摘要:

ANALYST, JUNE 1986, VOL. 111 695 Rapid Automatic Gas Chromatographic Method for the Continual Measurement of Hydrogen Cyanide and Cyanogen in Air Peter R. Fielden," Simon J. Smitht and John F. Alder Department of Instrumentation and Analytical Science, UMIST, P.O. Box 88, Manchester M60 IQD, UK Equipment for the continual measurement of hydrogen cyanide and cyanogen by gas chromatography is described and applied to a gas generation and testing apparatus for the characterisation of hydrogen cyanide-absorbing activated charcoals. Under automatic operation, two adsorbent samples may be assessed in parallel giving an analysis every 2 min, or a single sample every 1 min. A computer is used to collate the data generated and to present them as breakthrough profiles. Details for the generation of a stable hydrogen cyanide test atmosphere at a concentration of 2 mg min-1 and a flow-rate of 1 I min-1 are given, as are procedures for the generation of both hydrogen cyanide and cyanogen atmospheres for the calibration of the gas chromatograph.Keywords: Automatic method; gas chromatography; hydrogen cyanide and cyanogen determination; air samples In the course of testing the adsorptive characteristics of activated charcoal coated with chemisorptive reagents towards hydrogen cyanide (HCN) atmospheres, it has proved necessary for us to develop a dynamic testing facility. Reagent-impregnated charcoals, designed for use in respirator filters for the removal of HCN, have been assessed by subjecting standard charcoal sample bed-depths to calibrated HCN atmospheres.Some reagents were found to generate cyanogen [(CN),], upon exposure to HCN. A system was sought to generate, calibrate and monitor HCN atmospheres, to subject charcoal samples to HCN challenge and to measure the effluent stream from the filter beds for breakthrough of HCN and (CN)2, when generated. This paper describes a test facility for the generation of stable concentrations of HCN and the calibration of a gas chromatographic method, which was employed to monitor both influent and effluent HCN concentrations and to quantify any (CN)2 generated. Gas chromatography was chosen as a reliable means of quantifying mixtures containing HCN and (CN)2 over the ranges 1 pg 1-L-2 mg 1-1 of HCN and 1 pg 1-1-1 mg 1-1 of (CN)2, by taking discrete samples at regular intervals and effecting a rapid separation (<2 min).In this way, breakthrough profiles of charcoals could be obtained on a continual basis by automatic sampling of the atmo- spheres, data processing by computing integrators and data collection by a computer. The profiles were analysed, subsequently, to assess the performance of each charcoal - impregnated system. An activated charcoal cloth sample impregnated with copper and sodium dichromate is given as an example that shows the ability of the gas testing apparatus to follow the breakthrough process from low levels of HCN and (CN)2 up to the influent concentration. Fig. 1 shows a schematic diagram of the test apparatus and indicates the three units: ( a ) for gas generation; ( b ) for adsorbent exposure; and (c) for gas line analysis - calibration.Experimental The test facility performed three distinct functions: the generation of HCN at 2 mg 1-1 (at a 1 1 min-1 flow-rate) and the generation of HCN and (CN), at lower levels for the * To whom correspondence should be addressed. t Present address: Department of Chemistry and Chemical Engineering, Royal Military College of Canada, Kingston, Ontario, Canada K7L 2W3. calibration of the breakthrough concentration range (1-10 pg 1-1); the exposure of charcoal samples to 2 mg min-1 of HCN; and the analysis - calibration of the influent HCN level and the monitoring of the breakthrough of HCN and (CN)2. The apparatus for gas generation, adsorbent exposure and the connecting lines to the gas chromatograph were construc- ted from stainless steel and glass tubing (Y4 in 0.d.).Gas Generation HCN was generated over the concentration range of 1 pg 1-1- 2 mg 1-1 and (CN), over the concentration range 1-60 pg 1-1. Hydrogen cyanide generation Hydrogen cyanide, in a compressed gas cylinder (as a mixture with inert diluent gas) or as a liquid, can be used to generate atmospheres containing HCN at a steady concentration. The high cost of compressed gas