ISSN:0003-2654    

DOI:10.1039/AN98409FX013

出版商:RSC    

年代:1984

数据来源: RSC

ISSN:0003-2654    

DOI:10.1039/AN98409BX015

出版商:RSC    

年代:1984

数据来源: RSC

ISSN:0003-2654    

DOI:10.1039/AN98409BP029

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

ANALYST APRIL 1984 VOL. 109 413 Simultaneous Determinations in Flow Injection Analysis A Review Maria Dolores Luque de Castro and Miguel Valcarcel Cases Department of Analytical Chemistry Faculty of Sciences University of Cordoba Cordoba Spain Summary of Contents Introduction Conventional FIA methodology Simultaneous determinations with several detectors In series In parallel Single injection with splitting up of the sample Si m u ltaneous m u I t i-i n jecti o n Zone sampling Simultaneous determinations with a single detector Splitting up of the stream with two cells aligned in the same optical path With a pH gradient With ion exchange With sequential injection FIA methods based on differential kinetics FIA systems with two detectors FIA systems with multi-detection FIA systems with a single detector With single injection Combination of conventional FIA with stopped flow With splitting up of the flow and double path cell With splitting up of the flow in two different reactors and their subsequent confluence Different measurement times in the two bolus - reagent interphases FIA configuration with double injection Conclusions References Keywords Review; simultaneous determinations; flow injection analysis Introduction The concept of simultaneous determinations used in this paper is related to the determination of two or more species in the same sample with a single flow injection analysis (FIA) system.The need to measure several parameters rapidly in the same sample in areas such as clinical chemistry environmental pollution and industrial control has urged the development of automated methods of analysis which in both the continuous and discrete modes offer the possibility of carrying out simultaneous determinations.In the discrete mode the same sample must be present in as many locations as there are parameters that are going to be determined there being a single detection system. In the continuous mode or segmented flow analysis the sample is split up into as many channels as there are parameters that are going to be determined. Reaction and detection units exist for each of them. Although research on FIA has undergone great develop-ment in recent years commercialisation of the method is still 1imited.l One of the reasons for this gap can be attributed to the small number of simultaneous determinations that have been described.We are convinced that a broader develop-ment of this area relatively unexplored in FIA will stimulate the development of FIA instruments especially for applica-tion in clinical analysis where the need to know several parameters is very frequent principally in samples of blood and urine. In this paper a critical review of FIA methods that permit the simultaneous determination of several species is presen-ted. The most elementary classification of these methods can be made on the basis of taking the correlation between the detection unit and the species to be determined. Thus there is one group of methods in which a detector for each species to be analysed exists. Another group consists of methods in which the number of detectors is lower than the number of species to be determined.Figs. 1 and 2 show the most important manifolds used in both types of methods. However, this review has been constructed following the classification of these methods according to whether the conventional method-ology is used or whether they are based on differential kinetic analysis such as that which appears in Table 1 where the main simultaneous determinations developed by FIA are summar-ised. Table 1 also gives the detection and injection systems and the principle of each of them. Conventional FIA Methodology Simultaneous Determinations with Several Detectors The relative location of the detectors in simultaneous FIA systems permits a sub-classification according to whether their configuration is in series or in parallel.Their characteristics differ considerably. In series The most common examples are related to the use of the potentiometric technique [Fig. l(a)]. The use of ion-selectiv 414 ANALYST APRIL 1984 VOL. 109 I n parallel It is possible to make a division based on the manner in which the sample reaches each detector. The most usual division is that where splitting up of the sample bolus occurs in a regular and reproducible manner each part passing to each detector, or the use of several valves working simultaneously. Less usual but interesting is the case in which the two parts of a zone sampling are used a portion extracted from the original bolus is sent to one detector and the remainder of the same bolus is directed to the other detector.Each of these types is commented on below. Single injection with splitting up of the sample. These determinations (with slight modifications) correspond to the scheme in Fig. l ( b ) . A system with photometric and turbidimetric detection has been developed for the determina-tion of C1- and S042- by the displacement of the Hg(SCN)42-complex in the presence of Fe(II1) and the formation of barium sulphate,b respectively. It consists of a sampler that works in an 18-s cycle. so that it has a capacity of 200 samples per hour. The almost immediate splitting up of the sample bolus is replaced by the formation of two sub-systems a channel of the appropriate reagent merging with each of them. The length of the reactor of each sub-system is a function of the intrinsic characteristics of each reaction.To compensate for the decrease in pressure in these sub-systems tubing of suitable diameter and length is placed after each detector. In this way a homogeneous division of the sample is obtained. For the simultaneous determination of up to four species (Na K Mg and Ca) by the use of two optical detectors with different characteristics (emission and absorption) Basson and Van Standen7 designed an FIA system and applied it to surface ground and domestic waters with an automatic sampling frequency of 128 h-1. The injection is carried out in an aqueous carrier with which the reagent [ Li( I) La( 111) , Cs(I) C1- and NO3- solution] immediately merges. This solution is split up after the mixture coil reaches the corresponding detectors.One of the sub-boluses enters an atomic-absorption spectrophotometer (dual-channel ana-lyser) where Ca and Mg are determined while the other sub-bolus enters a flame photometer for the measurement of Na and K. Although these workers did not specify that the emission system is a double-channel one their claim for 500 analyses per hour permits us to conclude that the last detector holds this characteristic in common with the first one. The lack of carry-over between samples and the advantageous comparison with the automated segmented flow techniques confirm the usefulness of the proposed technique. A modification of the scheme in Fig. l(b) consists of an additional channel that merges with each of the auxiliaries R1 and R2 before their confluence with the main ones and was used for the simultaneous determination of nitrite and nitrate8 by the traditional modified Griess reaction (Shinn method).It consists of two almost identical FIA sub-systems the differ-ence being the inclusion of a reducing column and the aspiration of this waste to mitigate the effect of the compact-ness of the column on the flow In the higher sub-system the nitrite determination is carried out. In the lower one the existence of the reducing column [cadmium granules coated batch-wise with copper(I1) sulphate solution] causes nitrate reduction with an efficiency of 80-90% and permits after the confluence with the reagent channels the analysis of the total mixture the nitrate concentration being obtained by differ-ence.U U c w Fig. 1. Schematic diagrams of FIA manifolds for simultaneous determinations with several detectors (D). ( a ) In series configuration; and (b) parallel configuration with splitting up of the sample channel. C, Sample; R1 and R2 reagents; and W waste electrodes placed sequentially in an FIA system permits the determination of several species simultaneously. A typical case of this type of determination is that developed by Virtanen2 for the determination of Na(I) K(I) Ca(I1) and C1- in serum by the use of ion-selective electrodes in which the flow impinges laterally on both the sensors and the reference, which is located behind them and in the same position. Regression equations that correct for interferences from some species in the determination of others have been established.The over-all determination of Ca(I1) and pH in serum described by Hansen et al. ,3 in a paper in which they proposed several FIA methods for the determination of calcium in serum and water is carried out with the same scheme as in the above-mentioned example [Fig. l(a)] with the difference that the additional channel merges with the main channel in the measurement cell. The system consists of a flow-through capillary electrode suitable for the measurement of pH in blood and a calcium-selective electrode which the flow finally meets after passing through the tubular electrode. The reference electrode located at the bottom of the calcium cell, receives both the sample flow after impact with the calcium sensor and an additional channel of buffer whose purpose is to dilute the sample so that the composition of the solution that surrounds the reference electrode remains almost constant.The level which is also constant in the measurement cell is established with a differential pump. For other simultaneous determinations (nitrate and potassium or nitrate and sodium) with ion-selective electrodes the same workers utilised a configuration in which the sensors are not in series but with an angle between them of slightly under 180". A special manifold for simultaneous measurements that cannot be carried out in series but which can be included under this heading has been applied to the determination of glucose and urea using enzymatic electrodes for both species.4 The positions of the electrodes in the flow cell are diametric-ally opposed.Two recorders permit the signal corresponding to each sensor to be obtained. A peculiarity of the system is the absence of injection. The sample is aspirated for 1 s by a pump located behind the detection cell. A subsequent 2-s washing step permits the return to the base line. Also with differential characteristics but suitable for inclusion here is the simultaneous determination of pH, calcium potassium chloride etc. by ion-sensitive field effect transistors (ISFETs) which facilitate the in vivo measure-ments of these species when several ion-selective membranes are located on the different gates of the ISFET.5 These sensors have still not been sufficiently developed or applied to specific problems.Simultaneous multi-injection. Simultaneous multi-injection has been utilised by Slanina et al.9 for the determination of three species by the use of a multi-channel spectropho-tometer. The determination of chloride. ammonium and nitrate is carried out by utilising a three-injection valve system the loops of which are simultaneously filled by a singl ANALYST APRIL 1984. VOL. 109 415 sample stream. Each valve injects the corresponding sample in an aqueous stream that merges with a reagent suitable for the determination of the species to be analysed in that sub-system. The reactions that serve as a basis for the identification (and counting) of each species are methylene blue formation, displacement of the Hg( SCN)42- complex and non-chemical reaction for NH4+ CI- and NO3- respectively.Zone sampling. In their first paper on this technique Reis et al.1” outlined the possibility of its application to the simul-taneous determination of species that require different degrees of sample dispersion and later11 applied it to the determination of aluminium and iron in plant digests. The spectrophotometric determination of aluminium with Erio-chrome Cyanine R by FIA requires a high degree of sample dispersion. In contrast the determination of iron in acidic samples (nitric acid - perchloric acid) of plant digests is usually performed without sample dilution when an air - acetylene flame is employed. The most appropriate FIA system for this simultaneous determination with minimum consumption of reagents consists of a triple injector - commutator in which the sample is aspirated to fill the corresponding sample loop.When the injector is switched to the injection position the following occur simultaneously (a) the selected volume of sample is inserted into the carrier 1 (b) the second sample loop is moved towards the area of movement of the sample and (c) the reducing reagent starts to fill the corresponding loop. The injection of the sample gives a sample zone that is sent through the dispersion coil the second sample loop and the transmission line towards the atomic-absorption spectro-photometer. On switching the injector back to the filling position to start a new cycle the sample zone passes through the second loop.This simple movement causes the simul-taneous introduction of both the sample portion inside it and the selected volume of the reducing reagent into the corre-sponding carrier stream. The two zones flow together with the Eriochrome Cyanine R. One of the greatest possibilities for the future of simul-taneous determinations by FIA is the use of a single multi-detector such as an inductively coupled plasma (ICP). The capacity to utilise a spectrometer of this type in conjunction with FIA has been shown by Greenfield.12 Simultaneous Determinations with a Single Detector With sequential injection The Brazilian research group directed by Bergamin has published a technique for the determination of several species by the use of the merging zones principle and a triple injector -commutator which they termed “simultaneous determin-ations,” but which correspond more exactly to the character-istics of “Sequential determinations,” as they do not utilise a single injection or several simultaneous injections to deter-mine different species but by ingenious modifications of the injector - commutator they succeed in making each injected sample provide the form for measurement of a simple species.Representative examples are commented on below. For the determination of nitrite and nitrate,l3 the injector is modified in such a manner that one of the loops is used for the reagent injection [sulphanilamide + N-( 1-naphthy1)-ethylenediammonium chloride]; of the other two injectors, one is normal and the other has a reducing column incorpor-ated and they are used alternatively.The confluence of the reagent bolus with that of the sample from the normal loop permits the nitrite determination. The subsequent combina-tion of the reagent with the sample bolus from the loop with the reducing column incorporated makes the determination of both species possible. The use of a reducing pre-valve through which the sample passes or not also provides a similar means of determining these species.I4 The same group utilised a more complicated triple injector -commutator for the sequential determination of nitrogen and phosphorus in plant material. 15 The methods of determination are based on the Berthelot and molybdophosphate reactions, for which the measurement wavelength is 630 nm in both instances.The injector works in two positions. In one of them the sample and reagents for phosphorus determination are placed in the analytical system while the sample and the reagent involved in the nitrogen determination start to fill their corresponding loops. First the sample merges with the molybdate solution and the resulting bolus merges subse-quently with the ascorbic acid solution. When the injector is switched on again a new portion of sample now merges with the alkaline phenol reagent and this sample plus reagent mixture merges later with the hypochlorite solution. The volumes of sample injected are different for each species to be determined and the volumes of the reagents are also different, depending on the optimum characteristics of each determina-tion.The non-simultaneous nature of the determination is once again clearly established in the sense that the term “simul-taneous” is used in analytical terms meaning measurements of several species in the same sample or in different portions of the same samples but with measurements carried out at the same time. For this reason although they are included in this section as simultaneous determinations with a single detector, they can be considered as a particular variant of the other types included. Splitting up of the stream with two cells aligned in the same opticalpath. This type of simultaneous determination does not fit any of the proposed schemes and corresponds to an ingenious manifold designed by Stewart and RfiiiCka16 in the early days of the technique for the determination of nitrogen and phosphorus using a single spectrophotometer.It consists of a main channel through which an acidic stream circulates. The sample is injected into this and split into two channels, and the appropriate reagent merges with each of them. The two flow-through cuvettes are identical and aligned in the optical path of the sample beam. A coil inserted immediately after the injection point has the object of avoiding the effect of variation of the injection speed on sample splitting. This occurs at a ratio 4 1 (P N) as the split point of the sample proceeds through a short transmission line- ( 5 cm) into the phosphate branch and through the log (470 cm) phasing coil into the nitrogen line. The role of a coil of 470 cm in the nitrogen sub-system placed immediately after the splitting of the sample is to delay the passage of the sample plug in the nitrogen line until the sample zone in the phosphate line passes the flow cell and clears the optical path below the 1% absorbance level.Therefore in the common record the peaks corresponding to the two species do not overlap. With apHgradient. This restricted but interesting mode of simultaneous determination by conventional FIA involving a single detector and single injection is based on the use of a pH carrier that is different to the sample. If the volume of the latter is sufficiently large when it reaches the detector two zones of different characteristics exist that are close to the interphases with pH values different from that of the central zone of the plug.The characteristics of these regions can be used to determine several species in the same sample. An example of this type is the simultaneous determination of Pb(I1) and V(V)17.18 with 4-(2-pyridylazo)resorcinol (PAR). The complexes of PAR with these cations exist in the following pH ranges at pH below 3 only the V(V) complex exists in the pH range 3-9 the V(V) and Pb(I1) chelates co-exist and above pH 9 only the Pb(I1) complex is detected. When a PAR solution in NH4+ - NH3 buffer (pH 9.9) is used in an FIA system such as that shown in Fig. 2(a) but without an additional confluence channel and a sample of a mixture of those cations at a sufficiently acidic pH is injected at the moment at which the sample plug reaches the detector a pH gradient is obtained that allows the existence of plug zones in which only one of the complexes is formed 416 C S R I ANALYST APRIL 1984 VOL.109 volume D VW n n Fig. 2. Schematic diagrams of FIA manifolds for simultaneous determinations with a single detector. Several peaks are obtained for each volume injected. (a) Based on the establishment of a concentra-tion gradient; ( b ) with simultaneous double injection and an asymmetric merging configuration; and (c) based on the splitting up of the sample channel into two reactors which merge in front of the detector A pH gradient can also be obtained by the injection of an acidic or basic solution in a basic or acidic carrier respectively. The existence of these gradients together with the kinetic dissociation of the PAR - M(I1) complexes has been exploited by Betteridge and Fields19 for the simultaneous determination of Co(I1) - Mn(I1) and Ni(I1) - Cu(I1) mixtures in the presence of a similar amount of Co(I1).Determinations of Bi Th and Cu have been carried out with this FIA mode.2" With ion-exchange. FIA and ion-exchange association provide the possibility of carrying out simultaneous determi-nations such as that of zinc and cadmium based on their inhibitory effect on the luminol - hydrogen peroxide chemi-luminescence system catalysed by Co(I1) ,21 The FIA configur-ation used is that shown in Fig. l(a) with the previous confluence of an auxiliary channel with each of those shown. The four primary channels merge in the following manner: luminol and hydrogen peroxide on the one hand and the catalyst and the sample on the other.The subsequent confluence of the two resulting channels gives rise to the catalysis or inhibition of the monitored reaction. The carrier of the sample is formed by an acidic solution (0.1 M HC1 + 0.5 M NaCl) in which the sample and eluent solutions are injected immediately before the anion-exchange column and in a sequential manner in time to prevent overlapping of the eluates. The method is sensitive as it permits the determina-tion of zinc and cadmium in the ranges 10-100 and 2&200 ng ml-1 respectively. A time of 3 min per sample is required. FIA Methods Based on Differential Kinetics One of the advantages of kinetic analysis methods over equilibrium methods is the possibility of carrying out simul-taneous determinations based on the different rates of their reactions with a common reagent.In spite of this being a promising aspect it has several important limitations that restrict its applicability. Firstly it should be emphasised that it is not easy to find chemical systems in which significantly important differences can be established in the experimental conditions between two or more reaction rates. On the other hand. the differential kinetic methods described do not have a very high level of accuracy and/or reproducibility because slight disturbances produced by the diversity of samples or by slight changes in the working conditions lead to a low precision in comparison with other manual or kinetic techniques.FIA is an important alternative in the development of differential kinetic methods because it imparts technical advantages over manual procedures and also can be applied to chemical systems of reduced half-life (1-10 s). For extra-fast systems (half-life of the order lo-3-10-1 s) FIA is not suitable and it is necessary to apply stopped-flow technology which is very complex and is expensive owing to the high precision required in carrying the sample towards the detector and in the data processing systems. Few differential kinetic methods have been developed for FIA. For a description of the most significant contributions the systematisation in Table 1 has been adopted according to the number of detectors that the FIA system contains. FIA Systems with Two Detectors This consists of a differential kinetic mode with a simple principle.The signal produced by the reactant bolus is measured at two different times tl and t2 in each detector. A scheme of two determinations of this type proposed by Jensen and co-workers22.23 is shown in Fig. l(a). Both are based on the displacement reactions of the complexes (ligand inter-change). The first determination of this type is based on the different dissociation rates of the Mg(I1) and Sr(I1) complexes with CDTA (trans-l,2-diaminocyclohexane N N ","-tetra-acetate) : %YCD'* M-CDTA2- + - H+ HCDTA3- + M2 (1) M the dissociation of the Sr(I1) complex being 100 times faster than that of the Mg(I1) complex. If Cu(I1) exists in the solution it acts as a scavenger according to the fast reaction k F A HDCTA3- + CU*+ - Cu-CDTA2- + H+ If the concentration of Cu(I1) is high enough the following occurs: kCDTA cu [Cu2+] >> kzDTA [M2+] Reaction (1) determines the over-all process rate which is monitored spectrophotometrically by the absorption (hmax.= 320 nm) of the Cu-CDTA complex. The sample a mixture of Mg(I1) and Sr(I1) complexes with CDTA in excess merges with a Cu(I1) solution A short coil takes the reaction mixture to detector 1. By this time tl the dissociation of the complex formed initially has not started so that the signal is due to the Cu-CDTA complex formed with the free ligand. The total concentration (Mg + Sr) is deduced by difference. The sample is carried by a relatively long reactor and after a time t = t2 - tl the signal is measured in detector 2.At this moment the Mg-CDTA complex is not dissociated while the strontium complex is 50% dissociated. The Cu-CDTA complex formed is related directly to the Sr(I1) concentration and the Mg(I1) concentration is obtained by difference. The assembly is similar for the Mg(I1) + Ca(I1) mixture using a C2.2.11 cryptand24 as a ligand and sodium ion as a scavenger ANALYST APRIL 1984 VOL. 109 417 Single injection Table 1. Simultaneous determinations by FIA path cell 27 reactors and subsequent confluence 28 Splitting up the flow in twodifferent Different measurement times in the two Conventional FIA methods FIA methods based on differential kinetics Detection system In series In parallel With several detectors Injection system Single injection Single injection Mu1 ti-i nj ection Single multi-detector I Zone sampling Sequential injection With a single detector In series With several detectors With a single detector Principle References Several ion-selective electrodes with a single reference electrode 2.3,4,5 With splitting up of the flow after the With a valve for each parameter and Collection of part of the injected ICP Use of different reagents for different injection multi-channel detector bolus that is directed to other detector samples according to the parameter to be determined cells aligned in the same optical path Splitting up of the flow with two flow pH gradient Ion exchange 6,7,8 9 1 0 l l 12 13 14 15 16 17,18, 19,20 21 FIA Systems with Multi-detection Hooley and Dessy25 recently proposed a significant FIA system.In addition to developing a feedback system for the control and regularisation of flow that will certainly solve this critical aspect of FIA they proposed interesting schemes relating to photometric systems of detection based on LEDs (light emitting diode sources). The paper is orientated towards kinetic determinations based on multiple measurements. The multiple detection system consists of a quartz reactor tube with a series of independent detection units with an LED and a photodetector. Its signal is monitored by an electronic data-processing system. A plug of reactant sample passes successively through each measurement point at different times in such a manner that it is possible to process as many types of kinetic data as there are detectors.The relative locations of these detectors depends on the flow and rate reactions considered. It is an ideal system for kinetic determi-nations and very useful for the simple determination of rate constants. These workers proposed the simultaneous determi-nation of Mg(I1) and Zn(I1) by differential kinetics using the logarithmic extrapolation method. FIA Systems with a Single Detector This is the simplest mode and includes those systems which are classified as “differential kinetics” but which do not truly meet all the requirements for such methods. Basically a detector provides two different signals or a signal increase or both at two times (or at a time increase) that coincide(s) with different reaction times.In general the method of “proportional equations” can be used for the calculation of the concentration of a species in the mixture starting from the signals obtained and with a knowledge of the corresponding rate constants. When a signal and an increase are used [the FIA-gram peak and stopped-flow interval (delay time)] the method for the calculation is different. Several different FIA configurations for carrying out these determinations will be considered making a distinction between systems with a double and single injection. With single injection Combination of conventional FIA with stopped flow. Recently Kagenow and Jensen ,26 studying the application of FIA to the development of differential kinetic methods, carried out a simultaneous determination of Ca(I1) and Mg(I1) by the stopped-flow technique based on the same reactions of displacement and formation of coloured complexes as des-cribed above.In this instance the Mg - L complex undergoes rapid dissociation whereas the Ca - L complex dissociates more slowly when K+ is used as a scavenger. The FIA scheme is as shown in Fig. 2(a) automated in this instance. The sample contains Ca(II) Mg(II) the same buffer as the carrier and an excess of ligand C2.2.21 cryptand.24 In the other channel is a solution of the same buffer that contains the scavenger ion and o-cresolphthalein complexone. Both chan-nels merge at a confluence point in a micromixing chamber (ca. 3 pl) and after passing through a short reactor enter a photometric detector where the magnesium and calcium chelates with the complexone are detected.The former is completely formed when it arrives at the flow cell; the latter is developed during the delay time. With splitting up of the flow and double path cell. Betteridge and Fields27 recently proposed a two-point simultaneous kinetic determination of cobalt and nickel based on the different rates of ligand substitution of the citrate complexes of both metals using PAR as a scavenger. The stream containing the citrate complexes merges with another stream containing the scavenger. The stream is then split up into two channels and EDTA which halts the reaction is added to the shortest channel. In the other channel a coil measuring about 3 m is heated at 45 “C to complete the reaction.EDTA is also added after heating. Each stream passes through a double 418 ANALYST APRIL 1984 VOL. 109 path cell consisting of a Perspex block with two holes drilled in the light path of a single spectrophotometer. Two peaks are obtained. The application of conventional proportional methods allows the determination of cobalt and nickel in a mixture with slightly high relative errors. It is interesting that of the interferences from several ions such as copper are eliminated by masking with the EDTA after the formation of the Ni - PAR and Co - PAR complexes which are not dissociated by EDTA. With splitting up of the flow in two different reactors and their subsequent confluence. A very simple alternative for carrying out differential kinetic methods by FIA is based on the splitting up of the flow after the sample injection passing the two sub-boluses through two reaction tubes with different geometrical characteristics and subsequent confluence before reaching the detector [Fig.2(c)]. The different geometrical and hydrodynamic properties of the two channels provide different residence times for each of them and an FIA-gram with two peaks is obtained. The overlap of these peaks depends on the relative lengths of the channels. A manifold has been utilised for the analysis of mixtures of Co(I1) and Ni(I1) ions which have different rates of reaction with 2-hydroxybenzaldehyde thiosemicarbazone.28 The different contributions of the two complexes to the final absorbance in each peak permits the establishment of two equations for the determination of the individual ions.Different measurement times in the two bolus-reagent interphases. A new possibility for carrying out simultaneous determinations based on differential kinetics by FIA consists in the injection of an unusually large sample volume so that the mixture of the sample plug and reagent carrier is only produced in the two inter-phases. If there is a sufficient time interval between them before reaching the detector [depend-ing on the Vi/VR ratio ( Vi = injection volume and VR = reactor volume)] two signals or FIA peaks are obtained. This method has been applied to the determination of Co(I1) and Ni(I1) in mixtures based on the above-mentioned complexation reac-tion 29 FIA configuration with double injection In 1980 Kagenow and Jensen3O described an FIA system with a synchronised double injection valve and a single detector for the determination of Mg(I1) and Sr(I1) ions.The principle of the method means that two boluses of different composition are simultaneously injected and run through tubing of different lengths so that they reach the detector sequentially and give two peaks in the FIA-gram. The chemical principle of the determination is to use the traditional reactions of ligand displacement often utilised in differential kinetics. The complexes between [2.2.1] cryptand24 and alkaline earth metals have different stabilities. In this instance the dissociation constants KEL of the Mg(I1) and Sr(I1) complexes are very different KggL >> KSfL.To obtain a sufficiently high reaction rate a scavenger ion such as Kf is used which picks up the ligand at high speed the reverse reaction being insignificant when an excess of KN03 is present in the medium. To monitor the presence of alkaline earth metal ions in the system it is essential to use an additional ligand that forms colour complexes with both in a fast reaction. The FIA scheme with which this study has been carried out is shown in Fig. 2(6). Conclusions Few simultaneous FIA determinations have so far been described such methods accounting for only about 8% of papers published on FIA. This is surprising considering that FlA can easily be adapted to carrying out the determination of several species in a single sample.Undoubtedly when simul-taneous analyses of samples of great practical interest (in clinical analysis contamination environmental fields etc.) are carried out commercial FIA instrumentation will be developed especially automated apparatus. Such simultan-eous determinations are also of value in small control laboratories where automation is not essential. In view of the general characteristics of FIA in our opinion it is not difficult to design devices for simultaneous determina-tions in the same way as when the development of segmented continuous flow analysers began. When these systems are developed FIA will have clear advantages over other automated analytical methods owing to its intrinsic charac-teristics of rapidity simplicity versatility and low cost.Until these objectives have been attained other automated methods surpass FIA in this respect. Among all the systems described it should be emphasised that the methods based on differential kinetics suffer from the difficulties inherent in this methodology and therefore their application in routine analyses must be approached with caution. Similar considerations apply to the methods based on pH gradient and zone sampling the results of which are hardly influenced by the working conditions which can lead to problems of reproducibility. In our view the development of simultaneous routine determinations must be based on the use of several detection units or multi-detectors and by multiple injection or on simple splitting up of the injected bolus.Other developments have a markedly academic character which is why their adaptation to the routine analysis of real samples is very difficult. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 1.5. 16. 17. 18. 19. 20. References RGiitka J. Philos. Trans. R. SOC. London Ser. A 1982,305, 64.5. Virtanen R. “3rd Symposium on Ion-selective Electrodes,” Maison d’Editions de 1’Academie des Sciences de Hongrie, Budapest 1980 p. 37.5. Hansen E. H Rfiiitka J. and Ghose A. K. Anal. Chim. Acra 1978 100 1.51. Mascini M. and Palleshi G. Anal. Chim. Acra 1983 145, 213. Ramsing A. U. Janata. J. Rfiiitka J . and Levy M. Anal. Chim. Acra 1980 118 4.5. Basson W. D. and Van Standen J. F. Water Res. 1981 15, 333.Basson W. D. and Van Standen J. F. Fresenius 2. Anal. Chem. 1980 302 370. Anderson L. Anal. Chim. Acta 1979 110 123. Slanina J. Bakker F. Bruyn-Hes A. and Mols. J . J . Anal. Chim. Acta 1980 113 331. Reis B. F. Jacintho A. 0 Moratti J. Krug F. J. Zagatto, E. A. G. Bergamin F. H. and Pessendal. L. C. R. Anal. Chim. Acra 1981. 123 221. Zagatto E. A. G. Jacintho A. 0 Pessenda L. C. R . Krug, F. J . Reis B. F. and Bergamin F. H. Anal. Chim. Acra, 1981 125 37. Greenfield S Ind. Res. Dev. August 1981. Gin&. M. F. Bergamin F. H. Zagatto. E. A. G. and Reis. B. F. Anal. Chim. Acra 1980 114 191. Van Standen J. F. Anal. Chim. Acta 1982 138. 403. Reis B. F. Zagatto E. A. G. Jacintho A. O. Krug. F. J and Bergamin. F. H. Anal. Chim. Acra 1980 119 305. Stewart J. W. B and RfiiiEka J. Anal. Chim. A m 1976,82, 137. Betteridge D and Fields B. Anal. Chem. 1978. 50. 654. Fields B. Proc. Anal. Div. Chem. SOC. 1979 16 4. Betteridge D and Fields. B. AnaL Chzm. Acru 1981. 132, 139. Baban S. Anal. Proc. 1980. 17 53.5 ANALYST APRIL 1984 VOL. 109 419 21. 22. 23. 24. 25. 26. 27. Burguera J. L. Burguera M. and Townshend A. Anal. Chim. Acta 1981 127 199. Dahl J . H. and Jensen A. Anal. Chim. Acta 1979,105,327. Espersen. D. and Jensen A. Anal. Chim. Acta 1979 108, 241. Lehn J.-M. Struct. Bonding (Berlin) 1973 16 1. Hooley D. H. and Dessy R. E. Anal. Chem. 1983,55,313. Kagenow H. and Jensen A. Anal. Chim. Acta 1983 145, 125. Betteridge D. and Fields B. Fresenius Z . Anal. Chem. 1983, 314 386. 28. 29. 30. Valcarcel M. Luque de Castro M. D. Fernandez A. and Gomez-Nieto M. A. Paper presented at 9th International Symposium on Microchemical Techniques Amsterdam 1983. Valcarcel M. Luque de Castro M. D. Fernandez A. and Linares P. Paper presented at First International Symposium on Kinetics in Analytical Chemistry Cordoba 1983. Kagenow H. and Jensen A . Anal. Chim. Acta 1980 114, 227. Paper A3/265 Received August 16th 1983 Accepted October 31st I98

