ISSN:0003-2654    

DOI:10.1039/AN98712BP029

出版商:RSC    

年代:1987

数据来源: RSC

ISSN:0003-2654    

DOI:10.1039/AN98712FX033

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

ANALAO 112(9) 1197-1346 (1987) September 1987The Analyst11971205120912131217122312291233123712471253125712611265126912731279128512891293129913031309The Analytical Journal of The Royal Society of ChemistryCONTENTSComparison of Electrochemical and Ultraviolet Detection Methods in High-performance Liquid Chromatography forthe Determination. of Phenolic Compounds Commonly Found in Beers. Part 1. Optimisation of OperatingParameters-Patrick J. Hayes, Malcolm R. Smyth, Ian McMurroughComparison of Electrochemical and Ultraviolet Detection Methods in High-performance Liquid Chromatography forthe Determination of Phenolic Compounds Commonly Found in Beers. Part 2. Analysis of Beers-Patrick J. Hayes,Malcolm R. Smyth, Ian McMurroughModification of the High-performance Liquid Chromatographic Retention Behaviour of 2-Aminophenol by theInclusion of Metal Ions in the Mobile PhaseRoger M.Smith, Simon J. Bale, Stephen G. Westcott, M. Martin-SmithDetermination of Volatile Fatty Acids (C,C,) and Lactic Acid in Silage by Gas Chromatography-Richard J. Fussell,David V. McCalleyThe Potential of Fire Assay and Inductively Coupled Plasma Source Mass Spectrometry for the Determination ofPlatinum Group Elements in Geological Materials-Alan R. Date, Alan E. Davis, Yuk Ying CheungAccurate Measurement of Stable Isotopes of Lithium by Inductively Coupled Plasma Mass Spectrometry-Xia F. Sun,Bill T. G. Ting, Steve H. Zeisel, Morteza JanghorbaniDetermination of Aluminium in Serum by Atomic Absorption Spectrometry with the L'vov Platform at DifferentResonance Lines-Francesco Fag iol i, CI i nio Locatel I i, Paolo G i I I iPre-concentration and Determination of Trace Amounts of Lead in Water by Continuous Precipitation in anUnsegmented-flow Atomic Absorption Spectrometric System-Pilar Martinez-Jimenez, Mercedes Gallego,M ig uel Va Ica rcelSpectrophotometric Determination of Zinc in Potable Waters and Insulin with Methylglyoxal Bis(4-phenyl-3-thiosemicarbaz0ne)-M. A.Herrador, A. M. Jimenez, A. G. AsueroCatalytic Determination of Copper(l1) Using the Iron(ll1) - Thiosulphate Reaction and Its Application t o Metals andBiological Samples-Tsutomu Fukasawa, Susuma Kawakubo, Li TanSpectrophotometric Determination of Certain Local Anaesthetics Using 3-Methylbenzothiazolin-2-one Hydrazone-Michael E.El-Kommos, Kamia M. EmaraRapid Determination of Trace Amounts of Phosphate and Arsenate in Water by Spectrophotometric Detection of theirHeteropoly Acid - Malachite Green Aggregates Following Pre-concentration by Membrane Filtration-ChiyoMa tsu ba ra, Y a su ta ka Yam am oto, Ki yo ko Taka m u raSpectrophotometric Determination of Trace Amounts of Nitrite Based on the Nitrosation Reaction with N,N-Bis(P-hydroxypropy1)aniline and its Application t o Flow Injection Analysis-Shoji Motomizu, Shi Chen Rui, MitsukoOshima, Kyoji T6eiDetermination of Dissolved Inorganic Species of Iodine by Spectrophotometric Titration-Maria Pesavento, Antonel laProfumo, Raffaela Biesuzlodimetric Method for the Determination of Sorbic Acid in Soft Drinks-Oi-Wah Lau, Shiu-Fai LukSimultaneous Determination of Aluminium(lll) and Gallium(lll) with Lumogallion by Phase-resolved Fluorimetry-Study of the Formation of Some Substituted Triphenylmethane Reagent - Cationic Surfactant Associates-MaciejAmmonia-sensitive Fibre Optic Probe Utilising an lmmobilised Spectrophotometric Indicator-Peri han Caglar,Analysis of Chlorine - Oxygen Gas Mixtures-Andrew Mills, Anthony CookFlow-through Units for Solid-state, Liquid and PVC Matrix Membrane Ion-selective Electrodes t o Minimise StreamingDevelopment of Ion-selective Electrodes for Use in the Titration of Ionic Surfactants in Mixed Solvent Systems-C.J.Adsorptive Stripping Voltammetric Determination of Low Levels of Daunorubicin-Joseph Wang, Meng Lin, VinceQuality Assurance in Pesticide Formulation Analysis-Louis P.van Dyk, Dirk H. J. P. Reyskens, Debbie J. J. Goosen,Keith R. Vitense, Linda B. McGownJaroszRamaier NarayanaswamyPotentials-Theodore K. Christopoulos, Eleftherios P. DiamandisDowle, Brian G. Cooksey, (the late) John M. Ottaway, William C. CampbellVillaHenri Carstenscontinued inside back coverTypeset and printed by Black Bear Press Limited, Cambridge, EnglanREPORT OF THE ANALYTICAL METHODS COMMllTEE1315 Application of Gas - Liquid Chromatography t o the Analysis of Essential Oils. Part XII. Determination of PatchouliAlcohol in Oil of PatchouliSHORT PAPERSEffect of Phosphonium and Arsonium Salts on the Differential-pulse Polarograms of Three Permitted Synthetic Food 1319Colouring Matters-Arnold G.Fogg, Deepak Bhanot1323 Improved Picrate Method for the Spectrophotometric Determination of Non-ionic Surfactants-Maria de la LuzMerino-Teillet, Maria Eugenia Leon-Gonzalez, Maria Jesus Santos-Delgado, Luis Maria Polo-Diez1327 Determination of Cyanide Through its Reaction with Gelatin-stabilised Gold Sol in Air-Tarasankar Pal, Ashes Ganguly1331 Determination of Aspirin in Pharmaceutical Preparations by Spectrophotometry After Oxidation with PotassiumDichromate-Salah M. Sultan1335 Ultraviolet Spectrophotometric Determination of Phenols in Natural and Waste Waters with Iodine Monobromide-Francisco Bosch, Guillermina Font, Jorge Mafies1339 Determination of Carbon in Uranium Carbide Using Catalytic Oxidation by Manganese Dioxide-V.Chandramouli,Ram Briksha Yadav1343 BOOK REVIEWS" A N A L O I D "COMPRESSED ANALYTICALoffer a saving in the use of lab-oratory chemicals. A range of over30 chemicals includes Oxidizingand Reducing Agents, Reagents forPhotometric Analysis and Indicatorsfor Complexometric Titrations.For full particulars send for List No.51 3( R) to 1-RIDSDALE a COD LTD.Newham Hall, Newby,M iddlesbroug h,Cleveland TS8 9EAor telephone Middlesbrough 31 721 6(Telex: 587765 BASRID)DIFFUSIVEAN ALTERNATIVE APPROACH TOWORKPLACE AIR MONITORINGEDITED BY A BERLIN. R H BROWN, and K J SAUNDERSHardcover SOOppISBN 0 85186 343 3Rice €45.00$87.00RSC MembersPrice €27.00Diffusive Sampling is based ona symposium held inLuxembourg in September1986 and organised jointly bythe Commission of theEuropean Communities andthe United Kingdom Healthand Safety Executive incooperation with the WorldHealth Organization and theRoyal Society of Chemistry.0 Reviews the state of the artof diffusive sampler techniquesStimulates the exchange oftechnical information0 Assess the suitability andrange of applications forworkplace monitoring0 Promotes the furtherdevelopment of this techniqueand its wider use.*- RSC Members should send their orders toThe Royal Society of Chemistry. Membership Manager,30 Russell Sauare. London WClB 5DT U K Non-RSCROYAL & %!:T&Klnformat tonServicesmembers should send their orders to The Royal Societyof Chemistry, thstnbution Centre, Blackhorse Road,Letchworth, Herts SG6 1 HN, U KCircle 003 for further information Circle 005 for further informatio

