glass.The most obvious feature of the electropherogram (Fig. 3) is the large peak width, which corresponds to an efficiency of ~ 2500 plates m21 and was the same at all three applied potentials. This poor efficiency can be caused by a number of factors. For example, manual placement of the working electrode restricted the distance of closest approach for the electrode and led to significant band broadening.The effect of electrode placement on separation efficiency has been previously investigated for fused silica capillaries.17 The current Fig. 1 Schematic depicting the design and construction of the ceramic microchips. Fig. 2 Micrograph of a 100 mm wide channel in processed ceramic tape. The dark area is the recessed channel. Fig. 3 Representative electropherogram of catechol: 25 mM, 200 V cm21 separation potential. 306 Anal. Commun., 1999, 36, 305–307work focused on the evaluation of ceramic tape as a substrate for microfabricated separation systems. In the future, the system will be designed with screen-printed electrodes that are integrated into the system. This should significantly reduce or eliminate band broadening due to the detector. A second factor that can decrease the efficiency is the plug size. No attempt was made here to use “pinched flow” injection as has been described previously.14 The result was a large sample plug and decreased efficiency.Despite the poor efficiency, separation of two model analytes, dopamine and catechol, was accomplished using the ceramic microchip system. A representative electropherogram for the separation is shown in Fig. 4. The average migration time for dopamine in these separations was 25.3 ± 1.9 s. Dopamine is positive at pH 7 and, therefore, migrates faster than catechol (tm = 41.9 s).Again, the peaks are broad; however, the two peaks are clearly resolved, and the separation occurs in less than 1 min even at this low applied voltage. Conclusions The goal of this report is to establish the use of ceramic substrates for microchip CE systems. Figs. 3 and 4 clearly show the potential of the material for use in CE applications. The current limitation to the use of this system is the poor separation efficiency. Work is currently underway to construct and characterize a system that uses integrated screen-printed electrodes.These electrodes should permit much better separation efficiency, as they will be aligned much closer to the end of the capillary; this should substantially decrease the extracolumn effects. In addition to the use of screen-printed electrodes, work is underway to construct a system with a glass cover plate. This will allow better characterization of the plug shape and the use of optical detection. Acknowledgements M.K., H.B., and J.S.were supported, in part, by DARPA through grant N66001-97-1-8911. DuPont freely provided us with advice and materials. M.K. gratefully acknowledges a fellowship from the International Microelectronics and Packaging Society (IMAPS) Educational Foundation Grant. C.S.H. was supported by a NIH postdoctoral fellowship (F32 GM19889-01). S.L. gratefully acknowledges support from NSF Career grant #9702631. Part of this work was supported by the Center for Bioanalytical Research through the Kansas Technology Enterprise Corporation and a subcontract of NIH SBIR grant #R44GM52272 with Bioanalytical Systems.The authors gratefully acknowledge Nancy Harmony for editorial assistance with the manuscript. References 1 D. J. Harrison, A. Manz, Z. Fan, H. Ludi and H. M. Widmer, Anal. Chem., 1992, 64, 1926. 2 See, for example, Micro Total Analysis Systems ’98, Proceedings of the mTAS’98 Workshop, Banff, Canada, Oct 13–16, 1998, ed. D. J. Harrison and A. van den Berg, Kluwer, Boston, 1998. 3 C. S. Effenhauser, A. Paulus, A. Manz and H. M. Widmer, Anal. Chem., 1994, 66, 2949. 4 C. S. Effenhauser, G. J. M. Bruin and A. Paulus, Electrophoresis, 1997, 18, 2203. 5 C. S. Effenhauser, A. Manz and H. M. Widmer, Anal. Chem., 1993, 65, 2637. 6 D. J. Harrison, K. Fluri, K. Seiler, Z. Fan, C. S. Effenhauser and A. Manz, Science, 1993, 261, 895. 7 C. S. Effenhauser, G. J. M. Bruin, A. Paulus and M. Ehrat, Anal. Chem., 1997, 69, 3451. 8 J. R. Webster, M. A. Burns, D. T. Burke and C. H. Mastrangelo, in Micro Total Analysis Systems ’98, Proceedings of the mTAS’98 Workshop, Banff, Canada, Oct 13–16, 1998, ed. D. J. Harrison and A. van den Berg, Kluwer, Boston, 1998, pp. 249–252. 9 H. H. Bau, S. G. K. Anathasuresh, J. J. Santiago-Aviles, J. Zhong, M. Kim, M. Yi, P. Esponoza-Vallejos and L. Sola-Laguna, in Micro- Electro-Mechanical Systems (DSC-Vol. 66), 1998 International Mechanical Engineering Conference and Exposition, 1998, pp. 491– 498. 10 P. F. Gavin and A. G. Ewing, Anal. Chem., 1997, 69, 3838. 11 A. T. Woolley, L. Kaiqin, A. N. Glazer and R. A. Mathies, Anal. Chem., 1998, 70, 684. 12 L. A. Holland and S. M. Lunte, Anal. Commun., 1998, 35, 1H. 13 M. A. Roberts, J. S. Rossier, P. Bercier and H. Girault, Anal. Chem., 1997, 69, 2035. 14 S. C. Jacobsen, R. Hergenroder, L. B. Koutny, R. J. Warmack and J. M. Ramsey, Anal. Chem., 1994, 66, 1107. 15 X. Huang, R. N. Zare, S. Sloss and A. G. Ewing, Anal. Chem., 1991, 63, 189. 16 T. Tsuda, K. Nomura and G. Nakagawa, J. Chromatogr., 1983, 264, 385. 17 S. Park, S. M. Lunte and C. E. Lunte, Anal. Chem., 1995, 67, 911. Paper 9/04807C Fig. 4 Representative electropherogram of catechol and dopamine: 25 mM each, 200 V cm21. Anal. Commun., 1999, 36, 305–307 307

ISSN:1359-7337    

DOI:10.1039/a904807c

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Communication Electrostatic ion chromatography using hydroxide solutions as mobile phase with suppressed conductivity detection Wenzhi Hu,a Paul R. Haddad,*b Kyoshi Hasebec and Kazuhiko Tanakad a Division of Chemistry, Graduate School of Science, Hokkaido University, Sapporo 060-0810, Japan b Department of Chemistry, University of Tasmania, GPO Box 252-75, Hobart 7001, Australia c Division of Material Science, Graduate School of Environmental Earth Science, Hokkaido University, Sapporo 060-0810, Japan d National Industrial Research Institute of Nagoya, 1-1 Hirate-cho, Kita-ku, Nagoya 462-8510, Japan Received 11th June 1999, Accepted 6th July 1999 An electrostatic ion chromatography (EIC) method for the separation of inorganic anions with detection by suppressed conductivity has been developed, in which analyte retention times can be manipulated by variation of the composition of the mobile phase.A stationary phase prepared by coating silica-based octadecyl material with a sulfobetaine zwitterionic surfactant (Zwittergent 3-14) has been used in conjunction with aqueous hydroxide solutions as the mobile phase.Inorganic anions were eluted in the order SO4 22 < F2 < Cl2 < NO22 < Br2 < NO32 < ClO32 < I2, with retention times increasing with