ISSN:0003-2654    

DOI:10.1039/AN98914FX025

出版商:RSC    

年代:1989

数据来源: RSC

ISSN:0003-2654    

DOI:10.1039/AN98914BX027

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST, JULY 1989, VOL. 114 757 Editorial ~ Conference Report: 1989 Pittsburgh Conference and Exposition on Analytical Chemistry and Applied Spectroscopy March 6th-10th 1989, Atlanta, Georgia, USA The Pittsburgh Conterence, or to give the full title “The Pittsburgh Conference and Exposition on Analytical Che- mistry and Applied Spectroscopy” (Pittcon), which celebrated its 40th birthday in March 1989, is truly among the premier international meetings for the analytical chemist. In recent years it has been recognised by the Royal Society of Chemistry as one of the “key” meeting points for academics, industrial and instrumentation scientists and efforts have been made to have a strong presence there, to “fly the flag” of RSC Analytical Chemistry. In my role as United States Regional advisory Editor for The Analyst, it provides an opportunity to spend time with the UK editorial staff and also those from the RSC Information Services who nobly staff the “RSC booth” for four long days (and long they certainly are!).This year Atlanta, Georgia (the conference last met in Pittsburgh more than 20 years ago), provided an excellent venue to initiate Janet Dean. Editor of The Analyst, into the “delights” of contemporary US life. She had able tuition in this respect from Judith Egan, Editorial Manager of the Analytical Journals. They and other RSC staff hosted the highly regarded “RSC Reception” at which more than 250 guests and friends spent a very pleasant evening. Many now rate this as one of the ‘*highlight” events of the conference and it is certainly an excellent advertisement for the Society! I n many ways, the Pittsburgh Conference typifies the average British view of US scientific events; certainly it is a gigantic enterprise and no place for the faint-hearted or those in awe of crowds! However, to mitigate the sheer size, it is universally agreed that this is among the best organised exhibition/conference event anywhere (it really has to be so!).It is organised by volunteers, consisting of members of the two co-sponsoring groups, The Society for Analytical Chemists of Pittsburgh (SACP) and the Spectroscopy Society of Pittsburgh (SSP). Many other organisations stand in awe of their accumulated experience. Pittcon is many different things to different people. It is The analytical instrumentation show having in 1989,850 exhibitors and 2556 booth spaces.The wise know the frustration in attempting to contact people in the instrumentation business in the weeks before the meeting as they work, often frenetically, to prepare for the show, particularlq when new instruments are due to be unveiled. It regularly draws the highest number of conferees (26 741 this year) for any chemistry meeting in the USA and in recent years it has become increasingly recognised for technical presentations (1324 papers and 210 posters in 1989). While many papers are presented by instrument manufacturers, the scientific committee takes care to minimise the commercial nature of these and to ensure sound scientific content; this is often, although not always, successful. The quality of the technical sessions was emphasised this year by no less than 37 “Invited Speaker” symposia. Of course, the sheer magnitude of the programme can be daunting.At one time this year there were 17 simultaneous technical sessions in progress. The conferee must learn to be very selective, to attend just those papers that are of major interest, and most importantly to plan their sequence so that they can be reached in time (the Georgia World Conference Center covers a total of 15 acres). There is always a range of high-quality panel discussions, special seminars, etc., encompassing topics of wider interest. This year the annual ACS-sponsored conference breakfast discussion featured the Riopharmaceutical Industry in the 1990s. A special luncheon seminar and discussion held in collaboration with the US National Science Foundation (NSF) focused on the continuing under-representation of women in the science and engineering disciplines.The importance of a wider outreach was stressed as the week of the Conference was made “Science Week” in Atlanta and the state of Georgia and days were set aside as “Science Teachers” and “Science Students” days at the secondary school level. The conference always has good media news coverage wherever it is located. The conference also provides a flourishing “employment agency ,*’ to match job candidates and employer representa- tives. This year there were more than 850 each of candidate$ and openings. There can be little doubt of the international stature of Pittcon. This year there were 2518 overseas visitors.In addition to technical participants, many of these were associated with exhibitors from instrument companies. For some years, the British Consulate-General has co-ordinated exhibition facilities for smaller t o medium sized British companies who wish to exhibit at Pittcon; this year some 20 such availed themselves of this facility and formed a “UK aisle” in the Exhibition Hall. In contrast, some UK companies are now firmly established in the US marketplace with subsidiary enterprises or major North American marketing commitments. Pittcon is certainly the strongest opportunity for them in the analytical chemical market place. There was a time when academic analytical chemists tended to downrate Pittcon as merely an instrument show. Today, it is much more than this.We encourage our graduate (postgradu- ate) students to attend and often to present papers. For them the experience of the conference is extremely valuable. They can emulate their academic mentors and “cover” the exhibi- tion, wishing for all of the instrumentation that is unlikely ever to find its way into academic laboratories for purely economic reasons. Manufacturers know however that this is their next generation of customers and advisedly recognise them as such! Pittcon is now one of the best “advanced academic teaching tools that exist.” Academic analytical chemistry is in a “holding pattern” in colleges and universities during Pittcon, with a high proportion of faculty members and staff at the meeting. However, the return on the educational investment involved is very high. To pick out specific features of the instrumentation exhibi- tion as being of exceptional general interest is virtually impossible; individuals must be biased by their own interests.Some major companies have exhibits covering 4&50 booth spaces, but sometimes the small entrepreneur can create as much personal excitement. To me the areas of major interest ranged from the introduction of a major GC - microwave plasma emission spectral detector system and a viable HPLC/FID instrument (by a UK company), to the rapidly developing field of capillary zone electrophoresis; and also from the increased sophistication and availability of LC - MS758 ANALYST, JULY 1989, VOL. 114 instrumentation to the burgeoning fields of bioanalysis and biotechnology. One thing is certain, there is something for everyone and usually in a competitive sense as well. Scientific meetings and symposia of all kinds and sizes are a part of life in our expansive discipline. They provide a forum for broadening our contacts in all aspects of science; meetings such as Pittcon take a large amount of energy just to survive, but the effort is definitely worthwhile. To those in Europe who have not experienced it, I have this advice: if you ever have the opportunity to attend Pittcon, take it! Peter C. Uden US Regional Advisory Editor University of Massachusetts Amherst, MA, USA