ISSN:0003-2654    

DOI:10.1039/AN98712BX035

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

ANALYST, SEPTEMBER 1987, VOL. 112 1197 Comparison of Electrochemical and Ultraviolet Detection Methods in High-performance Liquid Chromatography for the Determination of Phenolic Compounds Commonly Found in Beers Part 1 Optimisation of Operating Parameters Patrick J. Hayes and Malcolm R. Smyth* School of Chemical Sciences, National Institute for Higher Education Dublin, Glasnevin, Dublin 9, Ireland and Ian McMurrough Research Centre, Arthur Guinness Son & Co. Ltd., St. James Gate, Dublin 2, Ireland The cyclic voltammetric behaviour of five benzoic acid derivatives, gallic acid, protocatechuic acid, p-hydroxybenzoic acid, vanillic acid and syringic acid, four cinnamic acid derivatives, caffeic acid, p-coumaric acid, ferulic acid and sinapic acid, and two flavanols, (+)-catechin and (-)-epicatechin, commonly found in beers has been studied at a glassy carbon electrode.This study was used to optimise the conditions for the electrochemical detection of these compounds following high-performance liquid chromatographic (HPLC) separation. The HPLC method described is a modification of a previously described method and was optimised with respect to gradient profile, mobile phase composition and mobile phase flow-rate. A comparison was made between an ultraviolet detector and the electrochemical detector operated in series. In general, the electrochemical detector offered lower limits of detection for most of the phenolic compounds studied and, in addition, was found to offer better stability for the determination of concentrations in the 0.1-10.0 p.p.m.range. Keywords: Phenolic compounds; beer; voltammetry; high-performance liquid chromatography; ultraviolet and electrochemical detection Phenolic compounds constitute an important group of natu- rally occurring compounds in plants.l.2 Their presence in alcoholic beverages such as beers arises from the use of barley in the brewing process. The phenolic compounds present in beers show considerable diversity in their structures and hence may be divided into several different classes of compounds (Fig. 1). Amongst the classes of simple monocyclic acids are the hydroxybenzoic and hydroxycinnamic acids. The impor- tance of the monocyclic acids arises from their ability to undergo decarboxylation , either by thermal fragmentation or through the activities of microorganisms.3.4 Consequently, highly flavour-active phenols are produced.5 Although these flavour-active phenols may be appreciated in certain beers,4 in others they may be regarded as distasteful.6 Another important class of phenolic compounds found in beers is flavanols.These polyphenols form a diverse range of compounds and have been classified into three sub-classes, based on their chromatographic behaviour.7 The first of the sub-classes, the simple flavanols, exist as monomers [e.g. , (+)-catechin and (-)-epicatechin], dimers or trimers. Oxida- tion and polymerisation of simple flavanols results in the formation of polymeric flavanols. These polymeric flavanols constitute the second sub-class. Finally, there are the com- plexed flavanols, which are formed on complexation of polyphenols with proteins.8 The over-all significance of flavanols stems from their influence on the quality of beers.Flavanols are assumed to be the precursors of haze formation in unstabilised beers.9 In particular, simple and complexed flavanols have been implicated because their presence in beers seems to be associated with the tendency towards instability.7 Further, the attainment of stability is possible only by reducing the content of simple flavanols below a currently unspecifi- able, but presumably very low, threshold. Considering the importance of phenolic compounds in ascribing flavour and quality to beers, monitoring of these compounds during brewing is necessary. The analytical methods most commonly used for qualitative and quantitative purposes are those based on modern chromatographic separa- tion techniques.In the past, paperlo and thin-layer chroma- tographyll were employed using non-specific detection reagents, rendering the methods inaccurate. Gas - liquid Benzoic acid derivatives: Gallic acid, R1 = R2 = R3 = OH Protocatechuic acid, R1 = R2 = OH, R3 = H pHydroxybenzoic acid, R1 = R3 = H, R2 = OH Vanillic acid, R1 = OMe, R2 = OH, R3 = H R3 R1 4 Syringic acid, R1 = R3 =OMe, R2 = OH R2 Cinnamic acid derivatives: ,COOH Caffeic acid, R1 = R2 = OH, R3 = H pCoumaric acid, R1 = R3 = H, R2 = OH Ferulic acid, R1 = OMe, R2 = OH, R3 = H Sinapic acid, R1 = R2 = OMe, R3 = OH R i R2 Flavanols: moH H o P ~ Y O H (+I-Catechin, R1 = H, R2 = OH (-)-Epicatechin, R1 = OH, R2 = H %. OH * To whom correspondence should be addressed.Fig. 1. Structures of phenolic compounds studied1198 ANALYST, SEPTEMBER 1987, VOL. 112 chromatography (GLC) has been used for the detection of phenols.12 One of the problems associated with GLC analysis is that non-volatile phenolic compounds require derivatisa- tion13 prior to the quantification step, thereby adding to the preliminary preparation of samples. However, the problem of derivatisation does not arise when using high-performance liquid chromatography (HPLC). Further, many different detection methods are available in HPLC analysis.14 In general, ultraviolet (UV) detectors are most popular and have been extensively used in the detection of phenols.lsJ6 However, electrochemical (EC) detectors are being used increasingly in HPLC analysis as they exhibit high sensitivity and selectivity.17 One of the limitations of EC detectors is that the analyte must be electroactive, otherwise it will not be detected. Phenols are electro-oxidisable compounds and therefore are amenable to EC detection.Several analytical methods based on HPLC with EC detection for phenolic compounds have been developed using isocratic con- ditions. 18719 Although both EC and UV detection methods have been separately applied to the determination of phenolic com- pounds in beer, a comparison of both detectors, under identical experimental conditions, has not been undertaken. The purpose of this study was to undertake such a comparison by placing the detectors in series. Furthermore, gradient rather than isocratic conditions were used in conjunction with reversed-phase HPLC, thereby reducing the analysis time.Experimental Instrumentation The liquid chromatograph used was an ACS Model 353 ternary solvent system equipped with an ACS Model 750-11 fixed-wavelength (254 nm) UV detector. An electrochemical detector, based on the wall-jet principle, was connected in series with the UV detector. The EC detector consisted of a Metrohm 656 electrochemical detector (detector cell) and a Metrohm 641 VA detector (electronic controller). The detector cell consisted of a glassy carbon working electrode, a silver - silver chloride reference electrode and a glassy carbon counter electrode. Outputs from both detectors were coupled to a Houston Instrument Omniscribe dual-channel chart recorder thereby allowing easy comparison of the chromato- grams.Chromatographic separations were carried out on a 25 cm x 4.6 mm i.d. reversed phase Nucleosil 10 CI8 stainless- steel column (Macherey - Nagel, Duren, FRG) with a short guard column packed with Nucleosil 10 CIS. Samples were injected on to the column through a Rheodyne Model 7125 injection valve with a 20-pl sample loop. Voltammetric studies of the phenolic compounds were performed using an EG & G Princeton Applied Research Corp. (PARC) Model 174A polarographic analyser. This was used in conjunction with an EG & G PARC Model 303 SMDE. The cell consisted of a glassy carbon working electrode, a platinum wire counter electrode and a silver - silver chloride reference electrode. Reagents The phenolic standards were of analytical-reagent grade, except for the following, which were of technical grade: protocatechuic acid, p-hydroxybenzoic acid, syringic acid and (-)-epicatechin.These four standards were not purified owing to the small initial amounts available. With the exception of ethanol, which was doubly distilled prior to use, all organic solvents used were of HPLC grade and did not require further purification. Both glacial acetic and hydro- chloric acids were of analytical-reagent grade and were diluted, when necessary, with Millipore-grade water. Procedures Preparation of standard solutions Individual standard solutions containing 1000 mg 1-1 of the phenolic compounds were prepared in methanol. A standard mixture containing 1.0 mg 1-1 of each phenolic compound in methanol was then prepared and used in the optimisation of the HPLC method.Standard mixtures of phenolic compounds in the range 0.1-103 mg 1-1 for calibration purposes were prepared from a standard mixture containing 1.0 g 1-1 of each phenolic compound prepared in 5% WV aqueous ethanol. HPLC analysis Prior to use, mobile phase solvents were filtered through a 0.45-pm filter (Millipore) and degassed for 50 min using an ultrasonic bath. During HPLC analysis, the mobile phase solvents were constantly degassed with helium, thereby reducing the possibility of gas bubbles forming in the HPLC pump. The experimental parameters used for HPLC analysis were as follows: electrochemical detector potential, + 1 .O V; ultraviolet detector wavelength, 254 nm; flow-rate, 2 ml min-1; mobile phase composition, A = 3.5% V/V aqueous acetic acid, B = 100% methanol; linear gradient from 0 to 50% V/V B in 30 min.Voltammetric studies Cyclic voltammograms of 1 x l o - 4 ~ solutions of the individual phenolic acid standards in 1 + 1 methanol - 2.5% V/V acetic acid were obtained using a scan rate of 50 mV s-1. All solutions were purged with oxygen-free nitrogen for 4 min prior to recording a voltammogram and a continuous stream of nitrogen was passed over the solutions while measurements were being undertaken. Results and Discussion Voltammetric Studies of Phenolic Compounds In order to establish a suitable amperometric detection potential for the phenolic compounds, preliminary studies were undertaken using cyclic voltammetry. These studies were performed at a glassy carbon working electrode with a scan rate of 50 mV s-1.The results obtained in this preliminary examination are given in Table 1. Only the flavanols gave two anodic peaks, Table 1. Electrochemical parameters measured by cyclic voltammetry for 1 x 10-4 M solutions of phenolic compounds in 1 + 1 methanol - 2.5% WV acetic acid. Scan rate, 50 mV s-l Peak potentialN vs. Ag - AgCl Phenolic compound Benzoic acid derivatives: Gallic acid . . . . . . . . Protocatechuicacid . . . . Syringic acid . . . . . . . . Vanillic acid . . . . . . . . p-Hydroxybenzoicacid . . . . Caffeic acid . . . . . . . . Sinapic acid . . . . . . . . Ferulic acid . . . . . . . . p-Coumaricacid . . . . . . (-)-Epicatechin . . . . . . Cinnamic acid derivatives: Flavanols: (+)-Catechin .. . . . . . . ~~ ~ Anodic peak 0.67 0.75 0.77 0.90 1.20 0.60 0.67 0.83 0.95 (1)0.49 (2)0.85 (1)0.51 (2)0.89 Cathodic peak 0.44 0.57 0.42 0.77 1 .oo 0.35 0.54 0.72 0.63 0.46 0.48 - -ANALYST, SEPTEMBER 1987, VOL. 112 1199 and only the first of these peaks represents a reversible couple. The second anodic peak at more positive potentials represents an irreversible electron transfer step. Single anodic and cathodic peaks were obtained for the benzoic and cinnamic acid derivatives. The difference between the anodic and cathodic peak potential values is considerable, indicating an irreversible electrode process. The variations in oxidation potential between the individual constituents of the benzoic and cinnamic acid groups reflect their structural differences.Although the number of electrons transferred in the various oxidation steps was not experimentally determined, it can be inferred from other studies that it involves two electrons.20,21 In acidic media non-ionised phenols undergo a two-electron irreversible oxidation step to give the phenoxonium ion, which may then undergo further chemical reactions. However, in basic media, phenol ionises to the phenoxide anion, which subsequently undergoes a reversible one-electron oxidation yielding a phenoxy radical. Therefore, given the irreversibility of the electrode process indicated by the cyclic voltammetric results, plus the acidic nature of the supporting electrolyte (pH 3 . 9 , the inference of the two-electron step seems reasonable.If the oxidation of p-hydroxybenzoic acid (I) is considered in terms of a two-electron step, then the following mechanism is possible: COOH COOH COOH Q OH -2e- -H' - + HooQ 0 I II Ill IV A number of possible phenoxonium ion intermediates may be formed on oxidation of I. Closer examination of these intermediary products reveals the improbability of IV ever existing. The COOH group attached to the benzene ring has an over-all electron-withdrawing effect. Hence, any charge on the carbon attached to this COOH substituent is especially unstable.22 Although the COOH group withdraws electrons from all positions of the benzene ring, it especially withdraws them from the carbon to which it is attached. Therefore, this carbon atom, which already suffers an electron deficit, is unable to accommodate this extra positive charge.Conse- quently, it is highly unlikely that IV is formed. Further chemical reaction of I1 and I11 follows the initial electrochemical step. In the presence of methanol, the phenoxonium ion is known to undergo nucleophilic attack resulting in the formation of a single major product.23 Under the present experimental conditions a similar reaction is predicted for phenoxonium ions I1 and 111: COOH CHqOH A 0 V COOH CH30H 111- -H- H3C0 @ H3C0 0 VI Comparison of the anodic peak potential value for p-hydroxybenzoic acid with those for the remaining benzoic acid derivatives reveals that p-hydroxybenzoic acid is the most difficult phenolic compound to oxidise. This is related to structural differences between the individual benzoic acid derivatives. All benzoic acid derivatives except p-hydroxyben- zoic acid have substituents meta to the COOH group.These meta substituents, e.g., OH and OCH3, are electron-releasing, thereby facilitating easy oxidation of the para-OH group. In addition, these electron-releasing substituents are also able to stabilise any intermediary cationic species produced during electrochemical oxidation. For example, if the oxidation of syringic acid (VII) is considered, then the following initial products are possible: COOH COOH COOH I I I H3C0 OCHS H3CO OCH3 OH 0 0 VII Vlll IX Products VIII and IX are the most probable because the electron-donating properties of the rneta-methoxy group stabilise the positive charge on the carbon to which it is attached.Because VIII and IX are only transient products, further chemical reactions, probably involving nucleophilic attack by methanol, will occur, resulting in final products similar to those postulated for p-hydroxybenzoic acid. Cinnamic acid derivatives also undergo an irreversible two-electron oxidation step. A number of possible oxidation sites exist for cinnamic acid derivatives. Chemical oxidation of the vinylic side-chain can occur under mild oxidation con- ditions .24 However, the likelihood of electrochemical oxida- tion at the same position when the side-chain contains the electron-withdrawing COOH group is remote. The other possibility is oxidation of the hydroxy group attached to the aromatic ring. Such a mechanism would be similar to that already postulated for the benzoic acid derivatives.The anodic peak potentials of the two flavanols are very similar, which is hardly surprising as both are isomeric forms of one another. The 30 mV difference between the first anodic and cathodic peaks indicates that the first step is a reversible two-electron transfer. This may be represented as follows: OH X mo R2 + 2e- + 2H+ 1 OH XI The presence of the second more positive anodic peak indicates further irreversible oxidation of XI. The similarity of the peak heights of the two anodic peaks suggests a further two-electron process, yielding XI1 as a possible product. This may then undergo a subsequent chemical reaction:1200 I T ANALYST, SEPTEMBER 1987, VOL. 112 XI - 2e- + H+ The appearance of a single cathodic wave indicates that sufficient product XI is present in the vicinity of the electrode to be reduced back to X.In summary, the cyclic voltammetric results indicate an irreversible two-electron oxidation step for benzoic and cinnamic acid derivatives. The intermediary cationic species probably undergo subsequent nucleophillic attack to give stable products. In contrast, flavanols undergo a reversible two-electron step followed by a further irreversible electro- chemical step. Based on the cyclic voltammetric results, a detection potential of +0.9 V appears suitable for the detection of all phenolic compounds studied except p-hydroxybenzoic acid. This value was subsequently used in the initial HPLC studies. However, because hydrodynamic rather than quiescent con- ditions prevail in the electrochemical detector, further optimi- sation of this detection potential was necessary.HPLC Separation of Phenolic Compounds Choice of gradient conditions The original analytical scheme on which this work is based was developed at Arthur Guinness & Co. Ltd., Dublin, and was applied to the determination of phenolic compounds in beers and worts.25 This method consisted of reversed-phase HPLC separation and incorporated a linearly increasing methanol gradient from 0 to 50% V/V in 2.5% V/V acetic acid in 30 min. When the original gradient conditions described above were used in this study, the same degree of resolution was not achieved. A further increase in acid concentration from 2.5 to 3.5% V/V was necessary before a comparable separation was achieved (Fig.2). The elution order is typical of reversed- phase chromatography, that is, polar compounds elute first, followed by those of decreasing polarity. The chromatogram corresponding to the EC detector does not contain any peak due to p-hydroxybenzoic acid because the applied detector potential was not sufficiently positive to oxidise it. A detector potential of approximately +1.20 V has been shown to be necessary to detect p-hydroxybenzoic acid. Nonetheless, this illustrates how varying the electrochemical detector potential achieves the selective determination of compounds. However, such selectivity is possible only if the anodic peak potentials are separated by a 100 mV difference. Chromatograms obtained with the UV detector contain 10 instead of the expected 11 peaks.This arises because p-hydroxybenzoic acid and (+)-catechin co-elute and appear as a single peak in the chromatogram. This fact was established by analysing binary mixtures of the phenolic compounds and also by injecting each compound separately on to the column to ensure that each compound was being detected by the UV detector. When injected separately, p-hydroxybenzoic acid and (+)-catechin each resulted in a single peak but their retention times were very similar (13.23 and 13.26 min, respectively). 10.01 A 0 15 Time/min 30 Fig. 2. Comparison of HPLC separation of a mixture of phenolic compounds (all at the 1.0 mg 1-l level) using (a) an ultraviolet and (b) an electrochemical detector connected in series. Conditions as stated under Procedures. (1) Gallic acid; (2) rotocatechuic acid; (3) p-hydroxybenzoic acid; (4) (+)-catechin; (55 vanillic acid; (6) caffeic acid; (7) syringic acid; (8) (-)-epicatechin; (9) p-coumaric acid; (10) ferulic acid; and (11) sinapic acid Occasionally, additional peaks were noticed in the later sections of the UV and EC chromatograms. At first it was thought that these peaks were caused by solvent impuri- ties,26,27 but it was later demonstrated that these additional peaks are due to the presence of cis-isomers of the cinnamic acid derivatives.28729 The possibility of cis- and trans-isomer forms of cinnamic acid derivatives arises because of a vinyl side-chain.On exposure of the trans-isomers to UV radiation or daylight, an equilibrium cis - trans mixture forms in methanolic solutions. However, equilibrium mixtures are formed more rapidly in solutions exposed to UV radiation than those exposed to daylight.This explains why it took approximately 1 week before the cis-isomer peaks appeared in UV and EC chromatograms. In all further studies samples were constantly stored in darkness; consequently, isomerisa- tion was prevented. Under the experimental conditions used in this study the cis- and trans-isomers were sufficiently resolved to allow quantifi- cation of the two forms. This illustrates the suitability of the column packing and mobile phase composition to separate geometric isomers of some cinnamic acid derivatives. Another feature of the separation was peak tailing asso- ciated with gallic and protocatechuic acids (peaks 1 and 2 in Fig.2). This tailing was very pronounced, rendering these peaks unsuitable for analytical purposes. The isolation and elimination of this phenomenon are difficult because tailing is caused by any of a variety of factors,30 e.g., column overloading, poor resolution, solute ionisation or poorly packed columns. Given that both compounds contain a COOH group, the possibility of the tailing arising from solute ionisation was checked by increasing the acetic acid concentra- tion from 3.5 to 10.0% V/V. Unfortunately, this did not eliminate the peak tailing; in fact, it tended to exacerbate theANALYST, SEPTEMBER 1987, VOL. 112 1201 problem. Furthermore, there was a considerable loss of resolution in the middle sections of chromatograms from both detectors.The increase in acid concentration caused vanillic, caffeic and syringic acids to co-elute. In general, increasing the acetic acid concentration caused a reduction in the retention times of the phenolic acids because the solvent strength was effectively decreased. The reduction in retention times was greater for the benzoic acid derivatives than it was for the cinnamic acid derivatives. Hence, in the middle section of the UV and EC chromatograms, where both benzoic and cin- namic acid derivatives elute, the relative separation between the two groups changes because their retention times are changing at different rates. As the tailing did not seem to result from solute ionisation, the possibility of it arising from poor resolution was examined. On decreasing the mobile phase flow-rate the appearance of a shoulder on both peaks became apparent.This type of behaviour is associated with a poor resolution between two components so that they appear as a single tailing peak and is referred to as “pseudo-tailing.”30 These pseudo-tails asso- ciated with gallic and protocatechuic acids are probably degradation products of the former phenolic acids. This was suggested by the increased severity of the tailing with increasing age of the standards. A problem usually encountered in reversed-phase HPLC, especially when using bonded-phase packings with low con- centrations (typically less than 10% VIV) of organic solvent, is peak broadening31 due to poor “wetting.” In order to check the “wetting” efficiency of the column by the mobile phase, methanol was added to the aqueous acetic acid.At the beginning of the gradient elution the concentration of organic solvent is minimal, and therefore the greatest improvement was expected in the early sections of the chromatograms. Unfortunately, because of the pseudo-tailing of gallic and protocatechuic acids, no improvement in the parent peaks was observed. For the remaining peaks the increase in methanol concentration did not cause any noticeable decrease in peak broadening. The only effect that the change in methanol concentration had was a reduction in resolution and retention times. Even with only a 1% V/V increase in methanol concentration a considerable loss of resolution in the middle section of the chromatograms occurred. A further increase in methanol concentration only further reduced the resolution in this middle section.In general, a decrease in the retention times of the phenolic compounds was observed as the methanol concentration increased. Furthermore the reten- tion times of the cinnamic acid derivatives decreased faster than those for the benzoic acid derivatives and consequently the separation between these two groups decreased. In order to explain this behaviour, the structures of these two groups must be considered. The presence of a vinyl side-chain in the cinnamic acids is the principal difference between these two groups of compounds. This non-polar side-chain will increase solubility of the cinnamic acid derivatives in methanol. Consequently, increasing the methanol content of the mobile phase will favour the solubility of the cinnamic acid derivatives and correspondingly decrease their retention times more than those of the benzoic acid derivatives.This accounts for the loss of resolution in the middle section of the chromatograms because the retention time of caffeic acid is decreasing faster than that for vanillic acid, so the peaks begin to overlap. Eventually these two acids co-elute and resolution between the two acids is lost. In summary, it appears that the middle sect;w of the UV and EC chromatograms, where benzoic and cinnamic acid derivatives elute, is very sensitive to any change in mobile phase composition. The most suitable mobile phase composi- tion appears to be a linearly increasing methanol gradient from 0 to 50% V/V in 3.5% acetic acid in 30 min.However, under these conditions p-hydroxybenzoic acid and (+)-cate- chin co-elute. An increase in either the acetic acid or methanol content of the mobile phase does not resolve these two compounds. Effect of mobile phase flow-rate Another possible option for improving the resolution in gradient elution is to change the mobile phase flow-rate through the column. Altering the flow-rate changes the column efficiency, thereby changing the number of theoretical plates (N) in the column. In this study the flow-rates examined were limited to values between 1.5 and 2.2 ml min-1. An upper limit of 2.2 ml min-l was imposed because higher flow-rates caused the column pressure to exceed its operating limit of 3000 lb in-2. Flow-rates of less than 1.5 ml min-1 require separation times in excess of 40 min, which is too long if this analytical method is to be useful in industrial applications.The dependence of the detector response, measured in terms of peak height, on the mobile phase flow-rate was markedly different for the UV and EC detectors. First the results for the UV detector are considered. UV chromato- grams of mixtures of phenolic compounds displayed an over-all average decrease Of 2.5% in peak height as the mobile phase flow-rate was increased from 1.5 to 2.0 ml min-1. However, further increase in flow-rate from 2.0 to 2.2 ml min-1 caused an average 4.5% decrease in peak height. Additional examination revealed the consistency of these results. In all instances there was a larger reduction in peak height using flow-rates greater than 2.0 ml min-1.Although these decreases in peak heights are small, the results do indicate some dependence on mobile phase flow-rate. The resolution of peaks was also affected by changing the flow-rate. On increasing the flow-rate from 1.5 to 2.2 ml min-1 a decrease in peak resolution resulted. However, despite this decrease, the peaks were sufficiently resolved even at 2.2 ml min-1 to be separately measured without any difficulty. Although the peaks were well resolved at 1.5 ml min-1, p-hydroxybenzoic acid and (+)-catechin still co-eluted. Because the response of UV detectors is independent of flow-rate, the decrease in peak height on increasing the mobile phase flow-rate cannot be attributed to the operating charac- teristics of the detector.Instead, it may be interpreted in terms of a reduction in the theoretical plate number (N) of the column on increasing the flow-rate. Increasing the flow-rate reduces the column efficiency in terms of N and hence the resolution between peaks decreases. At flow-rates above 2.0 ml min-l the decrease in effective plate number means that the optimium flow-rate is in the range 1.5-2.0 ml min-1 when the UV detector is used. Contrasted with the results for the UV detector are those obtained with the EC detector. The dependence of the peak height or peak current of (+)-catechin, vanillic acid and sinapic acid on mobile phase flow-rate is illustrated in Fig. 3. These results are representative of the trend for the other phenolic compounds not shown. A linear increase in peak currents results on increasing the flow-rate up to 2.0 ml min-1.Further increases in peak current result at higher flow-rates but these increases are not as large as those at the lower flow-rates. The fact that the increase was considerably reduced compared with lower flow-rates was initially inter- preted in terms of increasing electrode passivation by adsorp- tion of oxidation products on the electrode surface. The circumstances giving rise to this type of behaviour can be explained by considering the over-all system. In hydrodynamic voltammetry species are transported to the surface of the working electrode by convection and subsequently these species undergo electrochemical change at this electrode surface. Therefore, increasing the mobile phase flow-rate increases convection so more analyte molecules are transported to the electrode surface.This increase in concen- tration of analyte molecules at the electrode surface will resultANALYST, SEPTEMBER 1987, VOL. 112 1202 100 80 2 . Y t E? 3 60 0 Y ru CL 40 20 Flow-rate/ml min- Fig. 3. Dependence of peak current on flow-rate for (A) (+)- catechin; (B) sinapic acid; and ( C ) vanillic acid in an increased detector response, which is manifested as an increase in peak current in the corresponding chromatograms. For the phenolic compounds studied, this increase in their concentration at the electrode surface with increasing flow- rates was assumed to increase the amount of adsorption because phenols are known to be adsorbed at electrode surfaces.20J3 Consequently, the adsorbed material would reduce the electrode sensitivity.Furthermore, passivation by adsorbed products was thought to be cumulative, as each increase in flow-rate would correspondingly increase the amount of adsorbed material. This was considered to be so because the surface of the working electrode was not cleaned before each incremental change in flow-rate from 1.5 to 2.2 ml min-1. In order to check this hypothesis, the working electrode was polished before each increase in flow-rate. Care was taken not to polish the electrode excessively as this would degrade the surface of the electrode.32 An increase in peak currents was observed following the clearling of the electrode surface. For flow-rates up to 2.0 ml min-1 this increase was generally small, the average increase in peak current being 4 4 % .At the higher flow-rates of 2.1 and 2.2 ml min-1 the increase in peak currents was 7-9%. These results seem to indicate that electrode passivation is occurring but that it is not as significant as originally assumed, especially at higher flow- rates. Despite the increase in peak current, on cleaning the working electrode there is little or no change in the shape of graphs of peak current against flow-rate. At higher flow-rates (2.1 and 2.2 ml min-1) the peak currents do not increase as quickly as those at lower values. Consequently, the smaller increase in peak currents at flow-rates between 2.0 and 2.2 ml min-1 is attributable largely to a reduction in the theoretical plate number of the column. Indeed, this explana- tion would corroborate the results obtained using the UV detector.For both detectors the decrease in N becomes important at flow-rates greater than 2.0 ml min-1. Because the UV detector response is independent of flow-rate, the reduction of N above 2.0 ml min-1 immediately causes a reduction in the peak heights of eluting compounds. The EC detector response is dependent on flow-rate; therefore, increasing the flow-rate increases the resultant peak currents of detected compounds. However, above 2.0 ml min-1 any increase in peak current is largely negated by a decrease in N , and consequently the peak currents do not increase at the same rate as those at flow-rates below 2.0 ml min-1. Based on these results, it appears that a flow-rate of 2.0 ml min-1 is the optimum value for use when the EC and UV detectors are connected in series.Optimisation of electrochemical detector potential In HPLC analysis most electrochemical detectors are based on constant potential amperometry, i. e., the current is measured as a function of time at a constant applied potential. This immediately poses the problem of choosing a suitable detector potential. Usually the applied detector potential corresponds to the minimum potential at which the current reaches its limiting current plateau; it is known as the plateau potential (Eplateau) - By operating at Eplateau the maximum current response of the analyte is obtained at all times. This is especially important in trace analysis where the current responses will be greatly diminished owing to lower analyte concentrations.A number of methods are employed for determining Eplateau based on either direct or indirect measurement. Direct measurement of Eplateau values is made using hydrodynamic voltammetry (HDV). Using HDV the current is measured as a function of the applied detector potential. From the resulting chromatovoltammogram, which is similar in shape to a d.c. polarogram, the value of Eplateau can be measured directly. Indirect determination of Eplateau values is undertaken using techniques such as cyclic voltammetry (CV) or linear sweep voltammetry in a separate cell under quiescent conditions. In this study, preliminary investigations were undertaken using cyclic voltammetry in quiescent conditions, because the determination of Eplateau by CV is considerably faster than by HDV.From this work an arbitrary potential of +0.9 V was chosen for the detection of the phenolic compounds. However, values of Eplateau obtained by indirect methods are often not sufficiently accurate when applied to EC detectors operating under hydrodynamic conditions, because the shape of hydrodynamic voltammograms is dependent on flow-rate, cell geometry and electrode surface properties. Consequently, Eplateau values determined by HDV will differ from those determined by indirect measurement. For example, for species undergoing oxidation, Eplateau as determined by HDV is shifted to more positive values than the value obtained by CV.33 Because of this fact, further optimisation of the original applied detector potential of +0.9 V was necessary.This investigation was limited to potentials between +0.90 and + 1.10 V. When the detector potential was increased to + 1.20 V a significant increase in background current and noise level occurred. This was manifested on the chart recorder as an increase in the base line. It required almost 1 h before the base line settled to its original position and hence before any sample could be injected into the system. Further increases in the applied detector potential would presumably cause even larger background currents as the potential limit of the glassy carbon working electrode is approached. The peak currents measured at the different detector potentials are given in Table 2. Except for (+)-catechin, vanillic acid, p-coumaric acid and p-hydroxybenzoic acid, an increase in detector potential does not alter the response of the phenolic compounds.From the original cyclic voltam- metric studies, plateau potentials of +0.89, +0.90 and +0.95 V were calculated for (+)-catechin, vanillic acid and p-cou- maric acid, respectively. Under hydrodynamic conditions these Eplateau values are shifted to more positive potentials. Because of the large number of phenolic compounds it is impossible to measure each compound at its own Eplateau value as it would need too many measurements for a single sample. Instead, a compromise potential is sought that corresponds to the minimum potential to yield the maximum peak current for all the phenolic compounds. As seen from Table 2, this corresponds to a potential of +1.0 V.It is not possible to detect p-hydroxybenzoic acid at this potential. An EplateauANALYST, SEPTEMBER 1987, VOL. 112 1203 Table 2. Peak currents (nA) of phenolic compounds measured over a range of applied detector potentials Detector potential/V vs. Ag - AgCl Phenoliccompound 0.90 0.95 1.00 1.05 1.10 Gallicacid . . . . . . 60 62 Protocatechuicacid . . 40 42 p-Hydroxybenzoicacid . . ND* ND* (+)-Catechin . . . . 36 46 Vanillicacid . . . . . . 66 80 Caffeicacid . . . . . . 32 32 Syringicacid . . . . . . 92 94 (-)-Epicatechin . . . . 84 82 p-Coumaricacid . . . . 36 60 Ferulicacid . . . . . . 90 88 Sinapicacid . . . . . . 46 48 * Not detected. 64 42 10 54 88 32 92 84 110 90 49 64 64 42 42 24 38 54 54 88 86 32 32 92 92 88 84 112 110 90 90 48 48 Table 3. Comparison of the detection limits of phenolic acids using UV and EC detectors Detection limit/mg 1-1 Phenolic compound UV detector EC detector Gallic acid .. . . . . . . Protocatechuicacid . . . . . . (+)-Catechin . . . . . . . . Vanillic acid . . . . . . . . Caffeic acid . . . . . . . . Syringic acid . . . . . . . . p-Coumaricacid . . . . . . Ferulic acid . . . . . . . . Sinapic acid . . . . . . . . p-Hydroxybenzoicacid . . . . (-)-Epicatechin . . . . . . * Not detected. 0.20 0.20 - 0.08 0.10 0.08 0.08 0.10 0.10 0.10 0.20 0.20 ND* 0.050 0.05 0.05 0.05 0.05 0.05 0.05 0.05 value of + 1.20 V was calculated from CV measurements for p-hydroxybenzoic acid. Obviously the potential value from HDV measurements would be more positive than this. However, it is not feasible to use such a positive detector potential as it would only increase the background currents and noise levels, and consequently the response of the phenolic compounds would be diminished.In general, both background currents and noise levels tend to increase with increasing polarisation voltage and are usually greater in eluents based on mixed solvents than those based on purely aqueous solutions. Therefore, in the present mixed solvent system a detector potential of +1.0 V is probably the most suitable value if the aforementioned problems are to be avoided. In conclusion, it is advisable to use a combination of HDV and CV measurement methods when a suitable Eplateau value is being chosen for a large number of compounds. Using CV for initial measurements an approximate value for Eplateau can be quickly obtained.This value can then be optimised using HDV. In this way the number of measurements that are made by HDV are minimised. Comparison of UV and EC detector characteristics for the determination of phenolic compounds The detection limit is defined as the concentration of analyte that will produce a signal to noise ratio (S/N) of 2 and is considered to be the minimum concentration that can be detected.14 One of the primary limitations to the operation of detectors at high sensitivities is the noise associated with the entire HPLC system. The commonest types of noise14 are short-term noise, long-term noise and drift. Short-term noise is caused by pump pulsations and recorder and detector electronics, whereas long-term noise is produced by tempera- ture and pressure fluctuations.Drift or base-line drift results from either mobile phase or temperature variations. Some of these potential sources of noise were overcome by the choice of the HPLC equipment used in this study. The pump used ensured that pumping pulses were effectively eliminated, thereby reducing the possibility of short-term noise. This is especially important when using electrochemical detectors as they are known to be sensitive to pressure pulses within the detector ~ e l l . 3 ~ The presence of dissolved gases, particularly oxygen, has been found to affect not only flow stability but also UV absorption, causing base-line drift and random noise in the response from the UV detector.35.36 Therefore, constant helium sparging of the mobile phase solvents, preceded by degassing with an ultrasonic bath, ensured that dissolved gases did not cause any problems.From the results in Table 3, it is obvious that the best detection limits were achieved with the EC detector. One of the factors which severely limited the detection limit of the UV detector was the considerable up-scale base-line drift observed on the chromatograms. This up-scale movement at high detector sensitivities was observed over the entire duration of the gradient. Consequently, the peak heights of the later eluting compounds, e.g., p-coumaric, ferulic and sinapic acids, were difficult to measure because the base-line had moved almost off-scale as these compounds eluted. Surprisingly, very little base-line drift occurred in chromato- grams corresponding to the EC detector.This base-line drift is attributable to the continuously changing mobile phase composition inherent in gradient elution chromatography. A further factor that limited the sensitivity of the UV detector was the presence of noise. This was manifested in the chromatograms as a “fuzz,” which widened the base line. Small peaks were enveloped by this “fuzz,” making them indiscernible from the base line. It has been stated that using a fixed rather than a variable-wavelength detector increases the sensitivity because it produces less noise.37 Given the noise problems encountered with the fixed-wavelength detector in this study, it would seem inappropriate to use a variable- wavelength detector as only higher detection limits could be expected.Although some noise was present in chromatograms corre- sponding to the EC detector, it did not have the same deIeterious effects as encountered with the UV detector. The results obtained indicate that the EC detector is less affected than the UV detector by problems of noise and base-line drift. Consequently, it may be concluded that the EC detector is more suitable for quantitative determinations of phenolic compounds. Further, the EC detector response is less affected than a UV detector by gradient elution. If a detector is to be used in quantitative analysis, then the response should be linear with concentration. A wide linear dynamic range, typically 104-105, is desirable so that samples with a wide range of component concentrations may be measured in a single analysis.The linear range of each phenolic compound was obtained by plotting the logarithms of the response (peak height or current) against the logarithm of concentration. Linear UV and EC detector responses over the range 0.1-103 mg 1-1 were obtained for all phenolic com- pounds except gallic and protocatechuic acids. For these two phenolic acids, graphs were prepared over the range 0.2-103 mg 1-1 because they could not be detected below 0.2 mg 1-1 by either detector. A non-linear response was obtained for both phenolics, which is attributable to the pseudo-tailing asso- ciated with these two compounds. Because of this non-linear response, neither phenolic acid could be determined quantita- tively. The slopes of the graphs for those phenolic compounds with linear responses ranged between 0.97 and 1.03.These values were found not to be significantly different from unity for three degrees of freedom at the 95% confidence level.1204 ANALYST, SEPTEMBER 1987, VOL. 112 The linear responses obtained with both detectors indicate their suitability for quantitative determinations of phenolic compounds. Further, the linear dynamic ranges of the two detectors are of similar magnitude to values quoted in the literature.38.39 Precision of measurements The precision of the HPLC method employing the UV and the EC detector in series was investigated by calculating the coefficient of variation of peak heights of individual phenolic compounds following ten injections of a 1.0 mg 1-1 standard mixture on to the column.The coefficient of variation of the UV detector varied from 5.2% for vanillic acid to 9.5% for sinapic acid, whereas that of the electrochemical detector varied from 6.5% for vanillic acid to 9.8% for sinapic acid. The coefficient of variation was found to increase for longer eluting compounds, for compounds that are not completely resolved and for those compounds which give rise to tailing effects, i.e., gallic acid and protocatechuic acid. Although the coefficient of variation was found to be slightly better for the UV detector, it should be remembered that the EC detector was connected downstream of the UV detector and was hence disadvantaged in terms of peak broadening. 1. 2. 3. 4. 5 . 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. References Julkunen-Tiitto, R., J.Agric. Food Chern., 1985, 33, 213. Hahn, D. H., Faubion, J. M., and Rooney, L. W., Cereal Chern., 1983, 60, 255. Wackerbauer, K., Kossa, T., and Tressl, R., in “European Brewery Convention, Proceedings of the 16th Congress, Amsterdam,” 1977, p. 485. Wackerbauer, K., Kramer, P., and Siepert, J . , Brauwelt, 1982, 122, 618. Ryder, D. S., Murray, J. P., and Stewart, M., Tech. Q. Master Brew. Assoc. Am., 1978, 15, 79. Goodey, A. R. andTubb, R. S., J. Gen. Microbiol., 1982,128, 2516. McMurrough, I., Hennigan, G. P., and Loughrey, M. J., J. Inst. Brew., 1983, 89, 15. McMurrough, I., in “European Brewery Convention, Proceed- ings of the 17th Congress, Berlin (West),” 1979, p. 321. Gardner, R. J., and McGuinness, J. D., Tech. Q. Master Brew. Assoc. Am., 1977, 14, 250.Dadic, M., and Van Gheluwe, J. E. A., J. Inst. Brew., 1971,77, 376. Gramshaw, J. W., J. Inst. Brew., 1973, 79, 258. Keith, E. S . , and Powers, J. J., J. Food Sci., 1966, 31, 971. Dallas, F. C., Lautenback, A. F., and West, D. B., Proc. Am. SOC. Brew. Chern., 1967, 25, 103. White, P. C., Analyst, 1984, 109, 677. Garcia Barroso, C., Cela Torrijos, R., and Perez-Bustamante, J. A., Chrornatographia, 1983, 17, 249. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. Charalambous, G., Bruckner, K. J., Hardwick, W. A., and Linneback, A., Tech. Q. Master Brew. Assoc. Am., 1973, 10, 74. Fleet, B., and Little, C. J., J. Chromatogr. Sci., 1974, 12, 747. Kenyherz, T. M., and Kissinger, P. T., J. Agric Food Chern., 1977, 25, 959. Roston, D. A., and Kissinger, P. T., Anal. Chern., 1981, 53, 1695. Vermillion, F. J., and Pearl, I. A., J. Electrochern. SOC., 1964, 3, 1392. Fleischmann, M., and Pletcher, D., in Hush, N. S . , Editor, “Reactions of Molecules at Electrodes,” Wiley, New York, 1971, p. 371. Morrison, R. T., and Boyd, R. N., “Organic Chemistry,” Third Edition, Allyn and Bacon, Boston, 1978, pp. 361-363. Parker, V. D., in Baizer, M. M., Editor, “Organic Electro- chemistry,” Marcel Dekker, New York, 1973, pp. 532-540. Morrison, R. T., and Boyd, R. N., “Organic Chemistry”, Third Edition, Allyn and Bacon, Boston, 1978, p. 396. McMurrough, I . , Roche, G. R., and Cleary, K. G., J. Inst. Brew., 1984, 90, 181. Berry, V. V., J. Chrornatogr., 1982, 236, 279. McCown, S. M., Morrison, B. E., and Southern, D. L., Znt. Lab., 1984, 14, 76. Caccamese, S . , Assolina, R., and Davine, M., Chrornato- graphia, 1979, 12, 545. Conkerton, E. J., and Chapital, D. C., J. Chrornatogr., 1983, 281, 326. Snyder, L. R., and Kirkland, J. J., “Introduction to Modern Liquid Chromatography,” Second Edition, Wiley, New York, Snyder, L. R., and Kirkland, J. J., “Introduction to Modern Liquid Chromatography,” Second Edition, Wiley, New York, Chesney, D. J., Anderson, J. L., Weisshaar, E. E., and Tallman, D. E., Anal. Chim. Acta, 1981, 124, 321. Samuel, A. J., and Webber, T. J. N., in Ryan, T. H., Editor, “Electrochemical Detectors,” Plenum Press, New York, 1984, Ventura, D. A., and Nikelly, J. G., Anal. Chem., 1978, 50, 1017. Bakalyar, S. R., Bradley, M. P. T., and Honganen, R., J. Chrornatogr., 1978, 158, 277. Brown, J. M., Hewins, M., van der Linden, J. H. M., and Lynch, R. J., J . Chrornatogr., 1981, 204, 144. Laurence, J. S . , J. Chrornatogr., 1981, 211, 144. Stulik, K., and Pacakova, V., CRC Crit. Rev. Anal. Chern., 1984, 14, 297. Snyder, L. R., and Kirkland, J. J., “Introduction to Modern Liquid Chromatography,” Second Edition, Wiley, New York, 1979, p. 131. 1979, pp. 791-813. 1979, pp. 269-322. pp. 43-59. Paper A61464 Received December 8th, 1986 Accepted April loth, 1987