increasing concentration of hydroxide in the mobile phase. Retention times were also dependent on the nature of the counter-cation in the mobile phase, with divalent cations such as Ca2+ and Ba2+ showing longer retention times than monovalent cations such as Li+ and Na+.A retention mechanism involving formation of a binary electrical double layer is proposed, with the thickness of the double layer (and hence analyte retention) being dependent on the concentration of the mobile phase. The EIC system showed high sensitivity for the analyte ions due to the efficiency of the suppression reaction. Detection limits for SO4 22, F2, Cl2, NO22, Br2 and NO32 were less than 1.0 3 1027 mol L21, whilst those for ClO32 and I2 were 3.0 31027 mol L21 and 7.0 3 1027 mol L21, respectively, for a sample injection volume of 100 mL.Introduction Electrostatic ion chromatography (EIC), developed initially by Hu and co-workers, involves the use of a zwitterionic stationary phase formed by coating a silica-based octadecyl reversedphase material with a suitable hydrophobic zwitterionic surfactant (ZS), for example of the sulfobetaine type.EIC has been shown to be useful for the separation of inorganic anions using pure water as the mobile phase and direct conductometric detection.1 However, while this type of EIC gave unique separation selectivity and high sensitivity, the pure water mobile phases had two significant drawbacks. First, the analyte cations and anions were eluted as ion-pairs with the final chromatogram containing peaks for all combinations of the anions and cations in the sample, so that a single analyte anion could appear as several peaks.Second, there was no method available to manipulate the retention times of analytes, apart from varying the amount of surfactant coated onto the stationary phase. Inserting a short cation-exchange column between the sample injector and the separation column to convert the analyte cations to a common species can overcome the first drawback but also cause peak broadening.2 Exclusive partitioning of the analyte ions (i.e., production of a single peak for each analyte ion) without the need for sample pre-treatment has been achieved through the use of aqueous solutions of inorganic salts (e.g., NaCl, NaHCO3 and Na2SO4) as the mobile phase.3,4 The retention order for the analyte anions was identical to that observed using pure water as the mobile phase, but each analyte was eluted as a single, sharp peak.NaHCO3 and Na(CO3)2 mobile phases were particularly suitable because their conductance could be reduced using a conventional ion chromatographic (IC) suppressor.3 In this way, high detection sensitivity was maintained but again the separation selectivity could not be manipulated since the concentration of the salt in the mobile phase exerted only minor effects on retention times over the concentration ranges studied.In the present study we have examined the use of hydroxide solutions as mobile phases, using the rationale that such solutions should show similar separation selectivity to that of other electrolyte solutions, but suppressed conductivity detection should be especially suitable (since hydroxide is easily converted to water in the suppressor).Moreover, the weak anion-exchange affinity of hydroxide could create opportunities for manipulating retention times through variation of the composition of the mobile phase. In this way, the unique advantages of EIC could be maintained and the technique enhanced by the ability to vary analyte retention.Experimental Apparatus The HPLC system used throughout this study was a Shimadzu LC-10A system consisting of a Shimadzu (Kyoto, Japan) LC- 10AT pump, a manual sample injector (Rheodyne, Cotati, CA, USA) with a 2.0 mL injection loop, a CTO-10A column oven (the temperature was set at 30 °C during the analysis) and a CR- 6A Chromatopac data system (Shimadzu). An SPD-10A (Shimadzu) UV–visible detector (which was used to identify the UV-absorbing anions, e.g., NO22, Br2, NO32 and I2) and a CDD-6A (Shimadzu) conductivity detector were used in tandem for the analyte ion detection.An anion self-regenerating suppressor (Model ASRS, Dionex, Salt Lake City, UT, USA) was inserted between the UV–visible detector and the conductivity detector for reducing the background conductance of the mobile phase. The separation column used throughout this study was an ODS-packed column (250 3 4.6 mm, id, Chemical Inspection and Testing Institute, Tokyo, Japan) which had been modified with a sulfobetaine-type zwitterionic surfactant, as described below.Anal. Commun., 1999, 36, 309–312 309Reagents The sulfobetaine-type zwitterionic surfactant used for creating the stationary phase was 3-(N, N-dimethylmyristylammonio)- propanesulfonate (Zwittergent 3-14), obtained from Fluka (Buchs, Switzerland). Analytical-reagent grade inorganic salts used to form the test mixture of model analytes were obtained from Wako Chemicals (Osaka, Japan) and were used as received.Calcium hydroxide (purity 99.9%), barium hydroxide (purity 93%), sodium hydroxide and lithium hydroxide (analytical- reagent grade) were used to prepare the mobile phases and were also obtained from Wako Chemicals. Water used throughout this study was treated in the laboratory using a Millipore (Bedford, MA, USA) Milli-Q water purification system. Column preparation The separation column used throughout this study was obtained by modification of the ODS-packed column with Zwittergent 3-14 surfactant. The surfactant was dissolved using the mobile phase (the hydroxide solution) to give a 30 mm Zwittergent 3-14 solution.The ODS-packed column was then conditioned with the 30 mm C14N3S solution for about 50 min at a flow rate of 1.0 mL min21. The column was then rinsed thoroughly with the desired hydroxide mobile phase (without the surfactant) before being used for the separation of inorganic anions. Results and discussion Use of Ca(OH)2 and LiOH solutions as the mobile phase Fig. 1 shows the use of 1.0 mm Ca(OH)2 solution as the mobile phase for the separation of SO4 22, F2, Cl2, NO22, Br2, NO32, ClO32 and I2 (each 1.0 mm). Good separation was achieved with the observed elution order for these anions being SO4 22 < F2 < Cl2 < NO22 < Br2 < NO32 < ClO32 < I2. This elution order was identical to that observed in previous EIC systems using salt solutions3,4 or pure water1 as the mobile phase.The test mixture of anions was then analyzed using a lower concentration (0.5 mm) Ca(OH)2 solution as the mobile phase; the result is shown in Fig. 2, from which it can be seen that the degree of separation achieved by the more dilute mobile phase was poorer and the retention time for the analyte ions (except for SO4 22, which was eluted close to the void volume of the column) was shorter. Further