cylinders containing a calibrated dilution of HCN and the problems associated with the safe and secure storage of liquid in a busy laboratory have led us to generate HCN by a wet chemical method. Solutions of potassium cyanide (6,0.6 and 0.06% mlV) and sulphuric acid (34, 3.4 and 0.34% VW and used with the appropriate potassium cyanide solution concentration) were pumped (Watson Marlow 502 peristaltic pump with multi- channel pump head) into a reactor vessel (Quickfit FR240F with top MAFI/75) at a constant flow-rate (typically 0.2-0.3 ml min-l).Fine jets, placed below the reagent surface, were employed (0.01 in i.d. x 1/16 in 0.d. PTFE tubing, Phase Separations Limited, Queensferry , UK) to ensure rapid dispersion in the bulk liquor. Spent reagent was removed using a constant-head device within the generator. Air pumped at 0.5 1 min-1 (Charles Austen air pumps Type CAPEX2D) through a glass frit (sinter 0) was used to outgas the HCN formed according to the reaction KCN + H2S04 -+ KHS04 + HCN. An excess of acid was employed to ensure the complete reaction of the cyanide and so reduce the toxicity of the waste to aid disposal.At ambient temperature (20 "C), with 6% KCN - 34% HzS04 solution, a concentration of HCN at 6 mg 1-1 was produced at the output of the generator at a flow-rate of 0.5 1 min-1. This concentration could be adjusted by changing the concentration of the reagents, by controlling the reactor vessel temperature, by changing the feed rate of the liquid reagents or by changing the gas flow-rate. The latter two methods were employed by the authors for fine adjust- ment of the 6 mg 1-1 level on the grounds of simplicity; the laboratory conditions were such that the temperature nor-696 ANALYST, JUNE 1986, VOL. 111 - - * - - Humidifier Air Pump d KCN soh. -==+el H2S04 HCN generator - 1 Waste I I Flow selector h Peristaltic pump @ Flow controller and rotameter @ regulated bleed-off, via scrubbing tower @ Mixer Fig.1. Schematic diagram of gas ri indicatin calibration unit. *Ammoniacal copper(1f) solution fcolour change: blue + green in the presence of hydrogen cyanide) (a) gas preparation apparatus; ( b ) adsorbent exposure; and ( c ) gas line analysis - mally remained sufficiently constant over the working day so that thermostating was unnecessary. Hydrogen cyanide was diluted by mixing with a thoroughly humidified air stream to give HCN at ca. 2 mg 1-1 at a flow-rate of 2.5 1 min-1. The level of humidity was estimated to be ca. 80% relative humidity (RH) and remained constant providing that the temperature of the laboratory did not change significantly. The constancy of the humidity was checked by an in-line wet and dry-bulb hygrometer.This was useful as a comparative measurement, as accurate measure- ments would require a higher flow-rate than that used in the test rig. Type excess of HCN was diverted for adsorption by a scrubbing tower filled with impregnated charcoal (Sutcliffe Speakman, Type 1100, 14-28 mesh) such that a 1 1 min-1 flow-rate of 2 mg 1-1 of HCN was available. Hydrogen cyanide was generated at lower concentrations by using sulphuric acid at lower concentration (3.4 and 0.34% by V/V volume) with potassium cyanide solution diluted to 0.6 and 0.06% mlV, respectively. A range of concentrations could be generated in the gas phase by changing the relative flow-rates of air through the generation vessel and diluting stream. Cy anogen generation A method reported for the generation of cyanogen involved the reaction of sodium cyanide with copper(I1) sulphate.1 It is well known2 that HCN reacts with Cu(I1) to form (CN),: 2Cu(II) + 4HCN-+ 2Cu(I)(CN) + 4H+ + (CN), Although this is a reliable method for producing cyanogen in the HCN generator vessel, it has the disadvantage of forming a fine insoluble precipitate of Cu(I)CN, which blocks the sintered glass frit and leaves a deposit in the waste line.An alternative method was therefore employed in which HCN, generated as described, was converted into (CN), in a second reactor vessel. It was found that if a solution of C U ( I I ) ( N O ~ ) ~ . ~ H ~ O (10% mlV) mixed with twice the molar equivalent of ethylenediamine (en) was used, precipitation could be prevented.The reaction is described as follows: [Cu(II)(en),12+ + HCN(aq) -+ [Cu(I)(en),]+ + H+(aq) + 1/2(CN)2 The