ISSN:0003-2654    

DOI:10.1039/AN9891400757

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST. JULY 1989. VOL. 114 759 Denuder Tubes for Sampling of Gaseous Species A Review Zulfiqur Ali, C. L. Paul Thomas and John F. Alder Department of Instrumentation and Analytical Science, UMlST, P.O. Box 88, Manchester M60 IQD, UK Summary of Contents introduction Description of operation Theory Cyl i nd rica I systems Diffusion coefficients An nu lar systems Development of the Gormley I Kennedy solution Assessment of the empirically derived annular equations Organic atmospheric species Ammonia species Nitrate species Hydrogen chloride Sulphur species Mu It i-com ponent sa mpi i ng Applications to analysis Conclusions References Keywords: Denuder tubes; atmospheric acidity; environmental monitoring; gaseous sampling Introduction Denuder tubes were first considered on a theoretical basis in the 1890s during a study on the diffusion of ions into gases.’ Later, in the 1930s.studies of atmospheric condensation nuclei revealed significant particulate losses from gas flows in sample pipelines and, consequently, the diffusional removal of particulates in both cylindrical and rectangular section pipelines was considered .I Subsequent measurements of the diffusional proper ties of micro-particulates were made with multi-tubular assemblies and diffusion batteries, consisting of several flat plates with narrow gaps between them.3 In an appendix to this work. a mathematical treatment was pre- sented of the diffusion from a stream flowing through a thin rectangular tube. Diffusion measurements based on these methods con- tinued, and in 1949 Gormley and Kennedy4 derived a solution describing diffusion from a stream flowing through a cylin- drical tube, originally proposed by Townsend.1 Diffusion Coefficients for sulphur trioxide and ammonia were deter- mined by trapping these gases chemically in a glass tube lined with oleic acid or copper sulphate impregnated paper, re\pectively.s L,aminar flow subduction zones were incorpor- ated at the inlet to these devices, and the diffusion coefficients of sulphur trioxide and ammonia were measured by determin- ing the mixing ratios of these gases within the tubing. A study undertaken in the early 1960s was concerned with the effects of fluorides in the atmosphere.6 An annular unit was used to separate gaseous fluoride species from their particulate counterparts.The unit consisted of three concen- tric cylinders coated with sodium hydrogen carbonate or alutninium. No theoretical basis was given for this design, and annular systems were not studied again until the early 1980s. Following on from this, an analytical technique using an ”aerosol-passing gac adsorber” was described .7 Fabricated from a series of glass tubes lined with potassium hydroxide and magnesium perchlorate, it was employed to remove water vapour from an aerosol sampling system for a hydrogen flame chemihminescence detector used for aqueous sulphate ion determination. The removal of sulphur dioxide from a laminar flow gas stream by a diffusion denuder coated with lead(1V) oxide was shown to conform to the Gormley - Kennedy equation.8 Sulphur dioxide at a concentration of 1.5 p.p.m.in air was passed through the tube, which was then cut into 0.5-in sections and analysed radiochemically . The diffusion coeffi- cients for sulphur dioxide at 1 atm were found to be 0.117, 0.132 and 0.146 cm2 5-1 at 0, 20 and 37 “C, respectively. Lead-based sulphur dioxide denuders were then used to remove sulphur-containing gaseous interferences quantita- tively for measurements of the sulphate ion content in airborne aerosols.g--’2 The use of a heated sample tube to effect thermochemical changes in incoming sampled ammo- nium sulphates was first reported in 1978.1’ Other gases, notably ammonia, were removed from environmental gas samples by arrays of denuder systems,13 which allowed the use of higher gas flow-rates than previously encountered.However, it was not until 1979 that the role of the denuder tube was reversed and it was used as a gas collection device rather than an instrument to denude a sample of certain gaseous components.14 In this work, Ferm achieved a detec- tion limit of 0.4 nmol m-3 of ammonia in air (0.01 p.p.b.) after a 24-h sample period. Interference from particulate ammonium salts at “extremely high” (400 nmol m-3) concen- trations was of the same order as the detection limit. Description of Operation For a denuder tube to function, a series of criteria must be met.760 ANALYST, JULY 1989, VOL. 114 The removal of the analyte must have no global effect on the sample, the gas flow must be stable and laminar and the viscosity and temperature distribution within the gas flow must be homogeneous. Any axial diffusion of the sampled gas must be insignificant in comparison to the sample flow, and the collection surfaces should behave as infinitely large and perfect sinks towards the analyte.Adsorbate species should be neither created nor destroyed in the gas phase within the denuder tube. Schematic diagrams of the two types of denuder unit in common use are shown in Fig. 1. Air is drawn through either the centre of the tube, or the annulus, at a rate such that its flow is laminar [Reynold’s number (Re) < 21001.15 Laminar flow is achieved a short distance from the inlet. In a tube of diameter d, the length of this subduction zone can be calculated as follows: 1 = 0.07dRe . . . . * * (1) where 1 is the length of the laminar flow subduction zone.The establishment of laminar flow is important as this ensures that radial mixing can only take place via diffusion- based processes. It is essential that areas of collection surfaces where turbulent flow exists, and particulates therefore im- pinge on, do not contribute towards the analysis. Often a PTFE former or inlet is used to establish laminar flow prior to the collection surfaces. Molecular species diffuse through the laminar flow to the active collection surfaces, where they undergo irreversible adsorption or chemical bonding and are removed from the gas stream. Particulates with their much lower diffusion velocities cannot migrate to the walls during their residence time within the unit and hence, provided the above conditions are met, do not contribute to the final measurement. The deposition of particulates as a result of gravitational sedimentation would disrupt the performance of the unit.This problem is negated by using the tube in a vertical orientation, and employing cyclones and virtual impactors to remove large pieces of airborne detritus from the sample also ensures the effectiveness of this method. Theory Cylindrical Systems The behaviour of trace amounts of gases in a denuder system can be described in terms of Fickian diffusion. The expression in common usage was derived by Gormley and Kennedy.4 For a cylindrical denuder, their solution can be summarised as follows. For a laminar flow, the velocity of the gas at any point in the denuder tube (Fig. 1) is given by v = F ( R ~ - r 2 y 2 ~ ~ 4 .. . . . . (2) Sample flow Cylindrical denuder tube ___) - Laminar flow subduction zone Reactive coating Tube wal I Annular denuder tube - Laminar flow subduction zone I inner cylinder Reactive coating Outer tube wall Fig. 1. tubes Schematic diagrams of the cylindrical and annular denuder where V is the velocity of the gas flow at a radial displacement r from the axis, F is the sampled gas flow-rate and R is the radius of the tube. The diffusive motion of the gaseous analyte within the unit is resolved into its Cartesian components. The equations of motion that describe the transport of analyte parallel to the laminar flow and the denuder axis contain the appropriate correction : * (3) dP,Dldy = -P,v . . . . . - (4) dP,Dldx = -P,u . . . . dP,D/dz = -P,(w - V ) .. . . ( 5 ) where D is the diffusion coefficient of the analyte, Pa is the partial pressure of the analyte and u,v and w are the diffusive velocities of the analyte with respect to the x, y and z axes, respectively. The parameters u, v and w are then substituted into the continuity equation giving where d d d dx dY dz -(Pau) + -(Pa.) + -(Paw) = 0 . . * 1 1 d D dz V P , -- - ( P a y ) = 0 , . . . . , du dv dw v=-+- +- dx dy dz (7) A transfer from Cartesian to cylindrical co-ordinates pro- vides the following expression: where 0 is the angle of displacement from the x axis in the x - y plane. It is then assumed that dPa/dO = 0 for reasons of symmetry and d2Pa/dz2 = 0 for V >> w. The resulting equation forms the basis of the Gormley - Kennedy solution d’Pa 1 d P 2F dP, dr2 r dr z R ~ D dz -+-.>--.(R2- r2) --= 0 . . (9) for which the boundary condition is set such that PajR = 0 . . . . . . . . (no) The mean radial concentration C at a length Z along the tube axis is proportional to ( , J R Pa2n Vr.dr . . . . . . (1 1) The mixing ratio of the analyte at length 2 can be expressed as C / C ~ = [4/(R4P,,o)] oJ’Pa(R’-r2)r.dr . . (12) where Pa is the solution to equation (9), Pao the ambient vapour pressure of the analyte, co the ambient concentration of the analyte and C the mean radial concentration of the analyte. Gormley and Kennedy provided the solution to equation (12) as a series: dc0 = 0.819exp( -7.3144,) + 0.0975exp( -44.6AJ + where, at standard temperature and pressure, 0.0325exp( -212A,) .. . . . . (13) A, = 3tDoZ/2F . . . . . . (14) (subscript c refers to the cylindrical system). For most sampling applications, where A, d 0.05, only the first term of this series will be significant and equation (13) can be approximated to dco = 0.82exp(-7.3AC) . . . . (15) From these expressions it can be seen that the collection efficiency of a denuder tube increases with length, and alsoANALYST, JULY 1585, VOL. 114 A B C D E G H 761 I with increasing values for the diffusion coefficient of the analyte. The collection efficiency is enhanced further by decreasing the radius of the tube and reducing the mean axial velocity of the sampled gas. Davieslh has indicated that low values for the kinematic viscosity of the sample also lead to increased sampling efficiency; this is underlined further by Ferm’slA treatment of the Gormley - Kennedy solution.The performance of a denuder tube can be seen to be affected by variables such as pressure, temperature and relative humidity. Such detail is often not reported in experimental studies, and no mention has been made of procedures that enable data obtained under different environmental conditions to be corrected. In some situations ( i e . , at high altitude) these factors will become significant. Diffusion Coefficients A number of studies involved analysis of the contents of segmented denuder tubes to determine the diffusion coeffi- cient of the analyte, see Fig. 2. The sampling efficiency of each segment is defined by E = number o f moles adsorbed/ number of moles sampled .. . . (16) For the first segment E = aic . . . . . . . . (17) and for the second segment E = bi(c - a ) . . . . . . (18) where c is the number of moles sampled, a the number of moles adsorbed in the first segment and b the number of moles adsorbed in the second segment. Combining equations (17) and (18) gives E = (a - b)/a . . . . . . (19) hence CICO = bia . . . . . . (20) Repeating this analysis for further segments yields a series of values for A, with increasing values of 2. Then a plot of A, against Z is a straight line of slope nDd2F. Hence the diffusion coefficient Do can be determined. The values obtained for the diffusion coefficient then allow elucidation of the analyte form in the sampled gas. Hence the diffusive mechanism of the sampling process can be verified, and information about any hydration or other molecular aggregation of the analyte obtained.Annular Systems The Gormley - Kennedy solution discussed here [equation (13)] does not apply to annular systems. Possanzini et a1.17 proposed an empirically modified form: cIco = Aexp(- aAa) . . . . (21) where A and a are experimentally determined constants. Fig. 2. Schematic diagram showing the denuder divided into a series of equal sections for the measurement of diffusion coefficients. d c g = hla = c/b = . . . . = i/h. where a = number of moles in section A. b = number of moles in se’ction B, etc. Finally, d c g = (i + h + g + e)/ (a + b + c + d ) and d c g = 0.82exp(-7.3AC), A, = 0071 Z12F The form of Aa (subscript a refers to the annular system) was also modified to incorporate the equivalent diameter of the annular channel.This was defined as four times the hydraulic radius, which in its turn was defined as the ratio of the cross-sectional area of the annular channel to its perimeter. The final form of Aa was given as Aa=-(----) zDOZ d,+d, . . . . 2F d l - d z where d l is the internal diameter of the external cylinder and d2 the external diameter of the internal cylinder. Values for A and a were calculated by determining A, and the mixing ratio €or three denuder tubes of different sizes, hence giving the following expression: dcO = 0.82 f 0.1exp[(-11.27 k 0.61)AJ . . (23) The parameters A and a were determined with test atmos- pheres in the range 0.29-1.45 mg m-3 of sulphur dioxide in air and the relative humidity between 60 and 80%.The sampling rate ranged from 0.072 to 2.4 m3 h - 1 ; the temperature at which these experiments were run was not given. The determination of A and a could have been accomplished by sectional analysis of the denuder units, but no indication was given as to why such an approach was not adopted, despite its inherent precision. It should be noted that the reliability of the results obtained by this method is dependent on the accuracy of the value assigned to the diffusion coefficient of the analyte under study. As the hydraulic radius of the annular system increased, the velocity profile of the sampled gas tended towards that obtained between two infinite parallel sheets, i.e., A and a tended towards the limiting values of a parallel-plate system as the hydraulic radius increased. No reference was made to previous work, which investigated diffusional deposition in rectangular channels .3 A comparison between equations (15) and (23) revealed the advantages of the annular design over its cylindrical counterpart, For an equivalent sampling efficiency the annular system could be shorter and more compact, operate at higher sample flow-rates and have a larger sampling capacity, or a combination of any of these factors.These claims were subsequently verified in a comparative study,l8 which concluded that annular units were a substantial improvement over all other denuder tube designs. Development of the Gormley - Kennedy Solution Some of the assumptions underlying the Gormley - Kennedy solution have been challenged, and, subsequently, alterations to equation (13) have been proposed.Braman er al. 19 pointed out that the surfaces of a denuder tube would be depleted with increasing sample volume. The corollary to this was that the effective length of the denuder tube decreased with increasing sample volume. If the denuder was being operated at a constant sampling rate, then the collection efficiency would decrease with increasing sampling time. By assuming pseudo-first-order adsorption kinetics, it was proposed that the effective length of a denuder tube at any time after the commencement of sampling could be described Z, = Z,exp(-kt) . . . . . . (24) by where 2, is the effective length of the denuder at time t , ZO the length of the denuder at t = 0, k the depletion constant (feed rateitube capacity) and t the sample time.The sample size dependency of the length of the denuder was incorporated in the original expression, equation (15), which was modified to762 ANALYST. JLJLY 1989, VOL. 113 The significance of the increased precision obtained by this approach is dependent on the experimental conditions. Large depletion rate constants, due to high analyte concentrations or low surface capacities would result in measurable reductions in the effective length of the denuder. High analyte concentra- tions do not require long sampling times however, and an appropriate choice of reagent for the tube coating ensures an adequate surface capacity. For the analysis of sulphur dioxide in air using a denuder tube,20 the prediction given by equation (25) for the concentration of sulphur dioxide in the tube exhaust, after sampling for 17 h (a sample volume of 0.683 m3) is 0.32 pg m-3.This can be compared to the value of 0.29 pg m-3 obtained from equation (15). The measured analyte concentration was 30.5 pg m-3 and the difference between the two predictions is not significant. Measurements of the diffusion coefficients of gaseous species obtained by the denuder tube method would be more accurate if equation (25) were applied, but the practical advantages that could be carried over to the trace analysis of gas are limited. Other modifications to the Gormley - Kennedy equation have been suggested by Murphy and Fahey.21 The perfect sink criterion was challenged and the boundary condition that arises from it [equation (lo)] was shown to be an approxima- tion. Instead, a constant reaction probability y was proposed.This alternative treatment, based in part on previous work concerning haemodialysis,22 resulted in a new equation for the mixing ratio: dco = B,exp( -A,zZ*) . . . . (26) where Z* is a function of 2 [see equation (3011. As for its predecessors, equation (26) can be approximated to the first term: C/co = Blexp(-A12Z*) . . . . (27) The constants B, and A, are eigenvalues that arise out of the solution to the revised differential equation which incorpor- ates the new boundary condition. These two constants are also functions of a new dimensionless parameter the Sherwood number (Nshw), which can be likened to the Nusselt number used in heat-transfer theory.It is approximated by the following equation: II = = r1 = 1 where T is the temperature \ K l 7 rn the relative molecular mass of the analyte, To the standard temperature (K), R the radius of the tube (cm) and P the pressure (kPa). The exponent fi is derived from the expression describing the variation of the diffusional coefficient with temperature and pressure: where PO is the standard pressure. The term Z* is given by the expression Murphy and Faheyzt produced a table with values of B and A for a range of Sherwood numbers. This treatment models the physical processes that take place in a denuder tube more exactly than the Gormley - Kennedy approach, and hence can be seen as an improvement. The introduction of the Sherwood number enables parameters such as temperature and pressure to be incorporated in the denuder unit design. Murphy and Fahey demonstrated that at low pressures, where the Gormley - Kennedy solution fails, this new approach yielded data that fitted the experimental results with significantly greater precision.This method demands a priori knowledge of the reaction probability factor and the diffusional behaviour of the analyte. To obtain such information may require extensive experimental work. The assumption of a constant reaction probability over the entire collection surface is not necessarily valid. The point made by Braman et aZ.,lg that the effective iength decreases with increasing sample volume. is not included in this treatment. Such an approach would require that the reaction probability factor be expressed as a sample size dependent variable.As the Sherwood number approaches infinity, equation (27) tends towards a limiting value, that given by the Gormley - Kennedy solution, Under normal sampling conditions the differences between this approach and the Gormley - Ken- nedy expression are not significant and the usefulness of this method in the analysis of trace amounts of gas has yet to be established. Assessment of the Empirically Derived Annular Equations Based on previous work,4 14 Possanzini et al. 17 derived an expression describing diffusion in an annular system. The exponent term A was redefined in terms of the Reynolds number (Re) and the equivalent diameter of the system: 2 Do2 A=----- . . (31) where y is the kinematic viscosity, 6 the equivalent diamger (defined under Annular Systems) and Re = V6/y, where V is the mean axial velocity of the gas.For a cylindrical system yRe6 * ' ' * . * 6 = 4(ncR2/2xR) = d and - V = 4FIxd2 therefore Re = 4FIndy and hence A = ?1Do2/2F = Ac from equation (14). For an annular system the equivalent diameter changes: and - 4F v, = . . . " . . (33) n(dl2 - d2')>-2 therefore Gormley produced an expression describing diffusion through a laminar flow in a rectangular section pipeline (i.e., between infinite parallel planes3). The rectangular section had a width of 2a and a height of 2b, where b >> a. A treatment similar to that for the cylindrical system yielded the expression (34) Equations (34) and (23) may be compared, for an annulus can be considered to be a distorted rectangle (see Fig.3 ) . To a first approximation the two sections are equivalent with respect to gas flow behaviour within them. The centroid of the annulus is assigned to the height of the rectangle (2b) and the annulus gap to its width (2a). b, = n(dl + d2)/4 . . . . . . (35) a, = ( d , d2)/4 . . . . . . (36) where subscript a refers to the annular system.ANALYST, JULY 1989, VOL. 114 763 Annulus is opened out to give a trapezium Rectangular section 2b * Fig. 3. Schematic diagram of a section of the annulus and its corresponding rectangular section Substitution of equations (35) and (36) into equation (34) gives Equation (37) only differs from equation (23) in the values of the numerical coefficients used. The data obtained by Possanzini et al.17 have been used to generate graphs of collection efficiency versus length using equations (23) and (37), Fig.4. In a later study by the same group, nitrogen dioxide was removed from a gas sample by an annular denuder coated with potassium iodide in a Carbowax matrix.23 The procedure for deriving the empirical constants A and cx was repeated and another expression for the mixing ratio resulted: Graphs are also shown in Fig. 4(b) comparing the effici- encies predicted from equations (38) and (37) and based on the data supplied by Possanzini et al.23 The large difference betweeen the two empirically derived expressions [equations (38) and (23)] emphasises the drawbacks to such an approach. The explanation given by Possanzini et al. for this difference was the occurrence of non-quantitative adsorption of nitrogen dioxide on the potassium iodide impregnated Carbowax. This is a feasible explanation when the nature of the adsorbing surface is considered, but no other experimental evidence was offered.The distorted-rectangle approach produces an expression with an apparently large disparity between it and that derived empirically. The numerical constants in the exponent terms are -3.77 and -5.633 for equations (37) and (23), respec- tively. The difference between the two solutions is illustrated in Fig. 4. The divergence between the predicted collection efficiencies after a length of 20 cm is not significant, and is generally less than the experimental errors associated with the measurement of low gas concentrations.No work has been published that reproduces the result of Possanzini et al.,17 however, it would appear that their result is valid, provided quantitative or near quantitative adsorption takes place. An expression based on a solution to equation (9) for an annular system would provide an interesting result. However, in the context of gas sampling the derivation of such an expression would make little material difference. Applications to Analysis Organic Atmospheric Species Comparatively few results have been published on the collection of volatile organic compounds with denuder systems. Most of the work in this area has been concerned with species originating in automobile exhaust emissions. Annular denuders have been used to collect formaldehyde24 and, using 0 5 10 15 20 0 5 10 15 20 Lengthkm Fig.4. Graphs of predicted collection efficiency versus denuder length. A, Possanzini equation; B, Gormley equation. (a) Sulphur dioxide; ( b ) nitrogen dioxide a sodium hydrogen sulphate - triethanolamine coating, test atmospheres of 0.4 mg m-3 of formaldehyde in air were determined. A detection limit of 0.3 pg m-3 of formaldehyde in air was calculated for a sample time of 24 h and a flow-rate of 0.15 m3 h-1, although no supporting data were given to validate this claim. The analysis involved leaching the collection surfaces with water and using the chromotropic acid spectrophotometric technique24 to determine the formal- dehyde content of the washings. Phenol was found to interfere with the analysis. A 10-20% loss in apparent formaldehyde was reported for a phenol to formaldehyde ratio of 1 : 2.The effects of humidity and temperature on the analysis were not reported. A study of ageing indicated that coated exposed and unexposed units underwent no significant change after being sealed for 4 weeks. The conditions under which they were stored were not described. This work was subsequently extended to include other low boiling-point aldehydes, using high-performance liquid chro- matographic (HPLC) techniques.25 Annular denuders were coated with 1 % 2,4-dinitrophenylhydrazine and phosphoric acid in acetonitrile. Sampled aldehydes underwent derivatisa- tion on adsorption yielding the appropriate 2,4-dinitro- phenylhydrazone derivative. The products were eluted with acetonitrile, separated using HPLC and detected by UV - visible absorbance measurements or voltammetry.Test atmos- pheres of 1.2 mg m-3 of propionaldehyde in air and 1.6 mg m-3 of acetaldehyde in air were determined from sample volumes in the range 0.015-0.095 m3. Collection efficiencies observed for this system were significantly lower than predic- ted. Variations in solubility were advanced as an explanation, but no experimental evidence was given in support of such a mechanism. Variations in the stability and the kinetics of the derivatisation products and processes were not discussed, and the possibility of non-quantitative adsorption was not con- sidered either. The effect of humidity on the system was investigated and relative humidity levels of 50-90% were found to have no effect on the collection efficiency.Lower humidity levels were not reported. Samples taken from cigarette smoke, diesel exhaust and rural air were presented, showing that a number of aldehydes had been collected including acrolein and propionaldehyde. No data were given as to the stability of the unit with time, nor to the effects of chemical interference. Further studies with annular denuders resulted in gaseous organolead compounds being determined.26 Tetraalkyllead compounds were trapped on an iodine monochloride coating, stabilised in a polyethylene glycol - Carbowax 600 matrix. Elution was performed manually by washing with acidified hydrogen peroxide and the resulting washings were analysed using atomic absorption spectrometry at the 283.3-nm lead absorption line.The initial characterisation of this system was based on tetraethyllead atmospheres. However, the experimental details reported were sparse, with few details of concentration, calibration, sample volume or sensitivity being supplied. Interference from chemical or physical sources was not considered, but a study of ageing revealed that exposed764 ANALYST, JULY 1989, VOL. 114 tubes could be stored for 3 d prior to analysis without any significant effect on the final result. The storage conditions were not given. In initial field trials of this system, samples taken from underground garages were found to have concen- trations of organic lead as high as 1.5 pg m-3 in air, which was claimed to comprise 27% of the total lead content of the atmosphere. The technique could not distinguish between different organic moieties and the determination of a specific organolead compound was not possible.A glass tube coated to a depth of 0.1 pm with the soot from a benzene or toluene flame has been employed as a general purpose gas adsorber.27 The system was interfaced directly to a gas chromatograph and sample transfer was effected by thermal desorption. A temperature of 270 "C was maintained for 2.5 min to ensure quantitative desorption of all the trapped compounds. No other information was given as to the construction or operation of the thermal desorption unit. Diffusion coefficients calculated on the basis of the Gormley - Kennedy equation' were presented. As non-quantitative adsorption was reported, the validity of this method is questionable, for the perfect sink criterion of the Gormley - Kennedy solution was not fulfilled.Vapour profiles arising from the headspace analysis of foodstuffs, and from shipboard atmospheres were presented. Distinct vapour patterns were clearly obtained, but neither high-volatility compounds, nor low-volatility compounds normally associated with the par- ticulate phase, were retained. No calibration or sample volume data were presented and no claims were made as to the sensitivity of this technique. Ammonia Species Ammonia occur5 at low concentrations in the atmosphere and is thought to control atmospheric acidity owing to the formation of ammonium salts. which results in acid re- moval.28 zy The equilibrium between ammonia and its salts is complicated due to the large number of precursors and reaction products involved.The model may be simplified by considering ammonia in the vapour phase to be in equilibrium only with ammonium nitrate.?" Sampling using filtration techniques has led to the introduction of artifacts in the medsurement of this equilibrium. The analyte might be overestimated by the release of ammonia from ammonium nitrate: alternativel),, particle - particle reactions of ammo- nium salts with alkaline particles may result in the release of gaseous ammonia. The ammonia concentration tnight be underestimated as a result of the reaction of gaseous ammonia uith acids deposited on the filter. It is important, therefore, to employ a sampling \trategy that separates gaseous and particulate specie>.Ferml-' fir51 reported the use of a denuder system for the measurement of ambient ammonia. The separation of gaseous ammonia from its particulate phase was achieved by employ- ing a cylindrical denudcr tube ([ = 50 cm, i.d. = 3 mm) coated with oxalic acid. The analysis was achieved by dissolution of the sorption layer with 2 cm? of 0.1 M NaOH solution and the ammonium ions were determined by an ion-selective elec- trode with a detection limit of 8.5 mg m-3 of ammonia in air. Collection efficiency studies13 were carried out for ammonia in air in the concentration range 8.5-51 pg m-?, with a sampling time of24 h and a flow-rate of 2.9 1 min-1; collection efficiencies of 90.6% were achieved. An experimental value of 2.47 X m2 s-1 was obtained for the diffusion coefficient of ammonia, in agreement with a value of 2.36 x 10-5 m2 s-1 obtained by Coulson and Richardson" at 1 atm and 25 "C.Ferml4 also investigated particle deposition by sampling the ambient atmosphere with 1 m long uncoated tubes for a period of 2 weeks. The interference due to the particle phase was claimed to be of the same magnitude as the detection limit. The possibility of vaporisation of the ammonium ion, which may be significant with a sampling time of 2 weeks, was not considered. Further studies of particulate deposition have been carried out by Dimmock and Marshall-?' using the technique described by Ferm. Ammonium nitrate aerosols were generated by an ultrasonic nebuliser and characterised by light scattering. Their results showed that 2.84 pg m--3 of ammonium nitrate in air contributed 0.2 pg m--3 to a measured free ammonia concentration of 22.45 pg m-3.It was also reported that the concentration of ammonia recorded by this technique was dependent on the timing of the analytical procedure. Coated denuder units left to stand unsealed for 2.5 h prior to their analysis gave blank values ten times greater than normal. Similar units that were sealed before being left to stand showed no significant uptake of ammonia, even after 4 d. Further, exposed units that were sealed and stored before analysis gave elevated values for the ammonia concentration. The maximum increase in concentration reported was 20% rclativc to units analysed immediately after sampling. A 10-min period was sufficient to produce a measurable differ- ence.The authors emphasised the importance of prompt analysis of exposed denuder samplers, although an explana- tion for these effects was not offered. The collection of ammonia by the unexposed tubes left open to the atmosphere can be explained by a passive diffusive process. Ammonia present in the air within the denuder tube is removed rapidly by the collection surfaces. The concentration gradient that results between the ammonia-free atmosphere in the denuder and the ambient air causes continuous transport of ammonia to the surface of the denuder. The diffusional pumping continues until the collection surfaces are saturated. The reported increase of trapped ammonia in the sealed and exposed units can be attributed to the dissociation of trapped particulates within the sealed assembly.Particulate impacta- tion occurs in the laminar flow subduction zone during sampling. Normally this region is excluded from the subse- quent analysis; in a sealed denuder unit the natural equilib- rium between these solids and free ammonia is disrupted, causing dissociation of the solids. The liberated ammonia is promptly trapped by the collection surfaces. These observa- tions are important and any measurement technique for ammonia should take effects such as these into account. Annular denuder tubes ( I = 25 cm, annulus i.d. = 0.16 cm) coated with 1% oxalic acid in methanol have been used for sampling ammonia and ammonium ion species.3" A flow-rate of 20 1 min-1 and a sampling time of 30 min were used. Sampling artifacts were analysed by employing two denuders placed in series and separated by a filter and the concentration of gaseous ammonia was determined by analysing the contents of the first denuder.The ammonium ion concentration was determined as the sum of the ammonium in the filter and the ammonia released by the filter into the second denuder. The data showed that