experiments (Fig. 3) showed that the retention times for the analyte ions, especially for Cl2, NO22, Br2, NO32, ClO32 and I2, increased with increasing concentration of Ca(OH)2 in the mobile phase.Next, 2.0 mm LiOH was used as the mobile phase for the separation of the test mixture and Fig. 4 shows a typical result. Comparison of Figs. 1 and 4 (for which the hydroxide concentrations in the mobile phase were identical) reveals that the Ca(OH)2 mobile phase gave a better separation (and longer retention times) than the LiOH mobile phase. The relationship between analyte retention time and the concentration of LiOH in the mobile phase was also investigated and Fig. 5 shows that retention times increased with [LiOH], especially for Cl2, NO22, Br2, NO32, ClO32 and I2. Retention studies were also conducted using Ba(OH)2 and NaOH aqueous solutions as mobile phases and the results obtained for these mobile phases were very similar to those for the Ca(OH)2 and LiOH mobile phases, respectively, as discussed above. Proposed separation mechanism Three important observations lead to the conclusion that the EIC system introduced in this study involved a separation Fig. 1 Separation of the model analyte anions using C14N3S as the stationary phase and 1.0 mm Ca(OH)2 as the mobile phase. Column, ODSpacked column (250 3 4.6 mm, id) coated with C14N3S; flow-rate 1.0 mL min21; sample 1.0 mm each of SO4 22, F2, Cl2, NO22, Br2, NO32, ClO32, and I2; injection volume 100 mL; detection by suppressed conductivity. Peak identities: 1 = SO4 22, 2 = F2, 3 = Cl2, 4 = NO22, 5 = Br2, 6 = NO32, 7 = ClO32, and 8 = I2.Fig. 2 Separation of the model analyte anions using 0.5 mm Ca(OH)2 as the mobile phase. Chromatographic conditions and peak identities were the same as described in Fig. 1. Fig. 3 Plot of retention time versus concentration of Ca(OH)2 in the mobile phase. Analyte identities are the same as for Fig. 1 310 Anal. Commun., 1999, 36, 309–312mechanism which differed from that of conventional anionexchange IC. First, the divalent anion sulfate was bound with an affinity which was smaller than that for monovalent, hydrophilic anions such as F2 and Cl2.Second, the affinity for binding of the analyte anions (except for sulfate) by the stationary phase increased with increasing concentration of the hydroxide ion in the mobile phase. Third, for an identical concentration of OH2 in the mobile phase, the Ca(OH)2 and Ba(OH)2 mobile phases gave longer retention times and improved separation compared with those given by the LiOH and NaOH mobile phases. The above observations indicate that both the anion (OH2) and the counter-cation in the mobile phase played important roles in the mechanism by which the analyte anions were retained.A binary electrical double layer (EDL) model, proposed in a previous paper by the present authors,4 can be invoked to explain the retention mechanism involved in this EIC system as follows. When a hydroxide mobile phase is used in conjunction with the zwitterionic stationary phase, OH2 ions become bound to the positively charged functional group on the zwitterion (in this case, a quaternary ammonium group, QA), creating an anion-EDL comprising a Stern Layer close to the QA group and a diffuse layer more distant from the QA group.At the same time, the counter-cation in the mobile phase is bound by the negatively charged sulfonate group of the zwitterion, creating a cation-EDL comprising a Stern Layer and a diffuse layer.The uptake of both cations and anions onto the micellar surfaces is a unique ability of the zwitterionic micelles.5–8 The resultant binary-EDL recognizes the analyte anions in the same manner as described earlier for mobile phases comprising NaCl or NaHCO3, namely through a combination of ion-pair formation with cations in the cation- EDL and electrostatic repulsion from the anion-EDL.4 This explains why the retention order for the hydroxide-based mobile phases is the same as that for the NaCl or NaHCO3 mobile phases studied earlier.However, the most significant difference between the results observed for the NaCl or NaHCO3 mobile phases and those for the hydroxide-based mobile phases studied here was that varying the concentration of the mobile phase had little effect on analyte retention for the NaCl or NaHCO3 mobile phases but caused significant changes in retention times for the hydroxide mobile phases (see Figs. 3 and 5). The observed difference in behaviour for the hydroxide mobile phases might be explained by the establishment of a binary-EDL with a different composition to that established for the NaCl or NaHCO3 mobile phases. Since the hydroxide ion has a very low ion-exchange affinity for QA functional groups (in fact, the lowest affinity of any anion), the number of hydroxide ions in the anion-EDL for a given mobile phase concentration can be expected to be smaller than for mobile phases comprising other anions.For the purposes of electroneutrality, the cation-EDL for the above hydroxide mobile phase would also be less dense than for other mobile phases. The outcome of these effects is that the binary-EDL for hydroxide eluents is relatively thin and weaker retention of analytes results. In support of this hypothesis is the fact that the hydroxide mobile phases studied in this work have shown significantly lower retention times than those observed previously for mobile phases of the same concentration (and using the same stationary phase) but containing NaCl or NaHCO3.An increased thickness of the binary-EDL would be caused by an increase in the concentration of the mobile phase and would cause increased analyte retention, as observed in this study. Moreover, analyte retention should also be increased with an increase in the propensity for ion-pair formation of the cation used to form the cation-EDL. Thus, retention times should be longer for Ca(OH)2 and Ba(OH)2 than for LiOH and NaOH mobile phases.The experimental results are in agreement with this prediction. A final consideration in the choice of mobile phase is its compatibility with the detection system. In this case, suppressed conductivity detection has been used and rapid neutralization of the hydroxide by the suppressor is favoured by lower mobile phase concentrations. The Ca(OH)2 mobile phase is preferable in this regard since longer retention times and superior resolution were observed in comparison with equivalent concentrations of LiOH.Detection limits An ability to detect sub-mm levels of inorganic anions is a major advantage of this EIC system. Detection limits for the highly conducting anions (SO4 22, F2, Cl2, NO22, Br2 and NO32) were found to be better than 1.0 31027 mol L21, whilst for less conducting anions, such as ClO32 and I2, the detection limits were 3.0 31027 mol L21 and 7.0 31027mol