ISSN:0003-2654    

DOI:10.1039/AN9871201197

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

ANALYST, SEPTEMBER 1987, VOL. 112 1205 ComDarison of Electrochemical and Ultraviolet Detection Methods in Highlperformance Liquid Chromatography for the Determination of Phenolic Compounds Commonly Found in Beers Part 2.* Analysis of Beers Patrick J. Hayes and Malcolm R. Smytht School of Chemical Sciences, National Institute for Higher Education Dublin, Glasnevin, Dublin 9, Ireland and Ian McMurrough Research Centre, Arthur Guinness Son & Co. Ltd., St. James Gate, Dublin 2, Ireland The previously optimised HPLC method described in Part 1 was applied t o the analysis of some Irish-brewed beers. The results indicated that electrochemical detection was more suitable than UV detection with respect to sensitivity and selectivity. Keywords: Phenolic compounds; beer; high-performance liquid chromatography; ultraviolet and electrochemical detection A method was described in Part 11 for the high-performance liquid chromatographic (HPLC) separation of eleven phenolic compounds commonly found in beers.The method was optimised with respect to gradient profile, mobile phase composition and mobile phase flow-rate. Detection of the phenolic compounds was carried out using an ultraviolet (UV) and an electrochemical (EC) detector connected in series. The responses for both detectors were compared and the electro- chemical detector found to offer greater sensitivity and stability at low concentration levels. In this paper we report our results on the application of this method to the analysis of some Irish-brewed beers. Experimental Instrumentation and Reagents The chromatographic apparatus and reagents were as des- cribed in Part 1.1 Procedures Preparation of standard solutions Individual standard solutions containing 1000 mg 1-1 of the phenolic acid standards were prepared in methanol.Mixtures of the standards containing 1 mg 1-1 of the individual phenolic compounds were then prepared. Standard solutions of phenolic compounds for calibration purposes were also prepared. Stock solutions containing 1000 mg 1-1 of the phenolic standards were prepared in 5% V/V ethanol. From these stock solutions mixtures of phenolics were prepared in the concentration range 0.1-5 mg 1-1. Extraction of phenolic compounds from beers The following procedure is a modification of that proposed by McMurrough et al.2 and was used for extracting phenolic compounds from bottled beer samples and standard solutions prepared in 5% V/V ethanol.Duplimte extractions were carried out on all samples. Beer samples were degassed by simply transferring the sample from one container to another and removing the fob (frothy head) on each transfer. This was carried out for 25 * For Part 1 of this series, see p. 1197.1 f To whom correspondence should be addressed. transfers and was found to be more effective than subjecting the samples to sonication for 20 min. Samples (100 ml) of the degassed (“flat”) beer were then acidified to pH 2.0 with 2 M HC1 before being extracted with two 100-ml portions of isooctane. The remaining aqueous phase was extracted with four 100-ml portions of ethyl acetate and the organic layers were removed following centrifugation at 3000 rev min-1 for 5 min.The ethyl acetate extracts were evaporated to 1-2 ml under reduced pressure at 25°C and diluted to 10 ml with methanol. HPLC analysis The conditions and the procedure for HPLC analysis were as described in Part 1.1 Results and Discussion Standardisation In chromatographic analysis, calibration is usually achieved using either internal or external standardisation.3 Quantifi- cation using an internal standard requires the addition of a compound that elutes closely to the analyte band [ i . e . , has a capacity factor ( k ’ ) within f30% of that of the analyte]. However, when using external standardisation, the actual compounds of interest are used in the preparation of calibration graphs.Calibration methods based on internal standards are more accurate than external calibration methods because the former method compensates for any variations in the separation conditions. Unfortunately, it was not possible to use an internal standard calibration because of the large number of phenolic compounds studied in this investigation. An external standardisation procedure was therefore used. Calibration graphs for UV detection of peak height versus concentration were prepared for all phenolic compounds except gallic, protocatechuic and p-hydroxybenzoic acids. For EC detection it has been suggested that peak height times the peak width at half-height is a more reliable measure of concentration than peak height, as it takes into account the gradual poisoning of the electrode.4 Peak areas were therefore used for calibration graphs obtained with the EC detector. Preliminary attempts at direct injection of beer samples failed as the chromatograms were too complicated.Further- more, it took several hours before all compounds in theANALYST, SEPTEMBER 1987, VOL. 112 injected beer sample eluted from the column. Because of these problems, samples were extracted prior to analysis. Standard solutions containing all eleven phenolic com- pounds were prepared in 5% V/V ethanol, as this medium approximates beer. The concentration of these standard solutions ranged from 0.1 to 5.0 mg 1-1. Prior to extraction, beer samples and standard solutions were acidified to pH 2.0 with 2~ HCl, thereby preventing deprotonation of the phenolic compounds during extraction. Furthermore, the addition of acid releases any bound phenol- ics, ensuring that all compounds are present in the free state.Following acidification, the samples were twice extracted with isooctane. The very non-polar nature of isooctane ensured that only the non-polar constituents were extracted at this stage. The remaining aqueous phase was then extracted four times with ethyl acetate. Less than four extractions resulted in decreased recoveries of the phenolics. It was found to be necessary to centrifuge the ethyl acetate extracts in order to separate the aqueous and organic phases fully. Finally, the ethyl acetate extracts were evaporated under reduced pres- sure to approximately 2.0 ml and subsequently diluted to 10.0 ml with methanol.The use of extraction under reduced pressure enabled the sample to be concentrated at room temperature and avoided any possibility of thermal degrada- tion of the phenolics. Phenolic Content of Irish-brewed Beers Three beers were analysed for their phenolic content by HPLC using UV and EC detection (Table 1). Slight differ- ences were observed on comparing the results obtained with the two detectors, especially for the later eluting compounds. With ferulic and sinapic acids, the results from the EC detector were slightly lower than those obtained using the UV detector. These differences reflect the. difficulties associated with the measurement of peak heights in chromatograms obtained with the UV detector, stemming from the up-scale movement of the base line in these chromatograms.Ferulic and sinapic acids are the last of the phenolic compounds to elute [Fig. l ( a ) ] . By the time these compounds elute, the up-scale base-line movement has noticeably increased com- pared with earlier sections of the chromatogram. In contrast, this up-scale movement is not evident in the chromatogram obtained with the EC detector [Fig. l(b)]. Results were not obtained for gallic, protocatechuic and p-hydroxybenzoic acids, but they were identifi;d as being present in all three beers. For gallic and protocatechuic acids, non-linear responses were obtained on plotting the logarithm of peak height against the logarithm of concentration for standard solutions. This non-linearity was attributed to the pseudo- tailing associated with these two compounds1 and rendered the UV and EC detector responses unsuitable for quantitative purposes.The problem of co-elution between p-hydroxybenzoic acid and (+)-catechin meant that neither could be quantified using the UV detector. However, as the applied potential of the EC detector was sufficient for the electrochemical oxidation of (+)-catechin but not p-hydroxybenzoic acid, the former could be determined from the results with the EC detector and the latter could be determined by a difference calculation. The results in Table 1 are similar to those previously found for Irish-brewed beers.2 In general, the phenolic content of ( a ) 1 5 I 1 0 10 20 tlrnin D Fig. 1. High-performance liquid chromatogram of extract from stout using (a) ultraviolet and ( b ) electrochemical detection.(1) Gallic acid; protocatechuic acid; (3) p-hydroxybenzoic acid; (4) (+)-catechin; vanillic acid; (6) caffeic acid; (7) syringic acid; (8) (-)-epicatechin; p-coumaric acid; (10) ferulic acid; and (11) sinapic acid Table 1. Concentration of phenolic compounds determined in three Irish-brewed beers using HPLC with EC and UV detection. Results are expressed as the means of three determinations Concentration/mg 1- 1 Phenolic compound Benzoic acid derivatives: Vanillic acid . . . . . . . . Syringic acid . . . . . . . . Caffeic acid . . , . . . . . p-Coumaricacid . . . . . . Ferulic acid . . . . . . . . Sinapic acid . . . . . . . . (+)-Catechin . . . . . . (-)-Epicatechin . . . . . . Cinnamic acid derivatives: Flavanols: stout uv EC Ale uv EC Lager uv EC 1.79 1.79 1.16 1.17 0.29 0.30 0.73 0.75 1.35 1.32 0.24 0.20 ND* 0.28 0.10 0.11 1.42 1.41 0.68 0.68 0.13 0.13 0.92 0.91 1.07 1.05 0.15 0.10 ND 0.52 (0.10 (0.10 2.07 2.08 0.85 0.85 0.22 0.23 0.57 0.57 1.90 1.88 0.20 0.18 ND 0.82 0.25 0.25 * Not determinedANALYST, SEPTEMBER 1987, VOL.112 1207 these beers is low, with only minor differences between the individual beers. The origin of these phenolics is due to the raw materials used, i.e., barley and hops.4 It has been discovered that the phenolic content in the finished beer is dependent on the amounts extracted during the preparatory stages of the brewing process. However, it has also been noted that some phenolic acids present in beers may also be formed via the shikimic acid pathway.5 Sinapic acid is formed from ferulic acid, which is formed fromp-coumaric acid. Therefore, if such a reaction were to occur during the brewing process, the concentration of sinapic and ferulic acids would depend on the initial p-coumaric acid concentration.Consequently, there is a need to establish the exact source of these phenolic acids. Typical UV and EC detector responses for a typical extract of stout beer are shown in Fig. l(a) and ( b ) , respectively. Identification of the individual phenolic acids was achieved by comparing retention times measured from chromatograms of standard solutions with those of the beer extracts. Chromato- grams were obtained in an alternating standard-sample sequence. Such a sequence results in an improvement of the working electrode of the EC detector towards “fouling.” It also takes into account any slight changes in, the reference electrode potential and mobile phase composition.6 In addi- tion to the peaks corresponding to the phenolic compounds, some additional peaks were present but were not identified.Most of these additional peaks occurred in the early section of chromatograms and have been attributed to the presence of oligomeric flavanols.2 Some extra peaks were also observed in the middle and later sections of chromatograms and are probably due to derivatives of benzoic and cinnamic acids. In general, chromatograms corresponding to the EC detector contained fewer of these peaks in the middle and later sections. Conclusions The conclusion is that EC detection is more suitable than UV detection in HPLC analysis for the determination of phenolic compounds in beer extracts. In particular, EC detection has the advantage of higher sensitivity than UV detection, and also offers greater selectivity, which is very useful when analysing “real” samples as it reduces matrix effects and consequently improves the quantitation and identification of analyte peaks.Furthermore, the EC detector was almost insensitive to the changes in the mobile phase conditions associated with gradient elution. Consequently, steady base lines were achieved at high detector sensitivity settings. Unfortunately, severe base-line drift occurred under gradient conditions when the UV detector was operated at high sensitivities. This drift made the measurement of the peak heights of later eluting phenolics particularly difficult.During the period of this study some fouling of the working electrode by adsorbed phenols or their oxidation products was encountered, which necessitated the removal of the electrode for cleaning purposes. A number of disadvantages are associated with this type of procedure. Manual cleaning can often cause the electrode surface to be seriously damaged. There is also the possibility of the surface becoming contami- nated, either during the cleaning process or in handling operations subsequent to the cleaning step. Finally, the procedure of replacing the electrode in the detector cell can cause air bubbles to be trapped in the cell. Elimination of these problems would seem to be possible by using pulsed amperometric detection .7,8 Application of a triple-step potential waveform would electrochemically clean the surface of the working electrode. It also has the advantage of increasing the signal to noise ratio.8 1. 2. 3. 4. 5 -. 6. 7. 8. References Hayes, P. J . , Smyth, M. R., and McMurrough, I., Analyst, 1997, 112, 1197. McMurrough, I., Roche, G. P., and Cleary, K. G., J. Inst. Brew., 1984,90, 181. Dadgar, D., and Smyth, M. R., Trends Anal. Chem., 1986, 5 , 115. Albery, W. J., Beck, T. W., Brooks, W. N., and Fillenz, M., J. Efectroanal. Chem., 1981, 125, 205. Kenyhercz, T. M., and Kissinger, P. T., J. Agric. Food Chem., 1977, 25, 959. Roston, D. A., and Kissinger, P. T., Anal. Chem., 1981, 53, 1695. Edwards, P., and Haak, K. K., Int. Lab., 1983, June, 38. Johnson, D. C . , Polta, J. A., Polta, T. Z . , Neuburger, G. G., Johnson, J., Tang, A. P. C., Yeo, J.-H., and Baur, J., J. Chem. SOC., Faraday Trans. 1, 1986, 82, 1081. NOTE-Reference 1 is to Part 1 of this series. Paper A71118 Received March 25th, 1987 Accepted April 27th, 1987