there was a significant release of free ammonia from the filter. In one instance 6.98 pg m--? of free ammonia were measured in the second denuder. A second corresponding concentration of 7.9 pg m-3 of ammonium particulates was determined on a Gelman GA-4 type filter. The artifact appeared to vary not only with different types of filter but also with the same filter. Dasgupta32 has described a denuder tube with a thin cation-exchange membrane as the collecting element.Gaseous ammonia was collected on the perfluorosulphonate membrane as the ammonium ion, which diffused through the membrane and was carried off in a dilute acid stream for analysis by ion chromatography. The acidic solution also served to maintain the exchange sites in the H+ form. Collection efficiencies of greater than 99% were obtained for ammonia concentrations from 15 ng m-3 to 1.2 pg m-3, with flow-rates from 0.44 to 1.54 1 min-1. The cation-exchange membrane will allow a number of species to diffuse through; however, no interference studies were carried out. The gas phase limit of detection was predicted to be 45 ng m--? of ammonia in air.ANALYST, JULY 1989. VOL. 114 765 The use of two denuders in series separated by a filter, illustrates the extent of ammonia release associated with filters") and the importance of denuder tubes as sampling devices. The accurate determination of low levels of free ammonia in the atmosphere is made possible by the use of oxalic acid coated denuder tubes.However, immediate analysis of the oxalic acid coated denuder is required to avoid dissociation of ammonium particulates in the subduction zone. Particulate dissociation may be significant where long sampling times are required and, in addition, long sampling times do not permit diurnal studies. Annular denuder tubes, however, may be used to overcome many of these problems. The oxalic acid technique, although low-cost, is labour intensive and difficult to automate. The problem of ammonia adsorbing on to the uncoated glass sections of a denuder system has been neglected, despite the work of Dasgupta.32 Until the results of such a study are known, the error associated with such processes is difficult to estimate. Dasgupta's design of the denuder tube32 is an important development in the technique.A variety of gaseous atmos- pheric species may be sampled by using the appropriate membrane material and scrubber solution. Tf the scrubber solution is used as part of a continuous flow analysis system, then spectrophotometric methods of detection can be employed, although in this mode the concentrating ability of the device is largely lost. Nitrate Species Nitric acid is an important acidic atmospheric species. Its measurement and that of its salts in the particulate phase enable the role of nitrogen oxide species in atmospheric chemistry to be elucidated.Nitric acid and nitrate salts have been sampled using dual-filter techniques in which particulate nitrate was col- lected on the first filter and nitric acid on the second."J4 This method gives rise to positive and negative interferences arising from sampling artifacts. Nitric acid will be converted to the solid nitrate in the presence of sodium chloride or basic particulate species.35 Conversely, the dissociation of ammo- nium nitrate, or its reaction with sulphuric acid, gives rise to elevated nitric acid levels within the sampling assembly, with an accompanying reduction in the nitrate-containing species.36.37 Shaw et al.38 overcame these problems with a denuder difference experiment.This method employs two sampling assemblies. both consisting of a nylon filter that collects both gaseous HN03 and particulate nitrate; in one of the assemblies the filter was preceded by a denuder tube coated with magnesium oxide. The difference between the amounts of nitrate collected in the two assemblies was due to the removal of gaseous nitric acid from the sample. Experi- ments run for 23 h at a flow-rate of 3.4 1 min-1 showed that nitric acid concentrations exceeded those of particulate nitrates. Diurnal studies revealed that particulate nitrate concentrations remained fairly constant. whereas nitric acid levels increased during the day. In a comparative study, particulate nitrate and nitric acid were measured by the penetration, denuder difference and dual-filter methods.39 The sampler for the penetration method consisted of two nylon-lined denuder tubes (1 = 80 and 18 cm) separated by a Teflon filter.Sampling was carried out at a flow-rate of 1.5 1 min-1 with a sampling time of 24 h. Nitric acid and volatile particulate nitrates were analysed from the long upstream denuder. The short downstream denuder was used to collect the volatile particulate nitrate only. The denuder tube assembly used in the denuder difference technique allowed shorter sampling times. The assembly consisted of nine denuder tubes ( I = 50 cm, i.d. = 3 mm) coated with sodium hydrogen carbonate. The results of the two denuder methods and the dual-filter method for total inorganic nitrate agreed.39 However, the dual-filter method gave a very high value (4.52 pg m-3) for nitric acid in air; this could be due to the dissociation of particulate ammonium nitrate.A nitric acid concentration of 3.75 pg m-3 was obtained by the penetration method compared with an average of 3.11 pg m-3 for the denuder difference method. This overestimation is thought to be due to turbulent flow caused by the intersection of the flow stream with the leading edge of the nylon filter. Alternatively, it has also been suggested that the collection efficiencies of the two methods are incorrect. The penetration method has a rela- tively poor sensitivity and requires time-consuming sample preparation and it appears to have no advantage over the denuder difference method, which has been shown to give precise experimental data. The denuder tube methods described assume that the evaporation of particulate nitrates as they pass through the denuder tube is negligible.Larsen and Taylor4" examined the artifacts that may arise from the sampling of ammonium nitrate aerosols. Gaseous ammonia and nitric acid were removed from the ammonium nitrate aerosol by means of a diffusion stripper employing sodium hydroxide and sulphuric acid chemical sinks. The rate of evaporation was measured by following the changes in size distribution of the aerosol with time. Resistance to the transport of molecules across the vapour - liquid interface was small and sampling errors due to the evaporation of ammonium nitrate were calculated to be 2.5 and 1.8% for gaseous nitric acid and particulate ammonium nitrate, respectively. A number of factors will, however, cause these errors to decrease.Steady-state gas profiles in the denuder are not achieved instantaneously, hence in the time required to form steady-state profiles, the evaporation of particles would be less as a result of the higher concentration. Calculations performed by Stelson and Sienfelddl suggest that in the presence of ionic mixtures such as ammonium sulphate and ammonium nitrate the vapour pressures of ammonia and nitric acid would be lowered, and hence the sampling error would be lower. Ferm and Sjodin4I used a cylindrical denuder tube coated with sodium carbonate to sample nitrous acid. A flow-rate of 2 1 min-l yielded a collection efficiency of 95%. The sodium carbonate denuder was leached in water and the nitrite concentration was determined spectrophotornetrically.A number of sampling artifacts exist including the formation of nitrous acid in the presence of nitrogen monoxide, nitrogen dioxide, water and peroxyacetylnitratc (PAN), the last of which is also partly adsorbed and hydrolysed to the nitrite. Corrections for interferences may be made by placing two denuder tubes in series. Assuming a 96% collection efficiency the ambient nitrous acid concentration may be calculated from the following equations: I = 0.96[HN02] + 01 . . . . (39) J = 0.0384[HN02] + 02 . . . . (40) where 1 and J are the average nitrite contents in the denuders and on denotes the fraction of nitrite not originating from ambient nitrous acid. As oi and o2 are equal, solving equations (39) and (40) gives [HN02] = 1.085(1 - J ) .The applied corrections are not applicable where the concentrations of PAN are greater than those of nitrous acid. The method is therefore only useful in air close to NO, sources, such as polluted urban air. It cannot be used to measure natural background levels of nitrous acid. Automated monitoring of nitric acid and ammonia has been achieved by multi-stage selective thermal desorption from a tungsten(V1) oxide coated cylindrical denuder tube. 19,43 The analysis took 40 min with a sensitivity of sub-p.p.b. The adsorption of ammonia on tungstie acid is thought to be an acid - base reaction that is reversible at 350 "C 350 "C NH3 + HZW0;1 e N H 4 H W 0 4766 ANALYST, JULY 1989, VOL.113 The mechanism for nitric acid adsorption is unclear, although it has been suggested that irreversible adsorption is involved. Nitric acid desorbs as nitrogen dioxide. The denuder tubes are coated by the vacuum deposition of tungsten wire. Blue tungsten(1V) oxide is oxidised to yellow tungsten(V1) oxide by heating at 500 "C. Alkylamines, nitrogen dioxide and PAN were all shown to interfere. The amount of ammonia adsorbed decreased in the presence of 100 p.p.b. of ozone. Diurnal studies of nitric acid and ammonia showed that in most instances higher nitric acid concentration maxima correlated with lower ammonia concentration maxima. Diffusion coeffi- cients obtained by section analysis agreed with those reported by Wilke and Lee.44 Braman et al. 19 concluded that, although the nitric acid may be associated with water or other compounds, hence decreasing the diffusion coefficient, ammonia existed in the non-associated form.These findings contradict those of a parallel study,45 which showed that ammonia existed in its associated form. The tungstic acid denuder used in this study was prepared by coating six sections of tubing (1 = 15 cm, i.d. = 0.56 cm) with a thin film of sodium hydroxide (0.5 M) - sodium tungstate [0.05 M in water - propan-2-01 (50 + 50 V/V)]. By rolling the tubes on a horizontal roller the coating was air dried. Sodium tungstate was converted to tungstic acid by passing hydrogen chloride through the tubes, which were then washed with water to remove sodium chloride and then heated at 500 "C in a furnace for 8 h.The sample capacity was shown to be much greater than that of the tube coated by vacuum deposition of tungsten wire although the vacuum deposition method did produce blank values that were 20 times lower. Both nitrate and ammonia were extracted quantitatively with three successive 3-cm3 portions of 1 .O mM LiOH solution at 90 "C and then the nitrate and ammonium ions in the extract solution were determined by ion chromatography. The measurement of atmospheric nitric acid46 was carried out in a comparative study of the tunable diode laser absorption, tungstic acid denuder tube and Teflon - nylon filter-pack methods. The laser method gave higher nitric acid concentrations than the other methods and quantitative agreement between these techniques was not obtained, although general trends in the variation of the nitric acid concentration could be identified.The tungstic acid method measured higher nitric acid concentrations than those using the filter pack, the difference being 5% for daytime measure- ments and a factor of two lower during the night. The difference could be due to the retention of nitric acid on the particulate matter of the filter.47748 The tungstic acid method also showed higher particulate nitrate concentrations than the filter-pack method and this may be due to the dissociation of particulates on the filter. Nitric acid concentrations were investigated in a further comparative study of the tungstic acid and filter-pack methods.49 The former gave nitric acid concentrations three times higher than the latter and this difference was attributed to interference from organic nitrates in the denuder tube; however, no data were produced to confirm this.Interferences from nitrogen dioxide, PAN and propyl nitrate in the denuder tube were shown not to be significant. A number of denuder tube techniques have been developed for the measurement of nitric acid with the tungstic acid denuder tube being the most widely reported for this purpose. Since its introduction, a number of studies have compared the tungstic acid method with measurements performed using other techniques, the result being a great deal of scatter in the inter-comparison. However, it was shown that the tungstic acid denuder tube produced larger apparent nitric acid mixing ratios than the filter-pack technique and the source of this difference has not yet been identified.The dissociation of particulate nitrates in the subduction zone of the denuder tube, which is particularly likely if the subduction zone is heated during the thermal desorption cycle, has not been considered. In a recent study Roberts et ~ 1 . 5 0 concluded that tungstic acid coatings were subject to "slow evolutionary and occasionally catastrophic failure," indicating that tungstic acid coatings are not viable for long-term unattended use. There is a need for properly characterised adsorbing surfaces with high thermal stabilities and in this respect transition metal oxide surfaces may prove useful. Hydrogen Chloride Natural sources of hydrogen chloride include the reaction of marine-salt aerosols with atmospheric sulphuric and nitric acids, or volcanic emissions.The gas can also be released from refuse incineration plants and result from the combustion of certain types of coal. Dimmock and Marshall-5' described a manual method for the determination of hydrogen chloride using a cylindrical denuder tube coated with sodium fluoride. The extraction was carried out using distilled water with detection by a chloride ion-selective electrode. Adsorption efficiencies >go% were obtained and the room air sampled was found to contain hydrogen chloride concentrations in the range 0.16-0.55 pg m-3. Interferences from particulate chloride, sulphur dioxide and nitrogen dioxide were found to be negligible. Sulphur Species Sulphur dioxide has been sampled with a cylindrical denuder tube coated with a mixture of sodium tetrachloro- mercurate(I1) and malein buffer [the malein buffer was used to neutralise the hydrogen chloride produced during the reaction between tetrachloromercurate( TI) and sulphur dioxide] .*O The cylindrical denuder tube was coated by injecting 0.4 cm3 of 0.1 M sodium tetrachloromercurate(I1) - 0.1 M malein buffer in 1 + 1 methanol - water.The extraction was carried out using distilled water and the analysis by isotope dilution. Sampling was performed in a polluted industrial atmos- phere with a sampling time varying from 3 to 24 h with the relative humidity in all instances being >50"/0. Collection efficiencies of 90% were achieved except where the relative humidity approached 100% and the results obtained for the collection of sulphur dioxide in denuders and on potassium hydroxide filters agreed.Gas-phase concentrations of dimethyl sulphate and monomethyl hydrogen sulphate have been determined by collecting the species on cylindrical denuder tubes.52 A paper-lined denuder tube was used for the collection of monomethyl hydrogen sulphate and a nylon denuder was used for sampling total gas-phase alkyl sulphates. Dimethyl sul- phate was first collected on an XAD-2 resin bed and then determined by ion chromatography. The aqueous-extractable dimethyl sulphate hydrolyses to the monomethyl hydrogen sulphate and the increase in the monomethyl hydrogen sulphate concentration between the paper and the nylon denuder tubes was used as a measure of the dimethyl sulphate concentration.Agreement between the expected and calculated diffusion coefficients supported the identification of dimethyl sulphate and monomethyl hydrogen sulphate as the species adsorbed on the denuder tubes. Laboratory studies did, however, show that up to 10% of the sulphur dioxide passing through the nylon denuder may be adsorbed on the walls, and a correction procedure was developed to overcome this interference effect, In field trials, this denuder approach did not yield results that agreed with those obtained from filter-based techniques. The authors52 stated that the agreement was not good and they were unable to indicate which technique was the most reliable. Ambient sulphuric acid aerosols have been sampled by means of a heated denuder tube coated with manganese and palladium oxides53 with interferents being selectively removed from the sample by a series of pre-denuders. On completion ofANALYST, JULY 1989, VOL.114 767 sampling the denuder was placed in an oven and desorbed thermally at 800 "C into a convertor, which reduced any sulphur liberated to hydrogen sulphide, before being passed into a silver wool chemical pre-concentrator. Flash heating of the silver wool resulted in a concentrated injection of hydrogen sulphide into the gas chromatograph. A detection limit of 0.1 pg m-3 of sulphuric acid in air was obtained using this technique. A study of particulate dissociation at various denuder temperatures was undertaken and this underlined the possible dissociation of particulates in the subduction zone during thzrmal desorption.The results of this study indicated that dissociation of particulates trapped on the walls of the laminar flow subduction zone could occur during thermal desorption. Such findings'? have important ramifications €or the design criteria of automated denuder systems. Multi-component Sampling The criterion of laminar flow limits the sample flow-rate; therefore, in order to collect a measurable amount of analyte, long sampling times are required. To overcome this problem Stevens et al. 13 utilised 16 parallel tubes arranged in a circle to separate ammonia from ammonium salt particulates. No data relating to the amount of ammonia trapped were given in this report, and the problems of ensuring the reproducible handling of 16 tubes were not discussed.Forrest et d . 4 7 employed 48 sodium carbonate coated cylindrical tubes for sampling nitric acid, which allowed sampling flow-rates of 10-30 1 min-1. In this study, the unit was heated such that the relative humidity of the sample was lowered below the deliquescence point of sodium carbonate. The authors claimed that this did not significantly influence the results obtained, but the supporting data were not comprehensive. The results of a detailed study into such a procedure would establish whether heating the sample to permit better instru- ment operation is an acceptable practice in this area of application. Lewin and Hansen54 described a diffusion denuder assembly, which. by means of a vacuum and a pressurised air system, allows the individual coating and extraction of 15 quartz glass tubes.This semi-automated method enabled denuder tube assemblies to be coated with a number of selective absorbents, decreasing the risk of contamination compared with manual methods. Ammonia was sampled using tubes coated with 1 .S% oxalic acid in ethanoPJ and acidic gases were sampled using tubes coated with 1 M potassium hydroxide in methanol. Automatic spectrophotometric methods were employed for sulphate, nitrate and chloride determinations and the indophenol method for the determination of ammonia. The data supplied for this system were sparse and few details on sensitivity, reproducibility and calibration procedures were given. However, the advantages of using the device described for batch-processing denuder tubes are self evident.Liberti et al.55 employed three annular denuders in series for the determination of sulphur dioxide, ammonia, nitric acid and hydrogen chloride. Sulphur dioxide was sampled using a sodium tetrachloromercurate(I1) coated tube. Collection effi- ciencies >90Y0 were obtained for sulphur dioxide with a sampling flow-rate of 20 1 min-1. Ammonia was sampled with an oxalic acid coated tube, and hydrogen chloride and nitric acid on a sodium fluoride coated tube. Little experimental detail was reported, but data obtained from the denuder assembly over a 7-month period indicated that ammonia and sulphur dioxide could be monitored with such a system. Nitric acid and hydrogen chloride were detected only intermittently by the unit. No explanation was given for this behaviour, and complementary data from other techniques would be required to support any conclusions drawn from their results regarding the presence, or lack of it, of hydrogen chloride and nitric acid in the sample.A series o f coated cylindrical denuder tubes operated at different temperatures was used to collect free ammonia and to distinguish nitric acid from ammonium nitrate and sulphuric acid from ammonium sulphate by their different thermal behaviour.56 The tubes were extracted with doubly de-ionised water. Ammonium ions were detected by means of a gas-sensing membrane and nitrate and sulphate species were determined by ion chromatography. The tubes were coated for collecting nitric acid, free ammonia and sulphur dioxide.56 A tube operated between 390 and 410 K and coated with sodium fluoride was used for the collection of sulphuric acid and a sodium fluoride coated tube operated at 490-510 K was used to retain the sulphuric acid produced from the dissociation of ammonium wlphate and a further tube collected the liberated ammonia.This last technique appears to be a simple and low-cost method for carrying out multi-component analysis. However, the dissociation of ammonium nitrate and sulphate before they reach their target tubes causes a degree of uncertainty and it was noted that at ammonium nitrate concentrations of 20 pg m-3, half of the liberated nitric acid was found in the first tube. The selective collection and determination of airborne sulphuric acid and ammonium sulphate has also been achieved using two copper(I1) oxide coated denuder tubes at tempera- tures of 120 and 240 "C, This method relies on the adsorption of sulphuric acid on copper - copper(T1) oxide and the decomposition of ammonium sulphate at tenipera- tures above 220 "C. The resulting copper sulphate was converted to sulphur dioxide by heating at 800 "C and then measured using a flame photometric detector.Interferences due to inorganic and organic sulphur species were avoided by passing the sample stream through potassium carbonate and active carbon-coated denuders. The copper - copper(I1) oxide was prepared by filling the tube with 3% mlV hydrated copper(I1) nitrate in ethanol. The tube was emptied. dried first at room temperature and then at 200 "C to remove water of crystallisation.Nitrogen was passed through the tube at 400 "C for 5 min. Both ends of the tube were cleaned by inserting the tube in 1 M nitric acid and a black copper - copper(I1) oxide coating was obtained after flushing with nitrogen and heating at 900 "C for 5 min. The technique employed two parallel copper - copper(I1) oxide denuders. After completion of the sampling period in the first set of denuders, the sample tlow was switched to the second set of denuders and the first set was analysed individually. The data presented showed that very little ammonium sulphate was found in the first tube at the selected tempera- tures and at ammonium sulphate concentrations above 30 pg -m-3. No data were presented for possible dissociation artifacts at ammonium sulphate concentrations below 30 pg m-3.This may be significant as the measured ambient ammonium sulphate concentrations were less than this value. Earlier work with the sodium fluoride manual method showed that dissociation of ammonium sulphate becomes significant at lower concentrations.56 The copper - copper(l1) oxide denuder tube has also been employed for the measurement of sulphur dioxide .5X The effect of humidity on collection efficiencies and an investigation of sulphur-based interferents were carried out in the same study. A significant decrease in the collection efficiency with decreasing humidity was repor- ted, but no explanations for the reasons were given. The interference from organic sources was stated as "not impor- tant" and data were presented that demonstrated agreement between results obtained from denuder systems and those produced through filter techniques inscnsi tive to the presence of organic sulphur species.The reproducibility of the tech- niquc was claimed to be 3% and the available data showed the relative standard deviation to vary between 1 .S and 5%. Slaninasg have developed a "wet denuder" system for the sampling of ammonia, nitric acid, hydrochloric acid, sulphur dioxide and hydrogen peroxide. The wet denuder consisted of768 an annular denuder coated with an aqueous layer, which was rotated about its axis to keep the surfaces wet. Two parallel wet denuders were employed. In one denuder the absorption solution contained formaldehyde and p-hydroxyphenylacetic acid to collect and stabilise sulphur dioxide and hydrogen peroxide.The absorption solution in the second denuder contained a formic acid buffer of pH 4 for the collection of ammonia, nitric acid and hydrochloric acid. After sampling, the absorption solutions were pumped out of the denuders into two sampling tubes. Hydrogen peroxide and sulphate were determined using fluorimetric and photometric detec- tors, ammonium ions by spectrophotometric measurement and nitrate and chloride ions by ion chromatography. A cyclone was used to reduce interferences from particulates. The effects of rotation of the collection surfaces and moving liquids within the denuder unit on laminar flow and particulate transmission efficiency were investigated; no significant per- turbation of the system was reported. Clearly, the device described has other atmospheric monitoring applications and this approach may be expected to be extended to other areas of atmospheric analysis.Conclusions The determination of atmospheric species is fraught with problems; the dynamic equilibria between the various species make their determination particularly difficult. Separation of the gas and particulate phases by denuder tubes avoids the artifacts that may otherwise occur with filter-based methods. The denuder tube has the additional advantage of pre- concentrating the gaseous analyte, which is not possible using conventional filters. Denuder tubes have proved to be an important development in atmospheric sampling, allowing the daily patterns and long-term trends of atmospheric species to be discerned.The use of simple, cylindrical denuder tubes has several disadvantages. Their use is labour intensive, as both a coating procedure and extraction of the collected gases are involved and the low sampling flow-rates require long sampling times, hence causing a number of other artifacts. The introduction of the annular denuder tube allows much higher sampling flow-rates and hence shorter sampling times and the use of thermal desorption denuders eliminates the need for washing or re-coating. However, in this last method, the interference due to vaporisation of particulates in the subduction zone has not been clearly established. The wet denuder system has @aced a much older sampling device, the bubbler, into a denuder tube context and may have wide application for the monitoring of thermally unstable species.A degree of complexity has been introduced by the development of multi-component denuder tube systems. However, further fundamental work has to be carried out to realise the full potential of the technique. In particular, considerations have to be made about the nature and influence of non-adsorbing surfaces within the denuder tube. Consider- ations also have to be made about the employment of various sampling strategies such as the incorporation of iso-kinetic sampling from laminar gas flows. The use of a wide range of denuder tube systems has been demonstrated, although they have been employed mainly for the measurement of species involved in atmospheric pollution and acid rain studies. Few applications have been reported for organic vapours in the atmosphere.It is most likely that the applications to which denuder tubes are employed will increase in the future, as they are attractive by virtue of their simplicity, ruggedness and ability to undergo long-term exposure to particle-laden atmospheres with minimal inter- ference from the aerosol and particle content. ANALYST, JULY 1989, VOL. 114 Zulfiqur Ali was supported by the SERC and CEGB under the CASE Award scheme. Our work in this area is supported also by the Procurement Executive, Ministry of Defence, UK. 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. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. References Townsend, J. A., Philos. Trans. R. SOC. London, Ser. A , 1900, 193, 129. Nolan, J .J . , and Guerrini, V. H., Proc. R. Ir. Acad., Sect. A , 1935, 43, 5. Nolan, J. J., Nolan, P. J., and Gormley, P. G., Proc. R. Ir. Acad., Sect. A , 1938,45,47. Gormley, P. G., and Kennedy, M., Proc. R. Ir. Acad., Sect. A , 1949, 52, 103. Thomas, J . W., J. Colloid Sci., 1955, 10, 246. Pack, M. R., Hill, A. C., Thomas, M. D., and Transtrum, L. G., Am. SOC. Test. Muter. Spec. Tech. Publ., 1961, 281,27. Crider, W. L., Barkley, N. P., Knott, M. J . , and Slater, R. W., Anal. Chim. Acta, 1969, 47, 237. Fish, B. R., and Durham, J. L., Environ. Lett., 1971, 2, 13. Durham, J. L., Wilson, W. E., and Bailey, E. B., Atmos. Environ., 1978, 12, 883. Huntzicker, J . J., Hoffman, R. S . , and Ling, C. S . , Atmos. Environ., 1978, 12, 83. Durham, J . L., Wilson, W.E., and Bailey, E. B., US Environmental Protection Agency Report No. EPA 600/3- 761088, Washington, DC, 1976. Cobourn, W. G., Husar, J. D., and Husar, R. U., Atmos. Environ., 1978, 12, 89. Stevens, R. K.. Dzubay, T. G., Russwurm, G., and Rickel, D., Atmos. Environ., 1978, 12, 55. Ferm, M., Atmos. Environ., 1979, 13, 1385. Coulson, J . M., and Richardson, J. F., “Chemical Engineer- ing,” Volume 1, Second Edition, Revised, Pergamon Press, London, 1966. Fuchs, N. A., in Davies, C. N., Editor, “The Mechanics of Aerosols,” Pergamon Press, London, 1964. Possanzini, M., Febo, A., and Liberti, A., Atmos. Environ., 1983, 17, 260.5. Liberti, A., A q u a Aria, 1986, 2, 133. Braman, R. S . , Shelley, T. J . , and McClenny, W. A., Anal. Chem., 1982, 54, 358. Lewin, E.E., and Klokow, D., Comm. Eur. Communities, Phys-Chem. Behav. Atmos. Pollut., L. Rep. Eur 7624, pp. 54-61. Murphy, D. M., and Fahey, D. W., Anal. Chem., 1987, 59, 2753. Cooney, D. C., Kim, S . , and Davies, E. J., J . Chem. Eng. Sci., 1974, 29, 1731. Possanzini, M., Febo, A., and Cecchini, F., Anal. Lett., 1984, 17, 887. Cecchini, F., Febo, A., and Possanzini, M., Anal. Lett., 1985, 18, 681. Possanzini, M., Cicclioli, P., Di Palo, V., and Draisel, R., Chromatographia, 1987, 23, 829. Febo, A., Di Palo, V., and Possanzini, M., Sci. Total Environ., 1986, 48, 187. Cobb, G. P., Braman, R. S . , and Hua, K. M., Anal. Chem., 1986, 58, 2213. Gravehurst, G., and Bottger, A., in Versine, B., and Ott, H., Editors, “Proceedings of the 1st European Symposium on the Physico-Chemical Behaviour of Atmospheric Pollutants, Ispra, 16-18 October 1979,” Reidel, Dordrecht, Holland, 1980, p. 383. Georgii, H. W., and Muller, W. J., Tellus, 1974, 26, 180. Allegrini, I . , Santis, F. D., Palo, V. D., and Liberti, A., J. Aerosol Sci., 1984, 15, 465. Dimmock, N. A . , and Marshall, G. B., Anal. Chim. Acta, 1986. 185, 1.59. Dasgupta, P. K., Atmos. Environ., 1984, 18, 1.593. Spicer, C. W., and Schumacher, P., Atmos. Environ., 1977,11, 873. Appel, B. R., Tokiwa, Y., and Haik, M., Atmos. Environ., 1981, 15, 283. Appel, B. R., and Tokiwa, Y., Atmos. Environ., 1980,14,549. Stelson, A. W., Friedlander, S. K., and Seinfeld, J. H., Atmos. Environ., 1979, 13, 369.ANALYST, JULY 1989, VOL. 114 769 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. Harber, A. B., Richards, L. W., and Clarke, W. E., Atmos. Environ., 1977, 11, 87. Shaw, R. W., Stevens, R. K., Bowermaster, J., Tesch, J . W . , and Tew, E . , Atmos. Environ., 1982, 16, 845. Mulawa. M. A , , and Cadle, S. H., Atmos. Environ., 1985, 19, 1317. Larsen, T. V., and Taylor, G. S., Atmos. Environ., 1983, 17, 2489. Stelson, A. W., and Sienfeld, J. H., Atmos. Environ., 1982,16, 2507, Ferm, M., and Sjodin, A., Atmos. Environ., 1985, 19, 979. McClenny, W. A., Galley, P. C., Braman, R. S., and Shelley, T. J . , Anal. Chem., 1982, 54, 365. Wilke, C. R., and Lee, C. Y., Ind. Eng. Chem., 1955,47,1253. Eatough, D. J., White, F. V., Hansen, L. D., Eatough, N. L., and Ellis, E. C., Anal. Chem., 1985, 57,743. Anlauf, K. G., Fellin, P., Wiebe, H. A . , Schiff, H. I., Makay, G. I., Braman, R. S., and Gilbert, R., Amos. Environ., 1985, 19, 325. Forrest, J . , Spandau, D. J . , Tanner, R. L., and Newman, L., Atmos. Environ., 1982, 16, 1473. Spicer, C. W., Howes. J. E., Jr., Bishop, T. A . , Arnold. L. H., and Stevens, R. K . . Atmos. Environ., 1982, 16, 1487. Roberts, J. M., Norton, R. B., Goldan, P. D., andFehsenfeld, F. C., J. Atmos. Chem., 1987, 5, 217. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. Roberts, J. M., Longford, A. O., Golden, P. D., and Fehsenfeld, F. C., J. Atmos. Chem., 1988, 7, 137. Dimmock, N. A . , and Marshall, G. B., Anal. Chim. Acta, 1987, 202, 49. Eatough, D. J., Vernon, F. W., Hansen, L. D., Eatough, N. L., and Cheney, J . L., Environ. Sci. Technol., 1986,243,867. Lindqvist, F., Atmos. Environ., 1985, 19, 1671. Lewin, E. E., and Hansen, A. K . , Anal. Chem., 1984,sQ, 842. Liberti, A., Allegrini, I., Febo, A., and Possanzini, M., in Irgolic, K. J . , and Martell, A. E., Editors, “Envi-romntal Inorganic Chemistry,” VCH. Deerfield Beach, FL, 1985, pp. 419-430. Slanina, J., Doornenbal, L. V. L., Lingerak, A. W., and Meilof, W., Int. J. Environ. Anal. Chem., 1981, 9, 59. Slanina, J . , Schoonebeck, C. A. M., Klockow, D., and Niessner, R., Anal. Chem., 1985, 57, 1955. Slanina, J., Schoonebeek, C. A. M., and Keuken, M. P., Anal. Chem., 1987,59, 2764. Slanina, J., Report No. ECN-88- Verslag Van Project 4458 Droge-En Natte Depoisitie (Uitqebreide versie) door, Nether- lands Energy Research Foundation, Petten, Netherlands, May 1988. Paper 81044131 Received November 11 th, 1988 Accepted February 8th, 1989