ISSN:0003-2654    

DOI:10.1039/AN9840900413

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

ANALYST APRIL 1984 VOL. 109 42 1 Determination of Free Cyanide in Mineral Leachates Thomas P. Lynch Chemical Analysis Branch BP Research Centre Chertse y Road Sunbury-on-Thames Middlesex, Tw16 7LN UK Cyanide leaching is a common process by which metals e.g. silver and gold can be extracted from ores and related materials. This paper describes an automatic flow injection system for the determination of free cyanide in samples from leaching processes. The sample is injected into a carrier stream which is merged with a flowing reagent stream. The reagent stream contains sodium hydroxide together with a trace of cyanide and these optimise the solution composition for detection by a cyanide ion-selective electrode. A working range of 10-1 000 p.p.m. is covered with a sample rate of 120 h-1.Sample recoveries compared excellently with a titrimetric procedure and a short-term repeatability study on a sample gave a mean value of 306.5 k 7.4 p.p.m. at the 95% confidence level. The method selectivity with particular reference to metal - cyanide complexes is also discussed. The technique is applicable to process control as well as process development. Keywords Free cyanide determination; flow injection analysis; ion-selective electrode; leachate; minerals Leaching is a hydrometallurgical process in which some of the components of an ore are selectively dissolved. The suitability of such processes for the treatment of different ores is dependent on the relationship between the extent of dissolu-tion and the reagent consumption. Cyanide ion is a common leaching agent for the dissolution of silver and gold from ores and related materials and we were faced with a situation in which it would be necessary to analyse a large number of leachates containing from less than 10 to 1000 p.p.m.of free cyanide in order to study the cyanidation process initially on a laboratory scale and later on a production scale. The main requirements of any analytical method are precision specificity and speed and we decided that flow injection analysis (FIA) a simple flexible and rapid technique originally described by RfiiiEka and Hansenl and reviewed by Betteridge,2 provided the ideal solution. Our first task was to decide what form of detection we would employ and we consulted the literature for methods that would be suitable for conversion into an FIA system.Vogel3 described a titrimetric procedure with silver nitrate in the presence of ammonia but we discovered that many of the samples contained metal - cyano complexes with lower stability constants than the silver - cyanide complex and thus a true free cyanide concentration could not be obtained. Bark and Higson4 gave an excellent review on reported methods, which they categorised under two main headings namely colorimetric or titrimetric and since their review several other colorimetric procedures have been described.5-8 Colorimetric reactions have been extensively used in flow injection systems but a closer examination of the above procedures revealed that they are directed towards the determination of trace and ultra-trace amounts of cyanide and are therefore not applic-able to the levels of cyanide present in leachates.Colorimetric methods also involve the formation of a cyano complex and would be subject to the same problem encountered with the titrimetric procedure .3 The development of the cyanide ion-selective electrode has resulted in the publication of several procedures that either determine cyanide directly”l4 or use the electrode to indicate a titrimetric end-point. 1 0 ~ 3 ~ 5 Toth and Pungorlo reported that the electrode measures only cyanide ions and does not respond to hydrocyanic acid. They carried out a detailed study of metal - cyanide complexes and reported that those with a stability constant lower than that of the silver - dicyanide complex gave a response.The electrode was therefore ideally suited to our needs as we required to determine cyanide available to complex silver and this should include cyanide from complexes of lower stability. Ion-selective electrodes have been employed as selective detectors in flow injection systemsl6-18 and it appeared probable that the cyanide ion-selective electrode incorporated in such a system would satisfy our requirements. Experimental Reagents Cyanide stock standard solution. Dissolve 4.0 g of sodium hydroxide and 2.500 g of KCN in about 700 ml of de-ionised water and dilute to 1 1. The resulting solution contains 1000 p.p.m. of cyanide and can be standardised by silver nitrate titration.3 Standard cyanide solutions in the range 10-1 000 p.p.m. were prepared by dilution of this solution.Ionic strength adjuster. Dissolve 40 g of sodium hydroxide in about 800 ml of de-ionised water add 200 pl of cyanide stock solution and dilute to 1 1. Carrier solution. De-ionised water. Apparatus A flow diagram of the system is shown in Fig. 1. The reagents were pumped by a Watson Marlow 501s variable speed pump (10 rev min-l full speed) fitted with Tygon blue pump tubes and operated at 8 rev min-1 to give a measured flow stream of 3 ml min-1. Transmission tubing and mixing coils were constructed from 0.8 mm i.d. PTFE tubing using the Altex micro-plumbing system (supplied by Anachem Limited). The detector cell has three components. A cyanide ion-selective electrode (Orion Model 90-06) and a single-junction reference Sample carousel Ionic stre .’ a”’ -3 mi min-I xtion valve F 1L Mixing coil mgrn oj u sie r (o.9 m) Detectc-- .. ^ . . . - Pump Fig. 1. Flow injection manifol 422 ANALYST APRIL 1984 VOL. 109 Flow in ~ '/4-28 thread cd E E .-0 1 2 t /+ PTFE flow Fig. 2. 5min 3OC (a) I I- 2001 G Time Detector cell Table 1. Sample results from Fig. 3 Sample [CN-1 p.p.m. A . . . . . . 170 B . . . . . . 48 c . . . . . . 460 D . . . . . . 22 E . . . . . . 50 cel I Table 2. Method comparison [CN-I % m/V Orion cyanide electrode Flow out - Solution Titration FIA -1SE 1 . . . . 0.130 0.128 2 . . . . 0.128 0.128 3 . . . . 0.105 0.105 Orion reference electrode Table 3. Comparison of results obtained from two different leaching experiments [CN-I YO m/V 10 20 50 70 100 200 300 500 [CN-I p.p.m.Fig. 3. Typical detector response and corresponding calibration. (a) Reading from right to left the first 10 peaks are standard solutions. The cyanide concentration (in p.p.m.) is given by the number above each. (b) Calibration graph corresponding to standards described in (4 electrode (Orion Model 90-01) were fitted into our own design of PTFE flow-through holder (Fig. 2). Response of the detector cell was monitored on a chart recorder by way of an Orion Model 801 Ionalyzer millivolt meter. Sample injection was by a coupled autosampler injection valve assembly19 fitted with a 2 0 4 sample loop. Calibration graphs were prepared by processing standard solutions of sodium cyanide.The resulting peak height was plotted versus the logarithm of the cyanide concentration and a linear graph was obtained. A working range of 10-1 000 p.p.m. free cyanide was employed using two-decade graph paper and a typical calibration graph is shown in Fig. 3. The peaks labelled A-E in Fig. 3 represent leachate samples and the computed cyanide concentrations can be seen in Table 1. Results and Discussion Table 2 shows results for the determination of free cyanide in leachate solutions by both the flow injection procedure and the titrimetric method of Vogel.3 The .results show excellent Sample Feed . . . . Y2h . . . . l h . . . . 2 h . . . . 4 h . . . . 7 h . . . . 24h . . . . Wash . . Exp. A 0.130 0.115 0.120 0.120 0.114 0.116 0.114 0.011 Exp.B 0.129 0.024 0.022 0.015 0.008 0.008 0.008 0.004 agreement between the two independent methods. This was possible because unlike most of the samples to be analysed, the leachates chosen for this comparison did not contain cyanide complexes which would interfere with the titrimetric procedure. Table 3 shows results obtained for the analysis of samples from two leaching experiments and it can be seen that in both experiments the leaching process was effectively completed after 4 h. However the relative amounts of reagent consumed by the experiments differed markedly. The samples listed in Tables 1 and 3 contained metal - cyano complexes and could not be analysed by the titrimetric procedure. The short-term repeatability on a sample solution was as follows number of determinations (n) = 27; mean cyanide concentration = 306.5 p.p.m.; S,- (standard deviation) = 3.6 p.p.m.; and repeat-ability (95% confidence limit) = k7.4.Precision speed and specificity were the main features considered to be important for this method and flow injection analysis is ideally suited to such requirements. These systems can employ a wide variety of detectors and this combined with the facility for separation procedures such as solvent extraction can lead to a highly specific method. Another outstanding advantage of flow injection analysis is that system constants and timing sequences are maintained within far closer limits than is possible in manual processes which results in excellent repeatability for a process control procedure.Good recoveries are confirmed by the agreement with the titrimetric procedure (Table 2). The selectivity of the cyanide ion electrode with respect to its response to cyano species has already been discussed. There are however ions other than cyanide that can cause a response from the electrode and Table 4 lists the maximum allowable concentration of the more common interfering anions. Reference 20 states that sulphide must be absent from the analytical solution and as the majority of materials leached by cyanide are sulphide based this would seem a major problem. However during the cyanidation process oxygen is used for the dissolution of precious metals and under these conditions sulphide components are oxidised to thiosulphate, which does not interfere ANALYST APRIL 1984 VOL.109 423 Table 4. ion-selective electrode Common interferents and tolerable levels for the cyanide Maximum tolerable Maximum level in M CN-Interferent ratio (0.26 p.p.m.)/~ c1- . . . . . . 106 10 Br- . . . . . . 5 x 103 5 x 10-2 I- . . . . . . . . 0.1 1 x 10-6 S*- . . . . . . Mustbeabsent Must be absent I a) 1 min H 1 min c----l ( C) 1 min -4- Time Fig. 4. Effect on sample rate of the addition of trace amounts of cyanide to the reagent stream. ( a ) No cyanide; ( b ) 0.2 p.p.m.; and ( c ) 0.4 p.p.m. The conventional procedure for the determination of cyanide by an ion-selective electrode involves the immersion of the indicator and reference electrodes in a stirred pH buffered solution containing a known aliquot of sample.The experimental solution contains sodium hydroxide which maintains the pH above 10.5 ensuring that all uncomplexed cyanide is in the ionic form and provides a constant background ionic strength and therefore a constant cyanide activity coefficient allowing the e.m.f. to be equated to concentration. These conditions are also necessary in the flow injection system hence the ionic strength adjuster stream in the manifold assembly (Fig. 1). However initial experiments using only sodium hydroxide in this reagent revealed that it took 3 min to return to the base line after sample injection, i.e. an effective sample rate of 20 h-1. We suspected that the electrode was exhibiting a memory effect and a simple static test confirmed this.Memory effects occur on transferring from a high to a very low or zero concentration of analyte. We decided that we could reduce this effect by dosing the ionic strength adjuster solution with a trace amount of analyte ion. The results of these additions are illustrated in Fig. 4 and the addition of 0.4 p.p.m. of cyanide to the reagent reduced the base line return time to 30 s thus raising the sample rate from 20 to 120 h-1. The flow injection system has several other advantages over conventional procedures. For example the system is exten-sively automated very robust and reliable and as a result requires minimum operator skill and attention. Also from the time the sample is removed from the sample cup until it is taken to waste it is transported around a totally enclosed system.This greatly reduces the risk of sample contamination and virtually eliminates the possibility of the accidental release of HCN gas to the atmosphere. Conclusion The proposed method is rapid specific and precise and has been demonstrated to be suitable for the analysis of free cyanide in leachates over a working range of 10-1 000 p.p.m. The system can be applied to the analysis of samples from laboratory or pilot plant evaluations or as a process control method. The author thanks the British Petroleum Company plc for permission to publish this work. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. References Rfiiitka J. and Hansen E. H. Anal. Chim.Acta 1975 78, 145. Betteridge D. Anal. Chem. 1978 50 832A. Vogel A. I. “Quantitative Inorganic Analysis,” Third Edi-tion Longmans Harlow 1961 p. 271. Bark L. S. and Higson H. G. Analyst 1963 88 751. Nagashima S . Anal. Chim. Acta 1977 91 303. Nagashima S . Anal. Chim. Acta 1978 99 197. Sheng W. F . Yu-qin L. Fang Y . and Nai-Kui S. Talanta, 1981 28 694. Casapieri P. Scott R. and Simpson E. A. Anal. Chim. Acta 1970 49 188. Pungor E.; and Toth K. Analyst 1970,95 625. Tbth K. and Pungor E. Anal. Chim. Acta 1970,51,221. Sekerka I. and Lechner J. F . Anal. Chim. Acta 1977 93, 139. Lapatanick L. N. Anal. Chim. Acta 1974 72 430. Clysters H. Adams F. and Verbeek F. Anal. Chim. Acta, 1976 83,27. Gyorgy B . Andre R. and Pungor E. Anal. Chim. Acta, 1969 46 318. Conrad F . J. Talanta 1971 18 952. Hansen E. H. Ghose A. K. and RfiiiEka J. Analyst 1977, 102 705. Rfiiitka J. Hansen E. H. and Zagatto E. A. Anal. Chim. Acta 1977 88 1. Hansen E. H. RfiiiCka J. and Ghose A. K. Anal. Chim. Acta 1978 100 151. Lynch T. P. Taylor A. F. and Wilson J. N. Analyst 1983, 108 470. “Cyanide Ion Electrode Instruction Manual,” Orion Research, Cambridge MA 1977. Paper A3f250 Received August 8th 1983 Accepted November 7th 198