ISSN:0003-2654    

DOI:10.1039/AN9871201205

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

ANALYST, SEPTEMBER 1987, VOL. 112 1209 Modification of the High-performance Liquid Chromatographic Retention Behaviour of 2-Aminophenol by the Inclusion of Metal Ions in the Mobile Phase Roger M. Smith and Simon J. Bale Department of Chemistry, University of Technology, Loughborough, Leicestershire LE 7 7 3TU, UK and Stephen G. Westcott and M. Martin-Smith Glaxo Group Research, Ware, Hertfordshire SG 72 ODJ, UK The effect of transition metals as components of the mobile phase on the retention of 2-aminophenol in high-performance liquid chromatography was studied on both ODs-silica and porous polymer columns. 2-Aminophenol formed charged 1 : 1 complexes with the metal ions to give a reduction in retention through a mechanism analogous to ion-pair chromatography. Different metal ions and a range of concentrations and eluent pHs were investigated. Keywords: High-performance liquid chromatography; transition metal ions; mobile phase additives; metal chelates; 2-aminophenol High-performance liquid chromatography (HPLC) is being used increasingly for the analysis of inorganic compounds and considerable interest has been shown in the separation of organometallic and metal coordination compounds.1-3 Many methods for analyses of metal ions have been reported that involve the pre-column formation of metal chelates using ligands commonly used for metal extraction such as P-di- ketones4 and dithiocarbamates (e.g. , reference 5 ) , followed by the separation of the neutral complexes. A number of techniques have also evolved that use metal ions to improve or modify the separation of selected groups of compounds.In early work, normal-phase separations on silica columns loaded with silver ions (Ag+) were used for the separation of unsaturated compounds by argentation chro- matography6 and more recently Rh+- and Cdz+-coated columns have also been used.7 These methods suffered from a deterioration of the separation due to stripping of the metal ions from the silica surface. To overcome this problem, Chow and Grushka linked copper ions to the silica surface as complexes with propylamine ,8 6-diketones or dithiocarba- mates.9 Metal complexes have also been widely used in reversed- phase HPLC as mobile phase additives for the analysis of amino acids10 and the resolution of optically active com- pounds.11 In addition, zinc and nickel C12-dien complexes were used by Cooke et al.12 to alter the chromatography of peptides and sulphur drugs.However, very little use has been made of uncomplexed metal ions as mobile phase additives to modify selectively the separation of compounds, except for the use of silver nitrate to separate saturated and unsaturated fatty acids.13 Walters and Raghaven14 found that the addition of zinc chloride to the mobile phase would selectively alter the separation of isomeric 0-, m- and p-aminobenzoic acids. As part of a study of the degradation of arylamines, Sternson and De.Witte15 demonstrated that the addition of nickel salts to the eluent would reduce the retention time of 2-aminophenol but not that of 3- or 4-aminophenol. They subsequently examined this effect in greater detail in a range of methanol - buffer eluents and noted that increasing the nickel ion concentration caused a decrease in the retention time.16 These results were compared with the effect of nickel ions on the retention of a series of related compounds that would not undergo chelation and negligible effects were noted.The study also included some limited work with zinc and mercury(I1) ions, which suggested that these metals had only a small effect even on 2-aminophenol. However, in some of their work the influence of the metal ions did not appear as systematic as might have been expected from a simple ion-pair type of mechanism to form a 1 : 1 complex. This work is part of a series of studies looking at the general application of metal ions as mobile phase additives and follows earlier studies using metal ions in the eluent to alter the retentions of dithiocarbamate ions by on-column complexa- tion.17.18 This investigation extends the range of metals examined by Sternson et al.and examines in detail the role of the metal ions, their concentration and the mobile phase pH on the chromatography of 2-aminophenol, with the aim of developing a general method for altering selectively the retentions of compounds that can complex with metal ions. Experimental Reagents Laboratory-reagent grade ammonium acetate and transition metal acetates of cadmium, copper, cobalt, nickel, manganese and zinc and HPLC-grade methanol were obtained from Fisons Scientific Apparatus (Loughborough, Leks., UK). 2-Aminophenol (Aldrich Chemical, Gillingham, Dorset , UK) was recrystallised from water before use.Ammonium acetate solution, 0.26 M. A 2.6 M stock solution in water was diluted to 0 . 2 6 ~ before use with water or the inclusion of the appropriate 0.2 M metal acetate solution. Apparatus The liquid chromatograph consisted of a Pye Unicam PU 4010 pump, a Rheodyne 7125 injection valve fitted with a 20-yl sample loop and a Pye Unicam PU 4020 variable-wavelength UV detector set at 254 nm. Chromatograms were recorded on a Linseis chart recorder with a Hewlett-Packard 3390A integrator. The separations were carried out using Shandon stainless-steel columns (100 mm x 5 mm i.d.), which had been slurry packed with Hypersil-ODS (5 ym) (Shandon Southern, Runcorn, UK) or PLRP-S (porous polystyrene - divinylben- zene copolymer, 5 ym) (Polymer Laboratories, Church Stretton, UK).The temperature of the column was main- tained at 30 "C with a thermostated water-jacket. The mobile phase flow-rate was 1 ml min-1. pH measurements were made with a Pye Unicam 390 pH meter.1210 ANALYST, SEPTEMBER 1987, VOL. 112 Sample and Mobile Phase Preparation Solutions of 2-aminophenol at concentrations of 10-2, 10-3 and 10-4 M were freshly prepared in methanol and diluted to the required concentration with distilled water. Mobile phases for the silica column were methanol - ammonium acetate solution (0 + 100,5 + 95 and 15 + 85 VIV). The mobile phase for the polymer column was methanol - ammonium acetate solution (20 + 80 V/V). The pH was adjusted to 7.24 with concentrated hydrochloric acid or sodium hydroxide solution (4 M).The mobile phases were de-gassed under vacuum in an ultrasonic bath before use. Procedure Injections (5 1-11) of solutions of 2-aminophenol(lO-4-lO-2 M) were made on to the column and eluted with mobile phases containing metal ions (0.00-0.07~). The effect of the pH of the mobile phase was studied using the polymer column in the pH range 4-13 with each metal ion at a concentration of 0 . 0 2 ~ , except copper (5 x 10-5 M). Results and Discussion In conventional ion-pair chromatography, the presence of the ion-pairing reagent results in an increase in the retention of analytes by converting the polar ionic form of the analyte to a neutral paired complex, which is more hydrophobic and is more strongly retained on a non-polar reversed-phase col- umn.The extent of formation of the ion pair and hence the change in the retention depends on the concentration of the pairing reagent in the mobile phase and the formation constant of the ion-pair complex. A very similar mechanism is proposed to occur with groups that can chelate if a metal ion is present in the mobile phase, but in this instance the change in the retention depends on the final over-all charge on the complex. When nickel or cobalt ions were present in the mobile phase, dithiocarbamate analyte ions were rapidly converted to their 2 : 1 ligand - metal neutral complexes and the retentions increased.17J8 Because of the high formation constants the reaction was essentially complete on mixing and the retentions were independent of the concentration of the metal ions.In this instance the technique is more correctly described as on-column derivatisa- tion. However, other chelating groups often form much weaker complexes and, in the absence of a vast excess of ligand ions (i.e., analyte), incomplete complex formation would occur. In these instances the changes in retention should be related to the concentration of metal ions in a similar manner to ion-pair chromatography. Hence it should be possible to adjust the analyte retention to give the desired separation by altering the metal ion concentration. As metal chelate interactions can give both neutral and charged complexes, the retentions could either be extended or reduced, respectively, and by using different metal ions with different formation constants great flexibility should be possible.In the earlier work with 2-aminophenol (AP) the addition of nickel to the mobile phase caused a decrease in the retentions and it therefore appeared that a positively charged 1 : 1 metal - phenol complex (APM+) was being formed? AP+ AP $ AP- G APM+ AP2M As the metal ion concentration was increased, the shift in the retention also increased, suggesting that the complex forma- tion was an equilibrium process. The interaction between metal ions and 2-aminophenol therefore appeared to be a suitable system for further study to determine whether changes in retention could be related to the concentrations of different ions and to the formation constants of the complexes. As with ion-pair chromatography, the interaction occurs between the ionised analyte and the metal ions, so the effect should also be pH dependent.Separation on ODs-Silica Columns Initially, studies were carried out on a Hypersil-ODS column and methanol - ammonium acetate eluents using analogous conditions to those used by Sternson et aZ.16 However, the retention time of 2-aminophenol was found to be dependent on the amount of analyte injected. Smaller sample sizes gave broader peaks with longer retentions and greater asymmetry (Table 1). In contrast, the peak shapes for samples of 3- and 4-aminophenol and aniline were symmetrical and their reten- tion times were independent of the injection concentration. These changes in the aminophenol peak suggest that a specific interaction was occurring with the column, either with residual silanol groups on the surface of the stationary phase19 or with adsorbed trace metal ions, as the effects are seen only with o-aminophenol.A similar interaction between surface metal ions and P-diketones has been reported by Verzele et aZ.20 Addition of Metal Ions to the Eluent Although the injection of different amounts of 2-aminophenol gave different retention times, the values in each instance were reproducible and it was considered that if a constant sample size was used the influence of the addition of metal ions to the mobile phase could be studied. As found by Sternson et aZ., the incorporation of nickel acetate in the eluent caused a decrease in the capacity factor of 2-aminophe- no1 as the metal ion concentration increased (Fig.1). In order to ensure that the same proportion of the 2-aminophenol was available in the ionised form (AP-) for chelation, after addition of the nickel acetate solution the pH of the mobile phase was adjusted to 7.24. The shape of the graph suggested that, as expected from the formation equilibrium, there was a direct relationship between the changes in the capacity factors and the metal ion concentration from 0 to 0 . 0 8 ~ nickel acetate. These results contrasted with the earlier work of Sternson et aZ. ,I6 who had found that no further change in the retentions occurred with aqueous eluents when the nickel concentration was increased above 0.02 M. However, although their study was carried out using constant ionic strength, the pH was not Table 1.Effect of sample size on capacity factors of 2-aminophenol. Eluent, methanol - 0 . 2 6 ~ ammonium acetate solution (pH 7.24); column, Hypersil-ODS at ambient temperature Capacity factor Analyte concentration/ M 0% MeOH 5% MeOH 15% MeOH 10-2 10.02 (5.82)* 5.21 2.65 10-3 11.53 (8.75)* 5.88 2.84 10-4 - 6.24 2.99 * Asymmetry factor in parentheses. 4 1 I I I I I I I 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 Nickel ion concent ra t i o n h Fig. 1. Effect on the capacity factor of 2-aminophenol on an Hypersil-ODS column of the addition of nickel acetate to the mobile phase of 0 . 2 6 ~ ammonium acetate. A, Constant pH of 7.24; and B, without adjustment of pHANALYST, SEPTEMBER 1987, VOL. 112 1211 controlled. When the equivalent eluents were prepared in this study, it was found that the pH of the mobile phase had changed on the inclusion of 0.01 M nickel acetate from 7.24 to 6.68, for 0 .0 2 ~ nickel acetate to 6.61 and for 0 . 0 8 ~ nickel acetate to 6.40. When these unadjusted solutions were used for the chromatography of 2-aminophenol, the effects on the retentions were much smaller (Fig. 1). This would be expected, as a smaller proportion of the analyte would be present as the phenolate anion and thus available for complex formation. This graph tended to flatten much earlier as the increase in retention caused by the addition of the nickel ions is counteracted by a reduction in the effect because of the change in pH. Overall the results could well correspond to the apparently anomalous results observed by Sternson et al.16 To check if ionic strength had an influence on the separation, a series of chromatograms were measured using eluents containing nickel acetate, with adjustment of the ionic strength to 0.47 M with sodium perchlorate. In each instance the pH had first been adjusted to 7.24, but the results were identical with those obtained earlier with only adjustment of pH. However, because of the interactions with the silica surface and the consequential changes in retention times with sample size, it was felt that ODs-bonded silica columns might 0.04 4 'c. ? Q 0.02 1 I I I 5 10 15 Tirne/min Fig. 2. Chromatogram of 2- and 4-aminophenol on a PLRP-S column with different eluents: A, methanol - 0 . 2 6 ~ ammonium acetate solution (20 + 80 V/V); B, as A but including 0 .0 2 ~ nickel acetate in the aqueous buffer 12 I F T 4 D 2 I I I I I I I 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 Metal ion concentration/M Fig. 3. Effects of different concentrations of transition metal ions on the capacity factor of 2-aminophenol on a PLRP-S column with methanol - ammonium acetate solution (20 + 80 V/V) as the eluent: A, nickel; B, cobalt; C, zinc; D, copper; E, cadmium; and F, manganese not be suitable for a more detailed study of this separation or for routine application of this technique. Separations on a Polystyrene - Divinylbenzene Column In order to improve the peak shapes and avoid the shifts in retentions with sample size, the remainder of the study was carried out using a polystyrene - divinylbenzene (PLRP-S, 5 pm) column. This column material is not silica based so the analytes cannot suffer from secondary interactions with silanol groups and it has no active sites that could adsorb metal ions.Because of the greater retentive power of this column, in order to give capacity factors in a reasonable range, the eluent had to be changed to methanol - 0 . 2 6 ~ ammonium acetate solution (20 + 80 V/V). Samples of 2-aminophenol gave more symmetrical peaks (Fig. 2A) than on the Hypersil-ODS column and the retention was again reduced on the addition of nickel ions (Fig. 2B). The capacity factor was independent of the size of the sample, although the efficiency of the peaks was not very high (typically N = 300). Using this column, the effects of the addition of a number of different metal salts [copper(II), nickel(II), cobalt (11), zinc( 11) , cadmium( 11) and manganese( 11)] on the capacity factors were examined (Fig. 3).In each instance the addition resulted in a decrease in the retention proportional to the amount of metal ion present, although the effects of different metal ions differed. The largest change was observed with copper(I1) ions and a much lower concentration range was used for this metal ion than for the other salts. Some metal ions, such as manganese, had only a very slight effect. The extent of the change with the different metal salts, Cu > Ni > Co > Zn > Cd > Mn, could be correlated with the formation constant of the APM+ complex in aqueous solution (Table 2).21-23 The very high value for copper, the low value for manganese and over-all order of complexation agree well with the observed shifts in retention.Formation constants in 20% methanol in aqueous solutions could not be located in the literature and in further studies it is planned to derive experimental values that can be compared with the chromato- graphic changes in order to be able to predict the changes from different metals. A number of other metal ions were also examined but their application was limited by the stability of their oxidation state either towards the analyte or in solution in general. Iron(I1) ions were readily oxidised by air to iron(II1) in neutral and alkaline solutions and an attempt to form a neutral 1 : 1 chelate with silver(I), which would be expected to have a longer retention time, was not practical because the silver ions oxidised the 2-aminophenol. Effect of pH on Separations on PLRP-S Columns As noted earlier, the complexation with aminophenol occurs with the phenolate anion and the magnitude of any change on adding metal ions will therefore be dependent on the proportion of the analyte that is present as the anion.It was therefore of interest to determine the influence of the pH of the mobile phase on the retention of free and complexed 2-aminophenol, as this could potentially provide an alterna- tive method to adjust the retentions selectively instead of changing the metal ion concentration. Table 2. Formation constants of the 1 : 1 chelates of 2-aminophenol with different transition metal ions Metal ion Log K1* Metal ion Log K , * Cur1 .. . . . . 9.25 Zn" . . . . . . 5.99 Ni" . . . . . . 6.10 Cd'I . . . . . . 4.30 CO" . . , . . . 5.81 Mn" . . . . 3.60 * Values from references 21-23.1212 ANALYST, SEPTEMBER 1987, VOL. 112 Related Analytes to 2-Aminophenol Sternson et al. 16 demonstrated that the influence of the metals was restricted to 2-aminophenol and was not observed with closely related compounds that could not chelate, such as 3- and 4-aminophenol and anisidine. For confirmation, two further compounds, 2-hydroxyformanilide and 2-acetamido- phenol, were examined in this study using the Hypersil-ODS column, but their retentions were not altered by the additjon of nickel or copper ions. Conclusion The addition of metal ions to the mobile phase is a potentially useful technique which may be used to alter the selectivity of a separation by specifically changing the retention of com- pounds that can undergo chelation.The method was demon- strated with 2-aminophenol using a range of common transi- tion metal ions and it was shown that the magnitude of any change in retention could be adjusted by altering the concentration of the metal ions or the degree of ionisation of the analyte. Further work is being carried out to investigate the application of this method to the @-diketones25 and other chelating groups. 4 5 6 7 8 9 1 0 1 1 1 2 1 3 Eluent PH Fig. 4. Effect of pH on the capacity factor of 2-aminophenol on a PLRP-S column with methanol - ammonium acetate solution (20 + 80 VlV) containing 0 . 0 2 ~ transition metal salts. (a) A, No metal; B, cobalt; C, nickel; and D, copper (5 X 1 0 - 5 ~ ) .(b) A, No metal; B, cadmium; and C, zinc The capacity factors of 2-aminophenol were therefore measured over the pH range 4-13 (Fig. 4), which was possible using the PLRP-S column, because polystyrene - divinylben- zene columns, unlike silica-based bonded phases, are stable over a wide pH range.24 At low pH the retention of 2-aminophenol was reduced by the formation of the polar protonated amino salt and at high pH by the formation of the phenolate anion, whereas around pH 7 the non-polar neutral form predominated and a longer retention was observed. In the presence of the metal salts examined earlier, chelation with the phenolate anion can occur. Hence the metal ions have very little effect below pH 7.0 and an increasing effect as the pH increases (Fig.4). The range over which individual metals could be examined was restricted by their tendency to precipitate as insoluble hydroxides. This was particularly a problem with copper, cobalt and cadmium, which could not be used above pH 8.5, whereas nickel and zinc remained in solution up to pH 10. The relative magnitudes of the effects of the different metals at the increased pH were similar to those at pH 7.24, suggesting that the same equilibrium is involved. Manganese, which had little effect at neutral pH, had no significant effect at higher pH, whereas large shifts were always observed for copper ions. Therefore, over a limited range, which would be particu- larly restricted with bonded silica columns, the change in the retention could be adjusted by altering the pH.The authors thank the Science and Engineering Research Council for a CASE studentship to S. J. B. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. References O’Laughlin, J. W., J. Liq. Chromatogr., 1984, 7, S-1, 127. Veening, H., and Willeford, B. R., Adv. Chromatogr., 1983, 22, 117. Willeford, B. R., and Veening, H., J. Chromatogr., 1982,251, 61. Willett, J . D., and Knight, M. M., J . Chromatogr., 1982, 237, 99. Ichinoki, S . , and Yamazaki, M., Anal. Chem., 1985, 57, 2219. Vivilecchia, R., Thiebaub, M., and Frei, R. W., J . Chroma- togr. Sci., 1972, 10, 411. Mikes, F. , Schuring, V. , and Gil-Av, E., J. Chromatogr., 1973, 83, 91. Chow, F. K., and Gruskha, E., Anal. Chem., 1977,49, 1756. Chow, F. K., and Grushka, E., Anal. Chem., 1978, 50, 1346. Hare, P. E., and Gil-Av, E., Science, 1979, 204, 1226. Lindner, W., LePage, J. N., Davies, G., Seitz, D. E., and Karger, B. L., J. Chromatogr., 1979, 185, 323. Cooke, N. H. C., Viviattene, R. L., Eksteen, R., Wong, W. S., Davies, G., and Karger, B. L., J. Chromatogr., 1978,149,391. Vonach, B . , and Schomburg, G., J. Chromatogr., 1978, 149, 417. Walters, V., and Raghaven, N. V., J. Chromatogr., 1979,176, 470. Sternson, L. A., and DeWitte, W. J . , J. Chromatogr., 1977, 137, 305. Sternson, L. A. , Dixit, A. S., Riley, C. M., Siegler, R. W., and Schoech, D., J. Pharm. Biomed. Anal., 1983, 1, 105. Smith, R. M., Morarji, R. L., Salt, W. G., and Stretton, R. J., Analyst, 1980, 105, 184. Smith, R. M., Morarji, R. L., and Salt, W. G., Analyst, 1981, 106, 129. Eiceman, G. A , , and Janecka, F. A . , J. Chromatogr. Sci., 1983,21, 555. Verzele, M., De Potter, M., and Ghysels, J., J. High. Resolut. Chromatogr. Chromatogr. Commun., 1979, 1, 151. Sims, P., J. Chem. SOC., 1959, 3648. Charles, R. G., and Freiser, H., J. Am. Chem. SOC., 1952,74, 1385. Freiser, H., Charles, R. G., and Johnston, W. D., J . Am. Chem. SOC., 1952, 74, 1383. Lee, D. P., J. Chromatogr. Sci. , 1982, 20, 203. Smith, R. M., Bale, S. J . , Westcott, S. G., and Martin-Smith, M. , in preparation. Paper A7175 Received March 2nd, 1987 Accepted April loth, 1987