ISSN:0003-2654    

DOI:10.1039/AN9891400759

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST, JULY 1989. VOL. 114 77 1 Prediction of Retention Behaviour on Modification of the Mobile Phase in High-performance Liquid Chromatography Using Metal Ions: 2-Aminophenol as a Model System Roger M. Smith and Simon J. Bale* Department of Chemistry, University of Technology, Loughborough, Leicestershire LEI 7 3TU, UK Stephen G. Westcott and M. Martin-Smith Glaxo Group Research, Ware, Hertfordshire SG72 ODJ, UK A model has been developed for predicting the effect of transition metals as components of the mobile phase in high-performance liquid chromatography on the retention of an analyte containing a chelating group. It takes into account the secondary chemical equilibria in solution, including the formation constants of the complexes, the pH of the mobile phase and the concentration of the metal ions.The model was tested by comparison with data determined for the retention of 2-aminophenol on a porous polystyrene - divinylbenzene polymer column in the presence of metal ions. Keywords: High-performance liquid chromatography; transition metal ions; mobile phase additives; metal chelates; retention prediction Considerable use has been made of metal ions and complexes to alter the selectivity of separations carried out by high- performance liquid chromatography (HPLC) Metal ions provide a useful means of modifying retention as only specific chemical groups and ligands will interact with a given metal and the extent of the interaction can usually be regulated by adjusting the pH of the mobile phase. In addition, steric properties are important because metals have fixed co-ordina- tion sites and this may, under certain conditions, lead to stereoselectivity .The metal species used to effect a separation may be present as part of the stationary phase or, alterna- tively, may be incorporated in the mobile phase as a free ion or complex. Metal ions may be introduced on to the stationary phase by impregnation on the silica surface or by bonding via a suitable ligand attached to the surface.”-” Previous workers have made use of the reversible charge-transfer complexation between silver ions and x-donors in argentation chromatography.”* Columns loaded with ruthenium9 and cadmium ions10 have also been used for the separation of unsaturated compounds. Eiceman and Janecka” have used the binding properties of surface-active silanol groups on octadecylsilyl bonded silica to immobilise a range of transition metals, including copper(II), nickel(II), cobalt(I1) and zinc(1I).A mixture of aniline and substituted amines was used to demonstrate the different selectivities that could be achieved using these columns. Metal ions may also be incorporated in the mobile phase and there are several advantages in using this approach. Specialised stationary phases are difficult to prepare reprodu- cibly and in most instances are not stable due to leaching of the metal ions. Changing from one metal to another is difficult if not impossible. If metal ions are incorporated in the mobile phase, the change from one metal to another may be performed by washing the column with a few column volumes of the new mobile phase to give conditions for which the metal concentration is known.In general, when metal ions are present on the stationary phase analyte retentions are increased and the peaks are broad. However, when metal ions are present in the mobile phase, mass transfer is improved considerably and more efficient peaks are obtained. The complexes may also be either less or more polar depending on the charge and hence the retentions may either be increased or * Present address: Pfizer Central Research, Sandwich, Kent CT13 9NJ, UK. decreased. Selectivity is, in general, achieved through differ- ences in the stabilities of the complexes formed between the analytes and the metal ions in the mobile phase and the hydrophobicities of the resulting species.Metal chelates, such as nickel (3-diketonatesl2.13 and C12- dien complexes,-” have also been used as mobile phase additives in conjunction with reversed-phase columns. By using chiral metal chelates, enantioselectivity can also be achieved. 1 ~ 1 4 With the exception of silver ions15 relatively little use has been made of simple metal salts as mobile phase modifiers. Copper(I1) ions have been used to modify the separation of the colchicines16 and zinc(I1) ions to increase the separation of the isomeric aminobenzoic acids.17 Sternson and co- workers1*,19 found that the addition of nickel acetate to the mobile phase could be used to enhance the resolution of 2-aminophenol and its metabolites. We have recently exten- ded this last study and have reported a detailed examination of the effects of a range of metal ions on the retention of 2-aminophenol.20 The work was carried out using a poly- styrene - divinylbenzene column so that interactions with silanols which occur on a silica-based column would not interfere with the interpretation of the results.The aim was to use 2-aminaphenol as a model compound to gain a more detailed understanding of the retention modification caused by simple transition metal ions in the mobile phase. All the metal ions studied caused a reduction in retention due to the formation of a more polar 1 : 1 complex (ML+, L = ligand) between the ligand and the metal ions. The magnitudes of the changes were related to the metal ion concentration and depended on the pH of the mobile phase.In this paper, we have developed a general model tor the observed changes in the capacity factors of chelating groups in the presence of metal ions which takes into account the interactions taking place in these separations and involves the various equilibria present in solution. The model can be correlated with the effects of changing the metal ions and their concentrations and has been tested against the experimental results. Experimental Reagents Laboratory-reagent grade ammonium acetate, acetates of copper and nickel, and methanol (HPLC grade) were772 ANALYST, JULY 1989, VOL. 114 obtained from FSA Laboratory Supplies (Loughborough, Leicestershire, UK). 2-Aminophenol (Aldrich, Gillingham , Dorset, UK) was recrystallised from water before use. Apparatus The liquid chromatograph consisted of a Pye Unicam PU 4010 pump, a Rheodyne 7125 injection valve fitted with a 20-pl sample loop and a Pye Unicam PU 4020 variable-wavelength UV detector set at 280 nm. Chromatograms were recorded on a Linseis chart recorder and a Hewlett-Packard 3390A integrator.The separations were carried out using Shandon stainless-steel columns (100 x 5 mm i.d.), which had been slurry-packed with PLRP-S (porous polystyrene - divinylben- zene copolymer, 5 pm; Polymer Laboratories, Church Stret- ton, UK). The temperature of the column was maintained at 30 "C with a thermostated water-jacket. The mobile phase flow-rate was 1 ml min-'. Measurements of pH were made on a Pye Unicam 390 pH meter. Sample and Mobile Phase Preparation Solutions of 2-aminophenol were prepared fresh in methanol and diluted to the required concentrations with distilled water.The mobile phase was prepared from a stock solution of 2.6 M ammonium acetate by dilution with methanol and water to give a final concentration of ammonium acetate that was 0.26 M in methanol - water (20 + 80 VIV). Metal ions were added as required by the inclusion of the appropriate volume of 0.2 M metal acetate solutions in the buffer. In each instance the measured pH was adjusted to 7.24 with concentrated hydro- chloric acid or sodium hydroxide solution (4 M). The mobile phase was de-gassed under vacuum in an ultrasonic bath before use. Procedure Injections (20 pi) of solutions of 2-aminophenol were made on to the column and were eluted with mobile phases containing metal ions (0-0.07 M).The effect o f the pH of the mobile phase was studied using thc polymer column, in the pH range 4-13, with each metal ion at a concentration of 0.02 M except for copper (0.0001 M ) . Data Analysis Least-squares correlations were carried out on an Apple 11 computer using Curfit (Interactive Microware, State College, PA, USA) and non-linear least-squares using Genstat 4.03 from Lawes Agricultural Trust, Rothamsted Experimental Station, on a Htrneywell Multics computer. Results and Discussion The use of metal ions as part of the mobile phase to alter selectively the chromatography of 2-aminophenol is an example of the use of secondary chemical equilibria in HPLC.11 Such secondary equilibria are important in other modes of HPLC, such as ion-pair chromatography,22 in which the influence of the pairing ion and counter ion on the equilibria in solution is used to alter the distribution of the analyte between the mobile and stationary phases.The effect of the pH of the mobile phase on the retention of ionisable compounds is another example of the way in which secondary chemical equilibria may be used to alter the selectivity of a separation. The equilibria involved in the HPLC of acids, bases and ampholytes have been discussed by Horvath et al. ,23 who proposed equations to predict the capacity factors of ionisable compounds with a varying mobile phase pH. 12, 0 2 4 6 8 10 12 PH Fig. 1. Variation in the experimental capacity factor of 2-amino- phenol with the mobile phase pH in: H, the absence: and presence of nickel ions: a, 0.01 M; and A , 0.02 M on a PLRP-S column. Solid lines are fitted curves from equations (32) and (35).Mobile phase, methanol - water (20 + 80 VIV) containing 0.26 M ammonium acetate Secondary chemical equilibria have been reviewed by Karger et af.,24 who looked at all uses of such equilibria in liquid chromatography. Vespalec et af. 25 have also discussed equilibria involving metal ions in HPLC but were concerned more with the separation of metal species than with the use of metal ions to modify retentions. Retention Model The use of metal ions to adjust retention will depend on the formation constants of the complex and the concentrations of the metal ions and ligand. If the latter is ionic its concentration will be dependent on the ionisation constant and the pH of the mobile phase.The over-all retention of the analyte will thus depend on a series of interconnected equilibria. Each species in the system will have a retcntion time which will be directly dependent on its distribution between the stationary and mobile phases. If the interchange between the species is rapid, the over-all retention time of the analyte should be represen- ted by the ratio of the summation of the average time each of the species spends in the two phases. It should, therefore, be possible to generate a model of the chromatographic system which relates these different equilibria and which would be generally applicable to any metal - ligand combination. This model could then be used to predict the effects of using different metal ions or of changing the metal ion concentration or the pH of the mobile phase.The model should be capable of explaining the changes observed experimentally with 2-amino- phenol and the equilibrium constants should have values that can be related to those obtained with non-chromatographic methods. In developing a model based on the 2-aminophenol system it is first necessary to consider the situation in the absence of metal ions. 2-Aminophenol is an ampholyte in aqueous solution and can exist in three forms: the protonated amino cation (AH2+), the neutral molecule (AH) and the phenolate anion (A-). In mobile phases with a low pH, 2-aminophenol will be present as a protonated amino cation and, as expected, a low retention is observed.20 As the pH is increased the degree of protonation will decrease and an increase in the retention is observed until at intermediate pH values (7-9) the analyte is mostly neutral and the retention reaches a maximum (Fig.1). Finally, at higher pH values the phenol group will dissociate to give the negatively charged phenolate anion and again low retention times are observed. The well separated pK, values for the two groups suggest that at the intermediate pH values 2-aminophenol will not be present as a zwitterionANALYST, JULY 1989, VOL. 114 773 [AH,+]s [AH], [A-ls [MA+], [MAPIS Fig. 2. Equilibria involved in the HPLC of 2-aminophenol with metal ions in the mobile phase. M = mobile phase component; S = stationary phase component and this can be confirmed by calculation.Hence the relative proportions of the three species are pH dependent and the equilibria between the three forms are governed by the dissociation constants of the amino and phenolic groups, Kal and Ka2, respectively. Ka 1 Ka2 AH2+=AH+H+ . . . . . . (1) A H Z A - + H + . . . . . . (2) In the presence of metal ions (M2+) the phenolate anion can form chelates and two further species are introduced into the over-all equilibrium, the 1 : 1 (MA+) and the 2 : 1 (MA2) chelates. The proportions of these two species are dependent on their respective formation constants, K1 and K2, and on the relative concentration of the anion [A-] to that of the metal ions [M*+] . K1 K2 A-+M2+=MA+ . . . . . . (3) MA++A-=MA2 . . . . . . (4) Because of complexation with the acetate buffer com- ponents at high pH, only a proportion of the added concentra- tion of the metal ions will be available to complex with the anions.Hence the values of K1 and K2 will be the effective formation constants, which incorporate a factor for the acetate equilibria (which will be a constant at each pH and fixed acetate concentration), rather than the formation constants measured in aqueous solution. The distribution of 2-aminophenol among the five possible species may be written as Ka1 Ka2 K1 K2 AH?_+ AH A- e MA+ % MA2 . . ( 5 ) When 2-aminophenol is injected on to an HPLC column and eluted with an eluent containing metal ions, the same equilibria are set up in the mobile phase. Each of the species (X) present in the mobile phase is in turn in equilibrium with the stationary phase and is distributed between the mobile and stationary phases according to its distribution coefficient As the separation occurs, each of the species will have its own individual capacity factor.However, as long as the exchange of the ligand between the different species is rapid, a single peak will be observed with a capacity factor dependent on the over-all equilibrium of the analyte between the two phases. Hence the capacity factor under any set of conditions is dependent on the relative proportions of the individual ( D X = [x]S/[x]M) (Fig. 2). species and their individual capacity factors. At low pH the equilibrium shifts to the left and the observed capacity factor tends to that of the amino cation (AH2+). The introduction of metal ions, at above about pH 6, shifts the equilibrium to the right and the capacity factor tends to those of the chelates.Therefore, the capacity factor of 2-aminophenol may be controlled, and hence the selectivity of a separation of 2-aminophenol, by changing the position of the equilibrium by altering the pH, the metal ion concentration or by changing to a metal that forms either stronger or weaker chelates. Using these equilibria, a model that describes the retention can be generated. The capacity factor (k’) of a band of analyte passing through a column may be defined as amount in stationary phase amount in mobile phase k‘ = . . . . (ti) The amount of the analyte in each phase will be equal to the concentration of each species multiplied by the volume of that phase.Hence, based on the equilibria for 2-aminophenol (Fig. 2), equation (6) may be written as If q is the phase ratio (Vs/VM), i e . , the ratio of the volume of the stationary phase to that of the mobile phase, then k’ = The capacity factors of the individual species in the column may be defined as . . 4[AH2+ Is Protonated amine: k’,, = iAH2+IM (?[AHIS Neutral aminophenol: k’, = ~ . LAH1 M q [ W s Phenolate anion: k’, = ~ . . . . IA-1M (12) 1 : 1 chelate: klcl = ~ . . . . 4 [MA+ 1 s lMA+]M (13) 2 : 1 chelate: klc2 = ~ . . . . 4[MA2ls [MA2]M By rearranging equations (9)-( 13) to give expressions for the concentration of each species in the stationary phase, substituting these in equation (8) and cancelling the values for the phase ratio, equation (14) is obtained. Dividing both top and bottom by [AH]M gives equation (15) *774 ANALYST, JULY 1989, VOL.114 The equilibrium constant for each species in the mobile phase may be defined as (16) [H+]MLAHIM [ AH2+ 1 M . . . . . . Ka1 = Each of these expressions may be rearranged to give an equation for the equilibrium concentration of the analyte species in the mobile phase divided by the concentration of the neutral aminophenol ([AH],). Rearranging equation (16) gives Rearranging equation (17) gives Rearranging equation (18) gives [MA+]M = K , [ M 2 + ] ~ [ A - ] ~ . . . . (22) Rearranging equation (17) to give an expression for [A-IM, we obtain Substituting in equation (22) for [ k ] ~ from equation (23) and rearranging gives Rearranging equation (1 9) gives [MA~]M = K~[MA+]M[A-]M .. . . (25) Dividing both sides of equation (25) by [AH]M and substi- tuting for [MA+],/[AH], from equation (24) gives By substituting equations (20), (21), (24) and (26) in equation (15), and cancelling the [AH]M terms, the capacity factor can be expressed in terms of the variables of pH and metal ion concentration [equation (27)]. We can assume that protons and free metal ions have a negligible distribution into the stationary phase so that [M2+IM = [M2+] and [H+IM = [H+] and is, therefore, directly related to the pH. Equation (27) therefore describes the anticipated variation of the capacity factor of 2-aminophenol with the pH and metal ion concentration. It is a general model equation and could be applied to any similar metal ion - ligand system. For ligands that are not basic, the klb terms would be omitted.The equation also allows for the possibility that some of the 2 : 1 chelate may be formed during chromatography. However, this term contains an expression for the concentration of the Table 1. Variation of the capacity factor of 2-aminophenol with the concentration of analyte and nickel ions. Mobile phase: methanol - water (20 + 80 V/v) containing 0.26 M ammonium acetate, pH 7.24. Column: PLRP-S Capacity factor Ni*+ concentration/M 2-Aminophenol conccntrationh 0 0.04 0.06 10-2 11.70 7.02 4.05 10-3 11.77 7.27 4.11 10-4 11.73 7.33 3.11 lo-' __ 7.29 4.15 12 R 1 10 8 k 6 4 2 I I I 1 0 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 Nickel ion concentrationh Fig. 3. Experimentd and fitted data for the variation of the capacity factor ( k ' ) of 2-amfnophenol with the nickel ion concentration on a PLRP-S column.Mobile phase, methanol - water (20 + 80 VIV) containing 0.26 M ammonium acetate. Experimental data at: a, pH 6: H, pH 7.24; and V, pH 8. The lines are fitted curves from equation (33) anion [A-IM. As the anion concentration is dependent on the amount of analyte [AH] [equation (17)], the predicted retention would appear to be dependent on the sample size. However, in many instances K2 is much smaller than K1 so that the formation of any 2 : 1 chelate will be negligible. In addition, the metal ion concentration in the mobile phase (0.01-0.4 M) is in a large excess compared with the analyte concentration, which would be 0.001 M in the sample solution and will be diluted to 0.00001 M during elution, again favouring the 1 : 1 complex.At the other extreme if the formation of the 2 : 1 complex is highly favoured, all the analyte may be converted to the complex, effectively giving on-column derivatisation and the retention is then that of the neutral ML2 complex. This situation has been observed in the determination of dithio- carbamates with a mobile phase containing transition metal ions.26 In intermediate situations, when the formation of the neutral 2 : 1 complex is slightly preferred over the 1 : 1 complex, the over-all retention calculation would become more complex. With 2-aminophenol it was found, in practice, that the retention was independent of the sample size provided that the analyte concentration was low (10-3 M or less, Table 1).In this instance the over-all equation may therefore be simplified by omitting the final terms relating to [MA2] to give equation (28). In the absence of metal ions ( i e . , when [M2+] = 0) this equation may be reduced still further to an equation whichANALYST, JULY 1989, VOL. 114 775 describes the variation of the capacity factor of the ligand with pH [equation (29)]. k ’ , + k‘b[H+]/K,I + k’,K,?/[H+] 1 + [H+]/K,, + Ka2/[H+] k’ = . . (29) Equation (29) is similar to that derived by Horvath et al.” for the effect of pH on an ampholyte or zwitterion and contains the two equations derived independently by Miyake et ai.27 and Horvath et al.23 for the determination of dissociation constants by HPLC [equations (30) and (31)]. The derivation of these two equations was based on a similar but simpler model to give expressions for the variation of the capacity factor of a basic and an acidic compound with pH, respectively.These expressions describe sigmoidal curves (similar to those of pH titrations) as k’ varies between the extremes of the neutral (k’,) and fully ionised anionic (k’,) forms. Testing the Model Equation Equations (28) and (291, derived for the variation of the capacity factor of 2-aminophenol with the pH and metal ion concentration, were tested by comparison with the experimental observations using a non-linear least-squares computer program. The program fitted simplified versions of the equations to the experimental data and reported values of the fitted capacity factor for the experimental conditions of variation of pH or metal ion concentration.The accuracy of the fitted data was then assessed by plotting graphs of fitted capacity factors versus experimental capacity factors and obtaining correlation coefficients by linear least-squares calculation. In mobile phases free from metal ions equation (29) applies and can be expressed as A + Bx + Clx 1 + Dx + E/x k’ = . . . . . . (32) where x = [H+] and A , B, C, D and E are constants related to those in equation (29). This relationship was fitted to the experimental data obtained for the variation of the capacity factor of 2-aminophenol with the mobile phase pH and a close correlation was found between the experimental values and the fitted curve (Fig. 1). A linear correlation of the fitted data versus experimental data gave a close relationship with an intercept of 0.00429 and a correlation coefficient of 0.998, 1 2 , 1 lo\ * \ ---__ --- ’I--_ 4 6: 2 0 5.0 10.0 15.0 20.0 25.0 Copper ion c~ncentration/lO-~ M Fig.4. Experimental data for the variation of the capacity factor ( k ’ ) of 2-aminophenol with the copper ion concentration on a PLRP-S column. The broken line is the fitted curve. Mobile phase. methanol - water (20 + 80 VlV) contaking 0.26 M ammonium acetate at pH 7.24 showing that equation (29) had accurately described the experimental measurements. Using the values of A - E from the regression, it was possible to determine the values for the ionisation constants; pK,, = 5.13 and PK,,~ = 10.34. These values agree with those obtained from a spectroscopic study of pKa2 = 10.40 in methanol - water and are similar to those for aqueous solutions, pKaI = 4.79 and P K , ~ ~ = 9.97.2x The full model, equation (28), was then examined by considering the variation in the capacity factor of 2-amino- phenol with the metal ion concentration at a constant pH.In a previous paper20 we noted that the effect of the transition metal ions on the retention o f 2-aminophenol depended directly o n the metal ion concentration and also decreased in the order Cu2+ >>>Niz+ >Co2+ >Zn2+ >Cd2+ >Mn’+, which corresponded to the decrease in the magnitude of the formation constants ( K , ) for the 1 : 1 complexes. If the pH of the mobile phase is kept constant, equation (28) can be simplified to give F + G[M2+] 1 + N[M2+] . .. . . . k’ == (33) where F, C and H are constants related to the terms in equation (28). The formation constant, K , , of the 1 : 1 chelate can be calculated from H if pK,, and pKd2 are known. The capacity factor of the 1 : 1 chelate, k l c l , can be calculated from GIH = klcl . . . . * . (34) Equation (33) was fitted to four sets of experimental data obtained previously for the changes in the capacity factor of 2-aminophenol with the metal ion concentration.”) The correlations between the calculated values and the experimen- tal values for Ni2+ at pH 6,7.24 and 8, and for Cu2+ at pH 7.24 were all very close (Figs. 3 and 4 and Table 2). This showed that equation (28) was accurately describing the variation of k’ with the metal ion concentration. The formation constants, K 1 , for the nickel and copper chelates were calculated from the constants (Table 3) using the pK, value for the phenolic group obtained earlier (pK,2 = 10.34).Thc formation constants for nickel complexes at the different pH values were in reasonable agreement and the differences between them are probably due to the practical difficulties of obtaining values of k’ approaching that of k r c l . However, although as expected the derived values for the formation constant of copper (log K1 = 6.69 at pH 7.24) are relatively much greater than those for nickel (mean, log K1 = 4.59), both these effective formation constants are consider- ably lower than those reported in the literature for aqueous solutions (Cu2+, log K l = 9.25 and Ni2+, log K1 = 6.10).29 These differences are probably due to the presence of methanol in the mobile phase and to an interaction of the Table 2.Correlation of the calculated and experimental capacity factors of 2-aminophenol with changes in the metal ion concentration Correlation Metal ion pH Slope Intercept coefficient Ni’+ . . . . . . 6.0 0.973 0.229 0.994 Ni*+ . . . . . . 7.24 0.947 0.065 0.996 Nil+ . . . . . . 8.0 0.990 0.073 0.995 C U ~ + . . . . . . 7.24 0.995 0.048 0.999 Table 3. Formation constants of 2-aminophenol chelates determined by HPLC Metal ion pH Log1oK1 Ni2+ . . . . . . 6.0 4.20 Ni’+ . . . . . . 7.24 4.67 Ni2+ . . . . . . 8.0 4.85 Cu2+ . . . . . . 7.24 6.69776 ANALYST. JULY 1989, VOL. 114 metal ions with acetate ions in the mobile phase thus reducing their effective concentration.30 From the values of the empirical constants in equations (32) and (33) and the values of Kal, Ka2 and K 1 , it is possible to determine the individual capacity factors; k ’ , = 11.08, klh = 0.58, k’, = 0.22; k’,l(Ni2+) = 0.69; and klCl(Cu2+) = 1.86 [column, PLRP-S; mobile phase, methanol - water (20 + 80 VIV) containing 0.26 M ammonium acetate; pH, 7.241.As expected, the capacity factor of the neutral species ( k f n ) was large whereas those of the ionised species ( k J b ) and ( I c ’ ~ ) were low and corresponded to the values found experimentally (Fig. 1). The capacity factors of the ionised 1 : 1 complexes were also small. In a final comparison, the effect of changing the pH at a constant metal ion concentration was examined. Equation (28) can be simplified to give 3.4. 5. Chow. F. K., and Grushka, E., Anal. Chem., 1977, 49, 1756. Chow, F. K., and Grushka. E., Anal. Chem., 1978, 50. 1344. Cooke, N. H. C., Viavattene, R. L., Eksteen. R., Wong, W. S., Davies, G., and Karger. B. L.. J . Chrornatogr., 1978, 149, 391. Vivilecchia, R., Thikbaud, M., and Frei, R. W.. J . Chronza- togr. Sci., 1972, 10, 31 1 . Aigner, R.,‘Spitzy, H., and Frei, R. W.. Anal. Chern., 1976, 48, 2. Lam, S., and Grushka. E., J . Chronzarogr. Sci., 1977, 15,234. Mikes, F., Schurig, V., and Gil-Av. E., J. Chrornatogr., 1973, 83, 91. Kunzru, D . . and Frei, R. W., J . Chroniutogr. Sci., 1974, 12, 191. Eiceman, G . A., and Janecka, F. A., J . Chromatogr. Sci., 1983. 21. 555. Lochmiiller, C. H., and Hangac, H . H., J . Chromatogr. Sci., 1982.20, 171. Lochmuller, C. H . , Hill, W. B., Porter, R. M., Hangac, H. H., Culbcrson. C. F.. and Ryall, R. R., 1. Chromatogr. Sci., 1983, 21, 70. Lindner, W.. LePage, J. N.. Davies, G.. Seitz. D . E . . and Karger, B. L.. J . Chrornatogr., 1979, 185, 323. Vonach. B., and Schomburg, G . . .I. Chroniatogr., 1978, 149, 417. Lacey, E., and Brady, R. L.. J . Chrornatogr., 1983, 315, 233. Walters, V.. and Raghavan, N. V.. J . Chromatogr., 1979, 176, 47. Sternson, L. A , , and DeWitte. W. J . , J . Chromatogr., 1977, 137, 305. Sternson, L. A., Dixit, A. S.. Riley, C. M., Siegler, R . W., and Schoech, D . , J . Phurni. Biorned. Anal., 1983. 1, 105. Smith, R. M., Bale, S. J., Westcott, S. G . , and Martin-Smith, M., Analyst, 1987, 112, 1209. Foley, J . P., and May, W.E . , Anal. Chem., 1987, 59, 102. Hearn, M. J. W.. Editor, “Ion-pair Chromatography.” Chro- matographic Science Series, Volume 31, Marcel Dekker, New York, 1985. Horvath, C., Melander, W.. and Molnar, I.. Anal. Chem., 1977, 49, 142. Karger. B. L., IxPage. J . N.. and Tanaka, N., High-perfor- mance Liq. Chromatogr., Adv. Perspect., 1980, 1, 113. Vespalec. R., Vrchlabsky. M., arid Cigankova. M., J . Chro- matogr., 1977, 141, 313. Smith, R. M.. Morarji, R . L., and Salt, W. G.. Analysr. 1981, 106. 129. Miyake, K., Okumara. K., and Terada, H.. Chem. Phurm. B ~ d l . , 1985. 33, 769. Albert, A . , and Serjeant. E. P.. “The Determination of Ionization Constants,” Third Edition. Chapman and Hall, London, 1984. Sims, P., J . Chern. Soc., 1959, 3648. Fuentes, R., Morgan. L. O., and Matwiyoff. N. A . , Znorg. Chem., 1975, 14. 1837. 6. 7 . 8. 9. 10. 11. 12 13. A + Bx + Jlx 1 + D x + K/x . . . . . . . . . . . . k‘ = (35) 14. where A, B and C are the same constants as in equation (32) and J and K are constants which take into account the formation constant of the 1 : 1 chelate and the metal ion concentration. The equation was fitted to data obtained for mobile phases containing 0.01 or 0.02 M Ni2+ (Fig. 1). However, the correlation was poorer than obtained previously and at high pH values the expression levelled out, suggesting higher values for k ’ , of approximately 2. This anomaly probably arises because few experimental points could be obtained at pH values greater than 10 due to the insolubility of the nickel salts. 15. 16. 17. 18. 19. 20. 21. 22. Conclusion The model equation appears to give a good correlation between the experimental results and theory, suggesting that it correctly describes the experimental system. The values for the capacity factors and ionisation constants, which can be derived from the equation, agree well with those obtained by alternative techniques, but the formation constants of the complexes are much smaller than those reported for aqueous solutions, probably because of competing reactions from the buffer components and the presence of methanol in the mobile phase. 23. 24. 25. 26. 27 28. We thank the Science and Engineering Research Council for a CASE studentship to S. J. B. and Polymer Laboratories for a gift of the PLRP-S column packing material. 29. 30. References Pup er 8/04 76 9 C ReceiLied December 2nd, 1988 Accepted January 31st, 1989 1. 2. Davankov, V. A . , Adv. Chromatogr., 1980, 18, 139. Grushka. E., and Leshem, R . , Trends Anal. Chem., 1981, I , 95.