ISSN:0003-2654    

DOI:10.1039/AN9840900421

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

ANALYST. APRIL 1984 VOL. 109 425 Automatic Methods for the Determination of Total Inorganic Iodine and Free Iodide in Waters Richard E. Moxon Department of Industry Laboratory of the Government Chemist Cornwall House Stamford Street, London SE19NQ UK Methods for the determination of total inorganic iodine and free iodide based on the catalytic effect of iodide on the destruction of the thiocyanate ion by the nitrite ion have been developed and automated. These methods use a Technicon AutoAnalyzer system and a throughput of 20 samples per hour was achieved. Results obtained for a range of United Kingdom drinking waters showed that the total inorganic iodine method had a coefficient of variation of the order of 3% a detection limit of 0.2 pg I-' of I and a recovery of added iodine of 9&108%.The free iodide method had a coefficient of variation of the order of 1 O% a detection limit of 0.4 pg I-' of I and a recovery of added iodine of 8!3-109%. The effects of possible interfering substances on both methods have been investigated and shown to be negligiblefor normal drinking waters. The stability of dilute iodine solutions stored in containers made of different materials has also been evaluated. Keywords Automatic methods; total inorganic iodine determination; iodide determination; water analysis A project to develop and test methods for the determination of iodide species in fresh and potable waters was undertaken at the Laboratory of the Government Chemist for the Standing Committee of Analysts under a contract with the Department of the Environment.A survey of the literature revealed two methods with a detection limit of the order of 1 pg 1-1. These methods were those of DubravEiE,l which utilises the catalytic effect of iodide on the reduction of cerium(1V) sulphate by arsenious acid and of Sveikina,2 in which iodide catalyses the destruc-tion of the thiocyanate ion by the nitrite ion. Proskuryakova et al.3 have compared the two methods and concluded that the Sveikina method has marginally better sensitivity and preci-sion. Experimental The colorimetric finish of the Sveikina method in which iodide catalyses the destruction of the orange iron(II1) thiocyanate by the nitrite ion has been previously automated in this Laboratory by Moxon and Dixon4 using a Technicon AutoAnalyzer system.The final flow diagram used for this paper is a modification of this and is shown in Fig. 1. The reaction can be represented by 2CNS- + 3NO2- + 3N03- + 2H+ -+ 2CN- + 2S042- + 6 N 0 + H2O Using the range expansion setting of ~ 4 a suitable calibration graph for the range 0-5 pg 1-1 of I was obtained. When series of drinking waters were run the peaks were found to be irregular and erratic. It was found that there was a relationship between the hardness of the waters and the peak irregularity and this was removed when the alkaline earth metals were precipitated out with potassium carbonate. Subsequently samples and standards were made up in 0.3% potassium carbonate solution This addition of alkali pro-longed the stability of the standard solutions from 6 h to 8 d, but led to a decrease in the sensitivity of the method.However extra sensitivity was obtained by doubling the concentration of nitric acid in the ammonium iron(II1) sulphate reagent and halving the concentration of potassium thiocyanate reagent. Under these new conditions regular and precise peaks were obtained for both standard and sample solutions. When the effects of interfering ions were examined, it was found that chloride at a level of 400 mg 1-1 gave a 20% increase in the response to a 4 pg 1- I of I standard solution. To overcome this interference a large excess of chloride in the form of sodium chloride solution was introduced into the sample stream and it was found that as well as removing the interference effect of chloride the sensitivity of the method was greatly increased.This enhancement of the catalytic effect of iodide by chloride has also been noted by Dubraveit,' who also offered a theory as to its mode of action. Determination of Total Inorganic Iodine and Free Iodide A considerable proportion of the total inorganic iodine present in water can consist of iodate. System A described in Fig. 1 recovered iodate quantitatively and gave a measure of the total inorganic iodine in solution. In order to determine free iodide only the oxidation - reduction potential of the reaction mixture was adjusted so that iodate was not reduced to iodine or iodide. This was achieved by (a) reducing the concentration of nitric acid in the ammonium iron(II1) sulphate reagent (b) reducing the concentration of the sodium chloride reagent and (c) reducing the concentration of the potassium thiocyanate reagent.These changes caused a corresponding decrease in sensitivity. The final manifold system shown in Fig. 1 has two different sets of reagents. Set A was used for the determination of total inorganic iodine in water over the range 0.2-5.0 pg 1-1 of I and set B was used for the determination of free iodide in water over the range 0.4-5.0 pg 1-1 of I. A comparison of the concentrations of free iodide and total inorganic iodine in a range of drinking waters determined by the proposed methods is shown in Table 1. Table 1. Amounts of free iodide and total inorganic iodine in a range of United Kingdom drinking waters Sample source London (borehole) Harrogate .. . . Fife . . . . . . Bristol . . . . Nottingham . . Amesbury . . . . Oxford . . . . Gloucester . . Dunoon . . . . Nuneaton . . . . Total iodine/pg 1-1 Free iodide/pg 1 - I . . 28.0 28.0 . . 2.2 0.9 . . 7.8 7.8 . . 4.1 2.1 . . 14 3.5 . . 4.3 0.9 . . 4.3 1.3 . . 3.2 3.2 . . 1.1 0.9 . . 2.8 1.6 Crown Copyright 426 ANALYST APRIL 1984 VOL. 109 System A Flow-rateimi mi n - 1 3.4 2.0 0.6 0.3% KzC03 Waste 4 -5- Air 15 turn 1.2 * Sampler v (20 per hour 1.0 1 :2) A W 0.32 n U 0.01 15% KCNS 15 turn W 7.7% NH4Fe(S04)2.12H20 in 33.4% HN03 0.8 15 turn U 2.07% NaN02 0 . 3 2 n 2.5 15 turn Wasted I I - ' Spectrophotometer Delav coil 450 nm I ~ E x p ~ ~ i o n l Range expander Recorder System B Flow-rateiml min-1 3.4 A v 0.3% K2C03 2.0 Waste 4 0.005 8% KCNS 7.7% NH4Fe(S04)2.12H20 in 7.5% HN03 c c ~ E x p ~ ~ ~ o n l Range expander Recorder Fig.1. Flow diagrams for the determination of total inorganic iodine in waters (system A) and free iodide in waters (system B) Stability of Standard and Sample Solutions When trials were run on a series of test waters it was found that the total inorganic iodine values increased over a period of days especially when drinking water straight from the tap was used. This phenomenon was investigated by leaving standard blank and drinking water solutions made up in 0.3% potassium carbonate in containers made of different materials and analysing them over different periods of time. All the containers were previously thoroughly washed with concen-trated nitric acid and rinsed copiously with distilled water.Drinking water samples were centrifuged for 5 min at 50 Hz after addition of potassium carbonate to remove any precipi-tate and organic impurities. The results are summarised in Table 2. The results show a slight increase in the iodine concentra-tion of solutions stored in glass and polythene containers. The cause of this increase is unlikely to be contamination as all containers were washed with concentrated nitric acid and rinsed copiously with distilled water before use. In all solutions except that made up with laboratory drinking water taken straight from the tap no significant increase occurred when polystyrene bottles were used for storage. There may b ANALYST APRIL 1984 VOL.109 427 ~~ ~~ Table 2. Changes in the iodine concentration of standards blanks and drinking water solutions made up in 0.3% potassium carbonate in different containers over a period of days Iodine concentration/yg 1-Test solution 4 pg 1 - 1 solution . . . . . . . . Blank solution . . . . . . . . . . Laboratory drinking water straight from tap . . . . . . . . Cambridgedrinkingwater . . . . . . Hertforddrinkingwater . . . . . . Oxforddrinkingwater . . . . . . Container Glass calibrated flask Polyethylene bottle Polystyrene bottle Glass calibrated flask Polyethylene bottle Polystyrene bottle Glass calibrated flask Polyethylene bottle Polystyrene bottle Glass calibrated flask Polystyrene bottle Glass calibrated flask Polystyrene bottle Glass calibrated flask Polystyrene bottle After l h 3.9 3.9 3.9 0.0 0.0 0.0 4.7 4.5 4.5 5.4 5.4 4.4 4.5 5.6 5.7 After I d 4.0 3.9 3.9 0.2 0.1 0.0 5.5 5.5 5.4 -----_.After 3 d 4.0 4.0 3.9 0.4 0.2 0.0 5.7 5.8 5.7 5.8 5.5 5 .0 4.6 6.0 5.9 After 8d 4.2 4.0 3.9 0.5 0.3 0.0 5.8 5.6 5.7 6.1 5.4 5.0 4.5 6.0 5.8 surface effects occurring in glass and polyethylene containers but there is no direct evidence for this. The larger increase observed in laboratory tap water was thought to be due to the influence of dissolved gases such as chlorine carbon dioxide or oxygen. A series of experiments were run to test the effects of removing dissolved gases by boiling or purging with nitrogen and also monitoring pH changes over a period of time but the results were inconclusive.After 3 d the iodine level of the laboratory tap water remained constant. The results show that it is necessary to adopt a standardised sampling procedure. In this laboratory all water samples were at least 3 d old on receipt and these were made up in 0.3% potassium carbonate solution in polypropylene cali-brated flasks centrifuged and then stored in polystyrene bottles (30-ml universal containers obtained from Sterilin Ltd. Teddington Middlesex were found to be suitable). Fresh tap water was treated in the same way but allowed to stand for 3 d before analysis. Methods For the Determination of Total Inorganic Iodine and Free Iodide in Water All chemicals used should be of analytical-reagent grade and glass-distilled water should be used in preference to de-ionised water.Reagents for the determination of total inorganic iodine (system 4 Standard iodide solution 4 g 1-1 of I . Dissolve 0.523 2 g of potassium iodide previously dried in an oven at 105 "C for 2 h in distilled water and dilute to 100 ml in a calibrated flask. Standard iodide solution 40 mg 1-1 of I . Dilute 10 ml of the standard iodide solution (4 g 1-1 of I) to 1000 ml with distilled water in a calibrated flask (stable for 1 month). Standard iodide solution 200 pg 1-1 of I. Dilute 5 ml of the standard iodide solution (40 mg 1-1 of I) to 1000 ml with distilled water in a calibrated flask.Store in a polythene or polystyrene bottle (stable for 1 month). Working solutions. Into 200-ml calibrated flasks pipette 5 , 4 3 2 1 and 0 ml of standard iodide solution (200 pg 1-1 of I). Add 2 ml of 30% mlvpotassium carbonate solution and dilute to 200 ml with distilled water. These are the working standards. Store in polystyrene bottles and prepare freshly every 2 weeks. Potassium carbonate solution 30% mlV. Dissolve 300 g of potassium carbonate in distilled water and make up to 1 1. Potassium thiocyanate solution 0.01 1 5% ml V . Dissolve 0.115 g of potassium thiocyanate in distilled water and make Sodium nitrite solution. Dissolve 4.14 g of sodium nitrite in distilled water and dilute to 200 ml (stable for 1 d only). Sodium chloride solution 6% mlV.Dissolve 60 g of sodium chloride in distilled water and dilute to 1 1. Ammonium iron(III) sulphate reagent. Dissolve 77 g of ammonium iron(II1) sulphate [NH4Fe(S04). 12H20] in approximately 300 ml of distilled water. Add 334 ml of concentrated nitric acid (sp. gr. 1.42) and make up to 1 1. Heat on a hot-plate until all traces of solid dissolve. Reagents for the determination of free iodide (system B) Standard iodide solutions. These are exactly the same as those described in system A. Potassium carbonate solution 30% mlV. Dissolve 300 g of potassium carbonate in water and dilute to 1 1. Sodium nitrite solution 2.07% mlV. Dissolve 4.14 g of sodium nitrite in distilled water and dilute to 200 ml (stable for 1 d only). Sodium chloride solution 2.5% mlV.Dissolve 25 g of sodium chloride in distilled water and dilute to 1 1. Potassium thiocyanate solution 0.005 8% mlV. Dissolve 0.058 g of potassium thiocyanate in distilled water and dilute to 1 1. Ammonium iron(III) sulphate reagent. Dissolve 77 g of ammonium iron(II1) sulphate [NH4Fe(S04). 12H201 in approximately 400 ml of distilled water. Add 75 ml of concentrated nitric acid (sp. gr. 1.42) and make up to 1 1. Warm until all traces of solid dissolve. Sodium oxalate reagent. Dissolve 5 g of sodium oxalate in 100 ml of 5% VIV sulphuric acid (this reagent is toxic). Apparatus A centrifuge with a speed of 50 Hz and glass or polypropylene centrifuge tubes of 150-ml capacity were used. Polystyrene bottles of 30-50-ml capacity were employed. An AutoAnalyzer system for colorimetric analysis was utilised.The results shown in this paper were obtained using a Technicon AutoAnalyzer 1 system with a range expansion facility that was'operated and maintained in accordance with the instructions given in the Operator Instruction M a n ~ a l . ~ up to 11. Procedure Wash all glassware and polystyrene containers with concen-trated nitric acid and rinse copiously with distilled wate 428 ANALYST APRIL 1984 VOL. 109 Table 3. Effect of added ions on the determination of (A) total inorganic iodine concentration and (B) free iodide concentration in a 4 pg 1-1 standard iodide solution Element added Zn2+ . . . . . . . . . . . . cu2+ . . . . . . . . . . . . Li2+ . . . . . . . . . . . . Pb2+ . . . . . . . . . . . . Fe3+ .. . . . . . . . . . . Mn3 + . . . . . . . . . . . . Ni2+ . . . . . . . . . . . . Hg2+ . . . . . . . . . . . . co2+ . . . . . . . . . . . . Mg2+ . . . . . . . . . . . . CI- . . . . . . . . . . . . . . B r - . . . . . . . . . . . . . . F- . . . . . . . . . . . . . . 1 0 3 - . . . . . . . . . . . . sod*- . . . . . . . . . . . . Humic acid . . . . . . . . . . Iodoform . . . . . . . . . . . . Methyl iodide . . . . . . . . . . Concentration/ mgl-1 1 1 1 0.1 2 1 0.2 0.005 0.2 30 400 1 1 0.004 200 20 1 0.1 Maximum concentration in drinking waters/ mgl-1 0.69 0.36 0.01 0.046 1.5 0.06 0.013 0.001 0.011 23 245 -------( A ) Total I found/ 4.0 4.0 4.1 4.0 4.0 3.9 4.0 3.9 4.0 4.1 4.0 4.0 4.1 8.0 4.1 3.9 5.6 4.2 I-' ( B ) P8 1-' Free I found/ 3.9 3.8 3.8 3.9 3.8 3.9 4.2 4.2 4.2 4.0 4.0 4.1 4.3 4.0 4.5 0.0 5.6 4.2 Table 4.Variations in peak heights of standard solutions and total inorganic iodine concentrations of samples run on each of five consecutive days Standard solution peak heights (% full-scale deflection)-Concentration/ Mean peak Standard Coefficient of 5 62.9 0.4 0.6 4 52.2 0.6 1.1 3 41.6 0.9 2.2 2 28.1 1.2 4.2 1 15.3 0.6 3.7 pg1-l height deviation variation % Total inorganic iodine concentration of drinking water-Mean total I concentration/ Standard Coefficient of Sample source yg 1-1 deviatiodyg 1-1 variation Oh London (borehole) . . 28.0 0.76 2.7 Harrogate .. 2.2 0.06 2.7 Fife . . . . 7.8 0.54 6.9 Bristol . . . . 4.1 0.10 2.4 Nottingham . . 14.0 0.54 3.9 Amesbury . . 4.3 0.10 2.3 Oxford . . . . 4.3 0.10 2.3 Gloucester . . 3.2 0.08 2.5 Dunoon . . . . 1.1 0.07 6.4 Nuneaton . . 2.8 0.04 1.4 Table 5. Variations in peak heights of standard solutions and free iodide concentrations of samples run on each of five consecutive days Standard solution peak heights (% full-scale deflection)-Concentration/ Mean peak Standard Coefficient of 5 20.5 0.3 1.5 4 16.7 0.5 3.0 3 12.6 0.5 6.0 2 8.4 0.5 6.0 1 4.5 0.3 6.4 height deviation variation '/O Pg I-' Free iodide concentration of drinking waters-Mean free I concentration/ Standard Coefficient of Sample source pg 1-1 deviatiodpg 1-1 variation % London (borehole) .. Harrogate . . Fife . . . . Bristol . . . . Nottingham . . Amesbury . . Oxford . . . . Gloucester . . Dunoon . . . . Nuneaton . . 28.0 0.9 7.8 2.1 3.5 0.9 1.3 3.2 0.9 1.6 1.1 0.17 0.86 0.05 0.49 0.17 0.22 0.7 0.0 0.14 3.9 19.9 11.0 2.4 14.0 19.0 17.0 5.3 0.0 8.7 before use. Dispense 1.0 ml of 30% mlVpotassium carbonate solution into a 100-ml calibrated flask. Make up to 100 ml with a water sample and shake well. Centrifuge the resulting solution for 5 min at 50 Hz. Decant about half of the solution into polystyrene bottles. Allow fresh tap water to stabilise for 3 d before analysis. Set up the manifold system shown in the flow diagram (Fig. 1) and use the appropriate set of reagents for either total inorganic iodine determination (A) or free iodide determina-tion (B).Load the sample tray with a set of working standards followed by 20 samples interspersed with a working standard every fifth sample. Complete the series with another set of working standards and run at a rate of 20 per hour. If deposits of iron(II1) thiocyanate occur these may be removed by running two sample cups of sodium oxalate reagent through the system at the end of a run. Calculation Plot a calibration graph of the mean standard peak heights against their respective iodine concentrations. The iodine concentration of a sample is obtained by comparing its peak height with the calibration graph. Multiply the result by 1.01 to compensate for the addition of potassium carbonate.Results Effects of Interferences The effects of ions commonly occurring in drinking waters that could cause possible interference were tested and the results, representing a mean of three determinations are shown in Table 3. The maximum values for the concentrations of elements in drinking waters are those reported by Zoetman and Brinkmann. ANALYST APRIL 1984 VOL. 109 429 Table 6. Recovery of added iodide from water samples using the total inorganic iodine method Original iodine concentration/ Iodide added/ Calculated Sample source M l - ' CLg I-' total/yg 1-l London(borehole)* . . . . . . Harrogate . . . . . . . . . . Bristol . . . . . . . . . . Bristol* . . . . . . . . . . Nottingham* . . . . . . . . Amesbury * .. . . . . . . . . Oxford* . . . . . . . . . . Gloucester . . . . . . . . . . Nuneaton . . . . . . . . . . * These samples were diluted. 2.8 2.2 2.05 2.05 2.8 2.15 2.15 1.6 1.4 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 4.8 4.2 4.05 4.05 4.8 4.15 4.15 3.6 3.4 Iodine found/ 4.8 4.0 4.0 4.0 4.6 4.2 4.5 3.8 3.5 1.1gI-l Recovery Yo 100 90 99 99 96 101 108 106 103 Table 7. Recovery of added iodide from water samples using the free iodide method Original iodide concentration/ Iodide added/ Calculated Sample source I % - ' I % - ' total/pg 1-1 Slough . . . . . . . . . . 1.3 2.0 3.3 Catterick . . . . . . . . . . 0.7 2.0 2.7 Hexham . . . . . . . . . . 1.2 2.0 3.2 Royston . .. . . . . . . . 0.6 2.0 2.6 Braintree . . . . . . . . . . 1.2 2.0 3.2 Benson . . . . . . . . . . 1 .o 2.0 3.0 Iodide found/ I%-' Recovery YO 3.6 109 2.4 89 3.2 100 2.7 104 3.5 109 3.0 100 Table 8. Comparison of the proposed method for total inorganic iodine with that of Keller et a1.7 for a range of waters Iodine concentration/pg 1-~~ ~ Keller's et al. Sample source Proposed method method Slough . . . . . . . . 11.5 12.7 Oakington . . . . . . 5.5 6.0 Catterick . . . . . . 0.7 0.7 Hexham . . . . . . 2.3 2.9 Harrogate . . . . . . 2.2 1.7 Braintree . . . . . . 13.7 13.9 Benson . . . . . . . . 5.8 6.5 London(boreho1e) . . . . 28.0 30.0 Fife . . . . . . . . 7.8 5.8 Nottingham . . . . . . 14.0 15.4 Dunoon . . . . . . . . 1.1 1.1 Royston .. . . . . . . 4.5 5.2 Londonttap) . . . . . . 5.8 5.9 The results show that interferences will not present any major problems in the analysis of drinking waters for either total inorganic iodine or free iodide. The greater variability of the values for free iodide is due to the lower sensitivity and precision of the method. Precision of the Methods To obtain a measure of the precision of the methods a number of drinking waters from different areas in the United Kingdom were analysed on each of five consecutive days. The results are summarised in Table 4 for total inorganic iodine and Table 5 for free iodide. Variations in the peak heights of working standard solutions and individual iodine results are expressed by the standard deviation from the mean.This shows that the general precision of the total inorganic iodine method is of the order of 3% and that of the free iodide method is of the order of 10%. Accuracy of the Methods No certified water sample against which to test the accuracy of the methods could be found and no suitable referee method was available. An indication of the accuracy of the method was therefore obtained by adding known amounts of iodide to water samples that had been previously analysed for total inorganic iodine and free iodide. The results are summarised in Tables 6 and 7 where the results shown for spiked samples represent a mean of three separate determinations. As a further check the results obtained by the total inorganic iodine method were compared with those obtained using the method of Keller et a1.,7 which is based on the method of DubravWl [cerium(IV) sulphate - arsenious acid], modified for an AutoAnalyzer system.The results are sumarised in Table 8 where the values given are the means of three determinations. The mean recovery of iodide by the total inorganic iodine method and the free iodide method are 100 k 5.4% and 102 ? 7.570 respectively. The results show reasonable agreement with those of Keller et aL7 The effects of interferences on the method of Keller et al. were not determined and could explain some of the minor differences. Limit of Detection The limit of detection was taken to be the concentration at which the signal was three times greater than the noise of the base line. This gave a limit of detection of 0.2 pg 1-1 of I for the total inorganic iodine method and 0.4 pg 1-l of I for the free iodide method. Conclusion The automatic methods for the determination of total inorganic iodine and free iodide were shown to be applicable to normal drinking waters. The same manifold system was common to both methods and the sensitivity precision and accuracy were shown to be adequate. The effects of possible interfering ions were shown to be negligible for norma 430 ANALYST APRIL 1984 VOL. 109 drinking waters and once the system has been set up it is simple to operate and has a throughput of 20 samples per hour. This work was financed by the Department of the Environ-ment and published with their permission and that of the Government Chemist. References 1. 2. 3. Dubraveit M. Analyst 1955 80 295. Sveikina R. V. Gig. Sunit. 1975 1 80. Proskuryakova G. F. Sveikina R. V. andchernavina M. S . , Khim. Khim. Tekhnol. 1963 6 729. 4. 5 . Moxon R. D. and Dixon E. J. Analyst 1980 105 344. “Technicon AutoAnalyzer 1 Assembly and Operating Instruc-tion Manual,” Technicon Chromatography Corp. New York, 1962. Zoetman B. C. and Brinkmann F. J . “Hardness of Drinking Water and Public Health Proceedings of the European Scientific Colloquium Luxembourg May 1975,” Pergamon Press Oxford 1976 pp. 173-211. Keller H. E. Doenecke K. Weidler K. and Leppla W., Ann. N. Y. Acad. Sci. 1973 220 1. 6. 7. Paper A31306 Received September 6th I983 Accepted October 5th 198