ISSN:0003-2654    

DOI:10.1039/AN9871201209

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

ANALYST, SEPTEMBER 1987, VOL. 112 Determination of Volatile Fatty Acids (CrC5) and Lactic Acid Silage by Gas Chromatography Richard J. Fussell Ministry of Agriculture, Fisheries and Food, ADAS, Starcross, Exeter, Devon EX6 8PE, UK and David V. McCalley Department of Science, Bristol Polytechnic, Frenchay, Bristol BSI6 IQY, UK 1213 in A gas chromatographic procedure using a graphitised carbon black packing coated with Carbowax 20M has been developed for the determination of volatile fatty acids and lactic acid in silage. This packing gave improved results compared with porous polymer column materials. The method was precise (relative standard deviation for each acid 0.55%) and gave a quantitative recovery of acids spiked into silage extracts (9&109%). The results for the determination of lactic acid in silage obtained by the GC procedure were in good agreement with results obtained by a classical distillation procedure; however, the GC method was more selective, more precise and less time consuming. The method has been shown to be of value in the determination of volatile fatty acids and lactic acid in connection with the assessment of silage quality. Keywords: Volatile fatty acid determination; lactic acid determination; silage; gas chromatography Silage is the material produced by the controlled microbial fermentation of a crop of high moisture content in a process known as ensilage. Silage production is of major importance to the UK farming economy, permitting the conservation of forage to meet the feeding requirements of ruminants in winter.Ensilage is a complex biochemical and microbiological process in which the primary objective is the promotion of an anaerobic fermentation by lactic acid bacteria. Soluble plant carbohydrates are fermented to a mixture of carboxylic acids, predominantly lactic acid, which lowers the pH. A decrease in pH to approximately 4 restricts undesirable clostridial activity. Clostridia produce butyric acid and can break down amino acids to a variety of products of poor nutritional value. Butyric silages also possess poor keeping qualities and tend to be unpalatable to livestock. Thus a knowledge of the lactic and the volatile fatty acid (VFA) (CTCS acids) content in conjunction with results for pH, ethanol, protein and ammo- nia permits the evaluation of silage quality. High-performance liquid chromatography (HPLC) appears to be a promising technique for the determination of these high polarity, low volatility and sometimes thermally unstable compounds.For example, Vratny and Mudrik' separated these acids using a cation-exchange column. However, silage extracts yielded complex chromatograms with peaks for many non-volatile acids, making interpretation and quantitation difficult. Classical methods are time consuming and can lead to erroneous results. For example, the determination of lactic acid by oxidation to acetaldehyde followed by iodimetry2 can give high results as other substances present in silage extracts also yield acetaldehyde on oxidation.3 The derivatisation of fatty acids, followed by gas chromato- graphy (GC), has been employed to overcome some of the problems associated with GC of the free acids.For example, Jones and Kay4 chromatographed the benzyl esters, whereas Schooley et af.5 separated the tert-butyl dimethylsilyl deriva- tives. However, such methods usually require extraction of the acids prior to derivatisation. Ion-exchange chromato- graphy is time consuming for this purpose whereas solvent extraction, although relatively straightforward, is limited by the unfavourable partition coefficients which exist between these acids and water.6 Although direct derivatisation methods may be possible, derivatisation is not easily applic- able to the routine automatic analysis of large numbers of samples. Thus, GC of the free acids has been extensively studied.Serious difficulties can arise owing to the high polarity of the acids, particularly lactic acid, which can lead to an undesirable strong adsorption in the injector or column, causing asymmetric peaks and ghosting.7.8 A wide variety of stationary phases have been evaluated for the determination of VFAs, although few reports include lactic acid, emphasising the difficulties inherent in its determi- nation. Conventional liquid phases have been used with varying degrees of success.9 Porous polymers, however, have increased tolerance to aqueous injections, reduced column bleed and no support hydroxyl groups to contribute to adsorption and peak tailing.lo.11 A deactivated graphitised carbon black (GCB) support coated with Carbowax 20M has been proposed for the separation of free fatty acids12113 and is now commercially available.14 This material gave exellent separations of VFAs in foods,ls and was shown by Cosmi et al. 16 to be applicable to the determination of lactic acid in blood samples. Both the GCB Carbowax and porous polymer packings appeared to merit further investigation for the determination of free acids, especially lactic acid, in silage. Experimental Reagents Analytical-reagent grade propionic, isobutyric, butyric, iso- valeric and valeric acids were obtained from BDH Chemicals (Poole, UK), pivalic acid from Sigma (Poole, UK) and ethanol (A.R. quality) was obtained from James Burrough (Witham, UK). All other reagents used were of AnalaR grade and obtained from BDH Chemicals. Porapak Q (80-100 mesh), Porapak QS (80-100 mesh) and Chromosorb 101 (6@80 mesh) were obtained from Phase Separations (Queensferry, UK), Carbopack B-DA - 4% Carbowax 20M (80-120 mesh) from Supelchem (Sawbridge- worth, UK) and Graphpac GC - 0.3% Carbowax 20M - 0.1% H3P04 (6&80 mesh) from Alltech Associates (Carnforth, UK) .Quartz-wool was obtained from Alltech and silanised glass-wool from Phase Separations. Apparatus A Model 304 gas chromatograph was used in all experiments (Pye Unicam, Cambridge, UK) and was equipped with a flame-ionisation detector (FID). The columns were glass coils1214 ANALYST, SEPTEMBER 1987, VOL. 112 (1.5 m X 4 mm i.d.) packed with the appropriate stationary phase. Sample volumes ranging from 1 to 2 pl were injected using a PU 4700 autoinjector equipped with a 10-pl syringe of the plunger in glass barrel type (SGE, Melbourne, Australia).Injector and detector temperatures were maintained at 240 "C, as recommended by Bottcher.10 The carrier gas was nitrogen, at a flow-rate of 40 ml min-1, unless stated otherwise. The peak areas were measured using a Model C-R1B integrator (Shimadzu, Kyoto, Japan). Preparation of Columns All porous polymers were conditioned in a short, wide tube (12 cm x 2.5 cm i.d.) for 16 h at 240 "C prior to packing in the analytical column. Porous polymers may shrink in the conditioning process owing to the loss of monomer. The analytical columns were filled and plugged with quartz-wool such that injection was made into the wool just above the packing.10 The quartz-wool should be renewed regularly if samples giving large volumes of involatile residues are injected." The columns were further conditioned at 240 "C for a minimum of 2 h prior to use.Carbopack columns were similarly packed, purged for 30 min at room temperature to remove air and conditioned overnight using a flow-rate of 10 ml min-1 at a temperature of 245°C. The gas flow-rate was adjusted to 24 ml min-1, the oven temperature to 200 "C and ten 1-p1 injections of 0.03 M oxalic acid were made prior to use. This procedure was in accord with the recommendations of Cosmi et aZ.16 and the manufacturer's instructions. Graphpac columns were conditioned overnight at 180 "C using a flow-rate of 50 ml min-1. Procedure Transfer 20 g of fresh silage to a 250-ml wide-necked bottle and add 100 ml of distilled water.Cap and shake mechanically for 1 h, then filter through a Whatman No. 1 paper. Take 5 ml of filtrate, 1 ml of a 2.5 g 1-1 solution of pivalic acid (as internal standard, if required) and 2.5 ml of 0 . 1 2 ~ oxalic acid in a 10-ml calibrated flask and dilute to volume. Centrifuge at 2600 g for 5 min and inject the supernatant into the GC. Calculation of Results The GC results were calculated from the peak areas obtained from the integrator using the internal standard technique. The asymmetry factor, conventionally measured at 10% of the peak height, is defined as the width of the peak tail divided by the width of the peak front, when the peak is bisected by a perpendicular line drawn through its maximum.17 The silage extract solution referred to in the text is the solution obtained after adding 100 ml of distilled water to the 20 g of fresh silage.The analyte concentrations in the silage juice (the fluid associated with the fresh silage) can be derived by consideration of the dilution, taking into account the percentage of dry matter (% DM) in the sample. Results and Discussion The injection of standard solutions of VFAs in distilled water gave good separations using a column of the porous polymer Chromosorb 101 operated under isothermal conditions. However, tailing peaks were observed for lactic acid, a problem also noted by Playne.18 Porapak Q gave improved peak symmetry for lactic acid; however, it was not completely resolved from isovaleric acid. Porapak QS, a silanised version of Porapak Q, gave a much lower column efficiency and decreased resolution of the isomeric acids.Many authors have claimed that the addition of strong acids to the sample solution improves peak symmetry and inhibits ghosting.7.9J9 Accordingly, the effect of injections on to the Chromosorb 101 column of standard solutions made up in 5% formic acid, 0.5 M oxalic acid, 2.5 M HCl, 2% H3P04 and 2% HP03 was investigated. The results indicated that for standard solutions, added acids had little effect on the peak symmetry or peak-area response of the analytes. Bottcherlo has investigated the extraction of porous polymers with organic solvents to remove any excess monomer, and extraction or coating with acids to remove basic adsorption sites. Hence, we investigated extraction with mixtures of acetone and acetic acid, and extracting or coating the materials with mineral acids such as orthophosphoric acid.The treatment of porous polymers did appear to give some improvement in column performance, but in our work this improvement was not sufficient to merit consideration of their use in a routine, automatic analytical procedure. Cochraneg has noted that other workers have experienced difficulty in reproducing work involving porous polymer pre-treatments. As shown in Fig. 1, the Carbopack column gave excellent separations of VFAs and lactic acid in standard solutions. This packing is a GCB material which is acid-washed to remove impurities13 and coated with Carbowax 20M. Despite our results for the addition of acid to standard solutions obtained with porous polymers, 0 . 0 3 ~ oxalic acid was added to all injected solutions in accord with the manufacturer's instructions.By degradation to formic acid on injection, oxalic acid may maintain column performance and prevent the adsorption of acids on involatile deposits on the top of the column when biological samples are analysed. At the analysis temperature shown, ethanol eluted rather close to the solvent front although it was later apparent that quantitative results were still obtainable. Significant improvements in resolution, peak symmetry and sensitivity were obtained for all the acids of interest, particularly lactic acid which yielded an asymmetry factor of only 1.2. The separation was adequate to allow the use of pivalic acid as an internal standard, if required.This acid eluted between butyric and isovaleric acids. The Alltech Graphpac column produced comparable results to the Carbopack column although optimum separation occurred at a much lower column temperature (120°C). The difference may be caused by the lower stationary liquid phase loading of the Graphpac column. However, the Graphpac material also has a much lower surface area (ca. 10 m2 g-1) in comparison ( 1 2 3 4 0 25 tlmin Fig. 1. Chromatogram of VFAs, ethanol and lactic acid on a Carbopack column, Peak identities: (1) ethanol; (2) acetic acid; (3) propionic acid; (4) isobutyric acid; (5) butyric acid; (6) pivalic acid (internal standard); (7) isovaleric acid; (8) lactic acid; and (9) valeric acid. All standards 250 mg 1 - 1 , except lactic acid (2500 mg 1-I).Column temperature = 175 "C, injector and detector temperature = 240 "CANALYST, SEPTEMBER 1987, VOL. 112 1215 with the Carbopack material (ca. 100 m2 g-1). The surface of these GCB materials is highly non-polar and can give rise to gas - solid interactions in addition to the usual gas - liquid interactions; this may be another reason for the reduced retention on the Graphpac column. The over-all retention mechanism may be complex; in some instances increasing the stationary phase loading may decrease the retention of sample 0 1 2 5 25 t h i n Fig. 2. Chromatograms of extracts of ( a ) good silage and (b) poor butyric silage. Column and conditions as Fig. 1. The lactic and butyric acid contents of sample (a) (pH 4.1) were 6090 and <2 m l-l, res ectively, whereas the corresponding values for sample (by (pH 4.97 were 255 and 308 mg 1-1, respectively components due to shielding and deactivation of the carbon surface.At a particular point the surface will be saturated giving an increase in retention with increasing liquid phase loading, as normally found. The inclusion of H3P04 as a tail reducer in the stationary phase coating may also affect retention.20 Excellent peak shapes were again obtained for lactic acid (asymmetry factor 1.4). However, although all fatty acids were well separated on Graphpac, inferior resolution of isovaleric and lactic acid was obtained compared with the Carbopack column. Nevertheless, most silages examined in this study appeared to contain low levels of isovaleric acid, indicating that Graphpac may also be suitable for most samples.Simple adjustment of the stationary liquid phase loading of this column might improve the resolution of these two acids. The Carbopack column gave an excellent linear calibration response for VFAs and ethanol over the range 0-500 mg 1-1 and for lactic acid over the range 250-10000 mg 1-1. These concentration ranges were carefully selected to bracket the expected range of analyte concentrations found in the majority of diluted silage extracts. Fig. 2 shows chromatograms obtained from a good silage and a poor butyric silage by extraction with water. The differences between the two silages are clearly seen. It should be noted that the extraction method is a widely used procedure2 although small variations might be obtained if other pre-treatment steps such as maceration and homo- genisation of the silage were utilised.The proposed GC procedure was compared with the classical procedure involving conversion to acetaldehyde, distillation and iodimetry.3 Single extracts of six silage samples were subjected to analysis on the same day using both procedures. The entire procedure was repeated on the following day. The results are presented in Table 1. A paired t-test on the two sets of data showed no significant difference between the mean values at the 95% confidence level. Using the same data, the precision for each method was calculated as the pooled estimate of the coefficient of variation from rn pairs of results21 from the equation Z di2 Pooled estimate CV = dG where di is the difference between two duplicates expressed as a percentage of the mean for that sample.The GC procedure was more precise than the distillation procedure, with pooled estimates of the coefficient of variation of 6% and l8%, respectively. The six silages analysed gave no spurious Table 1. Comparison of the results for lactic acid obtained by a classical distillation procedure and the proposed GC method. All results expressed in mg 1-1 of extract solution Distillation GC (Carbopack) Dry matter, Sample Y O PH 1st analysis 2nd analysis Mean 1st analysis 2nd analysis Mean 1 23.2 4.5 3810 2380 3100 2780 2860 2820 2 28.8 5.2 2560 2480 2520 3140 2660 2900 3 30.5 5.3 3020 3150 3090 3550 3330 3440 4 23.2 4.5 2250 3300 2780 2660 2440 2560 5 30.2 4.7 1960 1860 1910 2060 2100 2080 6 21.4 4.1 3130 3440 3290 3100 2930 3020 Table 2.Determination of the precision of the GC analysis when applied to a standard solution and a silage extract. All results were calculated from quintuplicate injections of the acids and are expressed in mg I-' of extract solution Sample Ethanol Acetic Propionic Isobutyric Butyric Isovaleric Valeric Lactic xf u . . 251 f 2.07 249 k 4.45 249 k 2.06 251 k 0.32 249 k 1.35 249 k 1.36 249 k 1.72 2490 f 109 RSD,O/o . . 0.83 1.79 0.83 0.13 0.54 0.55 0.69 4.36 15 f a . . 85 f 0.77 362 f 6.15 70 f 0.33 9.74 k 0.23 218 k 0.84 10.4 k 0.48 25.5 f 0.90 965 f 33 Standard: Silage extract: RSD, O/o . . 0.91 1.70 0.47 2.36 0.39 4.60 3.51 3.471216 ANALYST, SEPTEMBER 1987, VOL. 112 Table 3. Recovery of acids spiked into two different silages.All results are the means of triplicate injections and are expressed in mg 1-1 of extract solution Sample Acetic Propionic Isobutyric Butyric Isovaleric Valeric Lactic Ethanol Sample I : (DM = 21.4Y0, pH = 4.1) Unspiked sample concentration . . 212 16 120 124 0 0 1820 - Theoretical concentration increase . . 500 500 500 500 500 500 2500 - Totalconcentrationfound . . . . 703 493 652 617 503 546 4414 Recovery, YO . . . . . . . . . . 98 95 106 99 101 109 104 - - Sample 2: (DM = 18.1Y0, pH = 4.9) Unspiked sample concentration . . 366 70 9 219 11 26 93 1 86 Theoretical concentration increase . . 250 250 250 250 250 250 2500 250 Totalconcentrationfound . . . . 611 323 270 47 1 268 286 3484 336 Recovery, YO . . . . . . . . . . 98 101 104 101 103 104 102 100 unidentified peaks on the chromatogram and no other peak co-eluted in the position of the internal standard.Further data on the precision of the analysis were obtained by the repeated injection of a single aqueous standard containing all of the analytes and a silage extract. Table 2 shows that excellent results were obtained indicating stable and consistent instrument performance. The injection of 20% formic acid after the injection of a sample gave no significant response of any analyte, showing that ghosting was not a problem. A thorough statistical analysis to determine detec- tion limits was not carried out, as very low values have little significance in indexing silage quality. However, an estimate of the limit of detection based on the mean value plus three times the standard deviation of replicate injections of silage extracts containing low levels of the appropriate acids would indicate a value of approximately 2 mg 1-1 for VFAs and approximately 60 mg 1-1 for lactic acid.The recovery of VFAs and lactic acid from silage extracts was determined by spiking the extracts with standard solutions as shown in Table 3. Excellent recoveries were obtained over the range 98-109%. Quantitative recovery of ethanol is also reported for one of the silages. It should be noted, however, that such results should be treated with caution as spiking may not exactly simulate the occurrence of the acids in the silage matrix. The Carbopack column showed no deterioration in perfor- mance after the injection of 150 :xtract samples.However, the quartz-wool insert should be regularly renewed if samples containing large amounts of involatile material are chromato- graphed. The first few centimetres of packing can also be replaced, although we have not as yet found this necessary. The GC procedure was developed with a view to 24-h continuous unattended operation. Preliminary experiments indicated that no significant changes occurred in the acid content of samples left to stand on an autosampler tray for 24 h at room temperature, although further work is necessary to confirm this result. In addition to the application to silage analysis, the method has been successfully employed for the determination of free acids in silage effluents, brewers grain and yoghurt wastes. The authors thank Mr. D.S. Farrington for his helpful suggestions and critical appraisal of the manuscript. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. References Vratny, P., and Mudrik, Z., J. Chromatogr., 1985, 322, 352. Jackson, E., Farrington, D. S., and Henderson, K., Editors, “The Analysis of Agricultural Materials,” HM Stationery Office, London, 1986. Wilson, R. F., and Terry, R. A., Analyst, 1977, 102,218. Jones, D. W., and Kay, J . J., J. Sci., Food Agric., 1976, 27, 1005. Schooley, D. L., Kubiak, F. M., and Evans, J. V., J. Chromatogr. Sci., 1985, 23, 385. Chalmers, R. A., and Lawson, A. M., “Organic Acids in Man,” Chapman and Hall, London, 1982. Geddes, D. A,, and Gilmour, M. N., J. Chromatogr. Sci., 1970, 8, 394. Van Eenaeme, C., Bienfait, J. M., Lambot, O., and Pondant, A., J. Chromatogr. Sci., 1974, 12, 398. Cochrane, G. C., J. Chrornatogr. Sci., 1975, 13,440. Bottcher, W., Arch. Tiernahr., 1982, 32, 287. Wesenberg, J., Laube, K., and Bottcher, W., Lebensm. Ind., 1985,32, 119. Di Corcia, A., and Samperi, R., Anal. Chem., 1974,46, 140. Di Corcia, A., Samperi, R., Sebastiani, E., and Severini, C., Anal. Chem., 1980, 52, 1345. Supelco Reporter International, 1985, 4, 10. Lamkin, W. M., Luginsland, N. D., and Pomeranz, Y., Cereal Chem., 1985, 62, 6. Cosmi, G., Di Corcia, A., Samperi, R., Vinci, G., Clin. Chem., 1983, 29, 319. Kirkland, J. J., Yau, W. W., Stoklosa, H. J., and Dilks, C. H., J. Chromatogr. Sci., 1977, 15, 303. Playne, M. J., J. Sci. Food Agric., 1985, 36, 638. Henderson, M. H., and Steedman, T. A., J. Chromatogr., 1982, 244, 337. Alltech Chromatography Catalogue, Alltech Associates, Carn- forth, 1987, p. 76. Cheeseman, R., and Wilson, A. L., Technical Report TR 66, Water Research Centre, Medmenham, 1978. Paper A7189 Received March loth, 1987 Accepted April 6th, 1987