ISSN:0003-2654    

DOI:10.1039/AN9891400771

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST, JULY 1989. VOL. 114 777 Design of a Computer-controlled Electroanalytical System Richard M. Miller Unilever Research Port Sunlight Laboratory, Quarry Road East, Bebington, Wirral, Merse yside L63 3JW, UK Kathryn E. Thomas Department of Instrumentation and Analytical Science, University of Manchester Institute of Science and Technology, P.O. Box 88, Manchester M60 IQD, UK A dedicated high-level computer language has been written for use with a computer-controlled electroanalytical system. The language can be compiled t o produce executable machine code modules, and provides considerable flexibility i n the design of experiments. The compiled modules can be linked t o programs in other high-level languages, allowing very complex strategies to be implemented. The language is sufficiently general t o allow extension t o the control of other types of analytical instrumentation.Keywords: Computer control; real-time control languages; electrochemistry The introduction of microprocessors and microcomputers over the last ten years has had an enormous impact on the design and use of analytical instrumentation. Nearly all new commercial analytical instruments of any complexity now include a microprocessor or microcomputer for automatic control of the measurement process or manipulation of the data, or both. The impact of these developments in instrument control and data analysis on the work of an analytical scientist has been large, significantly changing the scientist’s role from being a skilled manipulator of experimental apparatus to being a skilled designer and interpreter of experiments.However, the design of many computer-controlled analyti- cal instruments does not take full advantage of the flexibility available with computer control. Instead, effort has been concentrated on computerising existing experimental strat- egies and making the system accessible to the professional analytical scientist, without requiring any computer skills. This has led to the development of instruments that are close copies of earlier analogue or digital instrumentation. These systems are often “closed,” in that they cannot easily be modified. As a result, the user is dependent on the skill of the designer in selecting the types and range of experiments that will be permitted, and is prevented from adopting his traditional role in modifying, developing and improving experimental methodology. Instead of liberating the experi- menter, computer control may lead to a situation where the user can devise an experiment which is physically possible using the equipment available, but which cannot be performed because of the limitations of the software.To try and overcome these perceived difficulties, we have been exploring alternative control strategies in an attempt to find one that meets the practical requirements of the measurement scientist. A\ a model for a typical computer- controlled instrument, we have chosen an electroanalytical system. Analytical electrochemistry uses a very wide range of different perturbation experiments to characterise the system being studied and therefore requires a flexible control system.Many of the techniques involve the measurement of system responses to transient perturbations, requiring high speed and precisely timed interaction between the control system and the experiment. Further, as the inputs to and outputs from an electrochemical experiment are electrical , interfacing is not difficult and effort can be concentrated on the control system. Design Criteria for a Control System A computer control system for an analytical instrument should have the following characteristics. 1. Flexibility; it should be possible to specify any experimental protocol that the in- strumentation is physically capable of carrying out. 2. Com- patibility; once an experiment has been defined, it should be possible to execute the experiment from within another program.3. Convenience; the system should have a good “human interface” which allows a wide variety of scientists to take advantage of the inherent flexibility. The requirement for flexibility is paramount. All design choices place limits on what can be achieved within a particular system, but the constraints should be selected carefully to preserve flexibility. Without flexibility it is difficult to introduce new measurement methods and pro- cedures, leading to stagnation and a reduction in innovation. Inappropriate methods may be used simply because it is too difficult to change the measurement system. Compatibility is required to allow communication between the experimental system and other software, enabling a previously defined experiment to be executed from within another computer program.If this is possible, a much wider variety of experimental protocols can be used, significantly enhancing the flexibility of the system. For example, more sophisticated data analysis and interpretation procedures could be employed. A series of experiments could be carried out automatically using different experimental parameters to validate the data or further characterise the system. Pattern search techniques could be used to optimise the experimental conditions. Ideally, the defined experiment should exist as an executable module that may be called from within programs written in a wide range of languages and which can exchange data with the calling program. Finally, the control software must be structured so that users with a wide variety of levels of computer skill should be able to use it. There is very little point in producing a system of high flexibility, which appears so complex to the users that they are unable to use it with confidence. If they do not trust the measurement system, they will not trust the results and will seek alternative methods to obtain the information required.This also implies that extensive error checking is required in the control software to ensure that the user is specifying a valid experiment. The “human interface” between an operator and the control program of a laboratory computer has been studied by Ziegler2 who considered that there were three possible approaches to communication between the operator and the computer program.1. The user identifies a pre-defined “method” for which all required parameters have already been specified. 2. The program leads the user through a question and answer sequence to select the “method” and supply the required parameters. 3. The user defines the “method” and the required parameters by entering specific778 ANALYST, JULY 1989. VOL. 114 commands taken from an application-oriented command language. Each approach has distinct advantages and disadvantages. The use of predefined methods was thought to be convenient if a certain type of analysis with a given set of parameters were to be performed frequently, effectively a “turnkey” operation. However. if requirements change, this method of selection tends to reinforce the use of previously specified compromise “methods” which may not give optimum results in the new environment.‘This approach allows operation by very inex- perienced users, but at the cost of a considerable reduction in flexibility. Question and answer dialogues, or the use of “menus” from which selections are made, can be useful for the occasional user who can be guided through a hierarchical tree structure to select and define the method. This is the method most widely employed in commercial instrumentation, and has been used successfully for voltammetric measurements.3 5 It offers a useful compromise between flexibility and user conve- nience. A wide range of options can be catered for, but the highly structured command sequence ensures that the inex- perienccd user can only choose compatible options and reasonable parameters for the experiment.However, every type of experiment to be performed must be included in the original design for the system, so that the appropriate menus and parameter validation can be included. To that extent it suffers the same disadvantages as the turnkey system; modification or extension of the system is difficult and unlikely to be a practical operation for the user. Software upgrades may be available from the manufacturer, but will only be produced to meet the common needs of a number of users, rather than the specific needs of an individual. It has also been noted that menu-driven systems can be irritating for experi- enced u ~ e r s because of the time-consuming nature of such dialogues and the amount of extraneous information which is displayed.2 Application-oriented command languages speed up com- munications with the computer, as only those parameters which actually need to be defined are referred to by the user. However, the system is more complex to use, as the operator must be familiar with the syntax and operation of all of the commands necessary for the correct operation of the instrurnentation. Hence, the approach is more suited to the experienced daily user who is faced with the need to use a wider variety of experiments than can be accommodated in the turnkey approach. It is less appropriate for occasional users than the question and answer approach. One such application-oriented command language is the problem oriented language (POL,) devised by Finger68 for data analjsis.The use of an application-oriented command language enabled the programmer to develop a new language tailored to the specific problem. POL allowed a flexible order of entry for the input. More than one command could be placed on the same line and “filler” words could be included in the input line to aid understanding and give the language a more natural feel. These words would be ignored in process- ing. The POL system took these input lines and translated them into a series of calls to cpecific sub-routines to carry out the individual operations. Provision was also made for the system to use previously recorded command files and to be able to simulate a hierarchical question and answer user interface for less experienced users.One of the main disadvantages of POL was thc: large size of the program. The use of filler words and flexible input order meant that the part of the program dealing with input validation and keyword extraction was complex, large and slow. Although these features contributed to a flexible and comprehensible input format, there was a heavy price to pay in operational efficiency. A more concise form of input would Gmplify the translation process. Prendergast et ale9 have described the use of a general- purpose robot control language, using the plain language programming system SAWY. The robot required programming in Tiny-BASIC, which is not very user friendly. SAVVY was used to produce/generate a control system which took commands in plain language and converted them into the appropriate instructions in ‘Tiny-BAsic to program the robot.The system was able to determine the meaning of imprecise instructions so that it could deal with mis-spellings and alternative phraseolo- gies. An important feature of this system was that all inputs were broken down into a series of primitive operations which could actually be executed by the robot. This is a key feature of robot and instrument control strategies and it is widely implemented in both commercial and experimental control systems. ARTS is another robot control language, which was devel- oped by Schlieper ef al. 10 to control a ZYMATE 1 laboratory robot. Again, this language builds up a program to execute a given function from a series of elementary sub-operations.In this instance the language is designed to be used with programmable robots and laboratory instrumentation, and provides an interface for the user between the various programming languages and methods of different instru- ments, and a common descriptive system. When a program is exccutcd, the operations specified are broken down into the appropriate coded instructions which must be sent to the particular device being addressed and then transmitted. The ARIS system is also programmed to translate messages and data from the instruments into the ARTS environment. As each approach to the problem of the main-machine interface has specific applications, we decided to develop a control system which would be capable of generating applica- tion programs for each of the three approaches.This can be accomplished by producing a system based on an applications- oriented command language. Such a system would provide the desired flexibility and would meet the requirements of compatibility with other programs as the command list would be translated into an executable form which could be stored as a separate module and called from another executing pro- gram. Because experiments can be defined and stored in an executable form, it is relatively simple to provide a supervisor program which communicates with the user via a question and answer dialogue, selecting experiments from a library and passing the required parameters to the experiment module. Similarly, a turnkey system could be devised. Having decided that an application-oriented command language is required, the question of programming environ- ment must be faced.All computers execute programs in machine code; a series of primitive operations micro- programmed into the processor itself.” The micro- programmed instruction set is small and does not include many of the more complex operations which are useful in practical programming. Various high-level programming lan- guages have, therefore, been created which allow the conve- nient development of complex programs, and which are then translated into machine code for execution by the computer. An application-oriented command language would consist of a set of commands with associated parameters which cause certain specific operations to be carried out. As there will not be one to one mapping between the primitive operation required for the experiment and the primitive operations provided by the processor, it is clear that a translation step will be required.Machine code or assembler language would not be suitable for the creation of a control language, unless the microprogramming of the processor can be changed so that the required primitives are available. Specialist processors have been manufactured for a variety of applications, includ- ing sound generation and video display,’2%13 but they are only cost effective where a generic problem is involved, which can justify specialised fabrication. In the area of controllers for scientific instrumen tation such generic solutions are not feasible, as the primitive operations are different in eachANALYST, JULY 1989, VOL.114 r 7 I n terface 779 instance. Machine code and assembler language have been used successfully for the control of electrochemical experi- ments, but have been limited to simple turnkey systems where flexibility was not a requirement.lj.15 However, in order to achieve the goals set out above in a practical system, we feel that a high-level command must be developed, which will be translated into machine code. There are three different ways in which a sequence of commands in a high-level language can be translated into a sequence of machine code instructions which can be executed. The program can be compiled as with FORTRAN, interpreted as with BASIC or compiled and interpreted as with threaded interpreted languages such as FORTH. In a compiled language the sequence of commands is translated, or compiled, as one operation to generate the machine code program.This is then linked with machine code sub-routines from a library to carry out any specialised primitive operations required by the language definition. The result of these operations is an executable machine code module. Error checking and input validation is carried out during the compilation step. The advantages of this approach are separation of translation and execution, and the reproducible generation of an executable module. It is possible to generate an exact mapping between the high-level commands and the corresponding machine code, which means that exact timing of operation can be achieved for real-time control. Disadvantagcs are the size and complexity of compilers, and the fact that unlike interpreters and threaded interpreted languages, they cannot readily be used for interactive program development.In interpreted languages, command lines are translated and executed one at a time, following the logic flow defined in the program. Each line is checked for correct syntax and valid use of keywords and variable names. If there are any errors, an error message is generated and execution stops. If the line is valid. it is translated and executed. The next line is then interpreted in a similar way. This method has the advantages that interpreters are simpler and smaller programs than compilers, and that an interpreted language can be used interactively. However, because each instruction is validated and translated each time it is executed, interpreted languages are slow.Further, because of various housekeeping opera- tions which occur in computers running interactively, it is impossible to predict the precise timing of the operations except under very limited circumstances.5 One solution to this difficulty is to use machine code sub-routines for critically timed operations, called from supervisory programs written in an interpreted language. This approach has been used by Bond and co-workers16.17 and Ploegmakers et al. 18 However, using experiment designs which call for a sequence of timed operations is difficult, as the uncertainty introduced by a series of calls from the high-level language means that extensive re-writing of the software may be required to reduce the number of calls needed.This inevitably makes the develop- ment of new methodologies time consuming, and requires considerable expertise on the part of the user. Threaded interpreted languages (TILs) have found wide application for control of experimental equipment.I9 TlLs differ from both compilers and interpreters in the method of execution of the program. The functional unit in TILs is the “word.” Each word is defined in terms of a limited number of built-in primitive operations or other previously defined words. A threaded code interpreter produces a fully analysed internal form, consisting of a list of addresses of other previously defined words. The list is “threaded” together during the first, compilation stage. During the second stage the interpreter executes consecutive words with no further analysis or search being necessary.TlLs normally have a wide range of primary internal operations, and the compilation mode allows great flexibility in the creation and definition of secondary words. Data are handled by last in first out stacks to conserve memory space and reverse polish notation is used to allow easy analysis of the instruction in terms of stack opcrations. TILs are superficially very attractive for this application. They are very compact, flexible and fast. However, their use of stacks and reverse polish notation makes them difficult for the novice user. In addition, the time taken to execute a particular secondary word depends on the way in which it is defined. Precise timing is difficult to establish.Finally, the enormous flexibility of TILs makes it difficult to ensure that only valid operations are carried out. TILs are therefore only likdy to be of use to the experienced and competent user. Despite the apparent problems, a microcomputer-con- trolled polarograph has been designed using FORTH as the programming language .20 However, no rapid scan techniques were attempted so the accuracy and reliability of precisely timed sequences were not tested. From discussions such as the foregoing, we concluded that the most appropriate choice of control system would be an application-oriented command language designed to be com- piled into a directly executed machine code module. In the following sections the design and testing of a computer- controlled voltammetry system based on these principles are described.Experimental Instrumentation A schematic diagram of the instrumentation used is shown in Fig. 1. A commercial microcomputer was chosen to control the instrumentation system and to act as the software development environment. A BBC Model B computer (Acorn Computers) with dual disk drives for data and program storage was used. A feature of the BBC microcom- puter is the lMHz expansion bus, which can be used to control various peripheral devices. The various modules of the voltammetry system communicated with the BBC microcom- puter through a local system bus, which was interfaced to the 1 MHz bus through a buffer card. The potentiostat was a standard three-amplifier design21 modified to allow digital control of the current follower gain and connection/disconnec- tion to the electrochemical cell. A schematic diagram of this system is shown in Fig.2. The control voltage to the potentiostat was provided by a 12-bit digital to analogue converter interfaced to the system bus. The output from the current follower was passed to a 12-bit analogue to digital converter (ADC) card, which transferred digitised values to the computer. The electrode system was a static mercury drop electrode (Model 303, EG & G Instruments). A digital output microcomputer Disk drive I I I I I 1 I + ““G controller ~~ Fig. 1. Schematic diagram of thc experimental instrumentation780 ANALYST, JULY 1989, VOL. 114 Cell Computer Zontrol amplifier 1 A I I I ToADC Current follower Fig.2. Schematic diagram of the computer-controlled potentiostat Time - A: lnitialise system B: Set potential C: Create new droD1 D: Delay F: Step potential Fig. 3. operations Decomposition of a voltammetric experiment into primitive card was provided to control the various functions of the static mercury drop electrode. Full details of this system will be published elsewhere.22.23 Software In order to meet the requirement for a highly flexible and expandable system, it was necessary to identify the primitive operations, which would correspond to the various keywords of the high-level control language. A large number of existing voltammetric techniques were analysed to determine the minimum number of operations into which they could be decomposed. Fig. 3 shows how a typical voltammetric technique, in this instance staircase voltammetry, can be represented using a small group of operations.Common elements were identified across the set of techniques and from these the individual primitive operations were identified. In addition to the primitive operations required for directly driving the potentiostat and for the data acquisition process, there were a number of program control operations. Specific operations were also required to control the electrode system. These latter primitive operations were specific to the partic- ular instrumentation being used, whereas the others were more generic. Finally, primitive operations were also required for the storage and display of acquired data. A full list of the elementary operations used in this system is given in Table 1. Table 1.Elementary operations required for the storage and display of acquired data Keyword Function Start . . Purge . . Stir . . Drop . . Delay . . Set . . Step . . Loop . . End loop . . - - Sample Display stop . . . . Name control program . . . . Toggle stir switch . . for 0.3 s . . Initiate time delay . . . . Toggle gas purge switch Initiate mercury drop dislodge/dispense and delay Set applied potential (absolute value) Change applied potential (relative to existing Mark start of repeating group of operations Mark end of repeating group of operations 2. Manipulate data value) . . . . . . 1. Set variables . . Measure current . . Display results . . Finish experiment 0 0 StartCSV 1 Purge 2 Delay 100 3 Purge 4 Set0 5 Drop 6 Delay1 7 Drop 8 Delay1 9 Drop 10 Delay5 11 Loop 100 PotentialN 12 Delay 0.005 13 SampleA 14 Step -0.01 2 15 End loop 16 Loop100 17 Delay 0.005 18 SampleA 19 Step0.012 20 Endloop 21 DisplayA 22 s t o p - 1 .2 Fig.4. Cyclic staircase voltammetric determination of 100 p.p.m. of Cd2+ in 0.1 M nitric acid In addition to these keywords, the language implements some mathematical operators and a limited form of variable storage. A total of 15 variables are available, labelled A-0. Variables A-E are data streams consisting of up to 256 two-byte words. These streams can be used with pointers and are intended to store data acquired during experiments. Hence, the expression “Sample C” in a program instructs the computer to read the current value on the input of the ADC and to store the result in the next available location in data stream C.Variables F-0 are single values that may be used in programs for parameters which must be manipulated arith- metically during execution. Mathematical operators for addition, subtraction, multi- plication and division are provided. These can be applied to any variables including data streams; for example, in a differential technique, a result data stream can be calculated by subtracting two data streams representing pairs of data points collected at different positions on the perturbation cycle in an electrochemical experiment. All of the primitive operations which generate an action were coded as machine code sub-routines in a run-timeANALYST, JULY 1989, VOL. 114 78 1 library. The routines were optimised carefully for speed and the elapsed time for each operation was noted so that it could be allowed for in the compilation stage to preserve accurate timings.Where there was a parameter associated with the keyword for the primitive operation, indirect addressing was used to allow the parameters to be varied after compilation. This addressing scheme uses pointers embedded in the machine code which indicate specific memory locations where the current value of the parameter will be found. Primitive operations associated with program control were not coded as sub-routines, but formed part of the main program generated by compilation of the high-level control language. The minimum data acquisition time for a single point in a sweep was 55 us. This is equivalent to a data acquisition rate of 18 kHz.The maximum timing error for the acquisition of a single point at this acquisition rate was +4 ps. Owing to the way in which timing is carried out, the errors in timing are always positive and reflect the maximum time that may be taken to check various status flags. The minimum data acquisition rate was 2.3 x l0-4Hz. The maximum timing error for a single point over all data acquisition rates was +8 ps. Having defined the structure of the control language, a series of BASIC programs were written to provide a develop- men t and application environment. These programs covered the creation of a control program, its compilation and execution. The module for creation of the control program allowed the user to enter the high-level code line by line.Parameter syntax and range were checked for each line as it was input and if any errors were detected the program forced the line to be re-entered. As soon as the final “stop” command was entered, marking the end of the control program, the entire program was checked for syntactical and logical errors. Completed programs were then saved on disk. Programs could be edited at any stage with the same level of error checking to allow easy modification. The second module compiled the control programs into executable machine code. This was achieved by taking the primitive operations from the control program line by line and converting them into sub-routine calls to the appropriate machine code sub-routines from the run-time library. Parameters were also stored in the appropriate memory locations for each sub-routine.Programme control functions such as loops were converted into machine code operations within the calling machine code program. After compilation, the executable machine code program could be saved on disk for later execution. The final module loaded the specified machine code program into memory, together with the run-time library, executed an initialisation machine code routine to set up the registers and instrumentation and transferred control to the machine code program to execute the experiment. Results and Discussion In order to test the usefulness of the completed system, a number of simple trial experiments were run. Fig. 4 shows the results obtained from a cyclic staircase voltammetry experi- ment carried out on a solution containing 100p.p.m.of cadmium nitrate in 0.1 M nitric acid, together with the control program used. Note the two loops in lines 11-15 and 16-20 creating the forward and reverse scans, which are stored sequentially in data stream A. The system also records the potential at which each data point was recorded, allowing the “display” command simply to convert the sequential data set into the conventional display for a cyclic voltammogram. These results are broadly in agreement with those obtained by Bond et af.24 The experiment was carried out at an effective sweep rate of 2.4 V s-1. A slightly more complicated example is the differential- pulse anodic stripping voltammetric analysis of laboratory tap -1.2 0 PotentialN 0 StartDPASV2 13 Delay30 1 Purge 14 Loop200 2 Delay2 15 DelayO.l 3 Purge 16 SampleA 4 Stir 17 Step0.05 5 Set-1.2 18 DelayO.l 6 Drop 19 SampleB 7 Delay 1 8 Drop 21 Endloop 10 Drop 23 DisplayD 11 Delay150 24 Stop 12 Stir 20 Step -0.045 9 Delay 1 22 D = B - A Fig.5. laboratory tap water Differential-pulse anodic stripping voltammetric analysis of - 5 -10 - Step height/lO-3 V -15 Fig. 6. Variations in step height and step length with vertex number during simplex optimisation of the staircase sweep voltammetric dctcrmination of 100 p.p.m. of Cdz+ in 0.1 M nitric acid water. The results of this experiment are shown in Fig. 5 together with the control program. The main loop occurs between lines 14 and 21. Here the current is sampled either side of the differential pulse and stored in data streams A and B.The difference between the two data streams is obtained in line 22. As these arithmetic operators apply to the entire data stream, only one statement is required and this can be placed outside the main loop reducing the computational overhead. These results are comparable to those obtained for mixtures of various heavy metals.25 The signal to noise ratio obtained with this system is not good compared with conventional commercial instrumenta- tion. This is largely due to electronic noise pick-up within the potentiostat and signal conversion electronics. However, despite the limitations of the experimental data, the results indicate the capabilities of the concept and demonstrate the use of the control language. A large number of other simple voltammetric experiments have been programmed using this system including linear staircase voltammetry, sampled d.c.polarography, differen- tial-pulse polarography and chronoamperometry.’”26J7782 ANALYST, JULY 1989, VOL. 114 / I 150 C 0 .- .I- .- > 100 g P U S m 50 0 2 4 6 8 1 0 1 2 1 4 Simplex No. Fig. 7. simplex number Change in the mean response and standard deviation with -0.1 -0.5 -0.9 -0.1 -0.5 -1.0 PotentialN Fig. 8. ( a ) Initial and (6) final results for optimising the signal to noise ratio in the staircase voltammetric detcrmination of 100 p.p.m. of Cd*+ in 0.1 M nitric acid The ability to execute a previously compiled experiment from within another program was regarded as being a very important feature of the system. If parameters could be passed from the supervisory program to the experiment, a high level of automation would be possible.It has been widely recognised that automation of chemical analysis must extend beyond the unattended repetition of routine tasks to automatic method selection, development and optimisation. At present, setting up an analytical method often requires significant input from a highly trained analyst, even when the method is well established. The time spent in setting up the experiment is frequently less productive than other tasks the analyst might be engaged in, and, conse- quently, there is a high level of interest in the possibility of exploiting artificial intelligence and optimisation methodolo- gies to minimise the involvement of trained scientists.The area where most work has been carried out is in the automation of chromatography, particularly liquid chromato- graphy.28729 Because of the tedious nature of the multiple experiments necessary to optimise a separation, a variety of strategies have been adopted which range from simple stepwise searching of the available space, through directed search methods such as simplex optimisation, to the use of expert systems. Although fully automated method develop- ment is not yet available, all the necessary components now exist. Automatic optimisation has also been applied to the field of flow injection (FI). Betteridge et al.3" reported a computer- controlled system where the flow-rate of the various liquid streams was under software control. This allowed the determi- nation of isoprenaline to be optimised by varying thc pH, reagent concentration and flow-rate.All of the control software was written in BASIC and no critical timing was required. The software was configured as a series of indepen- dent modules which could be linked flexibly to create the desired experimental structure. The strategy of passing parameters to an external routine from the calling program does not seem to have been exploited in this study. This may be due to the use of a single-language environment, which considerably simplifies communications between different components oi the control system at the cost of greater limitations on the choice of language for the supervisory process. The system was able to operate unattended and the results obtained with the system were consistent with those obtained for manual optimisation.In this study a linear sweep voltammetry experiment was set up using the control language and saved as an executable machine code program. A supervisory program was written in BASIC to execute this experiment, calculate the peak signal to noise ratio and to use that result in the simplex optimisation of the experimental parameters. The step height and step length in the linear sweep experiment were specified in the control program as variables which would be modified by some other program. In effect, the control program was told that the values for the step height and step length would be found in specific memory locations which could be modified externally. The simplex optimisation performed was the modified procedure of Nelder and Mead31 with a variable step size.The experimental conditions for the three initial vertices of the simplex were selected randomly from within the experimental space that was available to this instrument. The BASIC program calculated the conditions for each new experiment and called the machine code program that executed the experiment, passing the experimental parameters through the specified memory locations. After each experiment was performed, the results were displayed and the operator identified the voltammetric peak and a region over which the background root mean square noise could be calculated. The signal to noise ratio was then calculated and used as the experimental response to be optimised in the simplex pro- cedure.A new simplex was calculated, generating a new experiment to be carried out. This continued until the optimisation criteria were met. The optimisation was performed on the staircase sweep voltammetric determination of 100 p.p.m. of cadmium nitrate solution in 0.1 M nitric acid. Each linear sweep contained 100 steps. The step height was allowed to vary from -0.006 to -0.015 V with a precision of 1 x 10-3 V and the step length was allowed to vary from 0.001 to 0.020 s with a precision of 5 X 10-4 s. The optimisation process was set to terminate when the standard deviation of the response of the vertices of the simplex with respect to the mean response reduced to 1.0. Fig. 6 shows the values of the variables selected for the optimisation procedure.Not all vertices are shown. If a reflection produced a vertex outside the experimental range, for example a vertex with a negative step length, the experiment was not performed. The vertex was instead allocated a zero response to force a contraction. These vertices have been omitted from the figure. It has been suggested that an error exists in the implementa- tion of the modified simplex algorithm that we use. We have been unable to find such an error, although it is possible that one exists. The simplex optimisation control program was written to test the usefulness of the compiled experiment code as a routine called from within a supervisory program and the feasibility of varying the experimental conditions by passing parameters from the supervisory program to the experimentalANALYST. JULY 1989.VOL. 114 783 routine. As such, the simplex program was not tested as extensively as it would have been were it a key feature of the study. However, whether or not this program contains an error, the experiment was sufficiently successful to illustrate the usefulness of the approach. Fig. 7 shows the change in mean response and standard deviation with increasing simplex number. The experimental results obtained under the optimum conditions are compared in Fig. 8 with those of experimental results obtained under the conditions of one of the initial vertices. In this experiment, there was operator intervention to determine the response factor for each set of experimental conditions; however, in experiments given elsewhere,z7 we have demonstrated that completely automatic operation is possible.Conclusions We have shown that it is possible to construct a computer- controlled instrument with a high degree of flexibility by creating a specialiscd high-level language. This concept is implemented through identification of the necessary primitive operations for the type of experiment to be carried out and the programming of these primitive operations in optimised machine code. The system is not confined to existing experimental strategies; instead, any experiment that is physically possible with the available instrumentation can be programmed. A range of existing voltammetric experiments have been programmed successfully, requiring precise, high speed, real-time control. We have also demonstrated that an experiment programmed using this high-level language can bc saved as a single executable module which can be called from within another program.We have illustrated this by conducting a semi-automatic simplex optimisation of a linear sweep voltam- metry experiment. It is important to note that this approach is not limited to voltammetric experiments or this specific collection of in- strumentation. For example, if it were necessary to use a different potentiostat and electrode combination, it would only be necessary to modify the appropriate machine code sub-routines that refer to the elementary operations control- ling those parts of the system. Similarly, there is no reason why the same concept could not be extended to other kinds of instrumental experiment, for example many forms of spectro- scopy.Each type of experiment will have different elementary operations, but the structure and concept of the language would be the same. Finally, the system is in principle portable from computer to computer. In this instance there would be a need for a greater degree of conversion work, but the system would retain the same look and feel to the user. It would therefore represent a route by which a common user interface for a wide variety of computer-controlled instruments could be created. Because of the ability to produce executable modules relatively simply, and to link these modules with other software structures, it is possible to create measurement systems with any degree of interaction that the user requires.The user may be allowed complete freedom to re-define the experiment, may be limited to the selection of a few parameters, for example through a menu system, or may be presented with a complete turnkey operation, which cannot be modified to any degree. All of these options can be achieved within the same environment, which allows the system developer considerable freedom and flexibility of approach. The authors thank H. E. Dennis and C. E. Oduoza for conducting numcrous experiments with this system and for providing valuable information on its performance and limitations. I . 2. 3 . 4. 5 . 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. References Bettcridge, D., and Goad. T. B., Analyst, 1981, 106, 257. Zicgler, E., Anal. Chim. Acta, 1980, 122, 315. Bruntlett, C. S . , Curr. Sep., 1983, 5, 21. He, P., Avery, J. P., and Faulkner, L. R., Anal. Chem.. 1982, 54, 1313A. He, P., and Faulkncr, L. R., J . Electroanul. Ctzem. Interfacial E1tx.troclzem.. 1987, 224, 277. Finger, M., Byte, 1982, 7(12), 314. Finger, M., Byle, 1983. 8( 1 ). 254. Finger, M., Byte, 1983, 8(2), 314. Prendcrgast, D., Sladc, W., and Winkless, N.. Byte, 1984,9( I ) , 122. Schlieper, W. A., Isenhour. ‘r. L., and Marshall, J . C.. J. Chem. Inf. Compt. Sci., 1987. 27, 137. Carter, L. R., and Huzan, E., “Computer Programming in BASIC,” Hodder and Stoughton, London, 1981, p. 7. Williams, G.. Edwards, J . , and Robinson, P., Byte, 1985, 10(8), 83. Nicholls, B., Byte, 1985, 10(12), 169. Gunasingham, H.. Anal. Chim. Acta, 1985, 169, 309. Bond, A. M., and Norris, A., Anal. Clzem., 1980. 52, 367. Bond, A. M., Greenhill, t I . B . , Heritage, I. D., and Keust, J . B., Anal. Chim. Acfu, 1984, 165, 209. Bond, A. M., Heritage, 1. D., and Briggs, M. H., Anal. Chem., 1984, 56, 1222. Ploegmakers. H . H. J . L., Mertens, M. J . M., and van Oort, W. J . , Anal. Chim. Acta, 1985, 174, 71. Loeliger. R. G., “Threaded Interpretive Languages; Their Design and Implcmcntation,” BYTE Books, Pcterborough, 1981. Zollinger, D. Ph., Bos, M., van Veen-Blaaw, A. M. W., and van der Linden, W. E., Anal. Chinr. Acta, 1985, 167, 89. Bond, A. M., “Modern Polarographic Methods in Analytical Chemistry,” Marcel Dekker, New York, 1980, p. 178. Miller, R. M., and Thomas, K. E., unpublished work. ’Thomas, K. E., PhD Thesis, University of Manchester, 1986. Bond, A. M., O’Halloran, R. J . , Ruzic, I., and Smith, D . E., Anal. Chem.. 1978, S O , 216. Graneli, A., and Jagner, D., Anal. Chim. Acta, 1976, 83, 19. Dennis, H., Masfers Dissertution. University of Manchester, 1984. Oduoza, C. E., PhD Thesis, University of Manchester, 1987. Bcrridgc, J. C., Anal. Chirn. Actu, 1986, 191, 243. Berridge, J . C., “Techniques for the Automated Optimisation of HPLC Separations,” Wilcy, Chichester, 1985. Bctteridge, D.. Sly, T. J., Wade, A. P., and Porter, D. G., Anal. Chern., 1986, 58, 2258. Nelder, J. A.. and Mead, R., Comput. J.. 1965, 7, 308. Paper 81042791 Received October 27th, 1988 Accepted March 13th, 1989