ISSN:0003-2654    

DOI:10.1039/AN9840900425

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

ANALYST APRIL 1984 VOL. 109 43 1 Accuracy of Determination of the Electrical Conductivity and the pH Value of River Waters Results of Water Authority Tests Made for the Harmonised Monitoring Scheme of the Department of the Environment Analytical Quality Control (Harmonised Monitoring) Committee* Water Research Centre P.O. Box 16 Henley Road Medmenham Marlow Buckinghamshire SL7 ZHD, UK The Department of the Environment in collaboration with the Regional Water Authorities has initiated a Scheme for the Harmonised Monitoring of Inland Fresh Waters in England and Wales. The Scottish Development Department has been closely associated with the development of this Scheme and has introduced a similar scheme in Scotland in collaboration with the River Purification Boards. To achieve the required comparability of results each of the 11 participating laboratories (drawn from 10 Regional Water Authorities and 1 River Purification Board) takes part in an analytical quality control (AQC) programme; this work is coordinated by the Water Research Centre.The general approach adopted to AQC has already been described and this paper presents the results of tests made on the determination of pH value and electrical conductivity in river waters. The accuracy requirements for both determinands (that the total error on a single result should be not larger than 0.2 unit for pH and not larger than 20% of sample conductivity or 5 pS cm-1, whichever is the greater for conductivity) were essentially achieved by all 1 1 participating laboratories. Inter-laboratory tests on the determination of pH were however restricted to the distribution of buffer solutions because river samples examined were not sufficiently stable.Keywords River water analysis; electrical conductivity determination; pH determination; accuracy of results; analytical quality control The Scheme for the Harmonised Monitoring of the Quality of Inland Fresh Water has been described in detail.' It is intended to provide objective data on river water quality so that accurate assessments can be made of long-term trends in the qualities of rivers and of the amount of materials discharged by them to the sea. The Scheme complements monitoring carried out for regional or local purposes and one of its essential aims is to achieve comparability of the results from all participating laboratories.To that end special investigations have been made to establish suitable sampling locations and to define the necessary sampling frequency. Sampling procedures have been recommended and each participating laboratory carries a specially designed pro-gramme of tests to ensure that its analytical results are of adequate accuracy for the Scheme. The Water Research Centre (WRC) is under contract to the Department of the Environment to advise on and to coordi-nate this analytical quality control (AQC) programme. The need for and details of an approach to a planned AQC system for this and similar monitoring schemes have been discussed elsewhere.* In view of the growing interest in achieving comparable results from a number of laboratories it was thought useful to describe the AQC work for the Harmonised Monitoring Scheme and to present the results for different determinands.This paper considers the determi-nation of pH and electrical conductivity; earlier papers describe the work for chloride,3 ammoniacal n i t r ~ g e n ~ total oxidised nitrogen and nitrites and suspended solids6 and subsequent papers will deal with other determinands of importance in rivers. Organisation of the Work A Committee was formed to plan the collaborative work and has representatives? from the Department of the Environ-ment (DOE) the Scottish Development Department each * Correspondence should be addressed to M. J . Gardner at the -i. The names of the representatives at the time the work reported Water Research Centre.here was carried out are given in the Appendix. Regional Water Authority (RWA) the Scottish River Purifi-cation Boards and the WRC. This Committee decided to adopt the approach to AQC described elsewhere,2 each determinand being studied in two phases. Phase (i). One laboratory in each of the 10 RWAs and one in Scotland participated the WRC acting as coordinating laboratory.2 Phase (ii). After satisfactory results have been obtained in phase (i) those laboratories act as coordinators of tests within RWAs and in Scotland. Certain RWAs are not involved in this stage because all analyses for Harmonised Monitoring are made by one laboratory. This paper deals only with phase (i). Required Analytical Accuracy The following requirements for analytical accuracy were agreed for the determination of electrical conductivity: maximum tolerable bias 10% of the electrical conductivity of the sample or 2.5 pS cm-1 whichever is the greater; maximum tolerable total standard deviation 5% of the electrical conductivity of the sample or 1.25 pS cm-1 whichever is the greater.For pH determination the following requirements were established maximum possible bias 0.1 pH unit; maximum possible standard deviation 0.05 pH unit. Analytical Quality Control The approach followed was exactly as presented previously2; no attempt is made here therefore to explain the reasons underlying the various activities described below. The partici-pating laboratories were as follows Anglian WA Regional Standards Laboratory Cambridge; Northumbrian WA How-don Laboratory Wallsend; North West WA Rivers Division Laboratory Warrington; Severn-Trent WA Regional Labora-tory Finham; Southern WA Resource Planning Laboratory, Otterbourne; South West WA Rivers and Marine Labora-tory Exeter; Thames WA Thames Conservatory Division Laboratory Reading; Welsh WA Chester Area Laboratory, Chester; Wessex WA Bristol Avon Division Laboratory 432 ANALYST APRIL 1984 VOL.109 Saltford; Yorkshire WA Headquarters Laboratory Leeds; Forth RPB Headquarters Laboratory Edinburgh. The sequence of participating laboratories in the above list does not relate to the order of numbering of laboratories in the tables. Analytical Method All laboratories employed the methods recommended by the DOE/NWC Standing Committee of analyst^.^ In one labora-tory (laboratory l l ) calibration buffers and samples were diluted (10 ml of sample/buffer + 1 ml of water) prior to pH determination as part of a procedure for the prevention of cross-contamination in an automated sample dispensing system.This practice was noted only after the laboratory had completed the tests described here. The results of this laboratory in these studies involving tests of precision on buffer solutions and river samples and tests of inter-laboratory bias using buffer solutions only were of adequate accuracy but the laboratory was informed that the dilution of river samples for pH measurement was likely to introduce unacceptable bias to analytical results for such samples. Effect of Temperature on the Measurement of pH and Conductivity For both determinands considered in this paper the tempera-ture of the sample at the time of measurement will affect any analytical result obtained.This effect is one in which the true pH value or conductivity of the sample rather than merely the error associated with measurement is dependent on sample temperature. Thus variations in the temperature of measure-ment will unless eliminated or corrected for contribute to between-laboratory bias. Such factors will adversely affect laboratories' comparability apart from any analytical errors that may be present. The approach to this question of sample temperature was slightly different for the two determinands. For conductivity measurements it was not possible, because of the practices in use in the various laboratories to agree that all samples should be measured at a single temperature.All but two laboratories (which measured at 25 "C) performed measurements at 20 "C. There was also the possibility that if certain samples proved inadequately stable, on-site measurement at ambient temperature would have to be considered. The likely bias introduced by the correction of measured conductivity to conductivity at some standard temperature was therefore examined. The temperature dependence of conductivity has been investigated by Wagner8 for various types of potable water and by Smith9 for samples from lakes and rivers. Values of the temperature correction factor (the fractional change in conductivity per degree) appropriate for correction of conductivity to 20 "C from 25 "C range from 0.019 to 0.026.The use of a correction value from Oile end of the range ( i e . 0.019) when the true value for the water in question was from the other end of the scale of values (Le. 0.026) would result in a bias in reported results of approximately 4%. This unlikely event was considered in relation to the target for maximum possible bias of 10% and it was decided that correction is not likely to introduce impor-tant between-laboratory bias. It is important to note, however that should the true correction factor of the water concerned lie outside the range given above or that correction is made over a range greater than 5 "C the correction of conductivity data for temperature might introduce unaccept-able bias.In pH measurement the option of correcting results for temperature of measurement is not available because the pH value of a sample does not behave in as predictable a manner as the conductivity. Here one laboratory (laboratory 10) performed determinations at 25°C and the remaining ten at 20 "C. It was agreed that the temperature of samples for pH determination would be controlled to within +1 "C. (For conductivity determinations it appeared that such stringent control of temperature is not essential; however knowledge of the sample temperature to within +l "C is necessary if an unacceptable bias is not to be introduced when temperature correction is applied.) Allowance has been made in the data archive for pH measurements to be made at ambient temperature should it prove necessary to make on-site determinations.Within-laboratory Precision Tests Each laboratory then carried out the same programme of tests to assess the precision of its analytical results. On each of ten days each laboratory made duplicate determinations in random order on appropriate standard solutions and portions of two river waters having respectively a determinand value near the lowest and highest routinely reported for the Harmonised Monitoring Scheme. For conductivity the standards were two solutions of potassium chloride of 0.1 and 0.9 of the upper concentration range of the laboratory's method. Duplicate conductivity blank determinations were also performed in each batch of analyses. The results for the standards were blank corrected to allow for possible between-batch variation in the conductivity of the water used to prepare the standard.Results for river samples were not blank corrected. For pH the standards consisted of three buffer solutions having nominal pH values of 4.0,7.0 and 9.0. In an attempt to ensure constant composition for the buffer solutions fresh samples were prepared for each batch of analyses. This eliminated the possibility of batch to batch variability caused by instability in the buffers. Potential differences between freshly prepared portions of the same buffer solution were minimised by their being made up from sachets of buffer powder selected at random from the same manufacturer's batch. Sachets of such buffer powder were distributed by the WRC to all laboratories.This procedure gave a prelimi-nary check on between-laboratory bias provided that por-tions of the distributed buffer powder were assumed to be identical. It was recognised that problems of drifting sample pH, caused by interaction with atmospheric carbon dioxide might be encountered if successive aliquots of a river sample were removed from the same container for analysis in any given batch. To overcome this potential problem a bulk sample of river water was collected and split into 20 separate containers of the type used routinely for such samples. Each determination of river water pH was then made on a freshly opened container every effort being made to ensure that the 20 sub-samples were identical. This approach was adopted because it was considered more likely that a sample could be split homogeneously than that a single large sample would remain unchanged once opened during a batch of analyses.On completion of the tests each laboratory analysed its results to obtain estimates of within-batch (s,) between-batch (sb) and total (st) standard deviations,Z where s = (s,2 + sb2)i. The values of s were compared with the appropriate target value using an F-test and were accepted as satisfactory provided st was not significantly greater (p = 0.05) than the appropriate target. Such a treatment of the results of tests on river waters could, if the sample were unstable from one batch to another give a falsely inflated estimate of between-batch variability and hence of total standard deviation.In the conductivity determi-nations as none of the estimates of laboratories' standard deviations was greater than the appropriate target value it can be concluded that such effects if they occur are not of importance. For pH determination where these effects are likely to be of greater magnitude it was agreed that for rea ~ ~~ ~ _ _ _ _ _ _ _ _ _ _ _ _ ~ ________~ ~ Table 1. Determination of electrical conductivity-results of precision tests. Target total standard deviation 5% or 1.25 pS cm-1 whichever is the Standard solution 1 Standard solution 2 Low-conductivity river water Relative Mean Standard total conduc- deviation/ standard Labora- tivity/ pS cm- devia-tory $3 tion, No. cm-1 sw sb st Yo 1 657.2 0.6 2.8 2.9 0.4 2 64.1 0.4 0.7 0.8 1.5 3 201.0 0.7 NS* 0.9 0.4 4 727.4 8.0 NS 9.1 1.2 5 200.8 0.4 1.0 1.1 0.5 6 176.7 2.1 4.1 4.6 2.6 7 101.5 3.5 4.1 5.4 5.37 8 102.5 3.3 NS 3.6 3.5 9 143.2 1.5 4.7 4.9 3.4 10 199.2 0.3 0.8 0.8 0.4 11 49.1 0.1 NS 0.2 0.4 * NS not significant.t Not significantly greater than the target value. Mean conduc-t ivi t y/ cm-1 6 113.8 620.5 1701.1 6 614.0 1801.9 798.9 906.2 881.7 1385.8 1766.0 489.8 CIS Relative Standard total standard deviation/ pScm-1 devia-tion, SW Sb st % 9.0 23.8 25.4 0.4 2.2 5.8 6.2 1.0 3.9 8.3 9.1 0.5 13.0 14.6 19.6 0.3 1.1 1.5 1.9 0.1 2.4 9.5 9.8 1.2 6.2 11.2 12.8 1.4 5.2 25.2 25.8 2.9 14.0 13.5 19.4 1.4 3.9 9.6 10.4 0.6 0.3 0.7 0.8 0.2 Mean conduc-tivity/ PS cm-1 513.4 42.5 218.8 750.1 824.6 164.5 665.5 132.3 674.1 256.0 52.0 Relative Standard total deviation/ standard pS cm- devia-tion, SW Sb st O/ 0.7 2.0 2.2 0.4 0.4 0.9 1.0 2.3 0.5 1.1 1.3 0.6 8.0 NS 9.5 1.3 0.7 1.4 1.6 0.2 4.5 6.1 7.6 4.6 5.0 6.1 7.9 1.2 1.6 3.8 4.1 3.1 5.3 11.4 12.6 1.9 1.7 NS 2.0 0.8 1.6 NS 1.7 3.3 Table 2.Determination of pH-results of precision tests. Target total standard deviation 0.05 pH unit. Where the estimate of total standard deviation with respect to pH exceeds the precision target a revised estimate in which the possible effect of sample instability between batches is not included, parentheses beneath the original estimate Buffer solutions PH4 Standard deviation pH units Laboratory Mean No.PH s w sb st 1 4.016 0.0055 0.0097 0.0110 2 4.010 0.0039 0.0067 0.0077 PH7 Standard deviation pH units Mean 6.995 0.0097 0.0356 0.0370 6.952 0.0070 0.0115 0.0135 PH s w sb st 3 4.015 0.0000 0.0412 0.0412 7.020 0.0224 NSJ- 0.0253 4 4.114 0.0267 0.0329 0.0424 7.046 0.020 1 0.03713 0.0428 5 3.999 0.0018 0.0089 0.009 1 7.015 0.009 1 0.001 2 0.0091 6 4.029 0.0032 0.0227 0.0226 6.978 0.0039 0.0432 0.0434 7 3.984 0.0217 0.0073 0.0229 6.972 0.0109 0.0226 0.0251 8 4.100 0.0230 NS 0.0278 6.930 0.0232 0.025 1 0.0342 9 3.983 0.0077 0.0204 0.021 8 6.901 0.0102 0.048 1 0.0492 10 3.882 0.0055 0.0251 0.0257 6.991 0.0050 0.0197 0.0203 11 4.056 0.050 0.0257 0.0262 6.983 0.0129 0.0208 0.0245 * Exceeds target. t NS not significant (F-test 0.05 probability level).PH 9 Low Mean 9,205 9.215 9.185 9.280 9.270 9.117 9.224 9.173 9.122 9.282 9.153 PH Standard deviation pH units S W 0.005 0 0.005 9 0.015 8 0.021 6 0.001 8 0.007 1 0.016 0 0.029 2 0.014 3 0.008 4 0.009 6 sb 0.006 7 0.013 0 0.017 9 0.041 4 0.016 3 0.022 2 0.019 5 NS 0.022 2 0.015 3 0.017 5 st 0.008 4 0.014 2 0.023 9 0.046 7 0.016 4 0.023 3 0.024 9 0.029 2 0.026 4 0.017 4 0.019 9 Standard deviation, Mean 7.465 0.0265 9.0306 6.792 0.0844 0.118 8.122 0.0193 0.0625 7.216 0.0332 0.0450 7.511 0.0189 0.0244 7.870 0.479 8.102 0.0217 0.0137 6.933 0.0102 7.600 0.0136 0.128 7.487 0.0335 0.035 6.573 0.0446 0.1661 pH s 434 ANALYST APRIL 1984 VOL. 109 samples where estimates of standard deviation that exceeded the target were obtained another estimate of total standard deviation would be calculated using a value sbp for the pooled between-batch standard deviation arrived at from the buffer solutions: sb2(pH 4) -k sb2(pH 7) + sb2(pH 9) 4 sbp = [ 3 1 The revised estimate of total standard deviation was obtained from If this estimate was not significantly greater than the target (F-test at thep = 0.05 level) then the precision was regarded as satisfactory.The results of the precision tests are summarised in Tables 1 and 2. The results of the tests on conductivity determination indicate that the precision is satisfactory. Those for pH show two instances of failure to meet the target (laboratories 2 and 6 for the low pH river water).An examination of the analytical profiles of these samples revealed that they were of low alkalinity and conductivity (12 and 15 mg 1-1 of CaC03 and 51 and 160 pS cm-1 respectively). Such samples are likely to present problems in the determination of pH owing to possible changes in true pH during measurement (by inter-action with atmospheric carbon dioxide) and difficulties involving the liquid junction potential of the reference electrode. It was decided to proceed with the programme of tests and to return to the problem of pH measurement in water of low ionic strength/poor buffer capacity at a later stage. Each laboratory then set up a statistical quality control chart2 based on the analysis of a standard solution in each batch of analyses.These charts are intended to aid the continuing long-term assessment of accuracy and are not considered further here. st2 = sw2 (sample) -!- sbp2 Tests for Between-laboratory Bias To complete this initial phase of AQC direct checks of between-laboratory bias were made as follows. For conductiv-ity determination the WRC distributed portions of one standard solution and two river water samples. For the determination of pH portions of two buffer solutions were circulated. Single determinations were made on each solution on each of five days and the results obtained are summarised in Tables 3 and 4 the 90% confidence limits for each mean being calculated from the results of the five determinations for each solution. All solutions distributed were examined for their stability prior to the test.The power of the stability test was such that a true change in determinand value (over a period of storage of 2 weeks in darkness at room temperature) of 2% for conductivity or of 0.025 pH unit for pH would have been detected at the 95% confidence level. These stability tests revealed that the river water samples examined were not adequately stable with respect to pH hence the use of buffer solutions only for this determinand. To assess whether or not the bias of a laboratory exceeded the target value the following procedure was adopted. Let the mean result and its 90% confidence interval of laboratory i be denoted by xi + Li. The value of the maximum possible bias of laboratory i (95% confidence level) was then calculated as ( x i + Li - X) if xi > X or (xi - Li - X ) if xi < X where X is the true value of the distributed standard solution.The results in Tables 3 and 4 indicate that all laboratories meet the target for maximum possible bias for both determi-nands (except for a wholly marginal failure by one laboratory for one of the buffer solutions). Laboratory 4 reported its conductivity results after measurement at 25 "C and its results were corrected to 20 "C using a factor of 0.020 pS cm-1 "C-1. Its results remain within target if a factor from the other end of the likely scale of values (i.e. 0.026) is used. Laboratory 10 provided information on the temperature dependence of the buffer solutions by performing determi-nations at both 20 and 25 "C.These results which are included in Table 4 indicate no important difference in values obtained. Table 3. Results for tests of inter-laboratory bias-conductivity. The target for maximum possible bias is 10% of the conductivity of the sample or 2.5 pS cm-1 (whichever is the larger) River water A Standard solution River water B Difference Difference from mean Maximum Difference Maximum from mean Maximum Labora- Standard of all possible Standard from WRC possible Standard of all possible tory Mean/ deviation/ results bias,* Mean/ deviation/ value bias,* Mean/ deviation/ results bias,* 1 650.0 2.55 1 .58 1.96 660.0 1.73 2.01 2.26 212.8 3.42 2.26 3.83 2 614.0 5.48 4.05 4.86 626.0 5.48 -3.25 -4.05 201.6 2.07 -3.12 -4.07 3 661.4 2.61 3.36 3.74 669.0 3.00 3.40 3.84 217.0 1.00 4.28 4.73 4 642.6 2.05 0.98 1.29 653.2 1.79 1.51 1.78 208.1 1.10 0.53 1.03 5 643.2 1.48 0.51 0.73 650.4 1.14 0.53 0.69 211.0 0.00 1.39 1.39 6 627.6 1.52 -1.92 -2.15 656.2 2.77 1.42 1.83 194.2 3.27 -6.68 -8.17 7 646.4 6.23 1.01 1.94 664.0 5.48 2.63 3.43 206.6 3.65 -0.68 -2.39 8 634.0 2.00 -0.98 -1.22 634.4 4.83 -0.56 -1.27 207.4 1.95 -0.33 -1.23 No.pScm-1 pScm-1 O/O YO pScm-1 pScm-l Y O % pScm-1 pScm-1 "/O O/O 9 t 638.8 2.28 -0.17 -0.51 655.4 7.27 0.51 1.57 211.6 3.58 1.68 3.32 lo$ 635.0 2.24 -0.77 -1.10 641.4 3.65 -0.87 -1.40 208.6 1.34 0.24 0.86 lo§ 650.3 6.18 1.63 2.55 652.6 6.14 0.09 0.99 212.9 1.82 2.31 3.14 11 645.2 2.49 0.83 1.20 657.2 4.82 1.58 2.29 213.4 1.14 2.55 3.07 laboratories 639.9 652.08 208.1 * The values given for maximum possible bias are quoted at a confidence level of 95%.+ Laboratory 9 was unable to participate in this test within the time available owing to the breakdown of its instruments. Its results are $ Results of measurements at 20 "C-used in calculation of the mean result of laboratories. § Results of measurements at 25 "C corrected to conductivity at 20 "C using (Y = 0.02 "C-1. Those for laboratory 4 were used in the Mean of all WRC value 647 reported but have not been used to calculate the mean result of laboratories. calculation of the mean result of laboratories but those for laboratory 10 were not because values obtained at 20 "C were available ANALYST APRIL 1984. VOL. 109 435 Table 4. Results of tests of inter-laboratory bias-pH. Target maximum possible bias 0.1 pH unit Buffer solution A Buffer solution B Difference Standard 95% from Maximum Laboratory Mean deviation confidence nominal possible Mean 1 6.874 0.018 0.017 -0.007 -0.024 7.404 2 6.872 0.008 0.008 -0.009 -0.017 7.424 3 6.960 0.019 0.018 0.079 0.097 7.420 4 6.930 0.019 0.018 0.049 0.067 7.468 5 6.978 0.004 0.004 0.097 0.101* 7.506 6 6.830 0.026 0.025 -0.051 -0.076 7.392 7 6.870 0.007 0.007 -0.011 -0.018 7.404 8 6.862 0.004 0.004 -0.019 -0.023 7.400 9 6.846 0.009 0.009 -0.035 -0.044 7.374 10(a)t 6.840 0.010 0.010 -0.041 -0.051 7.400 (b)$ 6.838 0.015 0.014 -0.043 -0.057 7.408 11 6.908 0.022 0.021 -0.027 0.048 7.430 laboratories 6.888 7.420 value .. . . 6.881 7.429 * Exceeds target. t (a) Results of 20 k 1 "C; used in calculation of the mean result of laboratories.$ (b) Results at 25 +- 1 "C. No. pH pH units limits value Yo bias o/o PH Mean of all Nominal Standard deviation, pH units 0.018 0.005 0.020 0.015 0.005 0.013 0.005 0.007 0.017 0.010 0.015 0.017 95 O/O confidence limits 0.017 0.005 0.019 0.014 0.005 0.012 0.005 0.007 0.016 0.010 0.014 0.017 Difference from nominal value o/o -0.025 -0.005 -0.009 0.039 0.077 -0.037 -0.025 -0.029 -0.055 - 0 .O29 -0.021 0.001 Maximum possible bias O/O -0.042 -0.010 -0.028 0.053 0.082 -0.049 -0.030 -0.036 -0.071 -0.039 -0.035 0.018 Table 5. Results of inter-laboratory follow-up test-conductivity. Values in pS cm-1 (corrected to 20 "C using a correction factor of 0.022 "C-l).The results of laboratories 3 and 7 can be rejected as statistical outliers (p = 0.05). This has not been done in the results shown here because such rejection makes no difference to the conclusions drawn regarding the ability of laboratories to meet the Harmonised Monitoring requirements for analytical accuracy for conductivity determination. Target maximum possible bias 10% Laboratory No. 1 2 3 4 5 6 7 8 9 10 11 Mean of all laboratories * Exceeds target. Mean 800.00 788.00 865.00 798.33 804.00 819.20 646.20 782.40 772.20 793.88 816.41 789.6 Standard deviation 0.00 4.47 0.00 1.26 2.92 0.84 1.30 1.34 4.44 0.00 1.28 95% confidence limits 0.00 4.26 0.00 1.20 2.78 0.80 1.24 1.28 4.23 0.00 1.22 Difference from mean of all results, O/O 1.32 -0.20 9.55 1.11 1.82 3.75 - 18.16* -0.91 -2.20 0.54 3.39 Maximum possible bias, O/O 1.32 -0.74 9.55 1.26 2.18 3.85 -18.32* -1.07 -2.74 0.54 3.55 It was concluded therefore that satisfactory freedom from between-laboratory bias had been achieved for conductivity determination on both the standard solution and on two real samples and for pH determination on the buffer solutions.Routine AQC To attempt to ensure that the required accuracy of results is maintained AQC is now an integral part of the routine analysis for the Harmonised Monitoring Scheme. As stated above primary reliance for this purpose is placed on within-laboratory AQC using statistical quality control charts.However to obtain direct checks on between-laboratory bias, portions of samples are distributed at intervals to all labora-tories. Since the completion of the preliminary tests only one such distribution (of a river sample for conductivity and a buffer solution for pH) has so far taken place. The results for these tests are shown in Tables 5 and 6. Not all laboratories adhered to the agreed procedure of temperature control for the measurement of pH (see note in Table 6). Despite this, results of adequate accuracy were obtained. Such freedom from important temperature dependence of pH value whilst being observed for a buffer solution cannot be relied upon where river samples are concerned.It appears from these results that reasonably satisfactory accuracy has been maintained although the occasional deviations emphasise the need for continuing AQC. Discussion Throughout this programme of AQC there were very few instances of failure to meet targets. It is therefore reasonable to assume that the preliminary tests of precision and between-laboratory bias demonstrated that laboratories were capable of achieving adequate accuracy for the determination of electrical conductivity. For pH unambiguous tests of inter-laboratory bias by means of the analysis of a distributed river sample proved impossible owing to sample instability. The use of buffer solutions in such tests means that there was no direc 436 ANALYST APRIL 1984 VOL. 109 Table 6. Results of inter-laboratory follow-up test-pH.Target maximum possible bias 0.1 pH unit Laboratory No. 1 2 3 4 5 6 7 8 9 10 11 Nominal pH value of buffer solution . . . . laboratories . . red) value. . . . Mean of all WRC (measu-* Exceeds target. Mean 6.218 6.264 6.330 6.192 6.248 6.244 6.228 6.200 6.266 6.168 6.309 PH Standard deviation, pH units 0.01 1 0.009 0.000 0.008 0.004 0.025 0.008 0.000 0.022 0.013 0.007 6.2 (from tables) 6.242 6.231 Difference from mean of 95% confidence all results, limits O/O 0.010 -0.024 0.009 0.022 0.000 0.088 0.008 -0.050 0.004 0.006 0.024 0.002 0.008 -0.014 0.000 -0.042 0.021 0.024 0.012 -0.074 0.007 0.067 Maximum possible bias Yo -0.035 0.030 0.088 -0.058 0.010 0.025 -0.022 -0.042 0.044 -0.087 0.074 demonstration that adequate accuracy can be achieved for the determination of pH.However inter-laboratory tests involv-ing buffer solutions showed that the effects of biased calibration were acceptably small in all laboratories. The participating laboratories demonstrated that for pH determi-nations made on most river water samples the required analytical precision was achieved. There is evidence that the determination of the pH value of poorly buffered waters and/or waters of low ionic strength poses special problems which will require additional investigation. 10311 It is intended that this subject will be considered further by the Committee. The sequential approach to AQC followed in this work involves a relatively large amount of work in each laboratory and a relatively long period to complete all tests.However, these very points provide many opportunities for unsuspected errors to be revealed thereby facilitating recognition and elimination of problems so that a permanently sound basis is established for routine achievement of the required accuracy. Where as with the determinands discussed here the accuracy requirements proved to be relatively lax compared with the accuracy that could be readily achieved the sequential approach to AQC allowed rapid progress to be made. Where the accuracy requirements are stringent relative to what is easily obtainable this approach provides the only systematic way in which these standards of accuracy may be attained.In making the tests described here not all laboratories followed the agreed procedure involving the measurement of sample pH at a fixed temperature (usually 20 "C) +1 "C. This did not lead to any difficulty in achieving the required accuracy when buffer solutions were analysed but no evidence is available that allows this conclusion to be extended to real samples. In the absence of further information on this subject, it has been agreed by the Committee that temperature control to k 1 "C is desirable and should be applied routinely to all pH determinations for the Harmonised Monitoring Scheme apart from any performed in situ. Conclusions The targets chosen for the accuracy of results for the determination of electrical conductivity appear to be suitable for the Harmonised Monitoring Scheme and capable of achievement for the river waters tested.It proved impossible to check fully compliance with accuracy requirements for the determination of pH but it was demonstrated that the analytical precision was adequate for most types of sample and that the calibration bias was acceptably small. There is, however a need for further work on the determination of pH on samples of low ionic strength and/or low buffer capacity. Continuing care is needed to ensure that adequate accuracy is maintained. Subsequent AQC is in addition to normal precautions based on the use of quality control charts and the analysis of solutions distributed at intervals by the WRC. These studies by the Committee have now demonstrated, for chloride ammonia total oxidised nitrogen nitrite sus-pended solids conductivity and pH a procedure for ensuring permanent comparability of results from a group of labora-tories.It is hoped to report the results of further studies on different determinands in subsequent papers. Although the analytical work reported here was performed by Regional Water Authority and River Purification Board laboratories the coordination of the work was carried out by the Water Research Centre under contract to the Department of the Environment whose permission to publish this paper is acknowledged. Appendix The following are or have been members of the Analytical Quality Control (Harmonised Monitoring) Committee in the period during which this work was performed Dr.P. R. Hinchcliffe Mr. J. G. Flint Mr. L. R. Pittwell Mr. R. Donachie Mrs. C. Brown Mr. N. Taylor Dr. R. J. D. Otter and Mr. P. H. Garnett (Department of the Environment); Mr. A. L. Wilson Mr. D. J. Dewey Dr. D. T. E. Hunt and Mr. M. J. Gardner (Water Research Centre); Mr. M. J. Beard (Southern Water Authority); Mr. B. E. P. Clement and Mr. R. Lamb (Welsh Water Authority); Mr. N. Croft Mr. M. G. Firth and Mr. D. Best (Yorkshire Water Authority); Dr. B. T. Croll (Anglian Water Authority); Mr. D. V. Hopkin (Thames Water Authority); Mr. J. G. Jones and Mr. A. Poole (Wessex Water Authority); Mr. B. Milford and Mr. B. Dale (South West Water Authority); Mr. T. Hooton (Scottish Develop-ment Department); Mr. J. B. Allcroft and Mr. A. Hollington (North West Water Authority); Mr. W. Wollers (North-umbrian Water Authority); Dr. K. C. Wheatstone and Mr. K. Bamford (Severn-Trent Water Authority); Mr. J. E. Saunders (Welsh Office); Mr. I. R. M. Black (Forth River Purification Board) ANALYST APRIL 1984 VOL. 109 437 References 1. 2. 3. 4. 5. 6. Simpson E. A J . Inst. Water Eng. Sci. 1978 32 45. Wilson A. L. Analyst 1979 104 273. Analytical Quality Control (Harmonised Monitoring) Commit-tee. Analyst 1979 104 290. Analytical Quality Control (Harmonised Monitoring) Commit-tee Analyst 1982 107 680. Analytical Quality Control (Harmonised Monitoring) Commit-tee Analyst 1982 107 1407. Analytical Quality Control (Harmonised Monitoring) Commit-tee Analyst 1983 108 1365. 7. DOE/NWC Standing Committee of Analysts “The Measure-ment of Electrical Conductivity and the Laboratory Determi-nation of the pH Value of Natural Treated and Waste Waters,” HM Stationery Office London 1978. Wagner R. Z. Wasser Abwasser Forsch. 1980 13 62. Smith S . Limnol. Oceanogr. 1962,7 330. Midgley D. and Torrance K. Analyst 1976 101 833. Midgley D. and Torrance K . Analyst 1979 104 63. 8. 9. 10. 11. Paper A31276 Received August 22nd 1983 Accepted October 6th 198