ISSN:0003-2654    

DOI:10.1039/AN9871201213

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

ANALYST, SEPTEMBER 1987, VOL. 112 1217 The Potential of Fire Assay and Inductively Coupled Plasma Source Mass Spectrometrv for the Determination of Platinum Group Elements in Geological Materials Alan R. Date, Alan E. Davis and Yuk Ying Cheung British Geological Survey, 64 Gray‘s Inn Road, London WClX 8NG, UK Preliminary work on the application of fire assay and inductively coupled plasma source mass spectrometry (ICP-MS) to the determination of platinum-group elements (PGEs) in geological materials is described. A combination of the neo-classical nickel sulphide fire assay collection procedure with final determination by ICP-MS is considered in relation to the simplified procedure required for dissolution of the nickel sulphide ”button“ and the wealth of information inherent in ICP-MS spectra.The method was applied to the determination of five PGEs and gold in the South African standard reference material PTO-1 (now known as SARM-7) and the Canadian standard reference material PTC-1. Isotope dilution analysis was applied to the determination of platinum. Keywords: Platinum-group elements; nickel sulphide fire assay; inductively coupled plasma source mass spectrometry; geological materials; isotope dilution The analytical chemistry of gold, silver and platinum-group elements (PGEs) has always reflected their common occur- rence in nature as rare, discrete, inhomogeneously distributed mineral species and their high financial value. Classical analytical techniques have been extended to the development of methods for their recovery from a wide variety of materials, including industrial waste products.These factors have resulted in a general requirement for complete dissolution of a representative sample allied with an accurate measurement technique. For many years these criteria were satisfied by fusion of large sample aliquots with gravimetric measurement of the various components after “partition,” and the classical lead assay reigned supreme. The modern requirement for accurate determination at low levels, particularly for indi- vidual PGEs, following parallel developments in geochemical research methods and industrial techniques, has resulted in a re-appraisal of noble metal analytical chemistry. A useful two-part review has been published.l.2 In general, the classical lead assay provides pre-concentra- tion for gold, platinum, palladium and rhodium, although their efficient collection on a silver cupellation button is critically dependent on the “flux” composition and assay conditions, and a low recovery for rhodium is not uncommon.The neo-classical fire assay procedure with nickel sulphide collection3 is gaining in popularity because it provides efficient collection for the complete platinum metal group, although special techniques are necessary to avoid loss of volatile osmium compounds at the button dissolution stage. This procedure is less efficient for gold collection. The most commonly used final measurement techniques include flame atomic absorption spectrometry (FAAS), instrumental neutron activation analysis (INAA) and induc- tively coupled plasma atomic emission spectrometry (ICP- AES).The first, and most widely used, technique, FAAS, is basically a single-element technique with pg g-1 (p.p.m.) sensitivity, it suffers from serious matrix interferences and only when used with graphite furnace atomisation (GFAAS) does it compete with the ng g-1 (p.p.b.) sensitivity of INAA. The multi-element capability of INAA and ICP-AES is offset by the need for sophisticated hardware (a reactor) and time for the former, and the lack of heavy-element sensitivity for the latter. The current technique of choice for the determina- tion of PGEs at low concentrations would therefore appear to lie with GFAAS. Early work on the development of inductively coupled plasma source mass spectrometry (ICP-MS) stressed the advantages of the technique in terms of detection capability, particularly for heavy elements.4 The detection limit for gold quoted by Date and Gray4 (0.06 ng ml-I), allied with the probability of fairly even sensitivity across much of the Periodic Table, suggested that the technique would be capable of excellent detection for the PGEs similar to that available with INAA and GFAAS, but without the requirement for very sophisticated hardware (INAA) or the need for careful control of solution chemistry before measurement (GFAAS).Preliminary work on the development of a combined fire assay and ICP-MS method for the determination of PGEs and gold is described here. Experimental Fire Assay Procedure The fire assay procedure used is based on that described by Robert et al.3 and Haines and Robert,s and only an outline is presented here.Samples (20-30 g) are transferred on to a sheet of glazed paper and fused borax (60 g), sodium carbonate (30 g), basic nickel carbonate (32 g) and flowers of sulphur (12.5 g) are added. The fusion charge is modified by the addition of silica sand (5 g) to prevent excessive attack of the fusion crucible. The fusion charge is rolled from corner to corner of the paper until it is thoroughly mixed. The charge is transferred into a large fireclay crucible (Battersea round, size J or K; Morgan Crucible). The surface of the glazed paper is wiped with a paper tissue, which is then placed on top of the charge. The fusion procedure is modified by placing the crucible in a cold furnace, which is then brought up to the required temperature (1150 “C), and held until a state of tranquillity is reached (approximately 1.5 h).The crucible is removed with the aid of furnace tongs and the contents are poured into a cast iron mould. After cooling for 30 min, the nickel sulphide button is removed and separated from the slag. The button is weighed and its mass recorded. If necessary, the procedure mav be modified by the preparation of a “cleaning charge” from the slag and a smaller mass of fusion mixture. The small nickel sulphide button so produced is added to the main button for subsequent treatment. General-purpose reagents were used throughout the fusion procedure.1218 ANALYST, SEPTEMBER 1987, VOL. 112 Button Dissolution The button is broken up by pressing it in a hydraulic press.The pieces are transferred into a tungsten carbide Tema mill and ground for 30 s. The ground material is re-weighed to determine grinding losses. The powder is brushed into a 1-1 beaker, 600 ml of concentrated hydrochloric acid are added and the beaker is covered with a watch-glass. In the absence of a fail-safe water-bath, dissolution of the nickel sulphide is carried out on a thermostatically controlled hot-plate. After several hours at a temperature setting of 120°C, and with dissolution complete, the lid and sides of the beaker are washed down with distilled water. The solution is stirred and the PGE sulphides are left to settle and cool overnight. The solution is saturated with H2S for 1 min before filtration. The solution is filtered under vacuum, using a Hartley funnel with a Whatman No.542 filter-paper (90 mm), the beaker is washed with cold 50% hydrochloric acid and the filter-paper is washed five times with distilled water to remove traces of nickel. The filtrate is normally discarded. A major modification is made to the published method in order to simplify the procedure for subsequent determination of PGEs by ICP-MS. The filter-paper is transferred into a small (150-ml) beaker. Concentrated hydrochloric acid (10 ml) and hydrogen peroxide (100 volume, 10 ml) are added to dissolve the noble metals from the filter-paper. The beaker is covered with a watch-glass and heated on a hot-plate until the hydrogen peroxide is completely decomposed and efferves- cence ceases. The watch-glass is removed and washed, and the solution is filtered to remove filter-paper pulp from the previous filtration.The filter-paper is washed several times with distilled water. The filtrate is evaporated to 10 ml of “constant boiling” HC1 (approximately 50%). The solution is transferred into a 100-ml calibrated flask and diluted to the mark with distilled water to give a final acid concentration of approximately 5% HCl. Analytical-reagent grade reagents were used throughout the dissolution procedure. Stable Isotope Spike Solution The stable isotope dilution experiment was carried out using a 198Pt-enriched spike solution containing platinum at 8 pg ml-1 in 4% V/VHCl. The isotopic composition of the spike solution is compared with the natural abundances for platinum isotopes: Source .. 190Pt 192Pt 194Pt 195Pt 196Pt 198Pt Natural, YO 0.13 0.78 32.9 33.8 25.3 7.21 Spike, YO . .--0.01 0.17 3.05 3.93 5.37 87.47 The isotope dilution technique may be applied effectively only if complete exchange between sample and spike occurs before any subsequent chemical treatment. The spike solution was therefore added to the samples by pipetting suitable aliquots into the middle of the fire assay charge before fusion. Aliquot sizes designed to give roughly equal atom abundance to the 198Pt and 195Pt isotopes, calculated on the assumption that previously published data for platinum in SARM-7 and PTC-1 were close to the “true” values, were as follows: SARM-7,25.5 g, 4 ml of spike solution; PTC-1,21.8 g, 3 ml of spike solution. Calibration Standard Solutions Reference standard solutions were prepared from Johnson Matthey Specpure metal powders or compounds.Stock reference solutions were prepared at 100 pg ml-1 in 10% V/V HCl. Calibration reference standard solutions were prepared by dilution to 1 pg ml-1 in 1’/0 V/V HC1. Reagent blank solutions containing 1 and 5% HCl were prepared. ICP-MS Instrument Operation The ICP-MS instrument used is the immediate successor to the original UK prototype system described by Gray and Date,6 and has been described recently by Date and Hutchison.7 Under the terms of the British Geological Survey (BGS) contract with the Commission of the European Communities for the development of ICP-MS, during the period 1983-85 the original prototype system operated by Gray at the University of Surrey was used for further instrument develop- ment, while the BGS instrument was operated in a more detailed applications study with its performance restricted to that described by Gray and Date.6 The work reported here was carried out under the operating conditions shown in Table 1.The main differences from the work reported by Date and Hutchison7 include the use of nickel apertures (0.5 mm) and optimisation using the maximum signal on either 115In+ or 208Pb+. Sample solutions were introduced into the ICP-MS instru- ment by conventional pneumatic nebulisation, with a solution uptake rate of 2.0 ml min-1. The mass spectrometer is currently limited to two modes of operation, single ion monitoring and mass scanning. After optimisation for maxi- mum signal using single ion monitoring, the system was operated in the mass scanning mode, covering either the complete range from ruthenium (96Ru+) to platinum (198Pt+), or a range suitable for each triad [Ru, Rh, Pd; Os, Ir, (Au), Table 1.ICP-MS operating conditions Inductively coupled plasma: Plasma . . . . . . . . . . . . . . Forward power . . . . . . . . . . Reflected power . . . . . . . . . . Coolant (outer) . . . . . . . . . . Auxiliary (intermediate) . . . . . . . . Carrier (inner) . . . . . . . . . . Nebuliser pressure . . . . . . . . . . Distance from load coil to sampling aperture . . Distance from end of torch to sampling aperture Sampling aperture . . . . . . . . . . Skimmer aperture . . . . . . . . . . Aperturetoskimmerseparation . . . . . . Optimisation .. . . . . . . . . . . Data acquisition . . . . . . . . . . . . Solution up-take rate (free aspiration) . . . . . . . . . . . . . . . . . . Allargon . . . . . . . . . . . . . . 1.25kW . . . . . . . . . . . . . . < 5 w . . . . . . . . . . . . . . 12.0lmin-1 . . . . . . . . . . . . . . 0.2lmin-1 . . . . . . . . . . . . . . 1.2lmin-I . . . . . . . . . . . . . . 281bin-2 . . . . . . . . . . . . . . 2.0mlmin-1 . . . . . . . . . . . . . . 10mm . . . . . . . . . . . . . . 5mm . . . . . . . . . . . . . . Nickel, diameter 0.5 mm Nickel, diameter 1 .O mm . . . . . . . . . . . . . . 7.0mm Maximum signal, llsIn+ or 2osPb+ Pulse counting; multi-channel scaling (MCS); . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1024 MCS channels; 1 ms dwell time; 60 sweepsANALYST, SEPTEMBER 1987, VOL.112 1219 Pt]. Data acquisition was carried out using pulse counting and multi-channel scaling (MCS). The Canberra Series 80 MCS was set with a data acquisition memory group of 1024 channels, a dwell time per channel of 1 ms and 60 separate sweeps. Under these conditions, a complete spectrum was accumulated in just over 1 min. Each sample and standard solution was aspirated for 0.5 min for system equilibration before data acquisition commenced. Results General Spectral Characteristics for PGE by ICP-MS The spectrum obtained in 1 min for a reference standard solution containing Ru, Rh, Pd, Re, Os, Ir, Pt and Au, each at 1 yg ml-1 in 5% V/V HCl, is illustrated in Fig. 1. The count rate obtained for rhodium, 643000 counts s-1, is associated with a background rate of between 30 and 50 counts s-l.This spectrum was obtained at an early stage in the development work when the quadrupole mass filter was incorrectly tuned to give relatively high count rates for heavy elements at the expense of light elements, but is used for illustration purposes because osmium and rhenium were later excluded from the reference standard solution. The osmium was omitted because no special techniques were used in this work to prevent its loss during the separation procedure. Rhenium is not normally included in the determination of PGEs. With the system re-tuned to provide fairly even mass sensitivity across the complete Periodic Table, the count rates fell to about 200 000 counts s-1, but were associated with background counts of between 10 and 20.In common with previous work,7 it was found that, in the absence of memory effects, the random (non-spectral) back- ground is identical for all isotopes. Further, the background count shows no variation for reagent blank solutions in the range from 1 to 5% HCl. Higher reagent blank solution concentrations were not required for this work and were not tested. In this example, the resolution for Ru, Rh and Pd is better than that for the heavier triad, but is more than adequate in all instances for quantitative measurement. Relative count rates between the two triads may be varied with resolution and with optimisation setting. For instance, it was found that higher count rates were obtained for the low-mass triad if optimisa- tion was carried out using the maximum signal on Il5In+.The reverse was true if optimisation was carried out using 208Pb+. In each instance the system remained stable after optimisa- tion, so no further investigation of this effect was carried out. It is clear from this example that unusually high count rates may be obtained for osmium. The most abundant isotope, 192Os+ (41.0%), exhibits a much higher count rate than lg7Re+ (62.6%) or 193Ir+ (62.6%). This feature of osmium behaviour, probably caused by the generation of volatile osmium compounds in parallel with the nebulisation process, is discussed later. 1 9 2 0 ~ + I I 103;h+ (643 000 counts s-l) lo8Pd+ (186 000 counts s-1) A Fig. 1. ICP-MS spectrum for a reference standard solution contain- ing Ru, Rh, Pd, Re, Os, Ir, Pt and Au, each at 1 pg ml-’ in 5% V/V HCl, covering the mass range mlz 95-205, and taken in approximately 1 min PGE Calibration Graphs Calibration graphs for IO~RU+, 103Rh+ and lO5Pd+ are shown in Fig.2 and for 193Ir+, 195Pt+, 197Au+ and 198Pt+ in Fig. 3, for element concentrations varying from 1 ng ml-1 to 10 yg ml-1. The two groups are shown independently because they represent two separate calibration exercises over limited mass ranges after optimisation using indium and lead, respectively. The signal levels for 103Rh+ and 197Au+ are similar and the two calibration graphs would be superimposed on a single plot. Calibration is normally carried out using only a single reference standard solution at 1 pg ml-1 in either 1 or 5% V/V HCl, and a corresponding reagent blank solution.PGE Detection Limits Detection limits for the six elements (ng ml-1, 30 blank) are given in the captions of Figs. 2 and 3. Similar performance for sample solutions, at the sample pre-concentration level used in this work (about 25 g to 100 ml), would represent a detection capability of about 0.5 ng 8-1. Detection limits well below 0.1 ng g-1 would appear to be easily attainable with a combination of fire assay and ICP-MS. Sample Spectra The spectrum obtained for SARM-7 after fire assay pre- concentration with nickel sulphide is illustrated in Fig. 4. In this example, the largest signal is obtained for tungsten, probably introduced during grinding of the nickel sulphide button in a tungsten carbide Tema mill and retained during subsequent dissolution of the button in HCl.There is some evidence for the presence of tantalum (181Ta+), possibly an impurity in the tungsten carbide. Other elements present, in addition to the expected ruthenium, rhodium, palladium, (silver), iridium, platinum and gold, include antimony, tellurium and small amounts of cadmium, tin and mercury. These were probably collected by the nickel sulphide during fire assay, but further research would be necessary to determine their exact source and collection efficiency. Plati- num is clearly the predominant PGE in this sample. 106 10‘ I I I I 0.001 0.01 0.1 1 .o 10.0 Element concentration/pg I-’ Fig. 2. Calibration graphs covering the range 1 ng ml-1 to 10 pg ml-l. Element detection limits (30 blank): Ru, 0.5; Rh, 0.1; and Pd, 0.7 ng ml-11220 ANALYST, SEPTEMBER 1987, VOL.112 Precision and Accuracy Data are presented in Table 2 for five PGEs and gold in two separate fire assay collections for SARM-7 and one for PTC-1. All six elements were below the detection limit in the fire assay reagent blank solutions, and these data were obtained using the reference standard solution blank (5% V/V HC1) for Y 10' I I I I I 0.001 0.01 0.1 1 .o 10.0 Element concentration/pg I-' Fig. 3. Calibration graphs covering the range 1 ng ml-I t o 10 vg ml-1. Element detection limits (30 blank): Ir, 0.11; Pt, 0.17; and Au, 0.07 ng ml-1 1mwt i l21Sb + (53830 counts s-l) 108pd + JlUi I Fig. 4. ICP-MS spectrum for SARM-7 (25.6 g) after nickel sulphide fire assay separation background correction. The quoted uncertainties were calcu- lated from single determinations on four separate days.There is reasonable agreement between ICP-MS data and previously published values8 for SARM-7, although ICP-MS values for palladium and platinum are low and those for gold are high. Although it is difficult to explain the high values for gold, the apparent loss of platinum and palladium relative to the other PGEs may result from their greater solubility at the button- dissolution stage. All ICP-MS values for PTC-1 are lower than previously published data.9 Isotope Dilution Data for Platinum With the small amount of 198Pt-enriched spike available, it was possible to carry out only a very limited isotope dilution experiment, but the data obtained for platinum in SARM-7 and PTC-1 are instructive.The complete PGE spectrum for SARM-7 after spiking with 198Pt is shown in Fig. 5. When compared with Fig. 4, the large peak at 198Pt+ is clearly visible, but in other respects the two spectra are very similar. It would be possible to determine all five PGEs and gold from this spectrum using external calibration, as before, although the data for platinum using 195Pt+ will be slightly in error, unless a correction is made for 195Pt added with the spike solution. For the isotope dilution determination of platinum in SARM-7 and PTC-1, however, a smaller mass range scan is used. The spectrum obtained over the range m/z 189-200 for a 10-fold dilution of the spike solution (0.8 pg ml-1 in Od% V/V HC1) is shown in Fig. 6, together with a spectrum for the reference standard solution containing iridium, platinum and gold, each at 1 pg ml-1.Similar spectra for SARM-7 with and without addition of the spike are illustrated in Fig. 7. The isotope dilution determination was carried out using the following equation: A,, ( 198Pt,, - R'95Pt,, A,, R195Ptsa - 198Ptsa M,, = M,, X - where M,, is the mass of Pt in the sample, Msp is the mass of Pt added in the spike, A,, is the relative atomic mass of platinum in the sample = 195.09, A,, is the relative atomic mass of platinum in the spike = 197.64, 198Pt,, is the abundance of 198Pt in the spike = 87.47%, 195Ptsp is the abundance of 195Pt in the spike = 3.9370, 195Ptsa is the sample (natural) abundance of 195Pt = 33.870, 198Ptsa is the sample (natural) abundance of 198Pt = 7.21% and R is the measured 198PW5Pt ratio in the solution (corrected for mass discrimination). The 19gPt/195Pt ratios for SARM-7, PTC-1 and the reference standard solution (REF) were measured three times on two separate occasions.The results are presented in Table 3. Taking SARM-7 as an example, and substituting the average ratio after correction for mass discrimination using Table 2. ICP-MS data for platinum group elements and gold in SARM-7 and PTC-1 SARM-71k.g g-1 ICP-MS Element Ion (button 1) Ruthenium . . . . lOlRu+ 0.41 f 0.01 Rhodium . . . . . . lO3Rh+ 0.19 f 0.01 Palladium . . . . . . lOSpd + 1.34 k 0.05 Iridium . . . . . . 193Ir+ 0.06 k 0.01 Platinum . . . . . . 19spt + 3.04 k 0.13 Gold . . . . . . . . 197Au+ 0.57 k 0.02 ICP-MS (button 2) Ref.8 0.47 fr 0.04 0.43 0.22 k 0.01 0.24 1.29 f 0.07 1.53 0.08 * 0.01 0.074 3.27 f 0.11 3.74 0.38 k 0.02 0.31 ICP-MS (button 1) Ref. 9 0.37 It 0.01 - 0.43 f 0.01 0.62 6.87 k 0.18 12.7 0.12 k 0.01 - 1.60 k 0.06 3.0 0.25 * 0.01 0.65ANALYST, SEPTEMBER 1987, VOL. 112 1221 ~~~~ ~ Table 3. Isotope ratio data (19*Pt+/195Pt+) used for isotope dilution analysis SARM-7 PTC- 1 REF SARM-7 1.064 1.324 0.2254 1.139 1.083 1.327 0.2258 1.118 1.065 1.337 0.2258 1.136 x . . . . . . . . 1.071 1.329 0.2257 1.131 (7 . . . . . . . . 0.011 0.007 0.0002 0.011 True . . . . . . 0.2130 Corrected . . . . 1.011 1.254 1.044 - PTC- 1 REF 1.418 0.2287 1.408 0.2316 1.435 0.2317 1.420 0.2307 0.014 0.0017 0.2130 1.311 184W + 121 1 Fig. 5. ICP-MS spectrum for SARM-7 (25.5 g) and 4 ml of a 198Pt-enriched spike solution containing 8 pg ml-l total Pt in 4% V/V HCI, after fire assay separation I ( a ) 1931r+ (93 167 counts s-1) I(b) I Fig.6. ICP-MS spectra for (a) a reference standard solution containin iridium, platinum and gold, each at 1 pg ml-1 in 5% V/V HCl and 6) the 198Pt spike solution at 0.8 pg ml-1 Pt in 0.4% V/VHCl the reference standard solution (1.027) and the mass of Pt spike added (32 pg) in the equation above, gives: I 195.09 87.47 - (1.027 X 3.93) 197.64 [ (1.027 X 33.8) - 7.21 M,, = 32X - = 95.8 pg (in a 25.5-g sample) This result for platinum, with an uncertainty of 0.08 pg g-1 (based on the six individual isotope ratio determinations), compares favourably with the reference value of 3.74 pg g-1, and suggests that complete exchange between the sample and spike did occur before the subsequent chemical separation.Taking the average ratio for PTC-1 (1.283), the mass of spike added (24 pg) and its sample mass (21.8 g) gives a concentration of 2.48 yg g-1 with an uncertainty of 0.06 pg g-l. This result is not in agreement with the previously published value of 3.0 pg 8-1 and perhaps confirms that low results for PTC-1 result from incomplete sample fusion. Discussion and Conclusions The data presented demonstrate that the combination of nickel sulphide fire assay and ICP-MS has great potential for 1 ( a ) 795Pt+ (26 166 counts s-l) 1 (b) (25 950 counts s-l) 19*Pt+ 1 I I 1 Fig. 7. ICP-MS spectra for SARM-7 after fire assay separation, (a) 25.6 ) without and ( b ) (25.5 g) with the addition of the 198Pt spike I 192Os+ (700 counts s-l) I 5 1 0 min 10-15 min 1 5 2 0 m i n I Fig.8. ICP-MS spectra for 1 ng of osmium, using the osmium tetroxide generator the simultaneous determination of five PGEs and gold in geological (and probably other) materials. The anomalous behaviour of osmium, illustrated in the spectrum obtained for a reference standard solution (Fig. l ) , and probably caused by the generation of volatile osmium tetraoxide during solution nebulisation, suggests that the determination of osmium by calibration would be difficult, even if it was possible to avoid its loss during the dissolution stage of the fire assay pre- concentration. The results presented for the South African standard reference ore SARM-7 are in reasonable agreement with previously published values, although there is some evidence for loss of palladium and platinum in this work.For the Canadian standard reference concentrate PTC-1, there is serious loss of all PGEs and gold. The data obtained for platinum by isotope dilution are instructive in this respect. As explained above, the isotope dilution procedure is effective only if complete exchange and mixing between the natural and enriched isotopes occur before any chemical separation procedure. The isotope dilution result for SARM- 7, although associated with a significant uncertainty, suggests that complete exchange has occurred in this instance, while that for PTC-1 is evidence for incomplete sample attack during the fire assay fusion. Further development of the fire assay procedure is essential.1222 ANALYST, SEPTEMBER 1987, VOL.112 Although the isotope dilution determination of platinum on a routine basis is not really feasible, there may be a case for the isotope dilution determination of osmium, an element occur- ring at lower concentrations. In this respect, parallel work carried out by one of the authors (A.R.D.) on the determina- tion of osmium isotope ratios, in collaboration with the Lawrence Livermore National Laboratory (LLNL) , Califor- nia, USA, is of interest. A series of ICP-MS spectra obtained for only 1 ng of natural osmium, using an osmium tetraoxide generator developed by LLNL, is illustrated in Fig. 8. Such an approach has already been used for the determination of osmium isotope ratios in a study of the half-lif? of 187Re, which decays to lS7Os.1OJ1 There is strong evidence that osmium is not lost during the early stages of dissolution of the nickel sulphide, because it may be determined by nuclear methods in the PGE sulphide precipitate.It may be possible to carry out a controlled “loss” of osmium into the ICP during dissolution of the PGE sulphides. An even more exciting possibility is the direct determination of all PGEs in the sulphide precipitate using laser ablation ICP-MS. The authors are grateful to Dr. H. Beens, Shell Research BV (Amsterdam), VG Isotopes Ltd. and Dr. Alan Gray, for supplying the 199Pt spike solution. This work was carried out with the support of the Directorate General for Science and Technology, Commission of the European Communities (Contract No. MSM104lUWH). The paper is published with the approval of the Director, British Geological Survey (NERC). 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. References Van Loon, J. C., Trends Anal. Chem., 1984, 3 , 272. Van Loon, J. C., Trends Anal. Chem., 1985, 4, 24. Robert, R. V. D., Van Wijk, E., and Palmer, R., Natl. Inst. Metall. Repub., S. Afr. Rep., No. 1371, 1971. Date, A. R., and Gray, A. L., Spectrochim. Acta, Part B , 1985, 40, 115. Haines, J . , and Robert, R. V. D., Counc. Miner. Technol. (MINTEK, S . Afr.) Rep., No. M34, 1982. Gray, A. L., and Date, A. R., Analyst, 1983, 108, 1033. Date, A. R., and Hutchison, D. G., J . Anal. At. Spectrom., 1987, 2, 269. Steele, T. W., Nat. Inst. Metall. Repub. S . Afr. Rep., No. 1696, 1975. Steger, H. F., Canadian Certified Reference Materials Project (CANMET), Report No. 83-3E, Canadian Government Pub- lishing Centre, Quebec, 1983. Lindner, M., Leich, D. A., Borg, R. J., Russ, G. P., Bazan, J . M., Simons, D. S . , and Date, A. R., Nature (London), 1986, 320,246. Russ, G . P., Bazan, J. M., and Date, A. R., Anal. Chem., 1987,59, 984. Paper A7146 Received February loth, 1987 Accepted April lst, 1987