ISSN:0003-2654    

DOI:10.1039/AN9891400777

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST, JULY 1989, VOL. 114 785 Biosensing With Coated-wire Electrodes Part 1. Glucose Sensors Hashim M. Abdulla," Gillian M. Greenwayt and Albert E. Platt Department of Science and Engineering, Humberside College of Higher Education, Cottingham Road, Hull HU6 7RT, UK Peter R. Fielden Department of Instrumentation and Analytical Science, UMlST, P.O. Box 88, Manchester M60 7QD, UK Enzyme electrodes based on fluoride- and iodide-selective coated-wire electrodes are described for the determination of glucose. The electrodes consist of a homogeneous poly(viny1 chloride) membrane containing both the electroactive species and immobilised glucose oxidase. Two enzyme electrodes based on iodide were investigated. The first of these utilised an ammonium molybdate catalyst, yielded a sub-Nernstian calibration with a mean slope of 32 k 1 mV decade-' and had a lifetime of 4d.The lifetime of the second electrode, for which peroxidase was also immobilised in the electrode membrane, was 6d. A super-Nernstian calibration was obtained for this electrode with a mean slope of 79 k 1 mV decade-'. Finally, the enzyme electrode based on fluoride gave a calibration with a mean slope of 32 & 1.5 mV decade-'; this also had a lifetime of 6 d. Keywords: Coated-wire electrode; glucose sensor; glucose oxidase In recent years there has been considerable interest in the use of enzyme electrodes, based either on potentiometric or voltammetric measurements. 1.2 Such electrodes usually con- sist of a sensor electrode, which monitors the change in concentration of a reactant or product of a reaction catalysed by the chosen enzyme; this enzyme is immobilised next to the surface of the sensor.Although these systems have been employed in clinical analysis,3 there is a need to develop a method that requires only small samples or one that can be utilised in vitro. The design of miniaturised ion-selective electrodes based on glass capillaries has been undertaken. These electrodes, however, are delicate and difficult to construct because they require an internal filling ~ o l u t i o n . ~ This paper describes an investigation into the possibility of using coated-wire electrodes that do not contain an internal reference solution to construct enzyme-based sensors. There are few instances of this type of electrode being utilised, although two urea electrodes based on pH measurements have been described.The basis of one was an antimony metal electrodes and the other an iridium oxide electrode.6 The coated-wire electrodes that we employed involved the use of ion exchangers as described by Helen et aZ.7 Both iodide- and fluoride-selective coated-wire electrodes were used to produce the enzyme sensors. For both sensors the initial enzyme reaction is the same: glucose oxidase Then, either the depletion of iodide ions is sensed8 glucose + 02-gluconic acid + H202 . . (1) MoVI catalyst or peroxidase HzOz + 21- + 2H+ 12 + 2H20 or the production of fluoride ions is monitored9 peroxidase H202 + XF- oxidised XF + where XF is an organofluorine compound. F- + H202 * Present address: Department of Pure and Applied Chemistry, Strathclyde University, Glasgow G1 lXL, UK.t To whom correspondence should be addressed. Present address: School o f Chemistry, University of Hull, Cottingham Road, Hull HU6 7RX, UK. Experimental Reagents Glucose oxidase (E.C. 1.1.3.4, 134 U mg-1) purified from AspergiZZus niger and peroxidase (E.C. 1.11.1.7,330 U mg-1) purified from Horseradish were obtained from Sigma (Poole, Dorset, UK). These were stored in a refrigerator at 4°C. Aliquat 336 (tricaprylylmethylammonium chloride), dodecyl- amine, dioctyl phenylphosphonate and poly(viny1 chloride) (PVC) low relative molecular mass were obtained from Aldrich (Gillingham, Dorset , UK). The required ion-associa- tion complex was obtained by shaking a 60% V/V solution of Aliquat 336 in decan-1-01 with an aqueous solution (0.5-1 M) of the halide salt.The fluoride and iodide standards were prepared from stock solutions of the analytical-reagent grade sodium salts in phosphate buffer (0.1 M, pH 5 ) . The glucose solutions were also prepared in phosphate buffer. Electrode Preparation The laboratory-manufactured electrodes consisted of plati- num wire (7 x 1 mm o.d., grade 1, Johnson Matthey, Royston, UK), soldered to a copper connecting wire and enclosed in glass tubing (6 mm i.d.). The bottom of the tube was sealed with epoxy resin. To construct the base electrodes (F- and I-) a solution of PVC with dioctyl phenylphosphonate plasticiser (20% mlV PVC, 0.3% mlV plasticiser in cyclohexane) was prepared and the platinum wire was coated by dipping it in this solution.After allowing the electrode to dry for 40min it was then dipped in the ion-association complex. The electrode was conditioned before use by soaking it for 15 min in a 100 mM halide solution. These electrodes were stored in air at room temperature and re-conditioned before use by repeating the conditioning process. Another fluoride electrode was pre- pared similarly, with dodecylamine in nitrobenzene as the ion exchanger. The enzyme electrodes were prepared by physical immobil- isation of the enzyme in PVC (see Table 1). Two enzyme electrode constructions were investigated. For membrane 1, the indicator electrode was constructed as described above, then dip-coated with a PVC layer (the same solution as for the base electrode in cyclohexane) and dry enzyme (ca.3 mg of786 ANALYST. JULY 1989, VOL. 114 Table 1. Construction and composition o f membranes for coated-wire glucose sensors Membrane No. 1 2 3 4 5 6 Base electrode . . . . . . . . 6 il- - F -~ Ion exchanger . . . . . . . . . . i Aliyuat 336 -------- Dodecylamine Membrane design . . . . . . . . Double layer i Homogeneous > - layer Glucoseoxidasc/mg . . . . . . 3 3 3 6 1 3 Catalyst for H202 cleavage . . . . Peroxidase Ammonium - Peroxidase > mol ybdate 5 4 3 2 1 -Log( [g iucosel/Mvl) Fig. 1. Calibration graphs for membranes: A, 2; B, 3 ; and (", 6 glucose oxidase and I m g of peroxidase for each base electrode) was pressed into the outer PVC layer before it hardened (typically 10 min after dip-coating). This resulted in an immobilised high-concentration layer of enzyme surround- ing the base coated-wire ion-selective electrode.In the second construction method (membranes 2-45), a homogeneous membrane was prepared. The electrode was first dipped into the PVC solution and allowed to dry for 10 min. The dry enzyme (1-6 mg of glucose oxidase) was then pressed into the soft PVC membrane to produce a single layer containing immobilised enzyme in contact with the platinum wire. After the PVC - enzyme coating had hardened the electrode was dipped into the ion-association complex as described previously. The electrodes were stored in pH 5 phosphate buffer solution at 4 "C. It was found that membrane permeation with the phosphate buffer solution was necessary for operation in order to ensure transportation of the species within the membrane.The electrodes consisting of both iodide and fluoride base electrodes were prepared in this manner. Electrodes were also constructed in which peroxidase was co-immobilised with glucose oxidase by the described proce- dures (see Table 1). Peroxidase was essential for the fluoride- based electrode in order to catalyse the liberation of free fluoride ions by hydrogen peroxide. Instrumentation and Measuring Procedure Measurements were made with a high-impedance voltmeter (Corning, PT16) using a silver - silver chloride double- junction reference electrode (Radiometer, KZ01). Solutions were stirred at a constant rate while results were recorded on a chart recorder (RBC Georz SE-120). All measurements were made in a water-bath (Grant JBl) maintained at 25°C and standard solutions were allowed to equilibrate for 1 h at this temperature, prior to use.For glucose determinations using the electrode based on the measurement of iodide (membranes 1-5), sodium halide solution was added to each glucose standard (the optimum iodide concentration was found to be LO-~M, which is consistent with the results of other workersh). For glucose determinations based on the measurement of fluoride, 2 cm3 of Hz02 (20% VlV) and 5 cm3 of 4-fluorophenol solution (3 g in 10 cni3 of methanol made up to 100 cm3 with distilled water) were added to 25 cm3 of glucose standard solution. Results and Discussion Optimisation of the Indicator Electrode Before preparing the enzyme electrodes it was essential to check the operation of the halide sensing (base) electrodes without enzyme.Of the base electrodes tested, the iodide- selective electrode was superior. This was found to have a near-Nernstian response for 0.1 mM-1 M iodide with a slope of SO k 1 mV decade-'. This type of electrode had a response time of l m i n (the time taken to reach 1 mV from the equilibrium potential) and its sensitivity did not change significantly over a period of 3 months. The effect of changing the temperature of a 1 mM solution of iodide on the e.m.f. was also studied and the electrode was shown to be insensitive to temperature changes over the range 10-40"C [slope = 0.08 mV decade-', relative standard deviation (RSD) = The fluoride electrode based on Aliquat 336 was found to be unresponsive, having a sensitivity of 8 mV decade-1.The electrode based on dodecylamine showed greater sensitivity (although well below the theoretical Nernstian response) for the range 1 mM-1 M with a slope of 32 k 1 mV decade-'. The deviation (from the log - linear relationship) at lower concentrations could be expected for the fluoride ion- exchange electrode as it is less selective than the iodide ion-exchange electrode according to the Hofmeister series. 10 The response time and lifetime were the same as those of the iodide electrode. 0.6%]. Optimisation of the Enzyme Electrode Membrane The first type of enzyme electrode (membrane 1) had a very low sensitivity (6 k 1.5 mV decade-'), which suggested that there was insufficient mobility between the layers. An enzyme electrode was then constructed in which a "cocktail" of all the active species was trapped in one layer of PVC.This type of construction was found to be superior for all membranes (2-6) and these are discussed in more detail below. The catalyst for the hydrogen peroxide cleavage of those sensors based on the iodide electrode was then studied by performing experiments on three electrodes. For the elec- trode with membrane 2, ammonium molybdate was used as the catalyst whilc oxygen was bubbled through the solution (Fig. 1j. A sub-Nernstian calibration was obtained for this electrode in the range 0.1-100 mM glucose (slope, 32 f: 1 mV decade-'). This electrode had a short lifetime of 4 d , a response time of 2-10min and a recovery time of 2-30min depending on the concentration (a 10-min response time and 30-min recovery time were needed only for very high concentrations, ie., 1 M).ANALYST, JULY 1989, VOL.114 787 Table 2. Characterisation of coated-wire glucose sensors Membrane 2 Membrane 3 Membrane 6 Linear range of Mean of calibration glucose/mM . . . . 0.1-100 1-100 1-100 Response time/min* . . 2-10 2-10 2-10 slope/mVdecade-l . . 32 k 1 7 9 k I 32 k 1.5 Washtime/min* . . . . 2-30 2-30 2-1s Lifetime/d . . . . . . 4 6 6 the highest value quoted is for 1 M glucose solution. +. Response and wash times increase with glucose concentration; I - I I I I 1 5 10 15 20 25 30 TI"C Fig. 2. solution, 10 mM Effect of temperature on the response of sensor 3. Glucose 3 2 1 -Log([glucosel/M) Fig. 3. 5 : and C. 6 d Effect of lifetime on the iodide sensor, membrane 2.A, 4; B, A glucose sensor was then prepared in which peroxidase was used as the catalyst (membrane 3). The peroxidase was co-immobilised with the glucose oxidase enzyme. With this design (Fig. 1) a super-Nernstian calibration was obtained over the range 1-100 mM (slope, 79 rfr 1 mV decade-'). The response time and recovery time of this electrode were the same as those of membrane 2 but the lifetime was 6 d . This enhanced sensitivity obtained with peroxidase was also observed by Al-Hitti et al.8 Electrodes with peroxidase were used for all further studies owing to their higher sensitivities and longer lifetimes. The optimum loading of the enzyme in the membranes was then investigated. Loadings of 1 , 3 and 6 mg were investigated (membranes 3-5).The main effect of changing the concentra- tion of the enzyme was found to be on the response time. Membrane 3 (3 mg, 400 U of glucose oxidase) was found to be the most useful analytically having an acceptable sensitivity and a reasonable response time (discussed above). A micro- meter, which was closed against the membrane until visual contact was observcd, was used to calculate the thickness of membrane 3 (0.24.3 mm). It should be noted that the amount of co-immobilised peroxidase was optimised in a similar manner and 1 mg (330 U) was used for all further studies. The enzyme electrode based on the fluoride sensor (mem- brane 6) was prepared with 3 mg of glucose oxidase and 1 mg of peroxidase (Fig. 1). The electrode response was found to be sub-Nernstian for 1-100m~ glucose with a slope of 32 k 1.5 mV decade-1.The properties of sensors based on mem- branes 2, 3 and 6 are summarised in Table 2 with the results quoted as the mean for three replicates of each type. Characterisation of the Sensors Efyect of sample temperature The change in e.m.f. on altering the temperature of a 10 mM solution of glucose was investigated. As can be seen in Fig. 2, the e.m.f. increases with temperature up to 20°C, at which point it begins to level off. If these results are compared with those obtained €or the iodide base electrode the increase in the electrode sensitivity must be due to the enzyme. As the temperature increases, the enzyme activity increases up to 20°C. Above this temperature a sharp levelling off is observed. Such behaviour is reproducible, but not yet understood.Sensor selectivity lnterferences in enzyme electrodes can be caused by two types of interferent, those that affect the base electrode and those that can be acted on by the enzyme. The first type of interference was studied for both the iodide and fluoride electrodes. The iodide electrode was found to be very selective and did not respond to S042-, CI-, NO3- or Br- at the 10 mM level. However, S2- did interfere at a concentration of 50 mM. The tluoride electrode was found to be much less sensitive and, using the mixed solutions method, was found to have a selectivity ratio of 30 for hydroxide ions. Glucose oxidase is not a completely specific enzyme and is known to affect sugars other than glucose.11 Both the iodide- and fluoride-based glucose sensors were investigated by the mixed solutions method with respect to their selectivity for glucose over maltose.A selectivity ratio of 0.89 was calculated in both instances. Sensor lijetime Fig. 3 shows the decrease in sensitivity of membrane 2 over several days. As can be seen, the lifetime of this electrode is short and the sensitivity and linearity of the curves decrease as the age of the sensor increases. This was because the enzyme was only physically immobilised. After several days stored in buffer solution, the yellow colour of the glucose oxidase could clearly be seen in solution. The simplicity of construction and low cost of manufacture (if a metal other than platinum were to be utilised) mean that these electrodes could be designed to be disposable.Alternatively, the enzyme could be immobi- lised more permanently by using a chemical immobilisation technique (e.g. , cross-linking with glutaraldehyde). Conclusions This work has investigated the operation of coated-wire electrode-based enzyme sensors. The simplicity of construc- tion, particularly with the homogeneous membranes contain- ing both enzyme and ion sensor, offers the possibility of the mass production of low-cost disposable spot-test sensors for glucose. Although, in terms of over-all performance, these sensors offer no immediate advantage over other designs, they should not be discounted. Their small size and inherent robustness make them ideal for clinical sensing applications. Further work is necessary to ensure reproducible performance and methods to improve shelf lifetime should be investigated.788 ANALYST, JULY 1989, VOL. 114 The authors acknowledge the Iraqi Government for providing financial support for Hashim M. Abdulla and also thank Martin Cawley for technical assistance. References Vadgama, P., in Covington. A. K . , Editor, “Ion-Selective Electrode Methodology,” Volume IT, CRC Press, FL, 1979. p. 23. Albery. W. J., and Bartlett, P. N . , J . Electroanal. Chem., 19x5, 194, 211. Home, P. D., and Alberti, K. G. M. M., in Turner, A. P. F., Karube, L.. and Wilson, G. S . , Editors, “Biosensors, Funda- mentals and Applications,” Oxford University Press, Oxford, 1987, p. 723. Silver, I . A., Philos. Trans. R. Soc. London, Ser. B , 1987,316, 161. 5 . 6. 7. 8. 9. 10. 11. Alexander, P. W.. and Joseph, J . P., Anal. Chim. Acta. 1981, 131, 103. Ianniello, K. M., and Yacynych, A. M., Anal. Chim. A c f u , 1983, 146, 249. Helen, J., Cormack. G., and Freiser, H . , A n d . Chem., 1972, 44. 857. Al-Hitti, I. K., Moody, G. J., and Thomas, J . D. R., Analyst, 1984, 109, 1205. Siddiqi, I . W., Clin. Clzem., 1982, 28, 1962. Buck, R. P., Sensors Actuafors, 1981, 1. 197. Nagy, G., von Storp. L. M., and Guilbault, G. G.. Anal. Chzm. Acta, 1973, 66, 443. Puper 8104759F Received December lst, 1988 Accepted March 15th, 1989