L21, respectively, determined at a signal+noise of 3 for a sample injection volume of 100 mL, using 1.0 mm Ca(OH)2 solution as the mobile phase.The linear calibration range (r2 > 0.9994) for all of the model analyte ions extended up to 1.0 mm. Relative standard deviations of retention time and peak heights and peak areas for a standard sample containing 1.0 mm each of the model analytes were better than 1.2%. Conclusions A zwitterionic stationary phase used in combination with a Ca(OH)2 mobile phase has provided a practical EIC system for the sensitive analysis of the anions using suppressed con- Fig. 4 Separation of the model analyte anions using 2.0 mm LiOH as the mobile phase. Chromatographic conditions and peak identities were the same as described in Fig. 1. Fig. 5 Plot of retention time versus concentration of LiOH in the mobile phase. Analyte identities are the same as for Fig. 1. Anal. Commun., 1999, 36, 309–312 311ductivity detection. Retention times can be manipulated by varying the concentration of Ca(OH)2 in the mobile phase and this is the first example of an EIC system in which such manipulation is possible. It should be noted here that absorption of carbon dioxide, leading to carbonate formation and resultant baseline drift in suppressed IC with NaOH eluents, is not a significant problem here because of the very low solubility of CaCO3 (Ksp = 4.8 3 1029). Acknowledgement Financial support from the Dionex Corporation is gratefully acknowledged. References 1 W. Hu, T. Takeuchi and H. Haraguchi, Anal. Chem., 1993, 65, 2204. 2 K. Hasebe, T. Sakuraba and W. Hu, J. Liq. Chromatogr. Rel. Technol., 1999, 22(4), 561. 3 W. Hu and P. R. Haddad, Anal. Commun., 1998, 35, 317. 4 W. Hu., P. R. Haddad, K. Hasebe, K. Tanaka, P. Tong and C. Khoo, Anal. Chem., 1999, 71, 1617. 5 Y. Chevalier, N. Kamenka, M. Chorro and R. Zana, Langmuir, 1996, 12, 3225. 6 N. Kamenka, M. Chorro, Y. Chevalier, H. Levy and R. Zana, Langmuir, 1995, 11, 4234. 7 N. Kamenk, Y. Chevalier and R. Zana, Langmuir, 1995, 11, 3351. 8 T. Okada and J. M. Patil, Langmuir, 1998, 14, 6241. Paper 9/04668B 312 Anal. Commun., 1999, 36, 309–312

ISSN:1359-7337    

DOI:10.1039/a904668b

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Communication Analysis of extended X-ray absorption fine structure spectra using annealing evolutionary algorithms Wensheng Cai,*a Liya Wang,a Xueguang Shaob and Zhongxiao Pana a Department of Applied Chemistry, University of Science and Technology of China, Hefei, Anhui, 230026, China b Department of Chemistry, University of Science and Technology of China, Hefei, Anhui, 230026, China Received 8th June 1999, Accepted 28th June 1999 An annealing evolutionary algorithm (AEA), which combines aspects of genetic algorithms and simulated annealing, is proposed to find the global minimum of a non-linear least square function for spectral fitting.By application of the algorithm to the fitting of structural parameters from experimental extended X-ray absorption fine structure (EXAFS) spectra of two Cu samples, it was found that reasonable results were obtained. Comparing with genetic algorithms and EXCURVE88, the AEA method is faster and more accurate in analysis of EXAFS spectra.A simulated annealing algorithm (SAA) is a stochastic optimization method based on the Monte Carlo importancesampling technique.1 It starts from an initial point and takes a single point iterative strategy. The mechanism that it can accept not only the evolved but also the degenerated solutions with the Metropolis acceptance criterion during its annealing procedure, allows the SAA the potential to find the global minimum instead of falling into the local optima.As a global optimization algorithm, the SAA has been widely used to fit non-convex cost-functions arising in a variety of problems, such as the fitting of curves,2 conformation analysis,3 the analysis of atom– atom interactions4 and the optimization of molecular clusters.5 A genetic algorithm (GA) is another global optimization method which starts from an initial population and allows evolution by selection, recombination and mutation.6–8 Compared with the SAA, the GA knows more about the explored areas of the whole space.In this paper a new algorithm combining aspects of population, selection and the simulated annealing procedure is proposed, named annealing evolutionary algorithm (AEA). Therefore the AEA has a larger possibility to escape from local optima than the SAA. Moreover, an improved SAA including a heating procedure for the determination of initial temperature and a local search procedure was implemented in the AEA. Extended X-ray absorption fine structure (EXAFS) has been extensively used as a useful tool in structural studies in recent years.9,10 Generally, standard sample comparison and least square curve fitting methods11,12 are used to obtain structural information from EXAFS oscillation, such as EXCURVE88 by the RC Daresbury laboratory in the UK.11 However, similar standard samples and reasonable initial parameters are needed in the analyzing procedures.Therefore, in our previous studies, the wavelet transform was introduced to process the EXAFS signal13 and also the GA was used to analyze the EXAFS spectrum.14 In this work, the AEA is applied to analyze the structural parameters from the EXAFS oscillation.It was completed by finding the global maximum fit of the experimental EXAFS spectra with their theoretical spectra starting from random initial parameters. EXAFS spectra of two Cu samples were investigated. The results showed that the fitting error is smaller than for the GA14 and EXCURVE8811 methods and the structural parameters obtained are reasonable.Methods Annealing evolutionary algorithms In order to guide the searching procedure to effectively fine the global optimum and escape from the local optimum, simulated annealing algorithms should allow some knowledge about the whole search space to be obtained from the explored results. This has been implemented in the AEA by an evolutionary strategy based on the idea of population and selection.The AEA improves the population of candidate solutions by mutation and selection operation instead of the single point iterative strategy used by the SAA. The algorithm includes three parts: (1) the heating procedure to determine the initial temperature, (2) the annealing procedure to obtain the optimized solution, (3) the local searching procedure to obtain the best solution in which the annealing procedure is the key part, and the heating part is similar to the annealing part except for the acceptance criterion, i.e., only the state with higher energy can be accepted. The whole procedure of the AEA used in this study can be described by the following steps in Fig. 1. (1) Initialize the population and do the heating procedure; the results of temperature and population will be used as the starting point of the annealing procedure. Fig. 1 Flow chart of the AEA. Anal. Commun., 1999, 36, 313–315 313(2) Start the annealing procedure with the initial temperature and population, and evaluate each individual of the population using an objective function.