ISSN:0003-2654    

DOI:10.1039/AN9840900431

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

ANALYST APRIL 1984 VOL. 109 439 Reference Electrodes for Use in the Potentiometric Determination of Chloride Part 1. Assessment of Mercury = Mercury(1) Sulphate Electrodes Derek Midgley Central Electricity Generating Board Central Electricity Research Laboratories Kelvin Avenue, Leatherhead Surrey KTZZ 7SE UK Mercury - mercury(1) sulphate reference electrodes are used in the potentiometric determination of chloride. There have been various reports of their erratic behaviour which is especially serious because chloride-selective electrodes are often working at the limits of their sensitivity in highly pure waters such as condensed steam and boiler water. The loss of precision in e.m.f. in such circumstances has a much larger effect on the precision in concentration terms than when electrodes are working in their usual (Nernstian) sensitivity range.Commercial mercury - mercury(1) sulphate electrodes have been tested over several weeks of continuous operation so that their performance can be assessed when used under conditions similar to those of chloride monitoring i.e. in pH 4.4 acetate buffer solution with silver chloride electrodes and 0.01 mol 1-1 nitric acid solution with mercury(1) chloride electrodes. Electrodes were obtained from two manufacturers and were of two types-one with ground-glass sleeve liquid junctions and one with a ceramic-frit junction. The latter was the best of those tested. The rate of change of its potential with respect to a calomel electrode was less than 0.5 mV over 100 h and the standard deviation of its potential over a week's operation was about 0.3-0.5 mV.Deviations of potential previously reported with this type of electrode could not be reproduced in this work. Some electrodes with ground-glass sleeve junctions became erratic after 2-4 d in the nitric acid solution normally used with mercury(1) chloride sensing electrodes. Their performance was better in the acetate buffer medium used with silver chloride sensing electrodes but inferior to that of similar electrodes with ceramic frit junctions. Keywords Chloride determination; potentiometry; mercury - mercury(/) sulphate electrodes; reference electrodes The determination of chloride is important for power station operations as chloride is one of the main causes of corrosion in boilers. In high-pressure drum boilers the chloride level should be below 200 yg 1-1 and should be known so that the boiler-water chemistry can be adjusted appropriately.In once-through boilers as in modern nuclear power stations, the feed water should contain less than 2 yg 1-1 of chloride. Chloride levels in condensed steam can indicate condenser leaks. Continuous monitoring of the chloride level is normal for once-through boilers and becoming more common for drum boilers. Most potentiometric methods of analysis use either silver -silver chloride or calomel reference electrodes but leakage of concentrated (23 mol 1-1) potassium chloride solution from these electrodes makes them unsuitable for use with chloride-selective electrodes. Mercury - mercury(1) sulphate reference electrodes are generally used for chloride determinations because leakage of their sodium or potassium sulphate filling solutions does not affect chloride-selective electrodes.Although the thermodynamics of the mercury - mercury(1) sulphate electrode are well characterised,' practical difficulties have been observed with commercially produced forms of reference half-cell which have not always exhibited the expected constancy of e.m.f.2-4 The difficulties experienced with mercury - mercury(1) sulphate electrodes are intermittent and depend on the configuration of the liquid junction and the experimental conditions; thus Torrance and Wilson2 and Marshall and Midgley3 found the ground-glass sleeve type of junction preferable to the ceramic-frit type while Richardson5 found the opposite.These problems were not investigated systematically but their origins may lie in the precipitation of mercury(1) chloride at the interface between the sample and reference solutions. Mercury(1) sulphate is relatively soluble (K = 6 x 10-7 mo12 1-2) and the concentrations of mercury(1) ions in 1 mol 1-1 sodium sulphate reference solution would cause precipitation to occur with yg 1-1 levels of chloride. On prolonged immersion of a mercury - mercury(1) sulphate reference electrode in a chloride solution a deposit may sometimes be seen around the liquid junction. The aim of this work was to assess the performance of commercial mercury - mercury( I) sulphate reference electrodes over several weeks of continuous use in media similar to those used in potentiometric analysis for chloride, i.e.pH 4.4 ammonium acetate - acetic acid buffer for silver chloride electrodes and 0.01 moll-' nitric acid for mercury(1) chloride electrodes. Analysis for chloride in condensed steam involves the least sensitive part of an ion-selective electrode's calibration that in which the e.m.f. is directly proportional to concentration4Jj and precision of the e.m.f. has a larger relative effect on the precision of analysis than at higher concentrations where the e.m.f. is proportional to the logarithm of the concentration. In this respect the performance of the mercury - mercury(1) sulphate electrode is more critical than that of the reference electrodes used for most potentiometric analyses. The products of the main manufacturers of mercury - mercury( I) sulphate electrodes available in the UK were surveyed and representative electrodes were selected.Experimental Apparatus The electrodes used are listed in Table 1. The glass sleeve of electrode Q fitted badly and was replaced by a silicone-rubber sleeve recommended by the manufacturers.7 The Pye 305 calomel electrode was taken as the master reference elec-trode which completed an electrochemical cell with each of the other electrodes and the Pye 360 was included as a check on the system these both being well characterised types of electrodes. ANALYST APRIL 1984 VOL. 109 440 Table 1. Electrodes tested Manufacturer’s Electrode Liquid identification type Filling solution junction Coding EIL 33 1370 230 . . . .. . . . Hg . Hg,S04 1 moll-’ Na2S04 Frit K L Pye 360 . . . . . . . . . . Ag . AgCl 3 moll-’ KCI Pye 305 . . . . . . . . . . Hg . Hg2S04 3 moll-1 KCI EIL 33 1380 230 . . . . . . . . Hg . Hg2S04 1 moll-1 Na2S04 GGS* M,N Beckman40455 . . . . . . . . Hg-Hg2S04 K2S04(saturated) GGS*.I- P Q Frit R Frit * GGS ground-glass sleeve. t The ground-glass sleeve on electrode Q was changed for one of silicone rubber after two weeks. Measurements were made as described previously.8.9 The electrodes were switched in turn to a Corning 110 pH meter by a modified signal multiplexer and the e.m.f.s were recorded by a data logger connected to the recorder output terminals of the pH meter. The electrodes were placed in a Perspex flow cell housed in a cabinet and the temperatures of both the flow cell and cabinet were controlled (at 25 “C for most of the tests) as described previously.8 Reagents All reagents were of AnalaR grade (obtained from BDH Chemicals).Nitric acid 5 moll-1. Concentrated nitric acid (317 ml) was diluted to 11. Standard chloride solution 10 000 mg 1-1. Sodium chloride (16.49 g) was dissolved in water and made up to 1 1 in a calibrated flask. Standard chloride solution 1 000 mg 1-1. Sodium chloride (1.649 g) was dissolved in water and made up to 1 1 in a calibrated flask. Nitric acid working solutions. These were prepared in 96-1 batches by dilution of 192 ml of 5 mol 1-1 nitric acid. Some batches also contained either 96 ml of 1 000 mg 1-1 chloride solution or 96 ml of 10 000 mg 1-1 chloride solution.Acetate buffer working solution. Concentrated acetic acid (114 ml) was added to about 9 1 of de-ionised water followed by 55 ml of concentrated ammonia solution (sp. gr. 0.88). The solution was allowed to cool and made up to 10 1 with de-ionised water. This solution was equivalent to the mixture of sample and buffer streams in the Technicon chloride monitor10 and had a pH of 4.4. Procedure The tests were carried out at 25 “C in the flow cell and cabinet described above. The test solution was pumped from the reservoir to a header above the cabinet. The solution then flowed under gravity into the flow cell at a rate of about 8 ml min-1 (controlled by a capillary restriction). The solution in the flow cell was stirred by a magnetic bar. Results Interpretation Because every measurement of e.m.f.involves two elec-trodes a strictly unambiguous assessment of the performance of individual electrodes is impossible. A practical determina-tion of performance characteristics can be made by taking as a “master” electrode one which is known to enable reprodu-cible e.m.f.s to be obtained when it is used in conjunction with a variety of other electrodes. All the e.m.f. measurements in this work were made with respect to a calomel electrode, which is widely accepted as having “good” properties. Over periods of 3-14 weeks however even calomel electrodes are subject to variations in e.m.f.8 and a further check on the results was desirable. Table 2. Short-term standard deviation in 0.01 mol 1-1 nitric acid Short-term standard deviation measured at close of week No.*/ mV Range over 22 weeks Electrode 1 K . . . . . . 0.067 L . . . . . . 0.084 M . . . . . . 0.110 N . . . . . . 0.071 P . . . . . . 0.047 Q . . . . . . t R . . . . . . 0.042 8 0.067 0.082 0.298 0.141 0.067 1.77 0.042 15 0.097 0.140 0.107 0.140 0.149 0.133 0.117 22 0.123 0.126 0.227 0.097 0.132 0.490 0.067 * Calculated from 10 readings at 1-min intervals. t Not available-see Table 1. Min. Max. 0.042 0.127 0.052 0.311 0.053 1.62 0.042 0.560 0.053 0.149 0.047 1.77 0.00 0.117 The check consisted of including in the tests a silver - silver chloride reference electrode of a type previously found satisfactory,g so that the performance of mercury - mercury(1) sulphate electrodes could also be judged against another type of well characterised electrode.The results obtained with the silver - silver chloride electrode are not treated in detail below but the silver - silver chloride calomel pair gave the most constant potential of all the electrode pairs thus confirming the choice of master electrode. Tests in 0.01 mol 1-1 Nitric Acid Medium for Use with Mercury(1) Chloride Electrodes Test conditions The recommended ionic medium for use with mercury(1) chloride electrodes is 0.01 mol 1-1 nitric acid.3 The test solution contained no chloride ion for weeks 1-3 1 mg 1-1 chloride for weeks 4-13 and 10 mg 1-1 chloride for weeks 14-22. The chloride ion was included at fairly high levels in an attempt to accelerate the appearance of the deviations previously reported.2-4 Short-term variations in potential The short-term variation in the e.m.f.s was determined from ten consecutive readings 1 min apart.The time interval was dictated by the requirements of the data logger but in terms of continuous monitoring 10 min is a suitably short period. An extract of the results is shown in Table 2. The standard deviations of electrodes with ceramic-frit junctions (K and L) varied over a comparatively small range and tended to change in parallel with one another indicating the influence of a factor common to all the electrodes e.g., temperature stirring rate or the behaviour of the master calomel reference electrode. Although the largest standard deviations occurred in weeks 15 17 20 and 22 there was no consistent change in the standard deviations over the period of test either with time or with chloride content.The electrodes with sleeve-type junctions (M N P and 0) varied considerably in performance; their standard deviations could be as small as those of the electrodes with frit junctions ANALYST APRIL 1984 VOL. 109 44 1 ~~ Table 3. Drift of standard potential relative to calomel electrode in nitric acid solution Drift rate during week No./mV per 100 h Electrode 1 4 7 K . . . . . . . . 0.3 -0.04* -0.1 L . . . . . . . . 0.3 -0.1* -0.1 M . . . . . . . -5.5 -0.4 3.0 N . . . . . . . . 0.9 -0.4 0.9 P . . . . . . . . -1.o* -0.8 -0.8* Q . . . . . . . . $ -4.0 -7.4 R . . . . . . . . 1.0t 0.41- 0.007 * The linear fit to the results had a correlation coefficient (/pi) >0.7.1- The linear fit to the results had a correlation coefficient (/pi) >0.9. j Not available-see Table 1. 11 0.2 -0.1 -3.3t 0.1 -1.l-k 2.6 1.61. 15 -0.2 -0.3 -3.51. -1.0 -0.8 3.9 0.5i 18 -0.2* -0.6 -3.2t -0.4 -1.1* 0.9 0.4t 22 -0.8t - 1 .o* -3.4-t -2.8f -1.2-l -2.1* 0.5* Table 4. Standard deviation of e.m.f. during 1 week in nitric acid solution Standard deviation during week No. */mV Electrode 1 4 K . . . . . . . . 0.3 0.2 L . . . . . . . . 0.3 0.2 M . . . . . . . . 4.1 2.5 N . . . . . . . . 2.6 2.0 P . . . . . . . . 0.4 0.4 Q . . . . _ . . . 1- 2.7 R . . . . . . . . 0.4 0.2 * Calculated from approximately 70 2-h readings in each week.i- Not available-see Table 1. 8 0.3 0.3 3.1 2.4 0.6 6.9 0.2 11 0.2 0.3 1.5 1.9 0.5 2.1 0.8 15 0.2 0.2 1.9 1.3 0.6 3.3 0.3 18 0.9 0.3 1.7 1.8 0.6 2.2 0.2 22 0.4 0.5 1.7 1.6 0.6 1.3 0.3 400 - -390 i-380 370 ui 350t 1 1 48 72 96 120 144 340 24 Time/h Fig. 1. Change of e.m.f. with time during seventh week of operation in 0.01 mol I-' nitric acid for electrodes with (L) ceramic-frit (M) ground-glass sleeve and (Q) silicone-rubber sleeve liquid junctions but at times were much larger. Changes in the standard deviations were not concerted in contrast to the electrodes with frit junctions. Electrode P with a ground-glass sleeve, was as good as the frit electrodes but electrode Q with a silicone-rubber sleeve was the worst of those tested.Drift in standard potential Drift is defined here as the tendency of the e.m.f. to change persistently in one direction over a given period of time. The rate of drift was determined from the gradient of a linear correlation of the 2-h readings of e.m.f. against time over a period of a week. The gradients obtained were almost always significantly different from zero (t-test 95% confidence limits). An extract of the results is shown in Table 3 and the drift over a typical week's operation is shown in Fig. 1 for three electrodes. Electrodes with ceramic-frit junctions (K L) had small rates of drift rarely exceeding 1 mV per 100 h in any one week. Electrode P with a ground-glass sleeve junction was almost as good but the other electrodes with sleeve junctions (M N and Q) had high rates of drift often changing in direction and magnitude from week to week.Fig. 1 shows that the linear drift rates for electrodes M and Q were only rough indicators of performance as potentials fluctuated widely within the period of test. Electrodes with low rates of drift did not show these fluctuations (e.g. curve L in Fig. 1). Fluctuations of electrode potential Fluctuations such as those for curves M and Q in Fig. 1 can lead to as much analytical error as does drift. As a measure of this source of error the standard deviations of the 70 or so 2-h readings taken each week were calculated. These results contain contributions from short-term (random) variations , from systematic changes (drift) and from long-term fluctua-tions which may not be random.The results in Table 4 confirm the impression given by the results for drift; the ground-glass sleeve electrodes M and N and electrode Q with the silicone-rubber sleeve showed much the largest variations. Electrode P with the ground-glass sleeve junction was very consistent while the smallest variations were shown by frit electrodes K and L. Effect of temperature The temperature in the flow cell was raised or lowered in 5 "C steps over the range 20-30 "C two cycles being completed. The steady e.m.f. of each electrode at each temperature was noted. The temperature coefficient of each type of electrode with respect to the 3 mol 1-1 calomel electrode is shown in Table 5.These results may be converted to the standard hydrogen electrode scale by adding -0.4 mV K-1. Fig. 2 shows that the e.m.f of the mercury - mercury(1) sulphate electrode follows the temperature change very closely 442 ANALYST APRIL 1984 VOL. 109 Table 5. Temperature coefficients at 25 "C Table 7. Drift of standard potential relative to calomel electrode in acetate buffer solution Electrode Temperature Standard Electrode solution mV K-1 mVK-l filling coefficient*/ deviationtl K-N . . . . 1mo11-1Na,S04 -0.39 0.02 P . . . . . . SaturatedK,SO -0.37 0.03 * Mean of all results for each type of electrode. t Standard deviation of the mean temperature coefficients. Drift rate during week No. /mV per 100 h Electrode 1 2 6 10 K . . . . . . -3.1* -0.2 0.4? -0.61-L .. . . . . -3.5* 0.01 0.6-1- -0.5* M . . . . . . -4.1* -0.6* 0.6 -0.1 N . . . . . . -3.8* -0.2 4.4* -0.8t R . . . . . . -3.9* -0.1 1.3* -0.01 Ms . . . . . . 7.73- -2.4 -2.1 -* Linear fit gave correlation coefficient (/pi) B0.9. 3- Linear fit gave correlation coefficient (\pi) >0.7. Table 6. Short-term deviations in acetate buffer solution Short-term standard deviation at close of week No. */mV Table 8. Standard deviation of e.m.f. during 1 week in acetate buffer solution Range over 10 weeks Min . Max. 0.047 0.110 0.042 0.162 0.052 0.106 0.048 0.157 0.067 0.216 0.032 0.105t intervals. Electrode 1 10 K . . . . . . 0.082 0.084 L . . . . . . 0.063 0.042 M . . . . . . 0.071 0.094 N . . . . . . 0.108 0.063 R . . . . . . 0.067 0.092 Ms .. . . . . 0.092 0.1051-* Calculated from 10 readings at 1-min t Tests terminated after 8 weeks. Standard deviation during week No. */mV Electrode 1 2 6 10 K . . . . . . 1.7 0.1 0.1 0.3 L . . . . . . 1.9 0.1 0.2 0.2 M . . . . . . 2.2 0.3 0.6 0.8 N . . . . . . 2.1 0.2 1.2 0.4 R . . . . . . 2.1 0.1 0.4 0.1 Ms . . . . . . 4.0 2.4 1.7 -* Calculated from approximately 70 2-h readings in each week. 390 I I 386 L 380 - -5 370 1 . 2 360 1 I I I 1 I I 24 48 72 96 120 144 168 330 oy Time/h L 1 1 I I I 0 2 4 6 8 10 Timeih Fig. 2. Effect of temperature on the mercury - mercury(1) sulphate electrode in nitric acid solution Fig. 3. sealed and (L) ceramic-frit electrodes Change of e.m.f. during the first week of the test for (M,) Tests in Acetate Buffer Solution for Use with Silver Chloride Electrodes The ionic medium generally used in the Central Electricity Generating Board (CEGB) for measurements with silver chloride electrodes is an ammonium acetate - acetic acid buffer solution at pH 4.0-5.0.2.6,lo The test solution was 0.2 mol 1-1 acetic acid half neutralised with ammonia solution.Only electrodes K-N were tested. The earlier performance of electrode Q was too poor to warrant further work and electrode P gave similar results in 0.01 mol 1-1 nitric acid solution to the ceramic-frit electrodes. In a separate series of tests the performance of electrode M was assessed with the breather hole closed so as to simulate a sealed reference electrode. For brevity the results with this electrode, designated Ms have been included in the following sections because the test conditions were the same.Short-term variations in potential The short-term variance in the e.m.f.s. was determined from ten consecutive readings 1 min apart. Table 6 shows an extract of the results which did not differ greatly in magnitude from those obtained in 0.01 moll-' nitric acid although the spread of results was smaller-The standard deviations obtained from electrodes with ground-glass sleeves (M N) were no worse than those from similar electrodes with ceramic frits (K L) in contrast to the results in the nitric acid medium. Drift of standard potential In general the same considerations apply as in nitric acid solutions. Table 7 gives an extract of the results obtained from a linear correlation of e.m.f.against time. The superior performance of the ceramic-frit junction over the ground-glass sleeve junction is less pronounced than in nitric acid solutions (Table 3). Electrode Ms however showed larger rates of drift than before it was sealed. Fig. 3 shows the change of e.m.f. with time for electrode Ms compared with that for electrode L. Fluctuations of electrode potential As for nitric acid medium the standard deviation of the 70 or so 2-h readings collected each week was taken as an indication of instability including drift random noise and long-term fluctuations. An extract of the results (Table 8) shows that all the mercury - mercury( I) sulphate electrodes and particularly those in ground-glass sleeves showed less variability than in nitric acid solutions.Comparison of Tables 3 and 7 suggests that changes in the rate of drift do not account for the reduction in variability (except for electrode M). Fluctuations such as those seen in graph M of Fig. 1 were not observed in the acetate medium and hence the variability was lower. The results for the sealed electrode Ms were much higher than for the others showing the effect of the slow fluctuations visible in Fig. 3 ANALYST APRIL 1984 VOL. 109 443 Effect of temperature The mercury - mercury(1) sulphate electrodes behaved similarly in the acetate and nitric acid media and the temperature coefficient in the acetate medium (-0.45 mV K-1) was close to that obtained in nitric acid (Table 5). Discussion Electrode Performance In all these tests electrodes K and L with ceramic-frit junctions performed very well; the rate of change of the standard potential with respect to a calomel electrode was less than 0.5 mV over 100 h and the standard deviation of the e.m.f.over a week's operation was about 0.3-0.5 mV. These results are almost as good as obtained previously for the best calomel and silver - silver chloride reference electrodes* and as good as those for the silver - silver chloride electrode included in these tests for comparison. The electrodes required no attention beyond topping up the internal filling solution every 6-8 weeks but there is a potential problem with the tendency of sodium sulphate decahydrate to precipitate from the solution. The concentration of 1 moll-1 sodium sulphate used in the EIL electrodes is close to saturation at 25 "C and over a period of weeks enough water evaporates through the breather hole for crystals to appear inside the electrode.At temperatures below about 18 "C sodium sulphate tends to precipitate from 1 mol 1-1 solution. It is possible for these crystals to form a plug that will impede the flow of solution through the frit leading to less stable e.m.f.s. (see below). If the plug is formed between the mercury - mercury(1) sulphate element and the frit it is possible for the electrical resistance to rise so much that the system is effectively open-circuit, although this did not occur during these tests. The use of 0.5 mol 1-1 sodium sulphate solution4 would avoid these problems. Electrodes M and N with ground-glass sleeve junctions did not perform so well especially in the nitric acid medium.These electrodes tended to give erratic e.m.f.s after 2-4 d of continuous operation. This may be because the internal filling solution emptied relatively quickly (needing to be replenished every week) so that towards the end of the week there was insufficient pressure to maintain a flow of solution through the junction. When the solution was prevented from flowing freely by closing the electrode's breather hole the perfor-mance was much worse than before and previous experience4 showed that good results were obtained in continuous operation if the electrode had a 50-cm head of filling solution so as to maintain the flow. The ground-glass sleeve on one of the electrodes (Q) fitted very poorly so that the electrode drained dry in about 2 d and went open-circuit .Replacement of the ground-glass sleeve with a silicone-rubber sleeve according to the manufacturer's instructions gave an electrode of markedly inferior perform-ance to all the others tested. This electrode had a low rate of loss of filling solution and behaved similarly to the EIL ground-glass sleeve electrode when it had its breather hole sealed (Ms). Electrode P with a properly fitting ground-glass sleeve was better than the other ground-glass sleeve elec-trodes although not as good as the electrodes with ceramic frits. At first it needed to be topped up weekly with filling solution but in the sixth week the level scarcely decreased. Crystals of potassium sulphate had formed round the spacer holding the mercury - mercury(1) sulphate element in place, and so sealed the electrode.No change in performance was apparent once the electrode had become sealed but this condition was not allowed to persist and the electrode was unblocked. The saturated filling solution used in electrode P is more likely to give trouble of this sort than the unsaturated solutions of electrodes K-M. Another cause of inconvenience with electrode P is that the sleeve can drop off. This did not occur when the electrode was in place in the flow cell but it needed to be handled more carefully than the other designs. The general performance of mercury - mercury(1) sulphate electrodes with ceramic-frit junctions contradicts previous experience in this laboratory where the ground-glass sleeve junction had been preferred.2.3 Experience with plant opera-tion however has favoured the ceramic-frit type.It is inferred that there are considerable variations between batches of commercially produced electrodes and also between individ-uals of the same batch (e.g. electrodes P and Q). Results with electrodes having a low rate of outflow of reference solution indicate that sealed reference electrodes would probably not be suitable for continuous analysis; similar conclusions were drawn8 for the calomel electrode commonly used in other potentiometric analyses. For continuous indus-trial analysis it would be convenient if mercury - mercury(1) sulphate reference electrodes could be fitted with side-arms for connection to a reservoir of sulphate solution.4Apart from reducing the frequency with which the electrode has to be refilled such a reservoir would help to maintain the flow of solution through the liquid junction which the present work shows to be important for optimising the performance of these reference electrodes.Badly manufactured ceramic frits or ground-glass sleeves can result in unstable liquid junctions and hence variable potentials.8 The potentials will vary more the larger is the difference between the mobility of anions and cations in solution as is apparent from a consideration of liquid-junction potentials. As the ratio of the mobilities of hydrogen and nitrate ions is about 5 1 while for ammonium and acetate ions it is less than 2 1 greater variability would be expected in nitric acid solution than in ammonium acetate solution and this was observed.The variation of the liquid-junction potential with pH did not agree very well with that calculated by the Henderson equation,ll e.g. in nitric acid medium the e.m.f. of an electrode with a ground-glass sleeve junction (M) changed almost linearly by -15 mV per pH unit in the pH range 1.8-2.2 compared with the predicted -6 mV per pH unit. Over ranges larger than about 0.4 pH unit the non-linearity of the relationship became obvious. The addi-tion of inert electrolytes such as potassium nitrate or potas-sium chloride to 0.01 mol 1-1 nitric acid solution caused increases of about 0.5 mV per 10 mmol 1-1 of salt which agreed with Henderson equation calculations.Consequences for Chloride Analysis The results show that reference electrodes are commercially available having acceptably constant e.m.f.s. for use in determining chloride. Of the electrodes tested those with ceramic-frit junctions were more suitable for use in continuous monitoring as those with ground-glass sleeve junctions needed frequent attention if errors were to be avoided. The erratic behaviour reported3 with both ceramic-frit and ground-glass sleeve electrodes in nitric acid solution was not reproduced in this work although the fluctuating e.m.f.s of some ground-glass sleeve electrodes in nitric acid solution seemed similar except for the lower frequency. Acceptable performance could be obtained in the nitric acid medium used with mercury( I) chloride sensing electrodes"4 and the acetate buffer which in CEGB practice6 is used with silver chloride sensing electrodes.Because the reference electrodes generally gave more constant e.m.f.s in the acetate buffer this medium should be preferred for use with silver chloride electrodes. It is not however suitable for use with the mercury(1) chloride electrodes which are generally used at the lowest levels of chloride. This work was carried out at the Central Electricity Research Laboratories and is published by permission of the Central Electricity Generating Board 444 ANALYST. APRIL 1984 VOL. 109 References 1. Ives D. J. G. and Smith F. R. in Ives D. J. G. and Janz, G. J. Editors “Reference Electrodes Theory and Practice,” Academic Press New York and London 1961 Chapter 8. Torrance K . and Wilson A. L. CERL Report No. RD/L/R 1517 Leatherhead 1968. Marshall G. B. and Midgley D. Analyst 1978 103 438. Marshall G. B. and Midgley D. Analyst 1979 104 55. Tomlinson K . and Torrance K. Analyst 1977 102 1. “Beckman Instructions 1075-A,” Beckman Instruments Inc., Fullerton California 1961. 2. 3. 4. 5. Richardson R. unpublished work. 6. 7. 8. 9. 10. Midgley D. and Torrance K. Analyst 1976 101 833. Midgley D. and Torrance K. Analyst 1979 104 63. “Operation Manual for the Technicon Trace Chloride Monitor 11,” Technicon Publication No. TICL 002 Technicon Instru-ments Co. Ltd. Basingstoke 1977. Ives D. J. G. and Janz G. J. Editors “Reference Electrodes, Theory and Practice,” Academic Press New York and London 1961 p. 54. 11. Paper A31372 Received October 27th 1983 Accepted November 21st 198