conversion of HCN into (CN)2 was accompanied by a colour change from purple to colourless; the [Cu(I)(en),]+ remained complexed even in the presence of CN-. It was found that spent reagent could be regenerated by the addition of hydrogen peroxide (30% mlv) such that [Cu(II)(en),]2+ was reformed from [C~(I)(en)~l+. Ethylenediamine was added to replace any excess lost owing to evaporation. Cyanogen concentrations of up to 60 pg 1-1 could be generated by this method. Any unreacted HCN was removed by passing through a scrubbing solution of [C~(II)(en)~l2+ (prepared by the same method as described for the reagent preparation) and airborne ethylenediamine removed by passing through a charcoal filter.Adsorbent Exposure Samples of impregnated charcoals were evaluated for their reactivity towards HCN in order to assess their usefulness as respirator charcoals for HCN removal. A standard sample bed-depth and cross-sectional area was ensured by filling stainless-steel tubes to a standard depth. These tubes contain- ing adsorbents were placed in-line by fitting into modified double socket tube and tube connectors (Quickfit SRD 4/75 and MF10/4B). The challenge gas flow-rate was monitored by a rotameter and regulated by a needle valve. Two samples could be subjected simultaneously to HCN atmospheres and conditioned or desorbed by switching from HCN - air mixtures to humidified air only. The influent and effluent gas streams from each sample tube were passed through a series ofANALYST, JUNE 1986, VOL.111 697 Negative pressure waste line Gas chromatograph -Column A Channel A Bleed valve A Influent Bleed valve B Channel B -Column B or influent Fig. 2. Schematic diagram showing details of gas line switching and flow splitting between the waste line and gas chromatograph sampling valves Hydrogen cyanide 120 80 40 0 Retention time/s Fig. 3. Separation of hydrogen cyanide and cyanogen within 2 min on a column packed with 25% glycerol triacetate on Chromosorb P (isothermal operation at 27 "C with an effective column length of ca. 5 cm) switching valves such that the two streams could be monitored simultaneously or parallel measurements made on a single gas stream (Fig. 2). Gas Line Analysis - Calibration It was necessary to monitor both the breakthrough charac- teristics of the adsorbed samples at low levels [ca.1 vg 1-1 for both HCN and (CN),] to assess their performance and the influent gas concentration [ca. 2 mg 1-1 for HCN] for calibration, so that the amount of HCN adsorbed by the charcoals could be accurately measured as a function of time. An analysis system was required that was capable of both measurement over a wide range of analyte concentration and the determination of HCN together with (CN),, so that the entire process of filter breakdown could be followed as the chemisorptive reagents were expended. A rapid gas chromatographic method was developed for the continual measurement of the gas streams. Gas chromato- graphic techniques and columns for the batch determination of HCN and (CN)2 have been well documented,3-6 but no rapid, continual method has yet been reported.Columns Hydrogen cyanide has been separated from air using columns packed with Porapak Q3 and Porapak QS.4 HCN and (CN), have been separated on a 2.8 m column containing 20% dinonyl phthalate solution on Dinokhrom N5 and on an 8 ft column with 25% glycerol triacetate (triacetin) on Chromo- sorb P.6 Initially 0.25 m X Y4 in i.d. glass columns, packed with ca. 5 cm of 25% glycerol triacetate on Chromosorb P (30-60 mesh), were used. The column dead space was packed with Chromosorb P without glycerol triacetate. It was necessary to employ this short, effective column length in order to reduce the separation time to below 2 min (Fig.3). Increasing the column temperature above 30 "C induced an unacceptably high column bleed rate of the glycerol triacetate. It was found after further investigation that the separation of HCN and (CN)2 effected on a Porapak Q column was more satisfactory; although the retention times of HCN and (CN)2 were similar at 1.3 and 1.1 min, respectively, column bleed was not a problem. An operating temperature of 100 "C was used to complete the separation in under 2 min on a 2 m x Y4 in i.d. glass column. Detectors The thermal conductivity detector (TCD) has