ISSN:0003-2654    

DOI:10.1039/AN9891400785

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST, JULY 1989, VOL. 114 789 Nylon Tube 0-Alkylation for lmmobilisation of Covalent Enzymes Francis N. Onyezili Department of Foundation Studies, University of Agriculture, Private Mail Bag 2373, Makurdi, Nigeria Nylon tube 0-alkylation and its effects on the catalytic efficiency of urease (E.C. 3.5.1.5) immobilised on the tube were examined using dimethyl sulphate (DMS) and diethyl sulphate (DES) as alkylating agents on Nylon 6 of different wall thicknesses. Effective and easy control of the alkylation process was possible as both reagents scarcely reacted with nylon at room temperature (25 "C) and as high-temperature alkylation, at 100 "C, could be stopped rapidly by immersing the tube in an ice-bath. The optimum incubation times for alkylation with DMS and DES were 3 and 10 min, respectively.Diethyl sulphate, a less toxic reagent than DMS, caused less damage t o the nylon tubes and produced more chemically stable 0-alkylated derivatives. However, it was less efficient than DMS in creating reactive sites for covalent attachment of catalytically active urease on the tube. Although thick-walled nylon tubes immobilised more active enzyme, such tubeswere less pliable than thin-walled tubes and could pose operational problems. Keywords : Nylon tube 0-a lk yla tion; u rease irn m obilisa tion Nylon tubes, because of their mechanical strength, pliability and ease of incorporation in automated continuous-flow assay systems, are popular support matrices for immobilised en- zymes.1,' However, this material is predominantly hydro- phobic and essentially chemically inert except at the few amide linkages interspersed in its structure.Therefore. covalent coupling of an enzyme to a nylon tube is effected by modifying the tube to alter its chemistry and so provide a suitable environment for binding the enzyme without denaturing it. Partial hydrolysis of the tube with hydrochloric acid3 and non-hydrolytic cleavage with dimethylaminopropylamine4 have been employed for this purpose, but high temperature 0-alkylation of the tube is often the method of choice because it yields the most catalytically active immobilised enzyme derivative.5 However, little is known both about the process of nylon tube 0-alkylation and the final effects of the process on the catalytic efficiency of the immobilised enzyme derivative. In this paper the process is examined further by using both dimethyl sulphate (DMS) and diethyl sulphate (DES) as alkylating agents and immobilising urease (E.C.3.5.1.5) on two different sizes of Nylon 6 tubes. Experimental Thin-walled (1.0 mm i.d., 0.25 mm wall thickness) and thick-walled (1.0 mm i.d., 0.75 mm wall thickness) nylon tubes were obtained from Portex (Hythe, UK). Dimethyl sulphate and DES were supplied by BDH (Poole, UK) and stored at 25 "C over calcium hydride. Glutaraldehyde (25"/0 mlV aqueous solution) was obtained from BDH and stored in small aliquots at -20 "C. When required, it was thawed out, diluted as appropriate and stored subsequently at 4 "C. Jack bean urease was obtained as a freeze-dried powder from Miles-Servac Laboratories (Maidenhead, UK) with a specific activity of 100 U mg-1 at pH 7.0 and 25 "C.Unless indicated otherwise. all the other reagents were of analytical-reagent grade and were used without further purification. Nylon Tube Alkylation This procedure, because of the toxicity of the alkylating agents, was performed in a fume cupboard using gloved hands and protective eye glasses and with the utmost caution. A 1.0-m length of the tube was filled with either DMS or DES, using a syringe, and sealed tightly at both ends with steel clamps. After incubating at 100 "C in a boiling water-bath for the desired time, to allow alkylation to occur, the tube was immersed in an ice-bath for 10 min. The excess of alkylating agent was then removed by perfusing the tube with 50 ml of re-distilled methanol at a flow-rate of 5 ml min-1 using a peristaltic pump.Determination of Cysteine Uptake by Alkylated Tubes The alkylated tube was filled with a freshly prepared solution of 0.5 M cysteine (pH 9.0) containing 0.001 M EDTA and incubated at 25 "C for 5 h. Unreacted cysteine was recovered by perfusing the tube with 50 ml of 0.001 M EDTA containing 0.1 M KC1. The cysteine content of the effluent was deter- mined with S,S'-dithiobis(2-nitrobenzoic acid) (DTNB) as described by Eliman.6 The uptake of cysteine by the alkylated tube was calculated from the difference between the cysteine concentration before and after incubation in the alkylated tube. Further Chemical Modification of Alkylated Tubes and Urease Immobilisation Chemical modifications of the alkylated tube with 1,h-diami- nohexane and glutaraldchyde were performed as described previously.7 'The 0-alkylated tubes were reacted first with 0.2 M 1,6-diaminohexane in methanol for 3 h at 25 "C and then with 2% mlV glutaraldehyde in 0.2 M borate buffer (pH 9.0) for 8 min at 25 "C.Urease was immobilised by filling the modified tube with a solution containing 2 mg ml-1 of the enzyme in 0.05 M EDTA buffer (pH 6.6) containing 0.001 M mercaptoethanol and incubating for 4 h at 4 "C. Unbound enzyme was removed by sequential perfusion with 1.0 M NaCl solution and distilled water. Assay of Immobilised Urease The immobilised urease activity was determined by measuring the ammonia in the effluent stream following perfusion of a solution of 0.05 M urea in 0.05 M EDTA (pH 6.6) through the immobilised enzyme tube at a flow-rate of 4 nil min-1.The ammonia assay procedure has been described by Chaney and Marbach. 8 Results and Discussion The esters resulting from 0-alkylation of nylon tubes form mono-substituted amidines with cysteine, as follows:790 ANALYST, JULY 1989, VOL. 114 Incubation timeimin Fig. 1. Cysteine uptake in DMS alkylated thin-walled nylon tubes. All alkylations were performed at 100 "C on 1.0-m lengths of tube. Cysteine uptake was determined as described under Experimental; 100% uptake was equivalent to 0.086 pmol of cysteine bound per metre of tube 100 8 s .- > 50 .I- .- .I- 2 0 4.0 8.0 Incubation timeimin Fig. 2. Diethyl sulphate incubation time and the 0-alkylation process. Alkylation with DES was performed at 100 "C on 1.0-m lengths of thin-walled nylon tube.A , Urease coupled directly to the alkylated tube; and B, tube modified further with 1 ,h-diaminohexane and glutaraldehyde before enzyme coupling. Urease activity was measured at 37 "C in the presence of 0.05 M urea in 0.05 M EDTA buffer (pH 6.6) perfused at a flow-rate of 4 ml min-1; 100% activity produced 12 pmol of ammonia per minute per meter of tube I OR"' 0-alkylated nylon Cyst e i n e w I I I NH + R"'0H COOH - C- H CH2SH Mono-substituted amidine derivative of nylon Hence, the uptake of cysteine by alkylated nylon tubes is a measure of the 0-alkylation process itself. Fig. 1 shows that the alkylation of thin-walled tubes with DMS at 100 "C was a fast process, with 80% of the maximum uptake of cysteine achieved after alkylating for 2 min.Thereafter, alkylation slowed down and, beyond 4 min, structural disintegration of the tube was observed. Esters such as those formed by alkylating nylon tubes have been reviewed9 and are reported to be sufficiently hygroscopic to bring about their own decomposition, particularly in the absence of stabilising aromatic groups. Work reported recently10 pro- vides experimental evidence of such hydrolysis of alkylated nylon tubes, which would explain their disintegration on prolonged exposure to the alkylating agent. As the reactive sites created by 0-alkylation ultimately provided the basis for covalent enzyme binding on the tube, the efficiency of 0-alkylation was monitored by exposing the 0 0.2 0.4 0.6 lncu bation timelmin Fig.3. Nylon tube wall thickness and thc 0-alkylation process. All alkylations were performed at 100 "C on 1.0-m lengths of tube with DMS. A, Thick-walled tubes; and B, thin-walled tubes, both modified (after alkylation) with 1,6-diaminohexane and glutaraldehyde before urease immobilisation, Urease activity was measured as described in Fig. 2; 100% activity produced 25 pmol of ammonia per minute per meter of tube [(XI = time at which disintegration of the tube was observed] alkylated tube, either directly or after further chemical modification, to urease and measuring its catalytic activity. Also, DES, which is a milder alkylating agent than DMS, was used thus permitting the alkylation process to be observed for a longer period. The results of these investigations are shown in Fig.2. The levels of enzymic activity in the tubes indicated that 0-alkyiation of thin-walled tubes with DES was slow for the first 2 min, then increased and continued for 10 min before structural disintegration began. Compared with alkylated tubes to which the enzyme was coupled directly, approxi- mately 24% less urease activity was detected in tubes that had been alkylated for the same optimum period (10 min), but modified subsequently with 1 ,6-diaminohexane and glutaral- dehyde before the enzyme was coupled. This finding (Fig. 2) is, presumably, an indication of the different chemistries governing urease immobilisation in each instance and suggests that subsequent modifications of 0-alkylated nylon tubes reduce the number of active binding sites available to the enzyme.These subsequent modifications were introduced to create reactive sites for the enzyme away from the proximity of the hydrophobic nylon surface, which could cause unfolding (and denaturation) of the enzyme protein. However, as the results suggest, such modifications of the 0-alkylated tube reduced its binding efficiency, at least with regard to catalytic- ally active urease. A comparison of DMS and DES alkylation of thin-walled nylon tubes was made by allowing the alkylation with each reagent to proceed for the same length of time (3 min) at 100 "C before the tubes, which had been similarly chemically modified,7 were exposed to an enzyme coupling solution of urease (2 mg ml-1) - mercaptoethanol (0.01 M) - EDTA pH 6.6 buffer (0.05 M). For the DMS - and DES - 1,6-diamo- nohexane - glutaraldehyde - urease tubes the urease activities were 9.6 and 0.9 U, respectively (the units of urease activity were equivalent to pmol of ammonia per minute per meter of the tube using the assay conditions described under Experimental).As DES alkylation could be performed for longer periods than alkylation with DMS, and as an increase in DES incubation time (within the limits beyond which disinte- gration of the tube occurred) resulted in a progressive increase in the activity of the final immobilised enzyme derivative (Fig. 2), it would appear that DES-alkylated tubes were more stable than those alkylated with DMS. This view is supported by a report9 that at elevated temperatures a degradative "0" to "N" migration of alkyl groups may follow the alkylation process.This migration, which is analogous to the Chapman rearrangement," would be less likely to occur in DESANALYST, JULY 1989, VOL. 114 791 alkylated tubes where the molecular size of the newly introduced alkyl group is larger. The wall thickness of the nylon tube was a significant factor in the O-alkylation process as thick-walled tubes survived DMS alkylation for longer periods than did the thin-walled type (Fig. 3). Further, DMS alkylation for the same length of time (3 min) produced a three-fold higher urease activity in the thick-walled tube compared with that in the thin-walled tube (Fig. 3). The nylon tube matrix exists in various degrees of order and disorder12 and has amorphous regions that are unsuitable for enzyme immobilisation.3 A lower proportion of such amorphous regions in the thick-walled tubes could explain the higher activities observed in these tubes.This suggestion is merely speculative as the microstructure of the nylon matrix used in this study was not investigated. Also, a surface area effect in which the deeper nylon matrix of the thick-walled tubes binds more catalytically active urease could account for the higher enzyme activities in these tubes. Hence, from the results obtained, thick-walled tubes are to be preferred to thin-walled tubes. However, the former are less pliable and could pose operational problems in large-scale applications. Also, the problem of completely washing-out the reagents, which diffuse into the tube matrix, is greater for thick-walled tubes.Conclusions The optimum incubation times for the O-alkylation of thin-walled nylon tubes with DMS and DES were 3 and 10 min, respectively. Thick-walled tubes produced more active immobilised enzyme derivatives than did thin-walled tubes but were less pliable, which could pose operational problems. Effective and easy control of the alkylation process was possible with both DMS and DES as these reagents scarcely reacted with the tubes at room temperature (25 “C) and as their action could be stopped rapidly by immersing the tubes in an ice-bath. Tubes alkylated with DES appeared to be more chemically stable than those alkylated with DMS. Also, DES, which is a less toxic reagent than DMS, caused less damage to the nylon tubes. However, DES was less efficient than DMS in creating reactive sites for enzyme immobilisation and different alkylating conditions may be required to realise the full potential of the former. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. References Sundaram, P. V., J . Solid-Phase Biochem., 1978, 3, 185. Werner, M., Mohrbacher, R . J . , Riendeau, M., Murador, E.. and Carnblaghi, S., Clin. Chem., 1979, 25, 20. Inman, D. J., andHornby, W. E., Biochem. J . , 1972,129,255. Hornby, W. E., Inrnan, D. J., and McDonald, A . , FEBS Lett,, 1972, 23, 114. Campbell, J., Hornby, W. E., and Morris, D. L., Blochim. Biophys. Acta, 1975, 384, 307. Eliman, G., Arch. Biochem. Biophys.. 1959, 82, 70. Onyezili, F. N., and Onitiri, A. C., Anal. Biochem., 1981, 117, 121. Chaney, A. L., and Marbach, E. P., Clin. Chem., 1962,8, 130. Roger, R., and Neilson, D., Chem. Rev., 1961, 61, 179. Onyezili, F. N., Biotechnol. Bioeng., 1987, 29, 399. Orchin, M., Kaplan, F., Macomber, R. S., Wilson, R. M., and Zirnrner, H., in “The Vocabulary of Organic Chemistry,” Wiley, New York, 1980, p. 351. Bawn, C. E., in “The Chemistry of High Polymers,” Butter- worths, London, 1948, p. 11. Paper 81041 981 Received October 24th, 1988 Accepted January 5th, 1989

ISSN:0003-2654    

DOI:10.1039/AN9891400789

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST, JULY 1989, VOL. 114 793 15th-Differential Polarographic Determination of Trace Amounts of Selenium(1V) and Selenium(V1) in Natural Waters at a Dropping Mercury Electrode Wu Dunhu" Changchun Institute of Geograph y, Academia Sinica, Changchun, Jilin Province, People's Republic of China Zhang Diyangt Changchun Institute of Physics, Academia Sinica, Changchun, Jilin Province, People's Republic of China Li Xiaoming Changchun College of Geolog y, Changchun, Jilin Province, People's Republic of China This paper describes a polarographic method for the determination of trace amounts of SelV and SeVl in natural waters using a dropping mercury electrode. In an HC104 - Na2S03 - NH3 - NH4CI - NH20H.HCI - K103 system (pH lo), the selenium complex SeS032- gave a peak potential at -0.57 V versus Ag.The peak current was directly proportional to Se in the concentration range 0.01-1 pg 1-1. The limit of detection was 10 ng 1-1. The proposed method is rapid, simple, sensitive and accurate. The recovery of Se from standard additions to samples of natural waters was between 93 and 105%. Keywords : Selenium; po la rograp h y; en vironm en tal chemistry; natural waters The role of Se as an important biological trace element in the human body has been investigated1 and close correlations between Kaschin back disease, heart disease, Kasham disease and various cancers and the Se content of waters, soils, foods, animals and plants found in certain areas have been re- ported.2.3 Owing to the presence of trace amounts of Se in the environment and in the human body, attempts were made to develop a method for its determination.A number of methods. including neutron activation analysis,4 atomic absorption spectrometry,5 fluorimetry,6 gas chromato- graphy,7 polarography8.9 and spectrophotometry'o have been employed for the determination of trace amounts of Se in the human body and in hair. However, some of these methods require expensive instrumentation and strictly controlled experimental conditions while others are not sufficiently sensitive to satisfy the requirements for determining trace amounts of Se in water. This paper describes a 1.5th-differential polarographic method for the determination of trace amounts of SeIV and SeVl in natural waters." The Se is measured using a very sensitive catalytic wave which gives a detection limit of 10 ng I-' in HC104 - Na2S03 - NH3 - NH4CI - NH20H.HCI - KI03 using an XJP-821 Neopolarograph.Experimental Apparatus and Reagents A laboratory-built XJP-821 Neopolarograph and an LZ3-100 recorder were used. In a three-electrode system, the dropping mercury electrode was the working electrode, an Ag electrode the reference and a Pt electrode the auxiliary. Na2S03 (10Y0), 0.8 M. Analytical-reagent grade anhydrous Na2S03 (10 g) was dissolved in 100 ml of distilled water (freshly prepared every 3 d). K103 (4%), 0.2 M. Analytical-reagent grade KI03 (4 g) was dissolved in 100 ml of distilled water. * Present address: Department of Applied Chemistry, Dalian Railway Institute, Dalian City, Liaoning Province, People's Republic of China.t To whom correspondence should be addressed. Present address: Design and Research Institute of Petrochemical Industry of Jilin Province, 77 Stalin Street, Changchun, Jilin Province, People's Republic of China. NH3 - NH4Cl buffer. Prepared by dissolving 50 g of NH4CI (analytical-reagent grade) in 190 ml of ammonia solution and diluting to 250 ml with distilled water (pH 10). NH20H. HCl, 5%. Analytical-reagent gradc NH20H.HCI ( 5 g) was dissolved in 100 ml of water. Standard solution of SeIV, 1.0 mg ml-1. Prepared by dissolving spectrometric grade Se02 (1.4053 g) in 1 1 of 0.1 M HCl and then stored in a refrigerator. Solutions containing 0.01 and 0.001 pg ml-1 of Se1V were prepared by appropriate dilution of the 1.0 mg ml-1 standard solution. Standard solution of SeVI, 1.0 mg ml-1.Prepared by dissolving Na2Se04 in distilled water. Stock solutions contain- ing 0.01 and 0.001 pg ml-1 of SeV1 were prepared by appropriate dilution of the standard solution. The water used throughout was distilled in a quartz still and the HC104, HN03 and HCI used were of super grade quality. Procedures Pipette 0.05,0.10,0.20,0.30,0.40 and 0.50 ml of SeIV solution (0.001 pg ml-1) into 10-ml electrolytic cells and to each cell, 0.2 0.3 0.4 0.5 0.6 0.7 - E N Fig. 1. Differential-pulse polarographic behaviour of SeIV. 1, Solution blank; 2 , O . O l ; and 3,0.05 pg I-' of SeIV. Measurement of the peak height is indicated by x794 ANALYST, JULY 1989, VOL. 114 Volume of HClO,/ml Fig. 2. solution containing SeIV Effect of volume of HCIOj added on the peak height of the t Y 9 0 5 10 15 20 d Na2S03, % Fig.3. solution containing SelV Effect of Na2S03 concentration (%) on the peak height of the including one blank, add 0.05 ml of HC104 and 1.0 ml of 10% Na2S03. Leave the solutions to stand for 15 min after mixing, then add 1.0 ml of NH3 - NH4C1 buffer (pH lo), 0.25 ml of 5% NH,OH.HCl and 0.50 ml of 4% KI03 in sequence to each cell. After mixing, dilute the contents of each cell to 5.0 ml with distilled water and leave to stand for about 1 h before performing the determination. Fig. 1 shows the 1.5th-differential polarographic curve of SetV obtained using the three-electrode system and scanning from -0.20 to -0.70 V. The peak occurred at -0.57 V. The Se present in natural waters appears almost entirely in the polarogram as SeIV and SeVI does not make a contribution to the wave obtained in a catalytic system.Therefore, for an accurate determination of the total Se present, it is necessary to reduce any SeVI to SeIV by the addition of HC1.12 The following procedure is recommended for the reduction of SeVI to SeIV. Add 0.05 ml of HC104 and 0.05 ml of HC1, in sequence, to electrolytic cells containing 0, 0.10, 0.20, 0.30, 0.40 and 0.50 ml of the 0.001 pg ml-1 stock solution of SeVI. Stand the cells on a hot-plate or in a sand-bath, allow the solution to evaporate slowly until fumes of HC104 appear, then cool the solution and add 1.0 ml of 10% Na2S03 as reductant. The procedure for the determination of SeVI is the same as that for SeIV. Results and Discussion Selection of the Catalytic System For the determination of SeIV, various compositions for the catalytic system including the components Na2S03, HC104, Na2S04, KHC03, K2C03, NH3 - NH4CI, KI03 and NH20H.HCI were studied. The HC104 - Na2S03 - NH3 - NH4CI - NH,OH.HCl - KI03 system was selected, having a high sensitivity for the SeIV catalytic wave.Effect 0.f the amount of He104 added In order to convert Na2S03 to H2S03 for use in the catalytic system, HC104 was added. As Fig. 2 shows, the maximum peak height for SeIV was achieved with the addition of 0.05 ml Selection of the Na2S03 concentration Fig. 3 shows the effect of the Na2S03 concentration (as reductant in acid medium) on the peak height of SeIV. The of HC104. 0 0.5 1.0 1.5 2.0 NH3 - NH,Cllml Fig. 4. peak height of the solution containing SeIV Effect of the volume of NH3 - NH4Cl buffer added on the I I I , 0 0.25 0.50 0.75 1.00 N H 0 H .H C I ( 5% )/m I Fig. 5. height of the solution containing SeIV Effect of the volume of NH20H.HCI added on the peak 0 2 4 6 8 K103, Yo Fig. 6. solution containing SeIV Effect of K I 0 3 concentration (%) on the peak height of the peak maximum occurred with the addition of 1.0 ml of 10% Na2S03 to the analytical system (final volume, 5 ml). Effect 0.f the amount of NH3 - NH4Cl buffer added The effect of the amount of NH3 - NH4CI buffer added on peak height is shown Fig. 4. The limiting value was reached when 1 ml of buffer solution was added. Eflect of the amount of NHzOH.HC1 added Fig. 5 shows the maximum peak height obtained when 0.25 ml of 5% NHzOH.HCI was added, in a total volume of 5.0 ml.The increase in the sensitivity of the Se peak height may be due to the presence of NHzOH.HCI, which influences adsorption at the surface of the Hg drop, and the correspond- ing increase in the peak height caused by the catalytic reaction occurring at the electrode. 13 Effect of the K I 0 3 concentration The Se peak height can be influenced considerably by a change in the concentration of KI03 (as a catalytic com- ponent). As shown in Fig. 6, the optimum choice was 0.5 ml of 4% KI03, in a total volume of 5.0 ml. Based on these results, the optimum catalytic system consisted of 0.05 ml of HC104, 1.0 ml of 10% NaZSO3, 1.0 mlANALYST, JULY 1989, VOL. 114 m 795 J B 8 10 12 14 16 18 Period of mercury dropis per drop Fig.7. Sweep rate, 60 mV s-1 Eftect of the period o f the mercury drop on the peak height. 20 40 60 80 100 120 Sweep rateimV s-1 Fig. 8. Effect of sweep rate on the peak height of the solution containing SetV. Period of mercury drop: 1, 9; 2, 11; and 3, 18 s per drop of"H3 - NHjCl (pH lo), 0.25 ml of 5% NH,OH.HCl and 0.5 ml of 4% KI03, diluted to 5.0 ml with distilled water. Optimisation of Instruniental Parameters Effect of the period of the mercury drop on the peak current The catalytic reaction at the electrode is influenced directly by the mercury flow-rate, hence the peak height was dependent on the length of the period of the mercury drop. The maximum peak height occurred with a 9-s period for a given sweep rate (Fig. 7). Effect of sweep rate on the peak height In order to obtain the maximum peak height, the sweep rate should be adjusted to fit the period of the mercury drop; 80 mV s-1 was chosen as the optimum sweep rate for a 9-s period (Fig.8). Irritiul idtage The more positive was the initial voltage, the lower was the Se peak height. However, at potentials more negative than -0.4 V, the Se peak height was also influenced by the period of the mercury drop; therefore, an initial voltage of -0.2 V was chosen. Effect of Standing Time In order to obtain a stable result, the solution to be analysed was required to stand for some time. Fig. 9 indicates that 4680 min was the optimum time; for times in the range 80 min-2 h, the peak height apparently decreased. Effect of Interfering Ions Generally, inorganic ions in natural waters do not interfere with the determination of Se1V.From experiments, it was found that Ga3+, Cu?+, AS^+ and Pb2+ (100 pg 1-I), Te4+ (50 vg I-'), Ca'+, Mg2+ and MnZ+ (20 mg l-l), Fe3+, NO3- and < 50' I 20 40 60 80 100 120 a , o Standing timeimin LL Fig. 9. containing SeIV Effect of standing time on the peak height of the solution m c + .- 250 z 5 200 2 100 m 2 150 m .- Y m a 50 [SeVIl/pg I - l 0.2 0.4 0.6 0.8 1.0 1.2 I '2 - 250 - 200 150 100 50 - - - - [SeV1l/pg 1-l [Se"Wpg I - I I 0 0.02 0.04 0.06 0.08 0.10 0.12 Fig. 10. 2, 100; and 3, 200 Peak height plotted against [SeIV] and [SeV1]. Slope: 1, 100; 9.2 a e a n 2 8.8 L. $ 8.6 E : 9.0 a, m 1 -0 Y- o U o 8.4 .- L I I I I 0.1 0.2 0.3 0.4 0.5 0.6 - E N Fig. 11. Electrocapillary curves. 1, Without NH,OH.HCl; 2,0.25 ml of 5% NH,OH.HCl added; and 3, 0.1 ml of 3% poly(viny1 alcohol) added 1 -0.2 -0.57 Eiv Fig.12. Cyclic voltammetric curves: 1, cathodic sweep; and 2, anodic sweep796 ANALYST, JULY 1989, VOL. 114 Table 1. Analytical results Standard Relative Mean value/ No. of deviation/ standard Standard Recovery/ Sample Parameter pg 1 - 1 replicates pg 1- 1 deviation.% added/pgl-' pg 1-1 Recovery,% Water supply from Changchuncity . . . . . . River water of the second Songhauriver . . . . . . River water of Bing county (Shanxi province) . . US Envirorimental Protection Agency Certified Values . . Total Se SeIV Se"1 Total Se Se'" Se"1 Total Se SeIV SC"' Se 1" 0.047 12 0.004 8.87 0.10 0.143 96 0.014 12 0.002 14.3 0.04 0.054 100 0.033 0.060 8 0.002 2.8 0.10 0.162 102 0.037 8 0.003 6.98 0.023 - 0.284" 5 0.017 5.75 0.30 0.567 94 0.009 5 0.275 0 227 4- 12 0.017 7.54 0.23 0.448 95 - - - - - - - - - - - - - __ - - - - - - - - - - - * Value determined by neutron activation analysis = 0.26 pg 1- I .i- Value given in US Environmental Protection Agency Manual, 600/4-79-020 = 0.231 pg 1 - I . - S042- (5 mg I - I ) , humic acid (10 mg 1 - 1 ) and 0.2 ml of 1% EDTA did not interfere. Mercury might interfere at levels greater than 25 pg 1 - 1 : however, a better result can be obtained from the river water sample by nitration of the organic matter in it. Relationship Between the Peak Current and the Concentra- tions of Sfi1.r' and Seb'I A linear relationship was obtained between the peak height and the Se concentration in the range 0.01-1.2 pg 1-1 under the optimum experimental conditions.As shown in Fig. 10, the graph for SeVI (reduced in the presence of HCI) was consistent with that oi SeIV. Study of the Rshwiour of the Catalytic Wave of Se Electrocapillury L'urb'e The drop time i s proportional to the surface tension of the dropping Hg. Therefore, as adsorption of NH,OH.HCI on the mercury surface would cause a decrease in the surface tension, the drop time should decrease correspondingly. Comparison of the curve:, chown in Fig. 11 clearly indicates adsorption of NH20H.HCl on the mercury drop over a wide range (-0.1 to -0.7 V). Effect of temperatiire on the peak current The height of the catalytic wave increased with decreasing temperature. Thc temperature coefficients were -2.5% per "C (0-16 OC) and -6.8% per "C (16-32 "C), indicating the adsorptive behaviour of the catalytic wave.EfJect of surface-active agent on the peak current The adsorptive behaviour of the catalytic wave was also revealed by the experimental result that the peak height of the catalytic wave decreases and ultimately disappears with the addition of surface-active agents, such as poly(viny1 alcohol) and gelatin. Cyclic voltummetric curve The irreversibility of the reaction occurring at the electrode is shown clearly by the cyclic voltammetric curves in Fig. 12. Curve 1 indicates the adsorptive behaviour of the catalytic reaction; curve 2 shows two unsymmetrical peaks. Catalytic mechanism The behaviour of Se'V in the catalytic reaction reported in the literature has been described as follows1~: H7Se04 + 2HCI-+ H2Se03 + CI2 + H20 .. (1) H2Se03 + 2H2S03 -+ Se + 2HS04- + 2H+ + H 2 0 . . (2) Se + SO,'- ++ SeSO32- . . . . ( 3 ) The high sensitivity of the catalytic wave in Na2S03 - NH3 - NH4Cl - K10, (pH 10) results from the SeS032- produced in the system. The reaction mechanism at the electrode is as follows: SeS03'- ---+ Se2- + SO,'- +2e- ,r w-, Kf I where K , is the rate constant (1.75 x 109 1 mol-1 s-1). The rate of the catalytic reaction is accelerated by the co-operative behaviour of NH,OH.HCI and K103. Analysis of Water Samples Two 5.0- or 10.0-ml aliquots (depending on the Se content of the sample) of each water sample were transferred into electrical cells. These samples were then concentrated to half the original volume by heating gently on a hot-plate and 0.054.1 ml of HC104 and 0.2 ml of HN03 (concentrated) were added.One solution was evaporated to a volume of 0.2 ml (fumes appeared) and the other to a volume of 0.5 ml; to the latter was added 0.05 ml of HCI (concentrated), after which it was evaporated to a volume of 0.2 ml (white fumes appeared). To the two treated samples was then added 0.5 ml of 10% Na2S03. After leaving the two solutions to stand for 20 min, 0.5 ml of NH3 - NH4Cl buffer, 0.1 ml of 5% NH20H.HCI, 0.25 ml of 4% K I 0 3 and 1.0 ml of water were added to each solution in sequence. After shaking and allowing the mixtures to stand for 1 h, the total Se and SeIV were determined. The reproducibility, accuracy and recovery of the experiment are indicated in Table 1.Conclusion Experimental results have demonstrated that the proposed 1.5th-differential polarographic method is simple, rapid, accurate and sensitive for the determination of Se in natural waters. The determination is carried out using the catalytic system HC104 - Na2S03 - NH3 - NH4CI - NH20H.HCI - KIO3 and suffers less interference from foreign ions under the selected conditions. The relationship between concentration and peak height (0.01-1.2 pg 1-1) is linear over a wide range with a correlation coefficient of 0.999, and the results are in good agreement with those achieved by neutron activation analysis and fluorimetry. Finally, the method can be applied toANALYST, JULY 1989, VOL. 114 the analysis of Se-containing water and soil samples from different regions.The authors thank Professor Wang ErKang, Changchun Institute of Applied Chemistry, Academia Sinica, Chang- chun, People’s Republic of China, for guidance and helpful discussions during the course of this work. 1. 2. 3. 4. 5. References Xu, H., “Biological Trace Elements-Selenium,” Huazhong University of Science and Technology, Wu Han, Hu Ba Province, 1983, pp. 1-5. “Selected Papers on Kasham Disease in China, Part I,” Chinese Medical Academy, Beijing, 1977, pp. 53-56. Schrauzer, G. N., Bioinorg. Chem., 1975, 5 , 275. Lieser, K. H., and Neitzert, V., J . Radioanal. Chem., 1976,31, 397. Ni, I . , Terashina, S., and Ando, A., Bunseki Kagaku, 1984,33, 288. 6. 7. 8. 9. 10. 11. 12. 13. 14. 797 Olson, 0. E., J . Assoc. Of5 Anal. Chem., 1969, 52, 627. Wang, S . , Xu, F., and Zhou, H . , Fenxi Huaxue, 1980, 8,236. Zhang, X., He, C., and Xu, C., “Acata Scientiarum Natural- ium Universitatis Pekinesis,” Beijing University Press, Beijing, 1985, No. 1. Wu. D., Fenxi Huaxue, 1988, 16, 159. Kawashima, T., Kai, S., and Takashima, S . , Anal. Chim. Ada, 1977, 89, 65. Goto, M., Hirano, T., and Ishii, D., Bull. Chem. Sac. J p n . , 1978, 51, 470. “Proceedings of the 1984 Electroanalysis Meeting, Changchun, China, July 9th, 1984, Part I,” 1984, p. 375. Gao, X., and Yao, X., “The Catalytic Polarographic Wave of the Platinum Group,” Scientific Press, Beijing, 1977. Jiao, K., He, C., and Zhang, X., “Acata Scientiarum Natural- ium Universitatis Pekinesis,” Beijing University Press, Beijing, 1984, No. 5. Paper 8101807C Received May 9th, 1988 Accepted January 6th, 1989