(3) Select a new population from the previous one based on the fitness evaluated. (4) Generate new solutions for each individual of the population which will be accepted or not according to the Metropolis criterion, and evaluate the new solutions. (5) Decrease the temperature and return (3) until the termination criterion is satisfied. (6) Do a local search to improve the final best solution.In Fig. 1 P(k) represents the population at the kth step, xij is the ith (i @ N) individual in the population at the jth (j @ L) step of each Mapkob chain, yij is a new solution generated from xij, f(xij) represents the evaluated value of xij, Dfij is f(yij) 2f(xij), T0 is the initial temperature evaluated by the heating procedure, a is the decreasing rate of temperature used in the annealing procedure, r is a real random number in [0,1], DT is the temperature increment in heating procedure and Temp is used to control the heating (Temp = 0) or annealing (Temp = 1) procedure.For a given point x, the new solution y is generated by eqn. (1): x r k x x r k x + - = - - = IIO y y ( , ) ( , ) UB random LB random 0 1 (1) in which r is a random real number n[0,1], LB and UB are the left and right boundary of the variable respectively. Function y is selected as follows: y h ( , ) k x e x k K = - E I II ¢� ¡Æ ¢«¢« - + EE 1 1 (2) where K is the total number of cooling steps and h is a descending parameter.It will return a value in the range (0, x), which will converge to zero along with an increase of k. Therefore the searching neighborhood can be automatically adjusted as the optimization process proceeds, and a local search is made at the end of the algorithms. Analysis of EXAFS spectrum using the AEA Based on its basic theory, the theoretical curve of EXAFS can be described by eqn. (3),9 c o l ( ) ( ) sin[ ( ) ./ ] / k N kr f k kr k r E k i j j j k r j j j j = + + A - - 2 2 0 2 2 2 0 2625 e e s D (3) where rj, Nj, s j and l are the structural parameters of the jth coordination shell, representing coordination distance (r), coordination number (N), Debye-Waller factor (s) and electron mean free path (l), respectively, o(k) is the phase displacement of scattering, DE0 is used to deduct the effect on the scattering phase displacement caused by the variation of chemical environment, |fj(k)| is the amplitude of scattering.The aim for analyzing the EXAFS curve is to obtain the values of rj, Nj, l, sj and DE0 in eqn. (3). Because the oscillation signal of a differing coordination shell is separated by FT filtering, generally only the five parameters for one coordination shell is considered. In order to use the AEA to optimize the five parameters, random initial values in given ranges were used. The span for parameters r, N, s, l and DE0 are bounded respectively as the following: 2.0 @ r @ 3.0, 8.0 @ N @ 12.0, 0.0 @ s @ 0.1, 4.0 @ l @ 10.0, and 230 @ DE0 @ 30.According to the least square principle, the objective function that is the fitted error for EXAFS spectrum fitting can be constructed as: FI NPT cal NPT = - [ ] = A 1 2 1 C k C k i i i ( ) ( ) exp (4) in which, C(k) = c(k) 3 k3 (5) where NPT is the number of points in the spectra and Ccal and Cexp correspond tohe calculated and experimental spectrum respectively. The AEA was used to optimize the five parameters by minimizing the objective function as in eqn.(4). The experimental EXAFS spectra of Cu samples were investigated. The experimental EXAFS spectra were obtained at the EXAFS station of the Beijing Synchotron Radiation Factory (BSRF) on beam 4W1B using an SiIII double-crystal monochromator. The computer program was written in C++ language and implemented on a Pentium-266. In all calculations, the population size is 40, the maximum heating step is 60, the maximum cooling step is 200, the length of the Mapkob chain is 30, the temperature declining factor a is 0.93, and the descending parameter h in eqn.(2) is 4. Results and discussion Analysis of simulated EXAFS spectrum In order to investigate the efficiency of the AEA in analyzing the spectrum of EXAFS, an oscillation was simulated by eqn. (3) from given structural parameters. Parameters used in the simulation and the results analyzed by the AEA are listed in Table 1.From Table 1, it is clear that parameters used to construct the simulated spectrum can be accurately obtained using the AEA to minimize the objective function FI as in eqn. (4). The comparison between the simulated spectrum and the fitted spectrum is shown in Fig. 2, and the FI obtained is also labeled in the figure. From both the figure and the value of FI, it can be seen that the fitting is satisfactory. Analysis of experimental EXAFS spectra In order to obtain the EXAFS oscillation signal, background removal, Fourier transform and Fourier filtering were con- Table 1 Comparison between the structural parameters obtained by the AEA and in simulation Parameter N r s l DE0 Simulated 8 2.52 0.0800 4.20 10.0 AEA 8 2.52 0.0799 4.16 9.30 Fig. 2 Comparison between the simulated EXAFS oscillation and the fitted result by the AEA. 314 Anal. Commun., 1999, 36, 313¡©315ducted on the measured spectra. Two spectra of Cu samples were investigated.For comparison, three methods, AEA, GA, and EXCURVE88 were used to analyze the spectra. The results obtained by the three methods for sample I and sample II are listed in Tables 2 and 3, respectively. From both the tables, it can be seen that the value of FI optimized by the AEA was much less than the values obtained by the other two methods, and parameter r is well fitted by the three methods; nevertheless the values of parameter s optimized by the first two methods are much different with EXCURVE88, but is closer to the value reported in ref. 12. The comparisons between filtered experimental EXAFS oscillation and the fitted results for the two samples are shown in Fig. 3 and 4, respectively. From Figs. 3 and 4, superior fitting between experimental and analyzed spectra can be found. Therefore the AEA can be used as a useful tool in the analysis of EXAFS spectra, and the advantage of it is that it does not need a standard sample and the results are independent of initial values of parameters.Acknowledgements This work was supported by the National Nature Science Foundation of China (No. 29775001). References 1 L. Ingber, Math. Comput. Modell., 1993, 18(11), 29. 2 D. Y. Chen, X. J. Yang, L. D. Lu and X. Wang, Spectrosc. Lett., 1998, 31, 1513. 