been used for HCN and (CN)2 monitoring,5 as has the flame ionisation detector (FID).6 More recently, a nitrogen-specific detector (RbC1) has been reported for HCN detection down to ng 1-1 levels.7 A nitrogen/phosphorus detector has also been repor- ted for the determination of HCN.6 We used an FID for the detection of HCN and (CN), down to levels below 1 pg 1-1.This choice also allowed for future versatility of the testing facility, whilst retaining sensitivities for HCN and (CN)2 that were acceptable for the requirements of the adsorbent tests. Apparatus for Automatic, Continual Analysis A gas chromatograph (Packard 437, Packard-Becker, The Netherlands) fitted with two identical analysis units consisting of a l-ml volume automatic gas sampling valve, a Porapak Q column and an FID was used. The conditions used were as follows: carrier gas, nitrogen (50 ml rnin-I), hydrogen flow- rate (FID), 50 ml min-1; air flow-rate (FID), 330 ml min-1; injector temperature, 120 "C; oven temperature (isothermal operation), 100 "C; detector temperature, 170 "C.A vacuum of 6 cmHg was applied to the exit sides of the gas sampling valves to enhance the flushing of the sample loop (Charles Austen air pumps, Type CAPEX2D) and to ensure the rapid sampling of the effluent gas stream. A needle valve was used to split the effluent, which was recombined with the698 Detector A output Detector B output ANALYST, JUNE 1986, VOL. 111 GC sample valve A GC sample valve B Integrator A-status Integrator B-status Sampling Inject - AA Integrate - Compute - Standby - Integrate - Compute - Standby - 11 Fig. 4. Timing sequence diagram for the dual-channel operation exit stream from the gas chromatograph before removing all of the HCN and (CN)Z from the exhaust by a charcoal tower.The FID output signals were processed by two computing integrators (Hewlett Packard 3390A) that in turn were linked via serial interfaces (HP 82939A) to a desk-top computer [Hewlett Packard, HP851. The GC was equipped with a programming facility so that a sequence of operations involving the switching of the sampling valves and starting of the integrators could be controlled automatically. The dura- tion of the integrator “on” time and the transmission of data to the computer were controlled from the integrator. The computer was used for data acquisition from the two integrators and subsequent data handling. Two modes of operation, parallel or staggered (illustrated by (Fig. 4), were available to the operator. In the parallel mode both sampling valves were switched at the same time so that an analysis of channels A and B was made simultaneously.By this method it was possible to monitor two samples at a time and so double the sample throughput, or to measure the influent and effluent concentrations for one sample alone. This latter option enabled an increase in the measurement precision to be made by following any changes in the influent level of hydrogen cyanide and normalising the breakthrough data relative to influent data by dividing each effluent concentration datum by an influent concentration datum. The staggered mode, where both GC channels were employed to monitor the effluent atmosphere of a single test sample at an interval of 1 rnin between channels, was employed in order to enhance the effluent data of one sample by doubling the resolution in time by increasing the apparent sampling rate to 1 sample min-1 such that rapid breakthrough could be followed more precisely.This was achieved by balancing both monitoring channels on the influent level and subsequently matching the response slopes of both column - FID detector channels. The minimum cycle time of 2 min, imposed by the retention time of hydrogen cyanide on the columns and the time required to complete an integrator report, could be reduced to 1 min by monitoring the one effluent stream with both sampling valves. A duty cycle of 2 min with a time interval of 1 min between the switching of sampling valves A and B realised an apparent cycle time of 1 min for one channel. It was necessary to overlap the computing period (ca.10 s) and data transmission period (<5 s) (“compute” period-Fig. 4) of the integrators with the beginning of the subsequent chromatograph. The required 2 min duty cycle time was achieved by disregarding the