ISSN:0003-2654    

DOI:10.1039/AN9840900439

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

ANALYST APRIL 1984. VOL. 109 445 Reference Electrodes for Use in the Potentiometric Determination of Chloride Part II.* Quinhydrone Electrodes Derek Midgley Central Electricity Generating Board Central Electricity Research Laboratories Kelvin Avenue, Leatherhead Surrey KT22 7SE UK Reference electrodes consisting of platinum or gold electrodes immersed in solutions of constant pH saturated with quinhydrone have been tested for use in potentiometric analysis for chloride. The solution inside the reference electrode is matched in composition with the main components of the treated sample solution (i.e. after addition of reagents to control pH and ionic strength but excluding the determinand). In this way the effect of the liquid-junction potential is minimised. The electrodes have a sealed construction and need no maintenance.The potentials of newly made electrodes change at a greater rate than those of the mercury - mercury(1) sulphate reference electrodes normally used in chloride analysis but after 3-4 weeks the rate of change of potential is similar to that obtained from the best commercial mercury - mercury(1) sulphate electrodes. Older electrodes develop anomalous tem peratu re responses. The quinhydrone reference electrode is comparable to the best mercury - mercury(1) sulphate electrodes in media used with mercury(1) chloride sensing electrodes (typically 0.01 mol 1-1 nitric acid). In the media of higher pH used with silver chloride sensing electrodes (typically pH 4.3-5.0 acetic acid - ammonium acetate buffer solution) the performance of the mercury - mercury(1) sulphate electrode is improved while that of the quinhydrone electrode deteriorates; the former would therefore generally be preferred.This work has provided the basis for similar but improved electrodes using more chemically stable constituents which will be described in Part Ill. Keywords Chloride determination; potentiometry; quinh ydrone electrodes; reference electrodes The reference electrode is a necessary part of any electrochemical cell used in potentiometry and its properties often have a very significant influence on the characteristics of a potentiometric method. Thus it has been demonstrated for mercury - mercury(1) sulphate electrodes1 and calomel and silver - silver chloride electrodes2J that the configuration of the reference electrode affects both the short-term precision and long-term stability of e.m.f.measurements. In addition, the reference electrode is generally as significant as the sensing electrode in contributing to the temperature coefficient of the cell. Conventional reference electrodes consist of metal - metal salt electrodes immersed in concentrated solutions of the corresponding sodium or potassium salts separated from the test solution by a physical barrier that permits minimum mass transport of solution out of the electrode and enables electrical contact to be maintained by electrolytic diffusion. This latter part of the electrode the liquid junction is the most common source of unreliable e.m.f.s; badly constructed junctions may allow excessive mass transport of solution across the junction or be of excessively high electrical resistance.In either instance the performance of the potentiometric cell will not be optimal and may be unacceptable. Even good junctions may deteriorate with use mainly because of chemical deposition in the pores; this has been demonstrated for silver - silver chloride3 and mercury - mercury( I) sulphate electrodes .43 Comparable problems rarely arise from the inner metal -metal salt electrode unless it is allowed to dry out. This, however means that the maintenance requirements of the reference electrode are generally greater than those of the sensing electrode at least for glass and solid-state electrodes. Sealed maintenance-free reference electrodes have been made but their performance has hitherto been markedly inferior to that of conventional designs which permit a small outflow of solution from the electrode.'-3 * For Part I of this series see p.439. The aim of this work was to devise reference electrodes that would avoid the problems associated with conventional reference electrodes without sacrificing their good points. The following properties were desirable. (i) The inner electrode should give a potential that can be established reproducibly whenever and wherever the electrode is made. (ii) The potential of this electrode should be stable in a solution of constant composition for several weeks if not months. (iii) The electrode once assembled as a whole should require no maintenance. (iv) The electrode as a whole should have a relatively low resistance i.e.4 0 kB. (v) The temperature dependence of the electrode poten-tial should be similar to that for the ion-selective electrode being used (typically 0.2-0.5 mV K-1) and should not suffer from serious hysteresis effects. (vi) The liquid-junction potential between the inner reference solution and the outer test solution should be constant and preferably zero. (vii) The inner reference solution should not contain substances that might interfere with the sensing electrode or react with substances in the test solution such that the e.m.f. would be influenced by that reaction. (viii) It should be possible to use either the same electrode with a very wide range of test solutions or it should be possible to adjust the electrode precisely to suit a specific type of test solution.The balance of advan-tage depends on the application. Versatility is often very important for the measurement of widely rang-ing parameters such as pH and redox potential and in potentiometric titrations because the sample often changes considerably during the course of the titra-tion. For direct potentiometry with ion-selective electrodes however it may be advantageous to have a reference electrode specifically designed for each type of analysis 446 ANALYST APRIL 1984 VOL. 109 In this work emphasis was placed on devising reference electrodes for use with chloride-selective electrodes. In analytical methods using ion-selective electrodes it is desirable for solutions to have a constant ionic strength and a reasonably high specific conductivity (typically > 100 pS cm-1).If the sample solutions do not meet these requirements it is normal to add an inert electrolyte in order to achieve the desired conditions. In many instances it is also necessary to control the pH of the solution in order to prevent direct interference by hydrogen or hydroxide ions or indirect interference caused by hydrolysis of metal ions or protonation of anions. In most instances the pH is also the most important factor in determining the liquid-junction potential. The systems proposed in this report rely on the control of the sample pH. Although this work was confined to reference electrodes for use in chloride determinations the principles involved should be applicable to almost any potentiometric system in which the pH is controlled.Chloride electrodes are used widely and problems with reference electrodes in these systems have been reported . 4 3 Because these electrodes are generally working in their limiting response ranges,6 the precision of the measurements is particularly important, The first system chosen was based on quinhydrone (benzo-quinhydrone) which has been extensively used for pH measurement.' It was expected to meet requirements (i) (iii), (iv) (vi) (vii) (viii) and the temperature coefficient part of (v) but the long-term stability (ii) was unknown as was the temperature hysteresis part of (v). This paper describes the results of tests with the new reference electrodes in conditions similar to those used in potentiometric analysis for chloride, i.e.in pH 4.5 ammonium acetate - acetic acid buffer solution for silver chloride electrodes and 0.01 mol 1-1 nitric acid for mercury(1) chloride electrodes. Theory Quinhydrone is a 1 1 molecular complex of p-benzoquinone (quinone) and its hydroquinone (quinol). When dissolved in water these two components form an electrochemically reversible oxidation - reduction system in which hydrogen ions also participate: C6H402 + 2H+ + 2e- $ C6H4(OH)2 If a platinum or gold electrode is immersed in the solution its potential is given by equation (1). =E'+-log2+-log-k.pH k fQ . . (1) 2 'QH2 2 fQH2 where the subscripts Q and QH2 refer to quinone and hydroquinone respectively and a c and f are activities, concentrations and activity coefficients respectively.As the two components are added as the complex quinhydrone cQ = cQH2 and as they are non-electrolytes f Q .= f Q H 2 = 1 so the second and third terms on the right-hand side of equation (1) are equal to zero. It can now be seen that the quinhydrone electrode acts as a pH electrode. A platinum electrode immersed in a solution of constant pH saturated with quinhydrone should therefore constitute an electrode of invariant potential. Over long periods of time, however the potential may change. The two factors most likely to cause such changes are inter-diffusion of the sample and reference solutions which would change the pH itself, and reaction of either quinone or hydroquinone with sub-stances dissolved in the reference solution so changing the ratio cQIcQHZ in equation (1).The first of these factors may be eliminated almost completely by matching the solution inside the reference electrode to that outside. The composition of the solution in which the electrodes are immersed is dominated by reagents added to control the pH and ionic strength of this solution. The solution inside the reference electrode should contain the same reagents in the same concentrations as in the external solution but with the addition of an excess of quinhydrone. Experimental Apparatus Experimental electrodes The electrodes used are listed in Table 1. In each instance a commercially produced electrode was the structural basis of the experimental electrode. EIL RJ23/1 electrodes (A and B) were modified by removing the mercury and mercury(1) chloride from the element and then cutting away the empty glass sheath from around the platinum contact.The other electrodes were modified Russell series SR electrodes the usual calomel element being replaced by a 1.25 mm diameter platinum or gold wire. In both types of electrode the breather hole in the seating of the element was sealed with silicone rubber and the liquid junction was formed at a ceramic frit. Commercial reference electrodes. The Pye 305 calomel electrode (3 mol 1-1 potassium chloride filling) was taken as the master reference electrode which completed an electro-chemical cell with each of the other electrodes and the Pye 360 silver - silver chloride electrode (3 moll-' potassium chloride filling) was included as a check on the system these being well characterised types of electrodes with ceramic-frit junctions.2 Commercial mercury - mercury(1) sulphate reference elec-trodes with both ceramic-frit and ground-glass sleeve junc-tions were tested simultaneously as described elsewhere ,1 so that direct comparisons of the two types of electrode could be made.Instrumentation. Measurements were made as described previously.2.8 The electrodes were switched in turn to a Corning 110 pH meter by a modified signal multiplexer and the e.m.f.s were recorded by a data logger connected to the recorder output terminals of the pH meter. The electrodes were placed in a Perspex flow cell housed in a cabinet and the temperatures of both the flow cell and cabinet were controlled (at 25 "C for most of the tests).Reagents All reagents were of AnalaR grade (obtained from BDH Chemicals). Nitric acid 5 moll-1. Concentrated nitric acid (317 ml) was diluted to 11. Standard chloride solution 10000 mg 1-1. Sodium chloride (16.49 g) was dissolved in water and made up to 1 1 in a calibrated flask. Standard chloride solution 1000 mg 1-1. Sodium chloride (1.649 g) was dissolved in water and made up to 1 1 in a calibrated flask. Table 1. Experimental reference electrodes Electrode Filling solution* material Coding 0.01 mol 1-1 nitric acid . . . . . . Pt A B V. 0.01 moll-' nitric acid . . . . . . Au U 0.1 moll-' acetic acid + 0.1 moll-' sodium acetate . . . . . . . Pt C D 0.1 rnol 1-1 acetic acid + 0.1 mol 1-1 ammoniumacetate .. . . . . Pt E F G 0.1 mol 1-1 acetic acid + 0.1 mol 1-1 ammoniumacetate . . . . . . Au w 0.15 mol 1-1 potassium hydrogen phthalate adjusted to pH 4.4 with potassiumhydroxide . . . . . . Pt H J * The solutions were saturated with quinhydrone. x ANALYST APRIL 1984 VOL. 109 447 Nitric acid working solutions. These were prepared in 96-1 batches from 192 ml of 5 mol 1-1 nitric acid made up with de-ionised water. Some solutions also contained 96 ml of 1000 mg 1-1 or 96 ml of 10000 mg 1-1 chloride solution. Acetate buffer working solution. Concentrated acetic acid (114 ml) was added to about 9 1 of de-ionised water followed by 55 ml of concentrated ammonia solution (sp.gr. 0.88). The solution was allowed to cool and made up to 10 1 with de-ionised water.This solution is equivalent to the mixture of sample and buffer streams in the Technicon chloride monitor .g Electrode filling solutions Nitric acid filling solution. An excess of quinhydrone was added to a portion of the chloride-free nitric acid working solution. Ammonium acetate filling solution. An excess of quin-hydrone was added to a portion of the acetate buffer working solution. Sodium acetate filling solution. Sodium acetate trihydrate (6.80 g) was dissolved in about 400 ml of de-ionised water. Glacial acetic acid (2.85 mi) was added and the whole was made up to 500 ml. A glass electrode was immersed in the solution and more glacial acetic acid was added dropwise until the pH was the same (4.3) as in the acetate buffer working solution.An excess of quinhydrone was then added. Phthalate filling solution. Potassium hydrogen phthalate (7.66 g) was dissolved in about 220 ml of de-ionised water. Solid potassium hydroxide was added until the pH of the solution was the same as that of the acetate buffer working solution. The solution was made up to 250 ml and an excess of quinhydrone was added. The concentration of potassium hydrogen phthalate was chosen so as to give a solution of approximately the same osmotic pressure as the acetate buffer working solution. Procedure Most of the tests were carried out in the flow cell and cabinet described above and any exceptions were noted. The test solution was pumped from the reservoir to a header above the cabinet. The solution then flowed under gravity into the flow cell at a rate of about 8 ml min-1 (controlled by a capillary restriction).This arrangement avoided the electrical noise associated with pumping. The solution in the flow cell was stirred by a magnetic bar. Results Interpretation Because every measurement of e.m.f. involves two elec-trodes a strictly unambiguous assessment of individual electrodes is impossible. A practical determination of perfor-mance characteristics was obtained by measuring the e.m.f. with respect to a calomel electrode that was known from previous studies172 to have good properties (as judged by several different criteria). A further check consisted of including in the tests a silver - silver chloride reference electrode previously found satisfactory ,2 so that the perfor-mance of the quinhydrone electrodes could be judged against that of a well characterised electrode.The results obtained with the silver - silver chloride electrode1 confirmed the choice of master electrode. Electrodes for Use With Mercury(1) Chloride Sensing Electrodes The recommended ionic medium for use with mercury(1) chloride electrodes is 0.01 mol 1-1 nitric acid.4.s The quin-hydrone electrodes tested were A B U V X and Y from Table 1. Effect of p H The effects of variation in pH were tested by injecting portions of 1 mol 1-1 nitric acid solution into 50-ml portions of de-ionised water in the cell containing electrode V and the calomel electrode. Fig. 1 shows that the e.m.f. of the quinhydrone electrode varied in accordance with the liquid-junction potential calculated from Henderson’s equation.’ Effect of ionic strength The effect of varying the ionic strength was tested by injecting portions of a solution containing 3 moll-1 potassium chloride and 0.01 rnol 1-1 nitric acid into 50 ml of 0.01 mol 1-1 nitric acid.The e.m.f. of electrode V was measured with respect to the calomel electrode. Changes in e.m.f. were not discernible (i.e. <0.1 mV) until more than 2.5 x 10-3 moll-’ of chloride had been added. The over-all change for the addition of 0.22 rnol 1-1 chloride was +0.6 mV compared with -0.4 mV calculated from the Henderson equation.7 Short-term variance The short-term variation in e.m.f. for electrodes A and B was determined from ten consecutive readings 1 min apart.The time interval was dictated by the requirements of the data logger but in terms of continuous monitoring 10 min is a suitably short period. An extract of the 22 weeks’ results is shown in Table 2. The standard deviations varied over a comparatively small range and tended to change in parallel with one another indicating some common sensitivity to external factors perhaps temperature stirring rate or the master calomel reference electrode. Although the largest standard deviations occurred in weeks 15,17,20 and 22 there was no consistent change in the standard deviations over the period of test. These standard deviations were as small as those obtained1 with the silver - silver chloride and the best of the mercury - mercury( I) sulphate reference electrodes.Drift of standard potential Drift is defined here as the tendency of the e.m.f. to change persistently in one direction over a given period of time. Comparing successive readings (2 h apart) might give one measure of drift but in no instance was the drift calculated in this way significantly different from zero (t-test 95% con-Table 2. Short-term standard deviations in 0.01 rnol I-* nitric acid Short-term standard deviation during week No. */mV 1-22 Electrode 1 8 15 22 Min. Max. A . . . . . . 0.067 0.067 0.106 0.170 0.048 0.170 B . . . . . . 0.053 0.074 0.129 0.088 0.042 0.129 * Calculated from 10 readings at 1-min intervals. 1 Fig. 1. Effect of pH on quinhydrone reference electrode with 0.01 rnol 1-1 nitric acid filling solution.The solid line shows the calculated change in liquid junction potentia 448 ANALYST APRIL 1983 VOL. 109 fidence limits) i.e. these differences reflect short-term noise rather than a long-term effect such as the drift. Even when readings 24 h apart were compared drifts significantly different from zero could not always be discerned and it was concluded that the interpretation of these 24-h results was still being obscured by random errors. When the e.m.f.s were linearly correlated against time over a week the gradient obtained was almost always significantly different from zero (t-test 95% confidence limits). This gradient was therefore taken as the best indication of the drift and an extract of the results is shown in Table 3. The potentials of the quinhydrone electrodes drifted at higher rates than those of the best mercury - mercury(1) sulphate electrodes1 for the first 5 weeks of the trial but thereafter the rates were similar.The worst mercury - mercury(1) sulphate electrodes had much higher and less consistent drift rates than the quinhydrone electrodes. Tests with quinhydrone reference electrodes incorporating organic acid solutions (below) showed much higher initial rates of drift than were observed in these tests with quin-hydrone - nitric acid electrodes. The quinhydrone - nitric acid electrodes were prepared about 6 weeks before the long-term tests started and so a period of rapidly changing standard potentials might have been missed. New quinhydrone - nitric acid reference electrodes U and V were prepared and immediately tested.The initial rate of drift of standard potential (which persisted steadily for 4 weeks) was -3.6 mV per 100 h for both electrodes which was much smaller than was observed with organic acid filling solutions (Tables 7 and The use of a gold inner electrode in electrode U made no difference to the rate of drift. This was tried because platinum has been reported7 as catalysing the oxidation of hydroquin-one which might have accounted for at least some of the drift in potential. In a further trial with electrodes in which the filling solutions had (X) or had not (Y) been purged with nitrogen before use the drift rates were again identical. 9). Fluctuation of electrode potentials When mercury - mercury( I) sulphate reference electrodes were tested,’ the standard deviation of the 70 or so 2-h readings was taken as a measure of the constancy of the electrode potential.This was done because those electrodes often showed not a consistent drift but one combined with fairly large and long-lasting fluctuations in potential. Because the potentials of the quinhydrone electrodes drifted con-sistently in one direction without large fluctuations this measure is not necessary to describe the electrode perform-ance. It is however included here for comparison with the mercury - mercury(1) sulphate electrodes. For electrodes A and B after week 1 of the results shown in Table 3 the standard deviation of the 2-h readings was in the range 0.2-0.5 mV throughout the test which is as good as that observed with the best mercury - mercury(1) sulphate electrodes and much better than with the worst.Effect of temperature The temperature in the flow cell was varied in steps over the range 15-35 “C two cycles being completed. The steady e.m.f. at each temperature was noted and the temperature coeffi-cients with respect to the 3 mol 1-1 calomel electrode are shown in Table 4. The results may be converted to the standard hydrogen scale by subtracting 0.4 mV K-1. The potentials of electrodes U and V followed the tempera-ture closely but the coefficient was larger than that calculated from literature values.7 Electrodes A and B however, exhibited anomalous behaviour. Although their potentials first changed rapidly in the expected direction once the temperature had steadied the change in e.m.f.reversed direction so that the starting potential was passed again and Table 3. Drift of standard potential relative to the calomel electrode in nitric acid solution Drift rate during week No. /mV per 100 h Electrode 1 4 8 11 15 18 22 A -1.6* -0.61- -0.81- -0.05 -0.41- -0.7* -0.4t B -1.6* -0.6 -0.81- -0.31 -0.4 -0.7* -0.4T * ICorrelation coefficient1 > 0.9. i. /Correlation coefficient1 B0.7. Table 4. Temperature coefficients (versus 3.0 mol 1-1 calomel electrode) in 0.01 mol 1-1 nitric acid Temperature Standard Electrode mV K-1 mV K-1 coefficient/ deviation*/ U,V . . . . . . . . . . . . -0.63 0.05 A B Peakt . . . . . . . . . . . . -0.53 0.03 Steady? . . . . . . . . . . +0.26 0.03 Calculated . . . .. . . . . . -0.38 -* Standard deviation of the mean of 2 determinations for each electrode. t See text. 272 27 1 > 270 I ’c: E 269 268 0 0 0 0 ~ 0 0 0 0 0 0 0 9 0 0 0 - 0 0 0 -I I I I I I 0 2 4 6 8 1 0 Time/h Fig. 2. Anomalous temperature response of aged electrode (B) in 0.01 rnol 1-l nitric acid medium the final change was opposite to the initial change. Fig. 2 shows the course of such a change for electrode B. The temperature coefficient calculated from the maximum change in the initial direction is reported as the “peak” coefficient in Table 4 and that calculated from the final potential is reported as the “steady” coefficient. The reason for the reversal in potential is not known. Temperature-dependent chemical equilibria involving quin-one hydroquinone or hydrogen ions could be the cause but would not be expected in the quinhydrone-saturated nitric acid solution unless some unknown impurities were present.Any reactions that did occur were apparently reversible as the e.m.f. returned to its original value when the initial tempera-ture was regained. Slow recovery after a change in temperat-ure was observed for sealed calomel reference electrodes containing saturated potassium chloride solutions .