3 M. Kinoshita, Y. Okamoto and F. Hirata, J. Chem. Phys., 1999, 110, 4090. 4 R. F. Gutterres, M. Argollo de Menezes, C. E. Fellows and O. Dulieu, Chem. Phys. Lett., 1999, 300, 131. 5 M. A. Moret, P. G. Pascutti, P. M. Bisch and K. C. Mundim, J. Comput. Chem., 1998, 19, 647. 6 J. H. Holland, Adaptation in Natural and Artificial Systems, University of Michigan Press, Ann Arbor, MI, USA, 1975. 7 C. B. Lucasius and G. Kateman, Chemom. Intell. Lab. Syst., 1993, 19, 1. 8 C. B. Lucasius and G. Kateman, Chemom. Intell. Lab. Syst., 1994, 25, 99. 9 K. Lu, Prog. Phys., 1985, 1, 26. 10 P. A. Lee, P. H. Citrin, P. Eisenberger and B. M. Kincaid, Rev. Mod. Phys., 1981, 53(4), Part I, 769. 11 S. J. Gurman, N. Binsted and I. Ross, J. Phys. C, 1984, 17, 143. 12 J. Mustre, Y. Yacoby, E. A. Stern and J. J. Rehr, Phys. Rev. B, 1990, 42, 10843. 13 X. Saho, L. Shao and G. Zhao, Anal. Commun., 1998, 35, 135. 14 X. Shao, G. Cui and G. Zhao, Chin. Spectrosc. Spectral Anal., 1998, 18(1), 106. Paper 9/04553H Table 2 Parameters of CuI obtained by the AEA, GAs and EXCURVE88 Parameter N r s l FI AEA 9 2.52 0.0829 4.17 0.0051 GAs 8 2.52 0.0753 4.38 0.0078 EXCURVE88 8 2.49 0.0165 3.2930 Table 3 Parameters of CuII obtained by the AEA, GAs and EXCURVE88 Parameter N r s l FI AEA 11 2.53 0.0869 4.59 0.0024 GAs 10 2.53 0.0784 4.56 0.0042 EXCURVE88 11 2.49 0.0180 0.0887 Fig. 3 Comparison between the experimental spectrum of CuI sample and the fitted result by the AEA. Fig. 4 Comparison between the experimental spectrum of CuII sample and the fitted result by the AEA. Anal. Commun., 1999, 36, 313–315 315

ISSN:1359-7337    

DOI:10.1039/a904553h

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Communication Cold vapour desorption of volatile organic compounds from an adsorbent trap M. E. Huxham and C. L. P. Thomas* Department of Instrumentation and Analytical Science UMIST, PO Box 88, Sackville St., Manchester, UK M60 1QD. E-mail: paul.thomas@umist.ac.uk Received 1st June 1999, Accepted 23rd July 1999 A computer simulation of active adsorbent sampling was used to investigate the possibility of effecting desorption of volatile organic compounds from an adsorbent trap both rapidly and efficiently via competitive adsorption with a concentrated solvent vapour.The simulation was then enacted in an experimental system to test the concept which used ion mobility spectrometry to follow the desorption of trapped pentan-2-one from a Tenax adsorbent bed, that was subjected to cold vapour desorption. The ion mobility spectra obtained indicated that pentan-2-one was rapidly released from the adsorbent trap with an accompanying enrichment of at least 600%.Desorption by the solvent vapour was shown to have an efficiency of greater than 99.8% in comparison to a thermal desorption method. Introduction There are various methods currently employed for the sampling and analysis of volatile organic compounds in air. One of the most common is trapping volatile organic compounds on a polymeric adsorbent, either actively or passively, followed by thermal desorption and analysis by gas chromatography. 1–8 Thermal desorption is often a two stage process involving a desorption step, generally carried out at temperatures in excess of 250 °C for between 5 and 30 min to ensure complete desorption,9–11 followed by cold trapping of desorbed species and flash heating or temperature controlled elution into the analytical system.The use of high temperatures over the time scale indicated can result in sample degradation either by the breakdown of the polymeric adsorbent to form artefacts, or by the destruction of thermally sensitive analytes.12 This study presents preliminary work concerning a method for desorbing volatile organic compounds from a Tenax adsorbent trap at ambient temperatures, using a high concentration of solvent vapour.The method studied here was based on predicted behaviours obtained from the use of a simple model to simulate the breakthrough behaviour of a binary mixture.13 The model is based on a finite element approach, which is a direct analogy of the separation system described by Craig in 1944.14 The adsorbent trap is divided into n elements.Each element contains a mobile and a stationary phase and analyte moves through the trap from one element to the next. At each time-step, a Type I isotherm is employed to effect distribution of analyte between mobile and stationary phases. The number of trap elements and the analyte partition parameter, related to the partition coefficient used in the Type I isotherm, will depend on a number of factors including the length of adsorbent trap, nature of the adsorbent and sampled flow.These must therefore be determined for each adsorbent system. This is achieved by adjusting the parameters until the model most closely predicts the experimentally observed breakthrough behaviour of the analyte. Experimental Modelling Using model parameters obtained from breakthrough experiments13 a computer simulation was run with a concentration of 16 mg m23 of pentan-2-one sampled at a rate of 15 cm3 min21 for 5 min.The gas-flow through the trap was then reversed and a high concentration of dichloromethane vapour was introduced to the adsorbent bed, sweeping the analyte from the trap. The features of this cold vapour desorption simulation are summarised in Table 1. Cold vapour desorption experiments Vapour sources. Pentan-2-one was selected as a suitable analyte for the cold vapour desorption study. It gives a wellcharacterised response in the positive ion mode of an ion mobility spectrometer and has a boiling point of 110 °C, comparable to that of many hydrocarbons of interest. A vapour concentration of 16.6 mg m23 was achieved using a permeation source, consisting of a sealed chromatography vial containing the analyte, housed in a stainless steel heating block maintained at 50 °C.Purified air was passed over the source at 15 cm3 min21, and the concentration of the vapour generated was determined by mass loss measurements of the permeation source.Dichloromethane was selected as the desorption solvent for this demonstration as it is highly volatile, minimising the possibility of instrument contamination, and does not give a response in the positive ion mode of an ion mobility spectrometer. It also has the attraction of being readily available. The high concentration dichloromethane vapour was generated by bubbling purified air through a reservoir containing the solvent with a gas flow of 15 cm3 min21.The concentration of dichloromethane in the resultant atmosphere was determined from mass loss measurements on the dichloromethane reservoir. A six-port valve allowed the analyte to be desorbed by back flushing the adsorbent trap, while a four-way valve allowed switching between clean air and dichloromethane vapour (Fig. 1). Table 1 Physical characteristics of components used in simulation, where K is partition parameter, C is concentration, Fs is sample flow and ts is sample time Sampling Kpentan-2-one 5.1 3 107 Cpentan-2-one/mg m23 16 Fs/cm3 min21 15 ts/min 5 Desorption Kdichloromethane 6.4 3 105 Cdichloromethane/g m23 10, 100, 1000, 3000 Fd/cm3 min21 15 Anal.Commun., 1999, 36, 317–320 317Instrumentation. Pentan-2-one levels in the eluent from the adsorbent were monitored using a heated ion mobility spectrometer, (Graseby Dynamics Ltd, Watford, Hertfordshire, UK) with a modified inlet to prevent saturation at high concentrations. 