first 20 s of each chromatograph, during which time the data from the previous chromatogram were processed. Once initiated, the number of repeat cycles was controlled from the GC programme. It was possible, however, to terminate data acquisition from the computer without affect- ing the operation of the GC timing sequence. The integrators could independently report peak area - retention data without the computer if the operator so desired. The computer data acquisition programme contained a facility for selectively monitoring any two peaks from either integrator by means of a variable width - position window.Further routines were available to calibrate and match the data from the two GC channels. The computer was used to display data graphically in the form of a breakthrough profile where the continual data from sequential chromatograms were interpolated to produce a continuous profile. The points were joined by straight lines for most of the samples tested, as further enhancement of the profiles through smooth curve fitting was not consideredANALYST, JUNE 1986, VOL. 111 699 Table 1. Gas rig calibration results Concentration Calibration Calibration range/ slope*/peak HCN . . . . . . Titrimetry 175-4000 6 2700 L 100 HCN . . . . . . Potentiometry 20-1750 14 2755 k 125 (CN), .. . , Potentiometry 2-60 9 1900 k 250 Gas method Pg I-' No. of data counts (p.p.m.)-' * Confidence interval of 95% for given slope of a regression line. The measured CN- solution concentration (mg 1-1) was related to the gas stream HCN concentration according to [CN-] = 2mxa/tv where rn is the relative molecular mass of HCN (g); x, the titre (cm3); a, the molarity of the silver nitrate solution (M); t , the collection time (min); and Y, the gas flow rate (1 min-1). At lower levels (<lo0 pg 1-1) it was necessary to employ a potentiometric method for the determination of CN-. A CN- electrode (Orion Research No. 940600) and a single-junction reference electrode (Orion Research No. 900100) were used with a computing potentiometric analyser (Orion Research, microprocessor Ionalyser 901).The collecting solution was 50 ml of 0.1 M sodium hydroxide solution. A standard cyanide solution was prepared by dissolving 2.5 g of potassium cyanide (AnalaR grade) with 10 ml of 10 M sodium hydroxide solution to 1 1 of solution. This was used for subsequent standard additions, where 5O-pl aliquots were added to the collection solution and the potential change noted. 20 40 60 80 100 120 Timeimin Fig. 5. Example of hydrogen cyanide and cyanogen breakthrough curves, produced by the gas testing rig, for a charcoal cloth sample impregnated with copper and sodium dichromate necessary for the assessment of performance of the charcoal samples tested. Calibration Calibration requirements were two-fold; firstly it was neces- sary to assess the concentration of the influent HCN or (CN), and secondly to calibrate the GC system for subsequent effluent - influent monitoring.A wet gas collection facility was built into the test rig so that the HCN or (CN), - air gas stream could be bubbled through a 100-ml Drechsel bottle (with glass sinter frit inlet) containing 50 ml of collecting solution; the gas mixture was passed through at 1 1 min-1 for 2 min. This was performed serially with sampling by the GC system, where three wet samples were taken with five GC samples between collections, so that absolute chemical calibration was possible. The collecting solution contained potassium iodide (2 g) sodium hydroxide (2 g) and ammonia solution (10 ml) in 1 1 of solution. Both HCN and (CN)2 dissolve and dissociate in aqueous alkali to liberate a molar equivalent of cyanide ions: HCN + OH- + CN- + H20 (CN), + 20H- + CN- + CNO- + H20 At high levels of HCN (>lo0 yg 1-1) an argentimetric titration was employed.Silver nitrate solution (0.02 M), which had been previously standardised against sodium chloride solution (0.01 M) with potassium chromate as indicator, was the titrant. The end-point was the appearance of an opales- cence, owing to the formation of silver iodide. The reaction is described by the following equations: Before end-point: Ag+ + 2CN- -+ Ag(CN),-,,,,,, At end point: Ag+ + I- -+ AgI 1 Results and Discussion Calibration The results of gas-rig calibration are summarised in Table 1. A high level