ISSN:0003-2654    

DOI:10.1039/AN9871201217

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

ANALYST, SEPTEMBER 1987, VOL. 112 1223 Accurate Measurement of Stable Isotopes of Lithium by Inductively Coupled Plasma Mass Spectrometry Xia F. Sun,* Bill T. G. Ting, Steve H. Zeisel and Morteza Janghorbanit Department of Pathology, Boston University School of Medicine, Boston, MA 021 18, USA A method based on inductively coupled plasma mass spectrometry (ICP-MS), for accurate isotopic analysis of lithium in human blood and urine is reported. The method permits the precise and accurate measurement of isotope ratios for 6LiPLi t o within kl0/o (RSD) for the lithium concentration range typical of urine and blood from patients on lithium therapy. The measured isotope ratio is independent of typical drifts in the ion beam intensities. The instrument response to increasing isotope ratio is highly linear for a t least an order of magnitude above the natural ratio.Complete isotope ratio data acquisition for a measurement precision of 1% (RSD) can be carried out in less than 7 min per sample. Applying the method of in vitro isotope dilution with 6Li a s the spike, the over-all accuracy of the method is better than 3% for any single sample, with a mean deviation in 12 measurements of better than 1%. Keywords : Inductive1 y coupled plasma mass spectrometry; lithium isotopes; stable isotope tracers Lithium is a therapeutic agent for manic-depressive illnesses. 1 The treatment is often life-long, consisting of maintenance doses of about 1-2 g of lithium carbonate (200400 mg of Li) daily.2 The clinical aim of the therapy is maintenance of blood concentration levels within the range 6-7 yg ml-1.3 Blood levels do not always correlate with response to therapy.4 Additionally, the maintenance doses required for therapeutic response are very close to those at which lithium toxicity becomes of concern.3 For instance, Amdisen3 recom- mends a 12-h serum lithium maintenance level of 0.8-1.0 mmol 1-I (5.6-6.9 pg ml-1) as the starting point. According to Amdisen, ".. . the critical level at which the vicious circle of lithium poisoning starts varies between patients in the range of 1.5-2.0 mmol l-1." The limitations of blood Li concentrations as reliable indices of therapeutic response are well recognised. A better alternative could be whole-body (or exchangeable) lithium. In vivo isotope dilution methods have previously been employed to measure accurately total body water5 and exchangeable body sodium and potassium contents.6 For these studies, either radioisotopes6 or stable isotope methods5 have been employed.Whereas there are no suitable radioiso- topes of lithium, the element naturally consists of two stable isotopes: 6Li (6.43% mlm), and 7Li (93.57% rnlrn). The less abundant isotope is available as a highly enriched preparation (94.88% mlrn). Therefore, it should be possible to develop in vivo isotope dilution methods to permit the accurate measure- ment of exchangeable lithium pool size(s) in man. Although the development of such a method appears feasible, it has not been reported previously. The successful implementation of such a method requires a suitable analytical technique for accurate measurement of stable isotopes of lithium in the relevant biological matrices of interest (blood, urine).Because of the need for accurate measurement of absolute isotope content, two general approaches are pos- sible: (1) measurement of isotope ratio combined with total elemental analysis or (2) direct quantitative isotopic analysis. In general, mass spectrometry does not permit the direct quantitative measurement of isotopes unless in vitro isotope dilution methods are employed.7 Because of the altered isotope ratio in samples derived from in vivo isotope dilution experiments, and the bi-isotopic nature of lithium, the application of in vitro isotope dilution requires respiking with * Visiting scholar from Tong Ji University of Medical Sciences, -t To whom correspondence should be addressed.People's Republic of China. the less abundant isotope (6Li). The only limitation of this procedure is the requirement for twice the number of analyses. We have recently reported the application of inductively coupled plasma mass spectrometry (ICP-MS) to the accurate measurement of stable isotope ratios and amounts for a number of trace elements including Zn and Cu8 and Fe.9 We have now evaluated this method for the accurate isotopic measurement of Li in specific relation to its future application to the problem of in vivo stable isotope dilution (in vivo SID) and report our results in this paper. Experimental All chemicals were of analytical-reagent grade and were used as purchased. Instrumentation The ICP-MS instrumentation was an Elan Model 250 system (SCIEX, Thornhill, Ontario, Canada).The nebuliser was of the Meinhard concentric glass type, TR-30C (Meinhard Associates, Santa Ana, CA, USA). The distance from the load coil to the sampler was 27 mm (not adjustable). The sample solution was aspirated into the argon plasma via a peristaltic pump (Rabbit; Rainin Instruments, Woburn, MA, USA) using an approximately 150-cm length of tubing [poly(vinyl chloride), i.d. 0.76 mm] (Rainin Instruments). The total calculated volume of sample introduction tubing was 0.68 ml. The solution flow-rate was approximately 1 ml min-1. Data acquisition was in the (peak hopping) isotope ratio mode. When operating in this mode, the instrument permits the control of a number of parameters via the software.For readers familiar with this instrument, the parameter set chosen was as follows: resolution, M; measurements per peak, 3; scanning mode, I; measurement mode, M; measurement time, 1.000 s; repeats per integration, 10; dwell time, 50 ms; and cycle time, 0.400 s. For readers not familiar with this instrument, the above parameters result in the following operation. On initiation of the data acquisition cycle, the instrument will start at the centre point less 0.05 a.m.u. for mlz = 6. It will acquire data for 50 ms, will then hop to the centre point, and, after 50 ms data acquisition, proceed to the point 0.05 a.m.u. to the right of the peak position, collecting ions for an additional 50 ms. This sequence is then repeated for the peak at mlz = 7, completing one cycle.The total time required for this is approximately 324 ms (4 ms overhead time per1224 ANALYST, SEPTEMBER 1987, VOL. 112 point). As the cycle time was set at 0.400 s, the second data acquisition cycle will then begin after 400 ms has elapsed from the start of the first cycle. The number of cycles is repeated until the measurement time of 1.000 s per point is reached. Following completion of a cycle, data are processed and a printout is given containing information on ion intensities (ions per second) and the corresponding isotope ratios. This constitutes one integration and is repeated as many times as required (ten for the present parameter set). The average value of the ten sequential measurements of ion intensities/ isotope ratios is taken as a single measurement and reported, where appropriate, together with +1 SD of the ten measure- ments.Unless indicated otherwise, the relative standard deviation (RSD) is taken as an indication of the measurement precision. Whereas the instrument has the capability for correcting for blank value/isobaric contributions prior to the calculation of isotope ratios, this capability was not used in this work. Prior to a day’s run, our operational practice is to optimise the instrument settings in the following order: lens settings, r.f. power, nebuliser pressure, plasma flow-rate and auxiliary gas flow-rate. The settings vary for different days, but typical values used in this work were as follows: r.f. power, about 1000 W; argon flow-rates (nebuliser pressure in lb in-2, auxiliary flow-rate in 1 min-1 and plasma flow-rate in 1 min-l, respectively) 42, 2.0, and 12.0.For these adjustments, we use a solution of 0.1 pg ml-1 Li (natural) and monitor the ion intensity at mlz = 7. Isotope Calibration Procedures The measured 6Lil7Li ratio varies from the expected isotopic ratio depending on instrument parameters (see Results and Discussion). As the accurate measurement of the ratio is of fundamental importance to our applications, we have devised a calibration procedure for converting the measured ratios in unknown samples (referred to as M&7) to the expected true isotope ratios expressed on the mass scale (referred to as the mass isotope ratio, MIR6n). We have adopted this terminol- ogy for similar studies with other trace elements.*T9 The calibration procedure consists in isotopic analyses of a set of isotope calibration standards prepared by incremental additions from a highly enriched 6Li solution to solutions of natural Li such as to provide standard solutions whose Li concentration is in the range 0.1-0.5 pg ml-1, but MIR6/7 values vary over the expected range of the ratios for the unknown samples.Isotope standards are analysed in an intermittent fashion with the unknown samples. Additionally, as there were initial concerns with respect to the possibility of matrix effects, we have also prepared similar isotope calibra- tion standards from natural matrices (human urine, plasma and red cells) spiked both with known amounts of natural Li and incremental additions of highly enriched 6Li (see Results and Discussion).Chemical Separation Scheme Following initial evaluation of a number of separation procedures, we adopted the procedure given in Scheme 1. 1 . 2. 3. 4. 5 . 6. For samples of red cells and serum (2-3 ml), digest using 10 ml of conc. HN03 and 10 ml of H202. For urine samples (5 ml) omit steps 1-3. Evaporate to a small drop; add water and repeat 2-3 times (to remove as much acid as possible). Adjust pH to 6.5-7.0 with KOH. Apply solution to anion - cation-exchange double column,* add 20 ml of water to the columns. Disconnect the anion-exchange column; elute Li from the cation-exchange column with about 40 ml of 0.9 M HCl. Evaporate the solution until the volume is about 25 ml; measure 6LiPLi by ICP-MS. 7. Also measure isotope calibration standards intermittently with the 8.Measure Li concentrations in the original sample by AAS. unknowns. *Column preparation- Anion-exchange column: Dowex 1-X8 (Bio-Rad Laboratories, Richmond, CA, USA), 100-200 mesh, chloride form, 5 x 1 cm i.d. column filled to about 3 cm height. Cation-exchange column: Dowex 5OW-X16, 200-400 mesh, hydrogen form, 20 x 1 cm i.d. column filled to about 17 cm height. Poured columns are connected then washed with 50 ml of 6~ HCl, followed by 50-100 ml of water until the pH is about 6.5. Scheme 1 Stable Isotopes and Natural Lithium Standards Highly enriched 6Li2C03 (6Li = 94.88% m/m) was obtained from Oak Ridge National Laboratory (Oak Ridge, TN, USA). An appropriate amount of the material was dissolved in de-ionised water and its resultant total Li concentration was measured by flame atomic absorption spectrometry (Perkin- Elmer Model 5000).The resultant stock solution was used in all experiments as the primary source of 6Li. Similarly, a natural lithium standard solution (1000 pg ml-1) (Traceables, MCB Reagents, EM Science, NJ, USA) was employed as the stock solution for all natural lithium used in this work. All working standards were derived from these two sources of lithium. This assured internal consistency regardless of any possible systematic errors in the measurements of Li concen- trations in the primary standards. Results and Discussion The development of an integrated analytical method for the accurate isotopic analysis of Li by ICP-MS requires careful scrutiny of a number of factors.We consider these aspects below. Instrumental Aspects The performance characteristics of ICP-MS have been ex- amined previously for isotopic analyses similar to the present application but for Fe,9 Zn and Cu.8 Instrument parameters of significance to the accurate measurement of isotope ratios are various background and interference problems, the achiev- able precision and stability of isotope ratio measurements, the linear dynamic range of the instrument with respect to isotope ratios and any matrix effects on the measured values of the isotope ratios. Background and interferences Fortunately, background ions such as those generated from the argon plasma,7 isobaric interferences9 and interferences related to interactions between reagents and the argon plasma7 appear not to play a major role with respect to stable isotopes of Li.This is fortunate because the limited number of stable isotopes of this element does not permit the selection of stable isotopes as for other elements such as Se7 or Fe.9 Typical ion intensities observed for mlz = 6 and 7 are listed in Table 1 for instances of relevance to this application. Based on these data, the general background intensities recorded at mlz = 6 and 7 correspond to 0.0003-0.0006 pg ml-1 of Li of natural composition (Table 1). The lithium concentration in samples resulting from studies with manic-depressive patients is in the pg ml-1 range3 so that the observed background intensities are negligible and do not necessitate corrections. The effect of background ions on the measured ratio of the isotope pair of interest could be important for some trace elements.7 For lithium, this does not appear to constitute a limitation at natural Li concentrations 20.05 pg ml-* (Table 1).The data in Table 1 have not been corrected for theANALYST, SEPTEMBER 1987, VOL. 112 1225 background. If this is done, it is clear that the resultant value of MR6/7 will be significantly different from the uncorrected value only for Li concentrations <0.05 pg ml-1. For instance, at lithium concentrations of 0.01 or 0.10 pg ml-1 the background-corrected values of MR6/7 are 0.0651 and 0.0631, respectively, which compare with the uncorrected values (Table 1) of 0.0667 k 0.0006 and 0.0632 k 0.0004, respec- tively. However, it is also clear from Table 1 that MR6/7 decreases with increasing lithium concentration.This is larger than the expected measurement precision for lithium concen- trations S0.05 pg ml-1. The reason for the observed change is not clear to us, but could be related to mass fractionation of the two isotopes. From a practical perspective, this is not a limitation provided that the lithium concentration is main- tained above 0.10 pg ml-1, and this can be readily accom- plished. Instrument stability An important issue related to the potential utility of this technique for accurate isotopic analysis is the stability of the measured ion beam ratios (MR6/7) during the entire measure- ment period (typically up to 8 h). Although the present version of the instrument is capable of reasonable ion beam stability (Table 2), continuous operation over several hours entails significant ion beam drift.As is evident from the data given in Table 2, the ion intensities at mlz = 7 underwent a continuous negative drift amounting to over 30% of the initial value during the 5-h observation period. This, coupled with unknown matrix effects, precludes the use of this method for accurate quantitative isotopic analysis at present (unless in Table 1. Ion intensity vs. Li concentration and MR6,7. Each set of data corresponds to a mean of replicate (n = 3-10) serial measurements, each consisting of ten sequential measurements (see Experimental) Li concn./ Time/ Id 171 pgml-1 II min ionss-* ions s- * MR6/7 0 10 0-30 4 2 2 2 396 2 5 0.106 f 0.004 0.01 5 30-45 743 2 31 11159 k 431 0.0667 If: 0.0006 0.05 5 45-61 3254 _+ 56 50056 _+ 1093 0.0650 _+ 0.0009 0.50 3 74-86 37289 f 398 586150 k 5409 0.0636 f 0.0002 0 5 86-94 4 8 2 2 479 5 18 0.0996 f 0.003 0.10 5 61-74 7444 f 70 117752 f 651 0.0632 f 0.0004 Table 2.Ion beam stability and its effect on M%,7. Data are for a 1 pg ml-1 Li solution run continuously for 5 h. Each data point corresponds to the mean k SD of ten sequential measurements Time/ Ion intensity for min m/z = 7/ions s-1 A, % M%,7 A , O/o 0-3 27-30 53-56 78-81 106-109 166-169 193-196 218-221 250-253 300-303 135-138 175280 5 7440 190415 f 4216 180266 f 8422 158481 2 6467 165770 2 2613 154366 f 21 18 141809 2 2413 133608 _+ 3331 116938 f 1127 121972 _+ 3450 120821 2 3144 0.0 +8.6 +2.8 -9.6 -5.4 -11.9 -19.1 -23.8 -33.3 -30.4 -31.1 0.0650 k 0.0006 0.0653 k 0.0011 0.0649 k 0.0006 0.0652 k 0.0004 0.0650 k 0.0006 0.0657 k 0.0006 0.0657 k 0.0011 0.0653 +- 0.0008 0.0654 k 0.0007 0.0649 k 0.0007 0.0644 k 0.0006 0.0 +0.5 -0.2 +0.2 0.0 +1.1 +1.1 +0.5 +0.6 -0.2 -0.9 Table 3.Linear regression parameters for the relationship between MR6/7 and MIR6,,. MR,,, = a(MI%,) - b. MIR6/, is the value of isotope ratio (mim) as determined from a knowledge of the total amounts of 6Li and 7Li introduced in each solution Spiked matrix a b r* Water . . . . . . . . 1.362 -0.0318 0.9998 Urine . . . . . . . . . . 1.336 -0.0307 0.9998 Serum . . . . . . . . 1.386 -0.0369 0.9984 RBC . . . . . , . . . . 1.332 -0.0248 0.9991 vitro isotope dilution analysis is employed). However, as clearly illustrated by the data in Table 2, the measured ratios (MR6/7) are independent of any instrument drift.The data given in the last column of Table 2 indicate that no systematic drift took place and that the values of MRa7 were always within the expected measurement precision of the technique. This independence has also been shown for other trace elements where instrument ion beam drift is substantially greater than shown in Table 2,s and appears, at least at present, to be a general characteristic of this method. Therefore, we conclude that the measured ion beam ratios (MR6/7) are constant within the expected measurement precision of about 1% (RSD). Dynamic linear range The numerical value of the measured ion beam ratio for any given isotope pair (M&7 for Li) is not necessarily identical with the corresponding true isotope ratio (MI&/, = 0.0687). The measured ratio varies significantly for different instru- ment operating conditions7-9 (cf., Tables 1, 3 and 4).The magnitude of the deviation between these two parameters depends on the specific element.7-9 For Li, this appears to be larger than for Zn, C U , ~ Fe9 or Se.7 However, for accurate quantitative analysis this does not constitute a difficulty provided that a quantitative relationship between the two parameters (MR6/7 vs. MIR617) can be demonstrated. If this relationship were linear over the range of isotope ratios of interest to these studies, then the task of converting the values of MR6/7 to their corresponding values of MIRsn would, of course, have been simplified. Data demonstrating the linear relationship between MR6/7 and MIRsn are given in Fig.1. Four sets of data are given, each obtained by spiking any given matrix of interest with a constant level of natural Li, but increasing levels of 6Li in order to obtain a known but variable The resultant spiked samples were then processed according to Scheme 1. The linear regression parameters for the four sets of standards are summarised in Table 3. The data clearly demonstrate an important feature of the method, viz., excellent dynamic linearity over the range of isotope ratio of interest to human metabolic studies. It should be noted that the spiked standards had been processed according to the separation procedure (Scheme 1), so that lack of matrix effect is contingent upon removal of the major matrix constituents.As will be shown later, the small differences observed in the linear regression parameters for different matrices (Tables 3 and 5 ) are not of quantitative consequence. This indicates that it should be a relatively easy task to convert the measured values of M&jI7 for any set of unknown samples to their corresponding values of MIR6/7 through a simple calibration procedure. For isotope calibration standards, water serves as a suitable matrix. This simplifies considerably the preparation of calibration standards. b I I 1 1 0 0.2 0.4 0.6 0.8 MJR617 Fig. 1. Isotope calibration graphs showing the linearity of the method and its independence from matrix composition. 0, Water; 0, red blood cells; X , urine; and A, serum1226 ANALYST, SEPTEMBER 1987, VOL.112 Memory effects In the present mode of sample introduction, the instrument utilises a peristaltic pump (total tubing volume 0.68 ml) with a solution flow-rate of approximately 1 ml min-1. Therefore, in the practical use of this method it is important to investigate the required length of time that must be allowed before the effects of a previous solution have been eliminated. We have called this “the memory effect.” We should emphasise that we do not mean to imply this to be related only, or even in large part, to the memory effect of the mass spectrometer alone. Further, we would expect this effect to vary for different elements if the extent of interaction between the sample introduction tubing and the element was significant.Data related to the “memory effect” for Li are summarised in Figs. 2 and 3. These data were acquired by sequential introduction of de-ionised water and Li-containing solutions (either 0.05 or 0.5 pg ml-1). Both ion intensities for 7Li (Fig. 2) and MRsi7 (Fig. 3) have been plotted for switching between de-ionised water and either concentration of Li. For both leading and trailing edges of the switch-over curve, a time delay of about 40-80 s is observed before the transient portion of the switch-over curve starts. At a solution flow-rate of approximately 1 ml min-1, the sample-introduction tubing is flushed within about 40 s. The remaining time is presumably related to the flushing of the nebuliser - torch assembly. In its 6 1 I I I I 0 50 100 150 200 Timeis 2 ‘ Fig.2. Data showing the over-all memory effect of the method for lithium. Abscissa indicates the time from introduction of the solution of interest. Ordinate indicates the logarithm of count rate (ions s-1 for ions at mlz = 7. (A) De-ionised water to 0.5 pg ml-l of Li; (B{ de-ionised water to 0.05 pg ml-1 of Li; (C) 0.50 pg ml-1 of Li to de-ionised water; and (D) 0.05 pg ml-1 of Li to de-ionised water 0.16 c A 1 I I I I 50 100 150 200 0.04 0 0.04 1 I I I 0 50 100 150 200 Ti me/s Fig. 3. Data showing the over-all memory effect on MR6,7. (A) De-ionised water to 0.5 pg ml-1 of Li; (B) de-ionised water to 0.05 pg ml-1 of Li; (C) 0.05 pg ml-1 of Li to de-ionised water; and (D) 0.5 pg ml-1 of Li to de-ionised water present mode, the transient portion of the curve requires an additional 40-80 s.Therefore, the total wait time in switching from de-ionised water to Li-containing solution is about 120 s. However , in going from Li-containing solution to de-ionised water, a longer time is needed because of the observed slight tailing (Fig. 2). In practice, this tailing effect does not introduce a significant limitation. Either analyte sequence can be arranged in increasing order of concentration or isotope ratio, or de-ionised water can be used to wash out the system prior to the introduction of next analyte solution. The memory effect is important with regard to the over-all analysis time. In the isotope ratio mode employed in this study, the following are realistic time segments required for complete data acquisition: 120 s for switch-over, 60-90 s for data acquisition (five data points) and 120 s for switch-over back to de-ionised water.The last time segment can be shortened if the analyte solutions are presented in ascending concentration or isotope ratio. Therefore, realistically, 5-7 min are required for the complete analysis of each sample. Matrix parameters Effect of Na concentration. The matrices of primary relevance to this application are urine, plasma and red cells. The separation of Li from the matrix constituents is relatively simple and can be carried out successfully with cation- exchange procedures (see below). Hence it is relatively easy to separate Li from organic constituents of the matrix and the anionic species. The major difficulty is the residual Na, whose elution overlaps that of Li.As the concentration ratio of Na to Li is generally high in these matrices (of the order of 1000 : l), it is important to evaluate any consequences of variations in the Na to Li concentration ratio in the final solution. Data relevant to this issue are summarised in Table 4. These data were obtained by 100 sequential measurements made on solutions of 0.10 pg ml-1 Li containing various concentrations of Na. Each block of 100 measurements corresponds to an actual run time of 30 min. Therefore, the observed changes in the measured parameters reflect a combination of effects of changes in the concentration of Na and any consequences of instrument drift. The data clearly demonstate that an increase in the concentration of Na may have a small effect on the ion intensity for 6Li.However, the effect appears to be small. In contrast, no changes are observed in the value of or its reproducibility. We conclude that the effect of Na concentra- tion, up to 100 pg ml-1, is not significant on the most important parameter of interest to these studies, v k . , M&7. We have not investigated this effect for concentrations higher than 100 pg ml-1 as such high concentrations of Na are undesirable. The chemical separation procedure outlined here readily reduces the Na concentration in the final analyte solution to well below this upper limit (Table 6). Table 4. Effect of sodium concentration on ion intensity and isotope ratio for lithium. Data correspond to mean f SD of 100 sequential measurements (30 min for each 100 measurements).The solution was 0.10 yg ml-* Li Na concentration/ Timehin pg ml-1 Idions s-l MR6/7 0-30 30-60 60-90 90- 120 120-150 150-180 180-240 240-300 300-360 360-420 420-480 0 2 2 2 100 100 100 10 10 10 0 11680 f 303 11550 f 187 11210 f 206 11350 f 153 10120 f 149 9943 f 172 9745 + 125 10750 f 185 10520 k 190 10340 f 195 10590 f 182 0.0681 f 0.0009 0.0680 f 0.0007 0.0682 k 0.0007 0.0681 f 0.0007 0.0681 f 0.0007 0.0680 f 0.0007 0.0683 f 0.0007 0.0686 k 0.0007 0.0686 f 0.0006 0.0686 k 0.0007 0.0684 f 0.0007ANALYST, SEPTEMBER 1987, VOL. 112 1227 Table 5. Matrix independence of MR6,7 Sample matrix MIR6/7* Water Serum Urine Red blood cells 0.0687 0.0610 k 0.0008 0.0616 4 0.0006 0.0598 k 0.0008 0.0608 f 0.0007 0.1439 0.1570 f 0.0016 0.1544 f 0.0020 0.1553 4 0.0015 0.1558 +_ 0.0025 0.2946 0.3517 k 0.0052 0.3506 f 0.0036 0.3422 + 0.0042 0.3514 4 0.0032 0.4526 0.5383 f 0.0037 0.5308 4 0.0050 0.5370 4 0.0070 0.5349 f 0.0080 0.5913 0.7225 k 0.0040 0.7181 4 0.0078 0.7005 4 0.0064 0.7135 k 0.0076 0.7954 0.9933 k 0.016 0.9963 f 0.013 0.9696 f 0.0094 0.9693 k 0.014 * MIR6,7: calculated value of isotope ratios (rnlrn) based on the measured amounts of isotopes used in preparing each solution.Table 6. Recovery of Li from urine subjected to Scheme 1 Sample No. Mass ratio, Li/Na* Recovery, YO 2.7 0.4 0.3 0.7 0.4 0.2 0.2 0.4 0.2 * The ratio before separation was 0.005. 100 98 95 93 92 85 96 94 97 ~ ~~ ___ Table 7. Recoveries for samples undergoing wet ashing Matrix Recovery, % n (k 1 SD) Serum . . . . . . . . . . . . 9 89 f 4 Red blood cells .. . . . . . . 10 88 4 5 ~~ Table 8. Recoveries of Li added to sub-samples of various matrices and analysed by isotope dilution analysis. Each sub-sample was spiked with 25.0 pg of natural Li and 7.325 pg of 6Li as an in vitro spike. Data given are means k 1 S.D. (pg) Matrix of 6Li-spiked standard De-ionised Red blood Samplematrix n water Urine Serum cells Urine . . . . 6 25.2k0.3 24.7f0.3 25.0k0.3 25.5f0.3 Red bloodcells 6 25.0 k 0.3 24.5 k 0.3 24.7 4 0.3 25.2 4 0.3 Serum . . . . 6 24.6f0.5 24.1 f 0 . 5 24.3 f O . 5 24.820.5 Overall matrix effects on MR6/7. Biological matrices of interest to these studies are complex, presenting many as yet unknown interactions with the argon plasma.7 These interac- tions could modify the measured ratios of any given isotope pair to the extent of introducing significant errors of measure- ment.7 We have investigated the over-all effect of the matrix on the measured ratio (M&7) at different isotope ratios for various matrices of interest to this project.The results of these investigations are summarised in Table 5. These data were obtained by spiking sub-samples of the matrices of interest with Li (natural) and 6Li in such a manner as to achieve the desired isotope ratio. These spiked samples were then processed according to Scheme 1. It is clear from these data that there are no major inter-matrix effects for MR617. As has also been discussed above in relation to isotope standards (Fig. l), the linear regression parameters of the isotope calibration plots appear to be largely independent of the matrix.In addition, as will be discussed later, the use of any of the matrices for the preparation of isotope calibration standards yields identical results with respect to the isotopic content of unknown samples. These observations taken collectively support the suggestion that the matrix effect is not an issue of concern to quantitatively accurate isotopic ratio analysis for Li provided that the proposed separation scheme (Scheme 1) is employed. Practical Aspects of Analysis A practical procedure for isotopic analysis of Li in the matrices of interest to this study is given in Scheme 1. A number of issues are important with regard to the merits of the proposed scheme: the effect of the matrix on the precision and accuracy of isotope ratio measurements, the effect of residual Na on the measurement parameters, the recovery and the accuracy of isotopic analyses.The first two issues have already been discussed. Data related to the latter two are given below. Recoveries Typical data for recovery of Li added to urine (30 pg ml-1 of Li) and the resultant Li to Na ratios are summarised in Table 6. The data demonstrate recoveries of >85%. The separation of Na from Li is sufficient for sample introduction into the ICP -MS system. Whereas for urine samples it is not necessary to remove the organic matrix components prior to application to the dual column, this becomes necessary for red blood cells and plasma. This can be accomplished in a variety of ways, such as deproteination with trichloroacetic acid, but we have found that the Li recoveries may be incomplete.Reasonably good recoveries can be obtained, however, if the samples are wet ashed with HN03 - H202. Typical results from these experi- ments are summarised in Table 7. The data clearly demonstrate that although the recoveries are reasonable, they are not complete for either serum or red blood cells. The consequences of the observed small losses in relation to accuracy of isotopic analyses depend on the particular methods used to obtain quantitative data. If only isotope ratio information is required, this is of no concern. If the quantitative analysis is carried out with in vitro isotope dilution (see below), this is also of no consequence. However, where the natural isotopic ratio has been modified, as would be the case in the important situation of in vivo tracer studies, application of the in vitro isotope dilution procedure will require re-spiking and measurement of both the re-spiked and the unspiked sub-samples.Although this can be done, it will involve twice the number of analyses. An alternative to this approach would entail the measurement of total lithium with an elemental analysis method such as atomic absorption spectrometry. In this situation, the incomplete recoveries resulting from the wet ashing procedure would introduce an underestimation, unless accurate elemental analyses were carried out on sample aliquots prior to chemical separation. Accuracy of isotopic analyses The over-all accuracy of the isotopic procedure was tested on sub-samples of urine, red blood cells and serum, each of which had been spiked with a known amount (25.9 pg) of natural lithium.Each spiked sub-sample was treated according to Scheme 1, with 6Li used as the in vitro spike. In order to test1228 ANALYST, SEPTEMBER 1987, VOL. 112 the effect of the matrix composition on the accuracy of measurements, 6Li-spiked standards were prepared from biological matrices and de-ionised water (Fig. 1). The elemental content of each test sub-sample was then deter- mined using each series of isotope-spiked calibration graphs separately. The results of these analyses are summarised in Table 8. The data demonstrate that the over-all reproducibil- ity of the analytical procedure is 1-2%. Further, absolute determinations are accurate to better than 3%.There are no systematic errors resulting from matrix effects, so that preparation of matrix-matched spiked standards does not appear to be necessary. This is consistent with similar studies carried out with this technique for other trace elements.7-9 It should be emphasised that the data in Table 8 are based on the concept of isotope dilution analysis and therefore the incom- plete recoveries (Table 6) inherent in the scheme are not reflected in the results, nor are they a cause for concern. Conclusions Accurate measurement of the two stable isotopes of Li is a prerequisite in the studies of Li pool sizes in human subjects. This requirement can now be met with a method that is also capable of a relatively high sample throughput. This is in clear contrast to other methods of isotope ratio mass spectrometry such as thermal ionisation mass spectrometry, which suffer from the major limitation of an unrealistically low sample throughput to permit human studies.10 In light of the importance of Li treatment1 in manic-depressive illnesses and the recognised limitations of current methods of Li assess- ment, the present method should prove an important develop- ment. This new method of in vivo stable isotope dilution (in vivo SID) should permit the investigation of whether blood lithium concentrations represent an accurate reflection of body lithium status. If not, will total (or exchangeable) body lithium be a better predictor of treatment outcome? Or is response to lithium therapy unrelated to body lithium status? This work was supported by a grant (DAMD17-85-G-5036) from the US Army Medical Research and Development Command, for which the authors are grateful. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. References Gershon, S., and Shopsin, B., Editors, “Lithium. Its Role in Psychiatric Research and Treatment,” Plenum Press, New York, 1973. Ehrlich, B. E., and Diamond, J. M., J. Membr. Biol., 1980,52, 187. Amdisen, A., Clin. Pharmacokinet., 1977,2, 73. Carroll, B. J., Arch. Gen. Psychiatry, 1979, 36, 870. Schoeller, D. A., vansanten, E., Peterson, D. W., Dietz, W., Jaspan, J., and Klein, P. D., Am. J. Clin. Nutr., 1980,33,2686. Moore, F. D., Olesew, K. H., McMurrey, J. D., Parker, H. V., Ball, M. R., and Baydew, C. M., “The Body Cell Mass and Its Supporting Environment,” W. B. Saunders, Philadelphia, 1963. Janghorbani, M., Ting, B. T. G., and Zeisel, S. H., in Prasad, A. S., Editor, “Trace Element Research in Humans,” Alan R. Liss, New York, 1987, in the press. Ting, B. T. G., and Janghorbani, M., Spectrochim. Acta, Part B , 1987, 41,21. Ting, B. T. G., and Janghorbani, M., Anal. Chem., 1986, 58, 1334. Janghorbani, M., Prog. Food Nutr. Sci., 1984, 8, 303. Paper A7132 Received February 3rd, 1987 Accepted April 13th, 1987