15 A sample flow of 15 cm3 min21 was used throughout, as larger flows disrupt the operation of the modified ion mobility spectrometer inlet, the operating parameters are summarised Table 2.Data were collected with a data acquisition card (Blue Chip Technology ADC-42) and a data acquisition program written in Turbo Pascal. The comparative thermal desorption studies were carried out using an Optic 100 injector port with a liquid nitrogen cold trap. Analysis of the cold trap contents was achieved by gas chromatography-mass spectrometry using a Carlo Erba GC 8035 and QMD 1000 mass spectrometer.The operating parameters are summarised in Table 2. Adsorbent traps. The adsorbent traps were constructed from 5 mm Optic 100 injection liners, filled with 60 mg Tenax TA, held in place by silanised glass wool. Heating at 250 °C in a stream of N2 for 8 h conditioned the tubes. Cold vapour desorption. The adsorbent tube was connected into the system described in Fig. 1, and purified air was passed through it for a few minutes. The six-port valve was then switched to ‘sample’ and pentan-2-one vapour passed into the trap for 5 min. During the sampling period the dichloromethane source was introduced by switching the four-port valve. At the end of the sampling period the gas lines up to the six-port valve were flooded with dichloromethane vapour at a concentration of 2.7 kg m23. At the end of the sampling period the six-port valve was switched to desorb.The response was monitored continuously throughout the sampling and desorption phases of the experiment. Efficiency studies. To evaluate the recovery efficiency of cold vapour desorption, the adsorbent trap contents were determined by thermal desorption-gas chromatography-mass spectrometry, both before and after the cold vapour desorption experiments. As pentan-2-one and dichloromethane have the same parent ion mass, the desorption efficiency was determined from the ion counts in the pentan-2-one spectrum of m/z 43, 58 and 71.These ions were not present in the dichloromethane spectrum. Results The model was used to predict the outcome of applying a high concentration solvent vapour to an adsorbent trap previously loaded with analyte. A model partition parameter, K, for pentan- 2-one was determined using a breakthrough curve for pentan- 2-one vapour at 19 mg m23 on a 60 mg Tenax adsorbent bed, shown in Fig. 2. The model was set up so that pentan-2-one vapour at a concentration of 16 mg m23 was sampled until breakthrough occurred.The adsorbent trap array was then reversed and dichloromethane vapour introduced. The simulation was repeated over a range of dichloromethane vapour concentrations, between 10 g m23 and 3 kg m23. Fig. 3 shows the predicted eluent concentration of the analyte following the start of cold vapour desorption. An intense transient spike in the eluent concentration of the analyte is observed.Within 0.5 cm3 the concentration has increased to Fig. 1 Cold vapour desorption apparatus during the desorption cycle: (i) mass-flow controller; (ii) permeation tube housing; (iii) four-port valve to switch between clean gas and the bubbler; (iv) six-port valve to switch between sampling and desorption; (v) adsorbent trap and (vi) solvent bubbler. Table 2 Instrument parameters during solvent desorption and thermal desorption experiments. Ion mobility spectrometer Drift flow 100 cm3 min21 Sample flow 15 cm3 min21 Sheath flow 35 cm3 min21 Temperature 150 °C Optic 100 injector Initial temperature 40 °C Temperature ramp 16° s21 Desorption temperature 230 °C Desorption time 300 s Cold trap temperature < 2140 °C Gas chromatograph Programme 50 to 200 °C @ 10° min21 Column DB-5 Length 30 m Internal diameter 0.25 mm Film thickness 0.25 mm Carrier Gas He Column flow 1 cm3 min21 Split 50 cm3 min21 Mass spectrometer Scan time 0.9 s Interscan time 0.1 s m/z range 14–250 u Fig. 2 A breakthrough curve of pentan-2-one vapour at 16 mg m23 on 60 mg Tenax trap, used to train model. It shows the response of a flame ionisation detector (FID) to pentan-2-one with increasing breakthrough volume. Fig. 3 Simulation of the desorption of pentan-2-one sampled onto a 60 mg Tenax trap, with a solvent vapour concentration of 3000 g m23. The pentan- 2-one is swept from the adsorbent in a sharp desorption profile. 318 Anal. Commun., 1999, 36, 317–320approximately 200 times the sampled concentration.Within 3 cm3 the analyte concentration has returned to below the sampled levels. The simulation was repeated at different desorption vapour concentrations. Table 3 presents both maximum concentration (given relative to the sampled concentration) and desorption volume (measured until the effluent concentration dropped below 1% of the applied concentration). The decrease in desorption time and corresponding increase in peak concentration is approximately linear with respect to desorption vapour concentration.Such results suggest that it should be possible to desorb analytes from an adsorbent bed using a highly concentrated solvent vapour, both efficiently and with a significant concentrating effect. On the basis of the above simulation, cold vapour desorption experiments were carried out in the laboratory. Monitoring the eluent gases presented significant experimental difficulties. Analysis by gas chromatography with samples taken by a gas sampling valve would be too slow to accurately define the desorption profile.Direct monitoring of the effluent by mass spectrometry was not a satisfactory approach because of the perceived adverse affect of the high dichloromethane concentration on the ionisation source. However, work with ion mobility spectrometry has demonstrated the feasibility of directly monitoring minor constituents in mixtures containing an excess of a potentially interfering compound.16 The high concentrations of dichloromethane used in this work (2.7 kg m23) did introduce artefacts into the positive mode mobility spectra.This interference may be due to trace contamination of the dichloromethane or disruption of ionisation process due to the high concentration vapour. However, this was overcome through the use of a background correction algorithm in the