* In these electrodes the concentration of dissolved chloride determines the potential so that a slowly attained solubility equilibrium in the interior of the electrode can explain the long recovery times. Because the potential of the quinhydrone electrode is determined by the ratio of quinone and hydroquinone concentrations it should not be affected by the solubility of quinhydrone in a way comparable to the effect of potassium chloride on the calomel electrode ANALYST APRIL 1984 VOL.109 449 Electrodes U and V which behaved as expected were 6-8 weeks old when the temperature coefficients were deter-mined. Electrodes A and B however were over 6 months old. When electrode A was disassembled the inner platinum electrode had a tarry coating. A possible explanation is that the quinone and hydroquinone formed electroactive poly-mers the existence of which might account for the slow reversal in potential either because of changes in solubility (cf. the saturated calomel electrode above) or because the electron-transfer reactions at the electrode are themselves slow.The accumulation of such products would have been much greater in the older electrodes A and B than in U and V. The error produced by making measurements at a temper-ature different from that at which calibration took place can only be assessed if the sensing and reference electrodes are considered simultaneously. With a mercury(1) chloride sens-ing electrode the errors with a new quinhydrone electrode such as V should be smaller than with a mercury - mercury(1) sulphate reference electrode. Compared with aged quin-hydrone electrodes however the mercury - mercury(1) sulphate electrode would be much better because of the time it takes the aged quinhydrone electrode to reach equilibrium after a change in temperature; during this time the precision of analysis of spot samples would be reduced and a spurious drift could be indicated in continuous analysis.With a silver chloride sensing electrode the errors caused by temperature changes would be less with a mercury - mer-cury(1) sulphate reference electrode because its temperature coefficient is closer to that of the sensing electrode. Precision in manual analysis The precision of measuring electrode potentials in flowing streams of constant composition has been discussed above. In manual analysis there is an additional source of variability arising from the need to establish a new liquid junction every time the electrode is removed from one solution rinsed and placed in another solution. The within-batch standard deviations of measurements with a mercury(1) chloride sensing electrode (Ionel SL-01) used in conjunction with both a quinhydrone electrode (B) and a mercury - mercury(1) sulphate electrode (M from the preced-ing paper') were determined at three chloride concentrations (0.1 0.5 and 1 mg 1-1).Measurements were made with five portions of each solution in random order. The procedure was that of Marshall and Midgley.4 The results in Table 5 show that the precision of measure-ments made with the quinhydrone reference electrode is at least as good as that obtained with mercury - mercury(1) sulphate electrodes. Electrodes for Use with Silver Chloride Sensing Electrodes The ionic medium generally usedgJ0 with silver chloride electrodes is an ammonium acetate - acetic acid buffer at pH 4.3-5.0.The electrodes tested were C-F and W in Table 1. The test solution was 0.2 mol 1-1 acetic acid half neutralised with ammonia solution. The same solution saturated with quinhydrone was therefore used as the internal solution of electrodes E F and W. Electrodes C and D which had sodium acetate instead of ammonium acetate were included in order to show whether any part of the e.m.f. changes for electrodes D and F could be attributed to the reaction of ammonia with quinhydrone . 7 Because the working solution was buffered and had a high ionic strength tests on the effect of variations in pH and ionic strength were considered to be unnecessary in contrast to the tests in nitric acid solutions. Because the acetate - quinhydrone electrodes had such a high initial rate of drift of e.m.f.a further series of tests was carried out using an acetate - quinhydrone electrode retained from the earlier tests (F) a freshly prepared acetate -Table 5. Precision in manual analysis Within-batch standard deviation/mV 0.1 mg I-' 0.5 mg 1-1 1 mg 1- * Electrode c1- c1- c1-Quinhydrone* . . . . . . 0.54 0.60 0.16 Mercury-mercury(1) sulphate* 1.04 0.38 0.30 Mercury- mercury(1) sulphate? 0.63 0.93 0.83 degrees of freedom (this work). freedom.4 * Measured with respect to Ionel SL-01 chloride electrode 4 1 Measured with respect to RfiiiEka Selectrode 5 degrees of Table 6. Short-term standard deviations in acetate buffer solution Short-term standard deviation during week No. */mV 1-10 Electrode 1 10 Min .Max. C . . . . . . 0.053 0.053 0.032 0.085 D . . . . . 0.052 0.095 0.052 0.095 E . . . . . . 0.097 0.097 0.032 0.097 F . I . . . . 0.057 0.070 0.032 0.125 * Calculated from 10 l-min readings. Table 7. Drift of standard potential relative to calomel electrode in acetate buffer solution Drift rate during week No./mV per 100 h Electrode 1 2 6 10 C . . . . . . . . . -20* -4.3* -0.31 -0.4t D . . . . . . . -20* -4.7* -0.61 -0.6t E . . . . . . . . . . -21* -4.6* -0.8* -1.6* F . . . . . . . . . . -20* -4.4* -0.2 -0.6* W . . . . . . . . -21* -6.6* -0.7 * /Correlation coefficient1 B0.9. 1 /Correlation coefficient1 >0.7. quinhydrone electrode (G) and phthalate - quinhydrone electrodes (H and J). The last type of electrode was chosen for comparison with the acetate - quinhydrone type because of the success reported by Cooper and Hand11 with a 0.05 mol 1-1 potassium hydrogen phthalate - quinhydrone electrode.Short-term variance The short-term variance in the e.m.f.s was determined from ten consecutive readings 1 min apart. Table 6 shows an extract of the 10 weeks' results. The standard deviations did not differ much from those obtained in 0.01 moll-' nitric acid although the spread of results was smaller. The results did not indicate any advantage for either the sodium acetate filling solution (C and D) or the ammonium acetate solution (E and F). Drift of standard potential In general the same considerations apply as in nitric acid solutions but the rates of drift were much larger. Table 7 gives an extract of the results obtained from a linear correlation of e.m.f.against time. The rates of drift of the quinhydrone electrodes were very high at the start of the test but they decreased progressively until after 5 weeks they were not consistently larger than those of mercury - mercury(1) sulphate electrodes.1 Fig. 3 shows typical traces during the first (G) and eleventh (F) weeks of operation. The drift was not linear during the first week but was so thereafter. The potentials always changed in the same direction whereas for mercury - mercury(1) sulphate electrodes the drifts changed erratically in direction from week to week 450 ANALYST APRIL 1984 VOL. 109 190 1 1 - F 140 24 48 72 96 120 144 Time/h Fig. 3. Changes of e.m.f. with time during first week of test for quinhydrone - acetate (G) and quinhydrone - phthalate (J) elec-trodes and during eleventh week of test for quinhydrone - acetate electrode (F) 145 1 or 2 +.135 r't 0 5 10 15 20 25 30 Time/h Fig. 4. Anomalous temperature response of aged electrode (D) in acetate buffer Table 7 also includes results from a later test of an electrode with a gold inner electrode (W). This was tried because platinum has been reported7 as catalysing the oxidation of hydroquinone which might account for at least some of the drift in potential. As with nitric acid-filled quinhydrone reference electrodes however the nature of the inner electrode did not affect the rate of drift. Fluctuations of electrode potential As above the standard deviation of the 70 or so 2-h readings collected each week was taken as an empirical indication of instability including drift random noise and long-term fluctuations mainly for comparison with mercury - mercury(1) sulphate electrodes.1 An extract of the results (Table 8) shows that the variations were initially much larger than those for nitric acid solutions but that after a few weeks they were about the same. It is inferred that these results are dominated by the rates of drift, which were high at first but decreased with time. Effect of temperature Qualitatively the electrodes showed similar features to those observed in nitric acid medium. The potential of the electrode that was about 6 weeks old (W) followed the temperature changes rapidly and reversibly but the temperature coeffi-cient (-1.10 _+ 0.05 mV K-1) was greater in magnitude than predicted (-0.58 mV K-1) from literature values of the standard electrode potential7 and the characteristics of acetate buffers.12 With older electrodes (3 months) the temperature responses were anomalous as in nitric acid medium but the effects were exaggerated for a decrease in temperature and reduced for an increase (cf. Figs. 2 and 4); the net result is that the temperature cofficients for the peak (-1.1 mV K-1) and steady potentials (-0.25 mV K-1) have the same sign. Table 8. Standard deviation of e.m.f. during 1 week in acetate buffer solution Standard deviation during week No. *JmV Electrode 1 2 6 10 c . . 11 1.8 0.1 0.2 D 11 1.9 0.2 0.3 E . . . . . . . . . . 11 1.9 0.2 0.7 F .. . . . . . . . . 10 1.8 0.1 0.3 . . . . . . . . . . . . . . . . * Calculated from approximately 70 2-h readings in each week. Table 9. Drift of standard potential relative to calomel electrode in acetate buffer solution Drift rate during week No./mV per 100 h Electrode 1 3 7 F . . . . . . . . . . -0.8* -0.8 -0.1 G . . . . . . . . -24T -3.lt -0.9t H . . . . . . . . -19t -1.4 -0.5 J . . . . . . . . . . -21t -3.9* -0.8 * (Correlation coefficient1 > 0.7. t JCorrelation coefficient) > 0.9. Table 10. Standard deviation of e.m.f. during 1 week in acetate buffer solution Standard deviation during week No. */mV Electrode 1 3 7 F . . . . . . . . . . 0.4 0.7 0.2 G . . . . . . . . 11 1.5 0.4 H . . . . . . . . 8.6 1.3 2.0 J . . . .. . . . . . 9.5 2.2 3.0 * Calculated from approximately 70 readings each week. Tests with the phthalate - quinhydrone system Electrodes H and J from Table 1 were used and for comparison both new (G) and old (F) acetate - quinhydrone electrodes were included in the tests. Short-term variance. The results for all electrodes were in the same range as before (standard deviations of 0.05-0.1 mV) the phthalate - quinhydrone electrodes showing no distinctive properties. Drift of standardpotential. Table 9 shows that the results for electrodes (H and J) with phthalate buffer filling solutions were very similar to those for new electrodes with an acetate filling both in this test (G of Table 9) and previously (Table 7). The well aged acetate - quinhydrone electrode (F) con-tinued to show the low drift rates found in the later stages of the tests above.Fig. 3 shows the course of the e.m.f. changes during the early part of the test. The similarity of curves G and J suggests that the composition of the buffer (at a given pH) is of little significance for the constancy of the e.m.f. Fluctuations of electrode potential. As before the standard deviation of 2-h readings collected each week was taken as an indication of the instability of the electrode potential. An extract of the results (Table 10) shows the same trends as before; the largest variations are associated with high rates of drift at the start of the test. The phthalate - quinhydrone electrodes H and J behaved very similarly to acetate -quinhydrone electrodes of the same age (G and also see Table 8) ANALYST APRIL 1984 VOL.109 45 1 Discussion Quinhydrone Reference Electrodes Electrodes with the nitric acid filling solution performed almost as well as the best mercury - mercury(1) sulphate electrodes although their standard potentials drifted more in the first few weeks after preparation (Table 3). Compared with mercury - mercury(1) sulphate electrodes the quinhy-drone reference electrodes were affected more by pH varia-tions in the test solution but less by the ionic strength. The precision of analysis with the two types of electrode was about the same. Over periods of more than a few days the quinhydrone electrodes were preferable to the standard EIL mercury - mercury(1) sulphate electrodes with ground-glass sleeves.The quinhydrone electrodes needed no maintenance com-pared with the frequent refilling required for mercury -mercury(1) sulphate electrodes. At the end of the tests the quinhydrone electrodes were stored for 18 months with their tips immersed in 0.01 mol 1-1 nitric acid solution before their potentials were checked again. As the e.m.f.s were within 2 mV of the readings 18 months previously storage presents no problem, Quinhydrone reference electrodes with organic acid filling solutions showed much higher rates of change of standard potential than either quinhydrone - nitric acid electrodes or mercury - mercury(1) sulphate electrodes particularly over the first few weeks after assembly. Practical application of the electrode would depend on the acceptability of a long period of conditioning.The reasons for the high rates of drift are not clear from the known reactions of quinhydrone.7 Oxygen does not oxidise hydroquinone in acid solution for kinetic rather than ther-modynamic reasons as milder reagents can effect oxidation. If a catalyst were present oxidation might occur but replacing the most obvious catalyst the platinum metal of the inner electrode by a gold electrode had no effect on the drift nor had purging the filling solution of oxygen. In alkaline solutions hydroquinone is oxidised by oxygen to form semiquinone radicals but polymeric products similar to humic acids have also been observed. The reactions depend on both the pH and the buffer system used to maintain that pH and their course in a new system cannot be confidently predicted.To check the stability of quinhydrone - buffer systems outside the electrode 0.005 mol 1-1 solutions of quinhydrone in 0.01 mol 1-1 nitric acid and 0.15 mol 1-1 potassium hydro-gen phthalate media were examined spectrophotometrically. The absorbance at 440 nm was measured every hour for 16 h. The pale yellow nitric acid solution increased in absorbance only slowly (0.0026 h-1) but the phthalate buffer solution changed more rapidly (0.019 h-1) and became redder. The absorbance of the phthalate - quinhydrone solutions increased across the entire visible spectrum the original peak at 440 nm becoming a shoulder. The absorbance increased linearly over the 16-h period for both solutions. The rates of increase were the same (within experimental error) in unstoppered and stoppered cuvettes contrary to the expected result if oxygen were being consumed.Other types of reaction such as those reported with ammonium salts and amino compounds did not appear to be significant because the electrodes prepared with ammonium acetate filling solutions changed at the same rate as those prepared with sodium acetate (compare electrodes C and D with E and F in Table 7). Quinhydrone solutions taken from the electrodes V and W after 6 months were analysed by normal-pulse polarography with a PAR Model 264 analyser. The cathodic wave (quinone reduction) was greatly diminished compared with that for a fresh solution more so for the acetate buffer filling than for the nitric acid filling.The anodic wave (hydroquinone oxidation) changed much less; in nitric acid solution the wave decreased and in the acetate buffer solution the wave increased. These results summarised in Table 11 are in accord with observations by other techniques but also reveal some anomalies. Inserting the polarographically determined concentrations in equation (1) enables the e.m.f. to be calculated and the results agree closely with observation. The direction of the change in e.m.f. is typical of a loss of the oxidised species (quinone) and this accords with the lack of effect from oxygen in both potentiometric and spectrophotometric studies as this would affect the reduced species. The response of the quinhydrone -water system to the loss of quinone is anomalous.In a saturated system the product ( K ) of the quinone and hydroquinone concentrations should be a constant which was calculated to be 3.1 x 10-4 mo12 1-2 from solubility and dissociation data.7 When freshly saturated solutions of quinhydrone were analysed polarographically the value of 3.2 x 10-4 mo12 1-2 was obtained with both 0.01 mol 1-l nitric acid and pH 4.5 acetate buffer solutions but the aged electrode filling solutions gave much lower values. The excess of solid quinhydrone present inside the electrodes failed, therefore to re-equilibrate as expected. The simplest explana-tion for this would have been that lack of mixing inside the electrode prevented saturation being reached and that possibly the deposition of polymeric products7 would further inhibit the dissolution of the quinhydrone.Adding an excess of fresh quinhydrone to the supernatant solution from the electrode however raised the product to 1.3-1.7 X mo12 1-2. It is inferred that in aged solutions the quinone and hydroquinone concentrations are no longer governed by equilibrium with solid quinhydrone. The nature of the equilibria involved is unknown and would require a study beyond the scope of this work. The temperature response of the aged quinhydrone refer-ence electrode remains problematical. The effect is worse (in showing larger changes and slower recoveries) with filling solutions of higher pH ( e . g . in acetate buffer compared with 0.01 mol 1-1 nitric acid). The electrodes with the worst temperature response are also those showing the highest rates of drift.It may be that the products of the reactions discussed above are implicated in the temperature response but no Table 11. Composition of electrode filling solution Quinone/ Electrode filling solution moll-' Freshly saturated . . 1.8 x 10- 2 Agedelectrode (V) . 3.1 x 10-3 Freshlysaturated . . 1.7 x 10-2 Aged electrode (W) . . 4.7 x 10-4 Nitric acid (0.01 moll-') Acetate buffer (pH 4.5) E.m.f. change from fresh solution/mV Hydroquinonei K / moll-' moI2 I-' Observed Calculated 1.8 x 10-2 3 . 2 x 10 - 4 - -1 . 1 X 3.8 X 10-5 17 17 - 1.9 x 10-2 3 . 2 X 10-4 -3.1 x 10-2 1.5 X 10-5 55 5 452 ANALYST APRIL 1984 VOL. 109 reports of this have been found in the literature. In continuous analysis the temperature of the electrode system is almost always controlled and the temperature response of the electrodes is then of only minor importance.Reference Electrodes for Chloride Determinations Mercury ( I ) chloride sensing electrodes For measurements in the nitric acid medium used with mercury(1) chloride electrodes the results in this work do not indicate a clear preference for a quinhydrone reference electrode over a good mercury - mercury(1) sulphate elec-trode.1 The quinhydrone electrode however has the advan-tage of requiring no maintenance and experience of mercury -mercury(1) sulphate electrodes in different laboratories indi-cates the uncertainty of finding a good electrode of this type. Silver chloride sensing electrodes The performance of mercury - mercury(1) sulphate elec-trodes is better in the acetate buffer solutions commonly used with silver chloride electrodes9310 than in nitric acid sohttions,l but the quinhydrone electrodes with acetate filling sc lutions show significant drifts in standard potential.For these conditions the mercury - mercury( I) sulphate electrode would normally be preferred. The convenience of using quinhydrone reference electrodes could best be achieved by changing the ionic medium to nitric acid as with the mercury(1) chloride electrode. Such a medium has been used successfully with silver chloride electrodes. 13 Another possibility is the sulphuric acid medium,l4 which should give results for a quinhydrone -sulphuric acid reference electrode similar to those obtained in this work with nitric acid.Further Developments Although the quinhydrone electrodes offer some advantages over mercury - mercury(1) sulphate electrodes the need for a conditioning period in order to avoid the initial high rates of drift is inconvenient and the temperature response is not fully explained. It is likely that these problems arise from the formation of products from as yet unidentified reactions. A reference electrode system based on a quinone -hydroquinone system of greater chemical stability was, therefore investigated. 15 This work was carried out at the Central Electricity Research Laboratories and is published with the permission of the Central Electricity Generating Board. I thank Mr. C. Gatford for doing the polarographic work. 1. 2. 3. 4. 5 . 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. References Midgley D. Analyst 1984 109. 439. Midgley D. and Torrance K. Analyst 1976 101 833. Brezinski D. P. Anal. Chim. Acta 1982 134 247. Marshall G. B. and Midgley D. Analyst 1978 103 438. Marshall G. B . and Midgley D. Analyst 1979 104 55. Midgley D. Ion Sel. Electrode Rev. 1981 3 43. Janz G. J. and Ives D. J. G. in Ives D. J. G. and Janz G. J. Editors “Reference Electrodes Theory and Practice,” Academic Press New York and London 1961, Chapter 6. Midgley D. and Torrance K. Analyst 1979 104 63. “Operation Manual for the Technicon Trace Chloride Monitor 11,” Technicon Publication No. TICL 002 Technicon Instru-ments Co. Ltd. Basingstoke 1977. Tomlinson K. and Torrance K. Analyst 1977 102 1. Cooper C. A. and Hand P. G. T. J . SOC. Chem. Znd. 1936, 55 341T. Bates R. G. “Determination of pH Theory and Practice,” Second Edition Wiley New York and London 1973. Florence T. M. J . Electroanal. Chem. Znterfac. Electrochem., 1971 31 77. Webber H. M. and Wheeler E. A. CERL Report No. RD/L/R 1369 Leatherhead 1966. Midgley D. Analyst 1984 in the press. NOTE-References 1 and 15 are to Parts I and I11 of this series, Paper A31373 Received October 27th 1983 Accepted November 21st 1983 respectively