ISSN:0003-2654    

DOI:10.1039/AN9871201223

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

ANALYST, SEPTEMBER 1987, VOL. 112 1229 Determination of Aluminium in Serum by Atomic Absorption Spectrometry with the L'vov Platform at Different Resonance Lines Francesco Fagioli" and Clinio Locatelli Laboratory for Analytical Chemistry, Department of Chemistry, University of Ferrara, via L. Borsari 46, 44 100 Ferrara, Italy and Paolo Gilli Department of Nephrology, Arcispedale S. Anna, Corso Giovecca 203, 44 100 Ferrara, Italy The determination of aluminium in sera in the concentrFtion range 2-1200 pg 1-1 was studied using a L'vov platform under optimum conditions in the presence of 0. r % mNof both Mg(NO& and Triton X-100. The use of less sensitive resonance lines instead of dilution of the sample with water was found to be advantageous, as aqueous dilution of sera to more than 1 + 1 caused a systematic positive error of 20%.Matrix interferences were investigated for all the resonance lines considered in order to evaluate the use of an analytical calibration function rather than the much more time consuming addition technique. Accuracy and precision data and the merits of the analytical procedure are also reported. The aluminium contents of the sera before and after desferrioxamine (DFO) administration to 48 chronic uraemic patients on periodical haemodialytic treatment permit some considerations of the clinical utility of the DFO test. Keywords: Serum aluminium determination; atomic absorption spectrometry; L'vov platform; matrix modification; desferrioxamine The syndrome of aluminium toxicity is now well recognised in patients with chronic renal failure, particularly in those treated by haemodialysis.1.2 However, the identification of a non-invasive reliable index for the aluminium body burden is still an open question.Many observations suggest that base-line serum A1 levels do not reflect the cumulative amount of A1 and would not readily help in predicting the risk of A1 intoxication. 1-4 Recently, two different groups4.5, proposed the administra- tion of the chelating agent desferrioxamine (DFO) as a test for the evaluation of aluminium stores. Indeed, as DFO is able to mobilise tissue aluminium, its administration usually induces a marked increase in serum A1 concentration and the rate of increase is considered to be a good index of aluminium overload. In view of these facts, it is extremely important to develop an analytical method that permits the reliable determination of serum aluminium varying from its limit of detection (ie., from 2 pg 1-1 at the most sensitive resonance line, namely at a wavelength of 309.3 nm) up to about 1300 pg 1-1.Atomic absorption spectrometry with a graphite tube furnace is the most frequently used technique for this purpose and there are many publications on this subject.5.6 Some of the recent advances of this technique concerning the determina- tion of metals in biological samples include the use of matrix modifiers738 such as magnesium nitrate,g the introduction of oxygen during ashing in order to facilitate the removal of carbonaceous residues10.11 and the addition of small amounts of surface-active substances, e . g ., Triton X-100.12 The last method alleviates the problem of the accumulation of organic residues in addition to facilitating sampling by decreasing the viscosity of the solution. It is also worth mentioning the use of pyrolytically coated graphite tubes and the use of the furnace at a temperature stabilised with a L'vov platform, which eliminates both chemical and physical interferences caused by the matrix.13.14 In this paper we report on the optimum thermal pro- grammes evaluated by means of the ashing - atomisation curves for the L'vov platform for aqueous solutions of aluminium and sera with the addition of 0.1% mlV of magnesium nitrate and Triton X-100. ~ ~~ * To whom correspondence should be addressed. Simple working conditions were elaborated that permit the use of an analytical calibration function instead of the addition technique, as the proposed method was intended for routine work.The possible use of resonance lines of lower sensitivity or the dilution of the sample with serum with a low known aluminium content were also investigated in order to avoid the dilution of the serum sample with water which, in our experience, causes a systematic positive error of 10-20% when a 1 + 1 dilution is exceeded. Precision and accuracy data were also evaluated for each resonance line considered in the aluminium concentration range of the respective analytical calibration functions. The method was employed for the determination of the aluminium content before and after the administration of DFO in the sera of 48 chronic uraemic patients on periodical haemodialytic treatment.Experimental Apparatus Measurements were made with a Perkin-Elmer Model 603 atomic absorption spectrometer equipped with an HGA-500 graphite furnace, a deuterium arc background corrector and a Model 050 strip-chart recorder. An AS-40 autosampler was used for automated operation and an Intensitron aluminium hollow-cathode lamp. All the measurements were carried out with new pyrolytically coated graphite tubes with a solid pyrolytic graphite platform. Reagents and Reference Solutions In order to prepare intermediate reference solutions of various concentrations, stock solutions of aluminium (1 g 1-1) (BDH Chemicals) were diluted with de-ionised water using Gilson micropipettes of the Pipetman P series.The matrix modifier, Mg(N03)2, was of Suprapur grade (Merck) and Triton X-100 was of scintillation grade (BDH Chemicals). Sample Collection All samples were prepared in the laboratory of the Depart- ment of Nephrology of the Saint Anna Hospital, Ferrara.1230 ANALYST, SEPTEMBER 1987, VOL. 112 Blood was taken with disposable plastic syringes with stain- less-steel needles and centrifuged in polyethylene tubes. Serum that was not used immediately was conserved in polyethylene tubes at 4°C for not more than 1 week. The pool of serum used for dilution in the measurements was prepared each time using sera from normal subjects. The pool of serum contained less than 10 pg 1-1 of aluminium. Sample Preparation All plastic tubes and flasks were stored overnight filled with a solution of 1% mlV disodium ethylenediaminetetraacetate (BDH Chemicals) and rinsed with copious amounts of de-ionised water before use.All the calibration solutions in water or in serum were prepared in a plastic tube by adding to 0.4 ml of water or serum, 50 pi of 1% n~lVMg(N03)~ and Triton X-100 and 50 pl of various intermediate reference solutions, the concentration of which varied from 50 pg 1-1 to 10 mg 1-1. The solutions thus prepared contained 0.1 YO m/V of both Mg(N03)* and Triton X-100 and aluminium at concentrations varying from 5 to 1000 pg 1-1; the serum was diluted to 20% VlV. The mixtures were stirred in a vortex mixer and placed in the cup of the automatic sampler. Identical experimental conditions were utilised for the studies of the ashing and atomisation curves.The aluminium contents of the sera from uraemic patients treated with DFO were determined using the resonance line that permitted the analysis by means of the analytical calibration function at a maximum 1 + 1 dilution with water in the presence of 0.1% mlV of both Mg(N03)2 and Triton X-100. The measured absorbance values correspond to means of five readings. Results and Discussion It must be stressed that the determination of aluminium in serum by means of atomic absorption spectrometry in a graphite tube furnace has suffered from various incon- veniences, and this is still so. The introduction of the use of Triton X-100 led to some improvement, as the viscosity of the solution is decreased and hence sampling of the serum is facilitated.The combustion of organic substances was also found to be rendered more complete as carbonaceous residues are not accumulated in the presence of Triton X-100. A major advantage resulted from the use of Mg(N03)2 as a matrix modifier, which permitted an increase in the ashing tempera- ture and hence the more convenient use of the pyrolytic L'vov platform. The latter has fundamental importance in the elimination of both physical and chemical types of matrix interferences. 13-17 Nevertheless, it must be admitted that the problems related to the frequent appearance of abnormal absorbance signals during repetitive measurements appear to be unresolvable, especially at the start of a series of tests with sera, even under the optimum working conditions. In addition the "mysteri- ous" appearance of contaminating aluminium could not be completely eliminated even by various methods of cleaning all the plastic materials, carried out mainly with disodium ethylenediaminetetraacetate and nitric acid at various concen- trations.Table 1 shows the optimum parameters of the graphite furnace obtained by studying the ashing and atomisation curves using aluminium reference solutions (50 pg 1-I), a serum pool with identical amounts of aluminium added and serum samples containing approximately the same amount of endogenic aluminium. The possible interferences of the serum matrix were evaluated by means of the ratio of the slope (sensitivity, S ) of the analytical calibration function obtained in the serum pool (S,) and that corresponding to the analytical calibration function in aqueous solution (&), as shown in Table 2.Table 3 contains some examples of the analytical calibration functions for the determination of aluminium in sera related to the three wavelengths considered and the respective sensitiv- ity ratios. The precision (syx, YO) is fairly good (k5% as an Table 1. Parameters of the graphite tube furnace and of the atomic absorption spectrometer for the determination of aluminium in sera. Background corrector, yes; signal, peak area; integration time, 6 s; chart recorder, 20 mm min-1; span, 10 mV full-scale; sample volume, 10 pl Step Parameter Rampis . . . . . . . . . . . Hold/s . . . . . . . . . . . Flow-rate of internal gas (Ar)/ml min-l Recorderls . . . . . . . . . . . Reads .. . . . . . . . . . Base line/s . . . . . . . . . . . TemperaturePC . . . . . . . . . 1 2 3 4 5 6 7 . . . . . . . . . . . 110 500 1500 1500 2400 2600 20 . . . . . . . . . . . 15 30 15 1 0 1 1 . . . . . . . . . . . 50 30 30 4 8 3 20 . . . . . . . . . . 300 300 300 0 0 300 300 . . . . . . . . . . . -3 . . . . . . . . . . . 38 -1 . . . . . . . . . . . 2 Table 2. Ratio of the slopes (sensitivity, S) of the analytical calibration function in serum pool (S,) and in aqueous solution (Sa). The error reported corresponds to the 95% probability level. The measurements in the serum pool and in the aqueous solutions were carried out in parallel for each resonance line No. of Equation of calibration Equation of calibration Sr(b)? Unm tests SJS, line for serum* sr(b), "1" line for aqueous solution* % r 309.3 10 1.01 x = (0.345 k 0.023) + (1.99 f 0.07) 10-2c 4.8 0.9989 x = (0.178 k 0.032) + (1.96 k 0.05) 10-2c 3.5 0.9994 308.2 10 1.02 x = (0.191 f 0.023) + (1.33 f 0.04) 10-*C 4.3 0.9990 X = (0.081 + 0.014) + (1.30 f 0.04) 10-2c 4.7 0.9993 257.5 10 0.99 x = (0.078 + 0.013) + (2.76 f 0.09) 10-3 c 6.5 0.9989 x = (0.070 + 0.017) + (2.80 f 0.06) c 3.2 0.9990 * The equation of the line is x = a + bc, where x is the peak area (A s), a is the intercept, b is the slope or sensitivity ( S ) and c is the aluminium concentration (pg 1-1).The null hypothesis, H,, on the intercept was rejected at a probability level of 95%.ANALYST, SEPTEMBER 1987, VOL. 112 123 1 order of magnitude) if causal spurious values and data of “mysteriously” contaminated samples are disregarded for all wavelengths considered.The accuracy of the measurements was evaluated by means of recovery tests in all instances. Recoveries between 90 and 110% are acceptable for the various amounts of aluminium added and in the lower range of the linear section of the calibration graph, while recoveries of 95-10570 have to be considered satisfactory for amounts of aluminium added at the upper end of the linear range of the calibration graph, in view of the practical aims of the method. It is apparent from Table 2 that the ratios of the slopes (Sp/sa) are almost equal to unity for the three resonance lines considered. This indicates that the serum matrix does not interfere in the measurements. Hence the determination of aluminium in sera can be carried out using the analytical calibration function instead of the more time- and labour- consuming addition technique.The day-to-day precision (expressed in terms of relative standard deviation, s,) of the sensitivity (S) is also good for both aqueous solutions and the serum pool within the limit of 150 firings of the platform. It is important to stress the merits of the suggestion mentioned above concerning the use of resonance lines of lower sensitivity instead of the appropriate dilution of the serum with water in order to set the aluminium concentration within the range of linearity of the calibration graph generally measured at the most sensitive resonance line (309.3 nm). The choice between the above-mentioned procedures could be optional.However, the method of dilution with water in excess of 1 + 1 should be discarded in view of the observations made in our laboratory regarding the unreliable analytical results obtained at various dilutions of the serum while maintaining constant the concentration of Mg(N0& and Triton X-100 (0.1% rnlV). In fact, Table 4 shows accuracy ( R , Yo) and precision (s,, “/o) data for aluminium determina- tions in sera for various dilutions made with water and the serum pool. Experiments were performed at various reso- nance lines. Various solutions were prepared by adding a known amount of aluminium to a serum pool of low aluminium content (the volume of the serum changed by less than 1%). These solutions were diluted with water or with the serum as specified in Table 4 and contained an equal amount of aluminium and 0.1% mlV of both Mg(NO& and Triton X-100.It is apparent from Table 4 that the precision for all wavelengths considered is almost equal to the value already given, while the accuracy is equal to the previously mentioned value when a 1 + 1 dilution with water is not exceeded. At greater dilutions a systematic error of 10-15% is observed, which is almost constant up to the maximum dilution studied (1 + 19). However, if dilutions are made with the serum pool, the error is around the value obtained for the given concentra- tion ( * 5 % ) . This experimental evidence makes it necessary to use a resonance line of lower sensitivity whenever the determina- tion of aluminium in serum requires a dilution with water exceeding 1 + 1.In such instances it is also possible to dilute the sample with a pool of serum of low aluminium content and to make the necessary correction for the latter. Table 5 gives results for the aluminium contents before and after DFO treatment of 48 chronic uraemic patients on periodical haemodialytic treatment. As far as the clinical utility of the DFO test is concerned, it should be noted that the Table 3. Analytical calibration functions and working parameters for the determination of aluminium in serum at various resonance lines. The error reported corresponds to the 95% probability level Wavelengthhm . . . . . . . . 309.3 308.2 257.5 Slit/nm . . . . . . . . . . Concentration range (c)/pg 1-1 . . Sensitivity ratio . . . . . . . . Analytical calibration function, Correlation coefficient ( I ) .. . . Mean standard residual deviation Detection limit, K = 3 (c)/pg I-’? . . x = Q + bc* . . . . . . . . syx, Yo . . . . . . . . . . 0.7 2.0-80 1 .o x = (0.096 f 0.021) + (1.91 f 0.05) 10-2 c 0.9998 1.6 2.3 0.7 3.G120 1.5 x = (0.095 f 0.084) + (1.30 k 0.08) c 0.9993 5.4 3.1 0.2 7.4 x = (0.103 f 0.064) + 13 .O-600 (2.60 * 0.21) 10-3 c 0.9990 3.7 13.4 * The null hypothesis, H,, on the intercept was rejected at a probability level of 95%. 7 The detection limit is expressed according to the IUPAC recommendation18 and corresponds to the 99% probability level. Table 4. Accuracy and precision tests on the determination of aluminium with various serum to diluent ratios at various resonance lines. The data were corrected for the endogenous aluminium content, which was determined separately. Each value represents the average for three samples Water as diluent Serum pool as diluent Amounts of serum and A1 added/ A1 found/ Recovery, A1 added/ A1 found/ Recovery, Unm diluent CLg I-’ vg1-l Y O s,, Yo Pg I-’ I-%-’ Y O s,, Yo - - - 309.3 0.9 + 0.1 50 50.5 101 3.6 50 l + l 50 48.2 96.4 4.2 50 1 + 4 50 56.5 113 4.5 50 48.7 97.4 2.9 1 + 9 50 57.6 115 2.3 50 49.0 98.0 3.2 1 + 19 50 56.9 114 2.7 50 51.4 103 3.8 308.2 0.9 + 0.1 50 50.3 101 5.1 50 1 + 1 50 48.6 97.2 3.3 50 1 + 4 50 58.4 117 2.8 50 52.3 105 3.9 1 + 9 50 57.9 116 2.7 50 47.1 94.2 3.6 1 + 19 50 58.2 116 4.2 50 47.8 95.6 1.9 257.5 0.9 + 1 200 208 104 4.8 200 1 + 1 200 205 103 3.2 200 1 + 4 200 235 118 3.8 200 207 104 3.5 1 + 9 200 227 114 2.6 200 194 97 2.0 1 + 19 200 238 119 2.7 200 189 94 5.1 - - - - - - - - - - - - - - -1232 ANALYST, SEPTEMBER 1987, VOL.112 Table 5. Aluminium contents of sera before ( T I ) and after ( T2) 48 h DFO infusion (40 mg kg-1) for patients on periodic haemodialytic treatment A1 content of serum/ Patient Patient A1 content of serum/ I % 1-' CLg1-l No. TI T2 No. TI 7-2 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 6 8 10 12 16 18 18 19 20 20 20 22 23 26 28 29 32 35 35 37 37 38 39 40 31 11 53 12 85 33 85 142 54 95 42 91 143 56 91 41 120 71 133 108 88 118 153 94 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 40 45 48 48 51 51 54 55 55 55 62 72 78 80 80 95 129 135 140 142 189 229 262 343 80 76 217 196 243 205 246 163 190 144 212 350 315 248 260 205 313 262 312 340 434 520 600 1200 test, considered to be positive when the aluminium content difference (T2 - T,) is greater than 180 yg 1-1, was always negative in patients with aluminium contents of sera at 7'1 of less than 40 yg 1-1; in contrast, it was almost always positive in patients with aluminium contents of sera at TI of greater than 100 yg 1-1. However, the result of the DFO test could not be predicted in patients whose aluminium contents of sera were in the range 40-100 yg 1-1, suggesting that the test is necessary, within this range, in order to evaluate aluminium intoxication. Conclusion The results support the applicability of the proposed pro- cedure for the determination of aluminium over a wide range of concentrations as encountered with the DFO test.The use of resonance lines of different sensitivity permits the routine determination of aluminium in sera with good precision and accuracy. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. References Fagioli, F., Scanavani, L., Locatelli, C., and Gilli, P., Anal. Lett., 1984, 17(A13), 1473. Gilli, P., Malacarne, F., and Fagioli, F., Lancet, 1983, 1, 956. Cundy, T., and Kanis, J. A., Lancet, 1983, 1, 1168. Milliner, D. S., Nebeker, H. G., Ott, S. A., Sherrad, D. J., Andress, D. L., Alfrey, A. C., and Coburn, J. W., Kidney Znt., 1984,25, 149. Fuchs, C., Brasche, M., Paschen, K., Nordbeck, H., and Quellhorst, E., Clin. Chim. Acta, 1974, 52, 71. Fagioli, F., Gilli, P., Betti, A., and Margutti, M., Ann. Univ. Ferrara (Nuova Ser.), Sez. 5, 1979, 4(4), 41. Ediger, R. D., At. Absorpt. Newsl., 1975, 14, 127. Ediger, R. D., Peterson, G., and Kerber, J. D., At. Absorpt. Newsl., 1974, 13, 61. Manning, D. C., and Slavin, W., Appl. Spectrosc., 1983,37,1. Delves, H. T., and Woodward, J., At. Spectrosc., 1981, 2, 65. Eaton, D. K., and Holcombe, J. A., Anal. Chem., 1983, 55, 946. Kaehny, W. D., Alfrey, A. C., Holman, R. E., and Shorr, W. J., Kidney Znt., 1977, 12, 361. L'vov, B. V., Spectrochim. Acta, Part B, 1978, 33, 153. Leung, F. Y., and Henderson, A. R., At. Spectrosc., 1983,4,1. Slavin, W., Manning, D. C., and Carnrick, G. R., Anal. Chem., 1981,53, 1504. Manning, D. C., Slavin, W., and Carnrick, G. R., Spectrochim. Acta, Part B , 1982,37, 331. Slavin, W., Manning, D. C., and Carnrick, G. R., At. Spectrosc., 1981, 2, 137. International Union of Pure and Applied Chemistry, Ana- lytical Chemistry Division, Spectrochim. Acta, Part B, 1978, 33, 219. Paper A 71 73 Received February 27th, 1987 Accepted April 8th, 1987

ISSN:0003-2654    

DOI:10.1039/AN9871201229

出版商:RSC    

年代:1987

数据来源: RSC