ISSN:0003-2654    

DOI:10.1039/AN9891400793

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST, JULY 1989, VOL. 114 799 Simulation of Flow Injection Transients With Application to Amperometric Detector Response Francis E. Powell School of Food and Fisheries Studies, Humberside College of Higher Education, Grimsby DN34 5BQ, UK Arnold G. Fogg Department of Chemistry, Lo ug h boro ugh University of Tech nolog y, Lo ug h boro ug h, L eicestersh ire LE77 3TU, UK An RC (resistance - capacitor) circuit has been built as an analogue simulator of the continuously stirred tank reactor model of the flow injection transport process. The analogue simulator was used as a fast-response time-calibration signal and t o evaluate the fidelity of the time response of amperometric detection systems. Further uses of the device are discussed. Keywords: Flow injection; continuously stirred tank reactor; analogue simulator; amperometric detection The important parameters that describe the shape of a flow injection (FI) response curve are the peak height (i.e., peak dispersion) and the peak variance (02).The latter is made up of three contributionsl: opeak2 = <Jinlection2 + otransport2 + Odetection2 * * (I) Excessive peak broadening due to the detector can present disadvantages in certain situations, e.g. , when limited trans- port dispersion is required in order to retain the original properties of the injected solution. Similar considerations apply in liquid chromatography, and here the detector variance has been divided further into a contribution due to the size and geometry of the detector cell, and contributions from the response times of the detector and recorderz: Odectection2 = 0 c e 1 1 ~ + oelectronic2 * * .(2) Cell volume dispersion is usually studied by introducing a step change in the concentration of the signal-producing com- ponent and following the resulting response curve.3-1° In this application it is important to have an estimate of the electronic time constants. In this study, the fidelity of the responses of an ampero- metric detector and a recorder have been evaluated using a signal generator as an analogue simulator for the FI-pulse and concentration-step transport processes. Continuous Stirred Tank Reactor (CSTR) Model of Transport Processes The CSTR represents an idealised mixing stage in a transport process." A solution containing a component of concentra- tion co flushes the solvent out of a mixing tank and the concentration rises with time t , given by the equation for the feed curve (F-curve) c/co = [l - exp(-th)] .. . . . . (3) where t is the residence time of the system. This response is symmetrically related to the curve [wash curve (W-curve)] obtained by washing out the vessel containing the solute at concentration co with pure solvent11 . . . . . . F + W = 1 . . (4) This model is used frequently to describe the effect of detector volume on solute dispersion.3" The single-tank concept has also been adopted to model the FI process,10 which in its simplest form consists of the transportation of the analyte from the point of injection to the detector without reaction [Fig. l(a)]. In this model, the dispersion in the transmission line is equivalent to the mixing process in a single CSTR stage [Fig.l(b)]. The concentration of analyte in the sample is co. A small volume of this solution is injected into the carrier stream as a slug and on reaching the detector produces the idealised FI characteristic shown in Fig. 2. The leading edge of the slug initiates the F-curve response but when the following edge reaches the detector at t,, a wash process takes over, producing a W-curve from the peak maximum at cp. The shape of this characteristic can be decribed by the following parameters: residence time (t); fraction of the peak signal (fl = c/c,; dispersion (D) = ccJcp; and peak width ( A t ) = tz - tl. The relationships existing between the parameters are obtained from the component F- and W-curves.The time tl is determined from the F-curve ( 5 ) c = co[l - exp(-tl/z)] . . . . . . or, in parametric form, At the peak maximum (7) The time t2 is determined from the W-curve c = c,exp[-(t2 - tp)/t] . . . . . . (8) or, in parametric form, (9) Hence This final expression is equivalent to that obtained by Tyson.12 Principle of the Analogue Circuit The F- and W-curves that form the basis of the mixing models described above are examples of first-order linear response functions, which have their counterparts in other physical systems. 13 In particular, RC (resistance - capacitor) electrical circuits can duplicate such behaviour and have been useful in the study of transport in a physiological context.14 A schematic diagram of such a circuit, which can generate a full F-curve or the FI model characteristic, is shown in Fig.3. The800 x0.21 - ANALYST, JULY 1989, VOL. 114 Bistable (4 Sample injection 0.42 ~ 1 - Multiplier o+ - 0.84 I 1.05 PI B - I c;:::::r Fig. 1. the hypothetical FI model system Schematic diagrams of (a) the single line FI manifold and ( b ) Fig. 2. Idealised FI response characteristic: cO, concentration of the injected sample; cp, peak maximum concentration at time fp; At, peak width at an arbitrary conccntration at times tl and t2 Auxiliary -1 voltage developed at the output mimics the concentration from the CSTR. The RC circuit uses a resistance decade box and five calibrated 1-pF capacitors to provide RC charge time constants (equivalent to r values) adjustable from 10 ms to 5 s. A fraction (0.21) of the input reference voltage is taken to a multiplier (actually a simple summing amplifier), which provides four equal steps from 0.21 to 0.84 and simulates the dispersion ( D ) in a reciprocal manner.Thus five dispersion values can be simulated. At the end of the charge cycle when the capacitor voltage (V,) equals the comparator reference voltage, the comparator switches, thus triggering the bistable. The bistable controls the two ganged CMOS analogue switches, switching from the charge to the discharge cycle. The resistance and leakage effect of these switches is assumed to be negligible. The push-button resets the bistable and switches to charge and the start of another sweep. In addition to these fractional input voltages, there is a 1.05 option, which, when operational, ensures that the bistable cannot trigger; a full F-curve with a plateau at the input reference value is thus produced.High-input impedence operational amplifiers mini- mise errors due to shunt currents by buffering the input reference voltage to the control circuit and ensuring the accuracy of the output. Apparatus and Experimental Complete Assembly The apparatus was centred around an electrochemical module that provided a voltage input to the cell or its analogue and also processed the response, which was in the form of a current flowing between the “test” input and a common virtual earth point within the module. This signal was converted internally into a voltage for display on the recorder (Fig. 4). Electrochemical Module and Signal-matching Box An EDT model ECPlOO polarograph unit (EDT Analytical) was used.This instrument has an optional filtering time constant facility calibrated at 1.2 s. The output from the module was 1 V full scale and for interfacing purposes this was Resistance decade box connections P Analogue C/O switches Reference (llMQ , C h ; L $ , 1 Input I t voltage V &r - - g e I VC Current output r test 1 MQ Common / Fig. 3. Schematic diagram of the analogue circuit for thc single-tank concentration step and the FI-pulse simulatorANALYST. JULY 1989, VOL. 114 801 . Response (current) Response b w (voltage, 1V maximum) Flow manifold Output Electrochemical . or analogue (voltage) module 4 Response box (10: 1) oscilloscope * Recorderor Signal matching (100 mV maximum) Fig.4. Schematic diagram of the assembly of response components I Auxiliary 1 MR 0 Analogue input (voltage) i Reference Test ---) Analogue output (voltage) 1 MS!. I Earth - JC - Fig. 5. Schematic diagram of the interface between flow analogue and EDT Model ECPlOO polarograph. “Auxiliary,” “reference” and “test” (or “working”), typically electrode connections from the polarograph to an electrochemical cell; R, resistance and C, capaci- tance components of the flow analogue circuit. C is connected to earth via the polarograph and “test” hcld to ground by the “virtual earth” of the current/voltage converter within the polarograph stepped down to 10OmV without loss of sensitivity via a laboratory-built operational amplifier signal impedance matching box. Recording Signals were recorded using either a Howe YTlOOO pen recorder, a Telequipment DM64 storage oscilloscope or a Telequipment S5 1B oscilloscope (in conjunction with a Polaroid CR-9 camera and an appropriate Polaroid 667 black and white film).Analogue Although the analogue simulator as shown in Fig. 3 can be used with any suitable voltage source and voltage recorder, additional interfacing is required when using the analogue simulator in conjunction with the electrochemical module. “Auxiliary” and “reference” inputs were isolated by a 1 MQ resistor functioning as the solution resistance. It was also necessary to convert the simulated voltage output into the current form using a voltage follower and a series 1-MQ current-limiting resistor between the RC circuit and the “test” input.The voltage input from the electrochemical module was 1 V. These features are shown in Fig. 5. A full circuit diagram can be obtained from the authors. Flow Manifold A concentration-step input experiment was performed by injecting a 0.5-ml sample of 0.75 mM potassium hexacyanofer- rate(I1) in 1 M potassium chloride solution using a Rheodyne 5020 valve. The solution was presented horizontally to a glassy carbon wall-jet electrode. The construction and flow charac- 1 .oo 0.75 *- 0.50 0 5 10 Time,% Fig. 6. Pulse trace from the analogue obtained with the Howe YTlOOO recorder. Chart speed, 10 mm s-1. Analogue output: z. 3 s; D , 1.359. Hatched lines joining experimental points (0) represent peak width At values calculated from equation (10) teristics of the laboratory-built wall-jet detector cell used have already been described.lS716 In previous studies the carrier stream has been presented vertically upwards to the glassy carbon electrode, but in the work described here the detector unit was incorporated through a hole in the wall of a plastic beaker so that the carrier stream could be presented horizon- tally to the electrode.The detector unit was fixed (using epoxy resin) in place in the wall at the point where it had been clamped in previous applications.15 The beaker contained 1 M potassium chloride solution in which was placed a silver - silver chloride reference electrode and a platinum auxiliary elec- trode to complete the three-electrode system. The potential of the working electrode was held at 0.70 V versus the reference to ensure diffusion-controlled electrolysis of the electroactive hexacyanoferrate(I1) species.The transport line between the injection valve and the detection cell was 8 cm of PTFE tubing (i.d. 0.8mm) and the solution was pumped by an Ismatec MS-4 RegloB peristaltic pump. A pulse damper was placed between the pump and the injection valve. Results and Discussion FI Characteristics Fig. 6 shows the display obtained with the pen recorder using t = 3 s and D = 1.359 from the RC analogue; the data points were calculated from equation (10). The trace demonstrates that accurate analogue signals can be generated and recorded under these conditions. Polarographic Module Filter Time Constant Many instruments have in-built exponential time constant circuitry to reduce high-frequency noise, but this can intro- duce over-all signal distortion. This effect can be demon- strated with the EDT module.The F-curves were obtained with and without operation of the filter circuit (Fig. 7). The convoluted arrangement used, viz., analogue function (1 - e-t’t, t = 1.2 s) * filter circuit response (e-f’., ‘t = 1.2s)802 ANALYST. JULY 1989, VOL. 114 1 .oo 0.75 0.63 0.59 0.50 0.25 Fig. 9. Oscilloscope (Telequipmcnt S.518) traces obtained. A, Experimental curve arising from the concentration step of 0 to 0.75 mM K4Fe(CN)6 in 1 M KCI in a glassy carbon wall-jet electrode detector cell. Flow manifold: 8cm x 0.8mm i.d.; flow-rate, 0.62 ml min-l. B, Calibration curve from the analogue. RC = 1 s; peak maximum sct at 0.63 full-scale response 1 0 8 6 4 2 0 Tirneis Fig.7. F-curves obtained from the analogue to the EDT module. I , Without filter; 11, with filter time constant T = 1.2 s in operation. The points shown were I, calculated from equation ( 3 ) and 11, calculated from equation (11) Time - Fig. 8. Pulse signal (T = 1.2 s) through the EDT Model ECPlOO polarograph. A, Without filtcring; B, with filtcr time constant (1.2 s) in operation corresponds in transport terms to the response from two tanks in series having equal volumes17: c/co = 1 - exp(-t/t)(t/a + 1) . . . . (11) The experimental results agree well with this equation (Fig. 7). which also predicts that the total residence time ( 2 ~ ) for the two mixing stages should occur at a c/cu value of 0.594; this is also confirmed experimentally in Fig.7 where c/co is seen to be located at 0.59 after 2.4 s. This value is significantly lower than the 0.632 fraction (based on a single tank) that is often used as an empirical measure of residence times in experimental situation+-6; hence such a practice should be viewed with some caution. When a simulated FI pulse (t = 1.2 s, D = 1.19) is passed through the same electronic path in the polarographic module as described, the output is modulated as shown in Fig. 8. The dispersion increases to 1.70 and the peak width broadens. In effect, the detection step introduces an extra mixing stage into the total response. Actual FI curves reported extensively in the literature (e.g., reference 18) bear a closer resemblance to the convoluted curve B than to curve A, suggesting that multi-mixing steps may more realistically represent actual manifold-detection systems than does the single mixing step.Another feature of the combined response is that the maximum occurs at the intersection with the original curve. In transport terms this follows from the mass balance on the second tank above, viz., t(dC,,,ldt) = C,, - Co,,. Here, C,,, represents the signal from the RC analogue and Gout the combined response; T is the effective residence time of the second stage (ie., in this instance, the filter time constant). Hence the output reaches a maximum (dC,,,ldt = 0) when C,, equals C,,,. A similar observation has been made on the effect of finite detector volume on Gaussian chromato- graphic elution curves.Ig Recorder Calibration The fidelity of the response through the polarograph module (with the filter circuit non-operational) and the signal-match- ing box was checked by sending an F-curve signal of known time constant (RC) from the analogue through the unit and measuring the response on a Telequipment DM64 storage oscilloscope, the time base of which was calibrated using mains frequency (SO Hz).A slow scan was used to establish the full-scale deflection and a fast scan was then employed to record the trace from which the experimental t value was interpolated (t = 0.632 f.s.d.). Theoretical time constant values from 2 to 0.1 s were faithfully reproduced. Hence, at an RC value of 0.1 s, t was found to be 0.101 fI 0.002 s (based on five replicates). However, when the pen recorder (Howe YT1000) was used in place of the oscilloscope and operated at its maximum chart speed (10 mm s-1) the response to the analogue simulator was faithful only down to approximately t = 1 s.Thereafter, the response time of the pen was of the same order as the time constant of the transient signal and distortions of residence time and peak parameters appeared and became progressively worse as the T value was reduced. As an independent check of this failure, the recorder response time was measured as 0.34s, by a method based on the modulation in amplitude of standard sine waves.20 Transport transients with short residence times are there- fore better captured with low-inertia recording. This is illustrated by the oscilloscope traces shown in Fig. 9, where the experimental F-curve (A) was generated by the step-wise change from 0 to 0 .7 S m ~ K4Fe(CN)6 (in I M KCl) whenANALYST, JULY 1989, VOL. 114 presented to the electrochemical detector through the short transport line of the flow manifold. The time axis of the oscilloscope recorder was calibrated from the superimposed pulse trace (curve B), obtained by replacing the flow line by the analogue circuit. For this, using a peak-maximum voltage of 0.63f.s.d., the peak time corresponded to the RC time constant (1 s) selected. Other experimental F-curves have been calibrated similarly using the versatility of the analogue device. This work is in progress. Conclusion Using an RC analogue circuit, the electronic components of an electrochemical detector have been shown to introduce negligible distortion to F-curve and FI peak transients down to residence times of 0.1 s unless the filter circuit is operational. However, a potentiometric pen recorder was found to be inadequate at residence times approximately three times the response time.Hence, in the recording of transport events having short residence times an oscilloscope or analogue to digital fast time capture microprocessor is essential. This is particularly relevant to response-curve studies of cell volume dispersion. The analogue circuit was also useful in the calibration of such a fast recording system. The authors thank Mr. Douglas Hankin of Humberside College of Higher Education for designing and constructing the analogue simulator used in this work. References 1.Reijn, J. M., van der Linden, W. E., and Poppe, H., Anal. Chim. Acta, 1980, 114, 105. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. Scott, R. P. W., “Liquid Chromatography Detectors,” Second Edition, Elsevier, Amsterdam, 1986, p. 27. Forina, M., Ann. Chim., 1973, 63, 763. Hanekamp, H. B., Bos, P., Brinkman, U. A. Th., and Frei, R. W., Fresenius 2. Anal. Chern., 1979, 297, 404. Brunt, K . , Analyst, 1982, 107, 1261. Stulik, K . , and Pacakova, V., Ann. Chim., 1986, 76, 315. Dieker, J . W., van der Linden, W. E., and Poppe, H. , Talanta, 1979, 26, 512. Kutner, W., Debrowski, J . , and Kemula, W., J. Chromatogr., 1980, 191, 47. Kristensen, E. W., Wilson, R. L., and Wightman, R. M., Anal. Chem., 1986, 58, 986. Tyson, J. F., and Idris, A. B., Analyst, 1981, 106, 1125. Nauman, E. B., and Buffham, B. A., “Mixing in Continuous Flow Systems,” Wiley, New York, 1982, p. 6. Tyson, J . F., Analyst, 1984, 109, 319. Pollard, A., “Process Control,” Heinemann, London, 1971, p. 23. Sheppard, C. W., “Basic Principles of the Tracer Method,” Wiley, New York, 1962, pp. 89-113. Fogg, A. G., and Summan, A. M., Analyst, 1984, 109, 1029. Powell, F. E., and Fogg, A. G., Analyst, 1988, 113, 483. Denbigh, K. G., and Turner, J. C. R., “Chemical Reactor Theory,” Third Edition, Cambridge University Press, 1984, p. 86. Valcarcel, M., and Luque De Castro, M. D., “Flow Injection Analysis,” Wiley, New York, 1987, p. 81. Vandenheuvel, F. A., Anal. Chem., 1963,35, 1193. McWilliam, I. G., and Bolton, H. G., Anal. Chem., 1969,41, 1762. Paper 81041 01 F Received October 17th, I988 Accepted February 16th, I989

ISSN:0003-2654    

DOI:10.1039/AN9891400799

出版商:RSC    

年代:1989

数据来源: RSC