ISSN:0003-2654    

DOI:10.1039/AN9840900445

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

ANALYST, APRIL 1984, VOL. 109 Amperometric Determination of Glycerol and Triglycerides Oxygen Electrode Tim A. Kelly* and Gary D. Christian Department of Chemistry, University of Washington, Seattle, Washington 98 195, USA 453 Using an A new triglyceride or glycerol measurement system is described based on the indirect electrochemical monitoring of NADH via its reaction with oxygen by horseradish peroxidase. Derivative amperometric signals due to oxygen depletion provide for a one-point kinetic analysis. Lipase hydrolyses triglycerides to glycerol and glycerol dehydrogenase catalyses the reaction of glycerol with NAD to produce NADH. Serum measurements are demonstrated. Keywords: Glycerol and triglyceride determination; oxygen electrode; lipase; NADH; horseradish peroxidase Serum triglyceride analysis is a crucial test in the clinical diagnosis and classification of hyperlipidaemia.132 Hyper- lipidaemia, primarily known as a coronary risk f a c t ~ r , l ~ ' ? ~ is also associated, directly and indirectly, with many other disorders. 194 Several methods for triglyceride determination have been developed.5-11 Recent determinations are primarily enzymatic and employ hydrolysis of triglycerides to glycerol with lipase .5-8 The generated glycerol is most often quantitatively related to NAD (NADH) by one of two enzymatic pathways, either by equation (l)5J,'JO or by equations (2)-(4) ,638711 Glycerol + NAD+ GDH\ dihydroxyacetone + NADH + H+ (1) Glycerol + ATP --$+ glycerol phosphate + ADP (2) Phosphoenolpyruvate + ADP* pyruvate + ATP (3) Pyruvate + NADH + H+ 5 lactate + NAD+ (4) where GDH represents glycerol dehydrogenase; GK, glycerol kinase; PK, pyruvate kinase; and LDH, lactate dehydro- genase. A proportional change in NADH absorbance at 340 nm5,63Jl or fluorescence at 450 nrn93) is measured.NADH has also been monitored electrochemically in clinical rneth0ds.7~12-14 Electrochemical measurements elimi- nate many of the difficulties encountered in photometric detection, such as the overlap of absorbing species and turbidity. Biamperometric measurements of hexacyanofer- rate(I1) in a flow system were used to monitor NADH by the following reaction7J2: Fe(CN)63- + NADH + H + A F ~ ( C N ) ~ ~ - + NAD+ ( 5 ) NADH may be indirectly monitored amperometricallyl5 using a membrane oxygen electrode16 according to the reaction14 Diaphorase NADH + H+ + 402% Mn2+ NAD+ + H20 (6) where HRP represents horseradish peroxidase.In this work, triglycerides were hydrolysed with lipase. The glycerol produced was oxidised with NAD in the presence of glycerol dehydrogenase. The NADH produced was then reoxidised by dissolved oxygen in the presence of horseradish peroxidase. The rate of decrease in the oxygen concentration was monitored amperometrically . Under the appropriate conditions, the maximum rate of change in the oxygen concentration was proportional to the initial triglyceride concentration. Preliminary studies with serum illustrate the feasibility of serum assays. * Present address: Department of Chemistry, Pacific Lutheran University, Tacoma, WA 98447, USA. Experimental Reagents All chemicals were of analytical-reagent grade unless other- wise specified.Glycine (ICN Pharmaceuticals), 0.1 M, and Tris (Sigma), 0.05 M, were prepared with de-ionised, distilled water and adjusted to a pH between 7 and 9 with 1 M HC1 and 1 M KOH solution. All solutions were freshly prepared daily in these buffers. Olive oil (0-1500), tristearin (T-6503), lipase (L- 3126, 56 U mg-I), GDH (G-3755, 2.68 U mg-I), NAD (N-7004) and HRP (P-8250, 150 U mg-1) were all obtained from Sigma. Apparatus A glucose analyser (Beckman Instruments) was used for all tests in conjunction with a Linear Instruments dual-pen strip-chart recorder. One pen input was connected to the d.c. amplifier potentiometer output for the direct signal, while the derivative signal was taken from the derivative amplifier potentiometer.Helena Laboratories Quickpettes (5-50 and 50-250 p1) and Beckman Lancer micropipettes (5, 10 and 50 pl) were used for make-up and injection of the solutions. Procedure For the determination of glycerol, the following solutions were sequentially added to the reaction cell: 750 pl of buffer and 50 pl each of NH4+ (2.4 M NH,Cl), Mn2+ (4.8 mM MnC12.4H20), GDH (0.65 U per run) and NAD (0.166 M) solutions. The NAD solution was kept on ice until ready for use. The standard or sample (50 pl) was then added and allowed to incubate at 33 "C for 300 s with stirring. Glycerol standards were prepared by serial dilution with buffer solutions or serum - buffer (1 + 4) solutions. Then 50 pl of HRP (30 U per run) solution was injected as the trigger and the rate of oxygen consumption was recorded.In determining triglyceride levels, two separate procedures were followed. A known amount of triglyceride was suspen- ded in buffer or 1 + 4 serum - buffer solution (with heating to facilitate the suspension) immediately before hydrolysis. In the first procedure, 750 pl of standard or sample suspension were added to a test-tube, followed by 250 pl of lipase suspension (500 U per run) and 50 pl of Ca2+ solution (1.0 M CaC12.2H20). The mixture was allowed to incubate at 37 "C for 15 min with occasional shaking. Aliquots of this solution (50 p1) were then analysed for glycerol as described above. In the second procedure, a 500-pi portion of buffer solution was added to the reaction chamber of the analyser along with454 ANALYST, APRIL 1984, VOL.109 Table 1. Simplex optimisation of the GDH - Vertex 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 NH,+/ 0.02 0.17 0.15 0.14 0.14 0.14 0.14 0.15 0.15 0.14 0.11 0.15 0.14 0.14 0.12 0.12 0.13 0.16 0.12 M HRP system Mn2+1 0.25 0.25 0.36 0.26 0.26 0.26 0.26 0.29 0.29 0.26 0.33 0.35 0.31 0.26 0.31 0.33 0.31 0.29 0.24 r n M GDHI U 0.50 0.50 0.50 0.80 0.50 0.50 0.50 0.20 0.40 0.50 0.40 0.40 0.50 0.40 0.70 0.40 0.50 0.60 0.65 NADI mg 5.0 5 .O 5 .O 4.5 8.3 5.2 5.2 6.9 2.2 8.3 5.1 5.1 8.4 7.5 5.5 5.0 2.6 5.4 5.5 HRP/ U 14 14 14 13 13 22 14 18 20 13 21 23 19 26 24 29 29 30 30 Incubation time/ min 4.0 4.0 4.0 4.0 4.0 4.0 5.5 4.5 4.7 4.0 4.9 5.2 5.2 5.8 5.7 4.8 4.9 5.2 5.0 Response, arbitrary units 240 235 250 205 210 340 270 255 225 210 315 320 285 360 390 345 370 405 415 250 pl of lipase suspension, 50 p1 each of NH4+ and Mn2+ solutions, 5 pl of Ca2+ solution, and finally 50 pl of standard or sample.This mixture was allowed to react for 10 min then 50 pl each of the GDH and NAD solutions were added and incubation was continued for another 300 s to produce NADH before the HRP trigger injection. Results and Discussion The reaction sequence was as follows: Triglyceride + 3H20 glycerol + 3 fatty acids (7) Glycerol + N A D + z dihydroxyacetone + NADH + H+ (8) NH4+ NADH + H+ + B025 NAD+ + H20 (9) The particular ions (NH4+, Mn2+ and Ca2+) associated with each reaction have been found to be the principal activators for the enzymes. 17 Other potentially influential compounds will be discussed later.Glycerol Determination Reactions (8) and (9) were performed simultaneously in the determination of glycerol. Simplex optimisation18 was em- ployed in determining optimum reagent concentrations in the two-step glycerol measurement system. The results are summarised in Table 1. Glycerol dehydrogenase has been reported to exhibit optimum activity at pH 8-10 and at lo-3-10-1 M NH4+, as well as being inhibited by Na+ and solutions of high ionic ~trength.19~20 In this work the following optimum reagent concentrations were found for reaction (8) [when combined with reaction (9)]: NH4+ = 0.12 M; and GDH = 0.65 U per run (Table 1). High ionic strength and Na+ concentration were avoided by diluting the serum samples. The variables of GDH activity and NH4+ concentration were also optimised individu- ally and independently (Figs.1 and 2). Differences from the simplex optimisation method were probably due to intervari- able dependencies. An optimum pH of 8.0 for the two-step system (Fig. 3) was the same as that previously reported for reaction (9).*4 At pH values above 8.0, Mn2+ tends to precipitate, causing irreproducible reaction rates for reaction (9).14 Of the two buffers tested, Tris resulted in signals 40% larger than those using glycine buffer. I 1 I 0 0.1 0.2 0.3 [NH4+]/mol I-' Fig. 1. Effect of NH4+ concentration on the GDH - HRP system 0.5 1 .o GDHl U Fig. 2. Effect of GDH activity on the GDH - HRP system The indicator (trigger) reaction (9) was that developed by Cheng and Christian.14 Considerable variation in the reaction rate and reproducibility has been found with respect to substrates, reaction conditions and possible mechanisms.1421 It was necessary that reaction conditions be slightly alkaline for adequate reproducibility. 14 Similar variation in reagent dependence was seen in this study, with the exception of the lack of an absolute requirement for Mn2+ (although the reaction was accelerated in the presence of Mn2+). In coupling the peroxidase reaction to the GDH reaction, optimumANALYST, APRIL 1984, VOL. 109 455 v) *-' 'c 250 3 > 2! 5 200 tu X m 2 150 9 0 - N -0 I 100 1 0 7 8 PH Fig. 3. Effect of pH on the GDH - HRP system 300 v) S 3 c .- L- 2" 200 -f! E .- X c F! - 100 -0 I I 0 I I I -5 -4 -3 -2 Log([glyceroll/~) Fig. 4. Semilogarithmic plot of the calibration graph for the determination of glycerol in standard aqueous samples reagent concentrations for the second reaction were Mn2+ = 0.24 mM and HRP = 30 U per run (Table 1).These concentrations were higher than previously employed. l4,22 It was observed that the GDH reaction (8) proceeded much more slowly than reaction (9), which adversely affected the analysis time and the sensitivity. This problem was approached by firstly increasing the NAD concentration (0.01 M NAD or 5.5 mg per run, from simplex optimisation), secondly, maintaining the reaction at 33 "C and thirdly, incubating for 5 min. In incubating, however, feedback inhibition by both NADH (competitively) and dihydroxyace- tone (uncompetitively) takes place,20 which resulted in a non-linear calibration graph (approximately a logarithmic response over wide concentration ranges, as shown in Fig.4) and reduced sensitivity compared with that of the NADH - oxygen reaction14 (detection limit 5 x 10-5 M glycerol for aqueous standards and 1 X 10-4 M glycerol for serum samples) (Fig. 5). A within-run precision of 3% was obtained. This could be reduced further by combining reagents in a single stock solution, thus decreasing the number of required pipette measurements. Note in Fig. 5 that the response for glycerol is reduced in serum, necessitating calibration using serum controls. I 1 I I 2 4 6 8 1 0 glycerol]/^ x lo4 Fig. 5. Calibration gra hs for glycerol determined in (A) standard aqueous samples and (€37 serum control samples 6 8 10 PH Fig. 6. Effect of pH on the lipase hydrolysis of triglycerides (by the external hydrolysis procedure) Triglyceride Determination Lipase catalyses the hydrolysis of triglycerides to glycerol and fatty acids.The hydrolysis of triglycerides [reaction (7)] was performed either in series with (internal hydrolysis) or separate from (external hydrolysis) the glycerol reaction. Complete hydrolysis is slow and not substrate specific.19J) The optimum pH and effect of several activators such as Ca2+, Na+, bile salts and emulsifiers are reported to vary signifi- cantly with different isoenzymes.23-25 With the external hydrolysis scheme, optimum activity for the hog pancreas isoenzyme was seen at pH 7.5 (Fig. 6). Ca2+ and bile salt activation of the lipase was obtained using 0.05 M Ca2+ externally or 0.005 M Ca2+ internally and 0.05% mlV sodium taurocholate (bile salt).Na+ and gum arabic (emulsi- fier) were found to have no significant effect on hydrolysis. a-Chymotrypsin, a non-specific esterase known to increase the rate of hydrolysis,6 was observed to have no effect in these studies. Of the two standard substrates used, olive oil was found to be more reactive and more convenient than solid tristearin. Olive oil was determinable down to 25 mg dl-1 (0.3 mM) (Fig. 7) with a relative standard deviation of 8%. Analysis by internal hydrolysis was faster and more sensitive than by external hydrolysis, by a factor of two in both456 80 Lo C - .- 2 60 I -E! Y .- X 40 +.I 9 9 - N U I - 20 I I 1 I I 0 100 200 300 400 500 Triglyceride/mg per 100 ml Fig.7. Calibration graph for the determination of triglycerides in standard aqueous samples (by the external hydrolysis procedure) instances. Sequential samples could be analysed more rapidly following external hydrolysis? however. Calibration must be carried out concurrently with sample analyses. The presence of lipase influenced the behaviour of reaction (9), at times giving a significant blank or slightly inhibiting HRP, and at other times substantially activating the catalysis. It was necessary to recharge the electrode and change the membrane on a weekly basis. It appears as though the presence of oils and suspensions has adverse surface effects on the PTFE membrane. Conclusions A new reaction sequence is proposed for the determination of glycerol and triglycerides.It utilises selective enzyme catalysis coupled with specific kinetic electrochemical detection. Opti- mum reagent concentrations and solution conditions were evaluated. Results demonstrate the feasibility of this new ANALYST. APRIL 1984, VOL. 109 approach to the difficult problem of effective triglyceride analysis and its application to serum. 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 Fredrickson, D. S . , Modern Concepts Cardiovasc. Dis., 1972, 41, 31. Fredrickson, D . S., Levy, R. I., and Lees, R. S . , N . Engl. J . Med., 1967,276,148. Schatz, E. J., J. Am. Med. Assoc., 1969,210 (4), 701. Ellefson, R. D., and Caray, W. T., in Tietz, N. W., Editor, “Fundamentals of Clinical Chemistry,” W.B. Saunders, Philadelphia, 1976, p. 496. Grossman, S. H., Mollo, E., and Ertingshausen, G., Clin. Chem., 1976,22,1310. BucoIo, G., and David, H., Clin. Chem., 1973,19,476. Attiyat, A. S., and Christian, G. D., Anal. Chim. Acta, 1979, 106,225. Lehnus, G., and Smith, L., Clin. Chem., 1978,24,27. Davidson, M. B., and Karjala, R.,J. Lipid Res., 1970.11,609. Gore, M. G., Anal. Biochem., 1976,75,604. Lauderdale, V. R., US Pat., 4 309 502, Jan. 1982. Thomas, L. C., and Christian, G. D., Anal. Chim. Acra, 1975, 78,271. Thomas, L. C., and Christian, G. D., Anal. Chim. Acta, 1976, 82,265. Cheng, F. S., and Christian, G. D., Anal. Chem., 1977, 49, 1785. Christian, G. D.,Adv. Biomed. Eng. Med. Phys., 1971,4,95. Clark, L. C., and Lyons, C., Ann. N . Y . Acad. Sci., 1962, 102, 29. Decker, L. A . , Editor, “Worthington Enzyme Manual,” Worthington Biochemical Corp., Freehold, NJ, 1977. Lond, D. E.,Anal. Chim. Acta, 1969,46,193. Strickland, J. E., and Miller, 0. N., Biochim. Biophys. Acta, 1968, 159, 221. McGregor, W. G., Phillips, J., and Suelter, C. H., J. Biol. Chem., 1974,249,3132. Akazawa, T., and Conn, E. E., J. Biol. Chem., 1958,232,403. Cheng, F. S., andChristian, G. D., Clin. Chem., 1978,24,621. Desnuelle, P., and Savary, P., J. Lipid Res., 1963,4,369. Wills, E. D.,Adv. Lipid Res., 1965,3, 197. Schifreen, R. S . , and Carr, P. W., Anal. Lett., 1979, 12 (Bl), 47. Paper A31247 Received August 8th, 1983 Accepted October 11 th, 1983

ISSN:0003-2654    

DOI:10.1039/AN9840900453

出版商:RSC    

年代:1984

数据来源: RSC