data processing. Also note that the non-linear responses associated with ion mobility spectrometry meant that the instrument gave a disproportionately high response to low analyte concentrations.Furthermore the dynamic range of the instrument extended to approximately 2 mg m23 pentan-2-one in air, significantly less than the anticipated maximum concentrations expected during the desorption process. Consequently, the signal from the ion mobility spectrometer would enable the time-scale of the desorption process to be determined; thus, data on the magnitude of the concentrations encountered would be essentially qualitative in nature.Pentan-2-one vapour at 16.6 mg m23 was sampled for 300 s, corresponding to a trapped mass of 1.24 mg. Desorption was then carried out using dichloromethane vapour at a concentration of 2.7 kg m23. Ion mobility spectra were collected every 10 s during the desorption process and these are shown in Fig. 4 as a response surface. Three main features are evident: The reactant ion peak and the protonated monomer and proton bound dimer of the pentan-2-one product ions.The reactant ions may be seen to deplete rapidly in 30 s accompanied by the formation of a proton bound dimer ion peak, indicating complete saturation of the ion mobility spectrometer. After 90 s the dimer depletes with the growth of a monomer ion peak and the reactant ion peak. This behaviour is consistent with the rapid release of pentan-2-one at concentrations greater than 0.1 g m23. The bridging observed between the monomer and dimer form of the ion is indicative of the high concentrations of neutrals in the reaction region of the instrument.Such phenomena have been observed previously with ethylacetate at concentrations greater than 0.1 g m23. Fig. 4 also shows that by 400 s the pentan-2-one concentrations had returned to trace levels, with the bridging phenomena now absent, and the protonated monomer dominating the ion mobility spectrum. The desorption profile obtained had a similar form to that predicted by the model, although the process took significantly longer indicating deficiencies in the model, the operation of the experiment or both.However, on reflection such a discrepancy is not surprising. Tenax is highly soluble in dichloromethane, as is pentan-2-one. Thus strong solvating effects between all three components at the high concentrations of dichloromethane used are probably the strongest factor in the observed retardation of the desorption process.Further, there are several assumptions within the Craig model and the Type I isotherm used within it which result in the predicted desorption being more efficient than that observed experimentally; for example the system is not in equilibrium during desorption. The efficiency of the cold vapour desorption was determined by comparing the quantity of analyte observed (for ions with m/z 43, 58 and 71) upon thermal desorption of a trap loaded with 1.24 mg of pentan-2-one both before and after application of the solvent vapour.The peak observed by thermal desorption after the solvent desorption step was found in both these cases to be less than 0.2% of that observed before solvent desorption. This suggests that the cold vapour desorption process was as efficient as thermal desorption in this instance. Conclusions This preliminary study has demonstrated a new way of desorbing material from an adsorbent bed at ambient temperatures using a high concentration solvent vapour. The extent of enrichment was difficult to quantify due to the interference of dichloromethane vapour and the narrow linear range of the instrument.However, the desorption efficiency was 99.8%, in the case of pentan-2-one, comparable to that obtained by a thermal desorption method. This work also demonstrates that a highly concentrated plume of interfering compound has the potential to effect rapidly, and grossly an analysis, based on active adsorbent sampling.Therefore, methods are needed to ensure that such phenomena are identified and prevented from adversely effecting environmental or workplace sampling, where sampling may occur over extended periods of time. Table 3 Simulated cold vapour desorption of pentan-2-one with dichloromethane indicating the extent of enrichment of pentan-2-one in the adsorbent eluent Desorption vapour conc./g m23 10 100 1000 3000 CCVD/C0 a 0.8 6.6 63 190 Desorption volume/cm3 440 75 11 4.2 a CCVD/C0 is the ratio of the desorbed analyte concentration divided by the concentration of the analyte in the sampled gas.Fig. 4 Ion mobility spectra recorded during the course of the solvent vapour desorption experiment are placed consecutively to create a response surface. The rapid reduction in reactant ion peak, A (4.5 ms) and simultaneous increase in monomer, B (5.3 ms) and dimer, C (6.7 ms) product ion peaks occurs as pentan-2-one is swept from the Tenax adsorbent and enters the instrument inlet.The response then dies away to that of the dichloromethane vapour, indicated by the re-appearance of the reactant ion peak. Anal. Commun., 1999, 36, 317–320 319Although this study has demonstrated the possibility of rapid analyte recovery by cold vapour desorption further research is underway to quantitate the extent of concentration enhancement and identify methodologies for eliminating the effects that have been attributed to solvent vapour-adsorbent interactions. References 1 R. Kostiainen, Atmos. Environ., 1995, 29, 693. 2 A. T. Hodgson, J. R. Girman and J. Binenboym, 79th Annual Meeting of the Air Pollution and Control Association, 1986. 3 H. Knöppel and H. Schauenburg, Environ. Int., 1989, 15, 413. 4 Y. Yokouchi, H. Bandow and H. Akimoto, J. Chromatogr., 1993, 642, 401. 5 J. Begerow and E. Jermann, Fresenius’ J. Anal. Chem., 1995, 351, 549. 6 H. P. König, U. Lahl and H. Kock, J. Aerosol Sci., 1965, 18, 660. 7 E. D. Pellizzari, Environ. Sci. Technol., 1982, 16, 781. 8 B. B. Kebbekus and J. W. Bozzelli, 75th Annual Meeting of the Air Pollution Control Association, 1982, 82–65.2. 9 M. de Bortoli, H. Knöppel, E. Pecchio and H. Vissers, Current trends in diffusive sampling, Comm. Eur. Communities, [rep.] EUR 10555, Diffusive sampling, 1987, 375. 10 J. A. Cucco, Anal. Lett., 1987, 20, 223. 11 C. A. McCaffrey, J. MacLachlan and B. I. Brookes, Analyst, 1994, 119, 897. 12 J. Rudolf, K. P. Muller and R. Koppmann, Anal. Chim. Acta, 1990, 236, 197. 13 M. E. Huxham and C. L. P. Thomas, Analyst, 1999, in the press. 14 L. C. Craig, J. Biol. Chem., 1944, 155, 519. 15 D. Young, C. L. P. Thomas, A. Brittain, J. Breach and G. Eiceman, Anal. Chim. Acta, 1999, 381, 69. 16 J. H. Cross, T. F. Limero, J. L. Love and F. Wang, Talanta, 1997, 45, 19. Paper 9/04353E 320 Anal. Commun., 1999, 36, 317–320

ISSN:1359-7337    

DOI:10.1039/a904353e

出版商:RSC    

年代:1999

数据来源: RSC