HCN calibration (175-4000 pg 1-1) was determined by the titrimetric method, whereas low level HCN (20-1750 pg 1-1) and low level (CN), (2-60 pg 1-1) calibrations were determined by potentiometry , Response slopes were found to be linear over the ranges tested.This has shown that the data generated from adsorbent tests do not need further linearis- ation for levels of HCN and (CN), within the calibration ranges. The influent test level of HCN (2 mg 1-1) was calibrated both before and after a breakthrough test. Any changes in the two measured levels were averaged and the mean value was taken as the influent level. This was found to be sufficiently precise for most tests carried out as the major source of error was in the preparation and sampling of the adsorbent materials. After switch-on a settling period of 40-60 min ensured a typical stability of ca.1% RSD with respect to the mass of HCN delivered as a function of time. This remained stable for the duration of the tests, which could last for as long as 120 min. The columns have remained stable, giving reproducible results for over a year in our laboratory and have been used continually over the working day. Adsorbent Test The results of an adsorbent test are typified by a hydrogen cyanide filter material consisting of a combination of copper and sodium dichromate impregnated on to activated charcoal cloth (supplied by Chemical Defence Establishment, Porton Down, Wiltshire). This impregnant system is similar to those used in commercially available charcoals for hydrogen cyanide removal. The breakthrough curves (Fig. 5 ) for hydrogen cyanide and cyanogen (a by-product of the removal process) were obtained by using the test rig described.The adsorbent sample (20 mm700 ANALYST, JUNE 1986, VOL. 111 in diameter) was subjected to a challenge of 2 mg min-1 at a flow-rate of 1 1 min-1 of HCN in air at 1 1 min-1 and 80% RH. The effluent gas levels were monitored by the GC system at intervals of 2 min for a duration of 2 h. The HCN breakthrough curve was a typical sigmoid with a rapid initial increase in concentration followed by a much slower asymptotic approach to the influent HCN level. Owing to a breakdown of the removal process, the (CN), curve displayed a premature breakthrough with respect to that of HCN and peaked at about 925 pg 1-1, followed by a decay to about 150 yg 1-1. It is thought that the removal process is a two-stage reaction involving the conversion of HCN to (CN), by Cu(II), followed by the oxidation of (CN), into oxamide by the dichromate. As the dichromate oxidation step fails, (CN)* breakthrough occurs and rises until the (CN), generation step also loses efficiency. At this point, HCN breakthrough is detected and the (CN), level drops, accounting for the formation of a peak profile. Further examination of the shape of the profiles will reveal important information about the nature and kinetics of the removal process. Such data will allow the optimisation of existing impregnant systems and the assessment of novel systems. This work was supported by the Procurement Executive, Ministry of Defence. References 1. Janz, G. J., Znorg. Synth., 1957, 5 , 43. 2. Cotton, F. A., and Wilkinson, G., “Advanced Inorganic Chemistry,” Fourth Edition, Wiley-Interscience, New York, 1980, p. 802. Nota, G., Miraglia, V. R., Improta, C., and Acampora, A., J . Chromatogr., 1981, 207, 47. Darr, R. W., Capson, T. L., and Hileman, F. D., Anal. Chem., 1980,52, 1379. Steba, V. K., Zrazheskii, V. I., Parkhomenko, V. D., and Pivduarov, A. A., Khim. Promst., Ser. Melody Anal. Kont- volga Kach. Prod. Khim. Promsti., 1979, 11, 6. 6. Isbell, R. E., Anal. Chem., 1963, 35, 255. 7. Schmit, K . , Lebensm. Gerichtl. Chem., 1977, 31(b), 110. 3. 4. 5 . Paper A51410 Received November 8th) 1985 Accepted January 8th) 1986

ISSN:0003-2654    

DOI:10.1039/AN9861100695

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

ANALYST, JUNE 1986, VOL. 111 701 High-performance Liquid Chromatographic Analysis of Preservative- Treated Timber for 2-(Thiocyanomethylthio)benzothiazole and Methylene Bisthiocyanate Michael J. Kennedy Wood Chemistry and Preservation Section, Queensland Department of Forestry, G. P.O. Box 944, Brisbane 400 I , Australia Preservative-treated timber was quantitatively analysed for 2-(thiocyanomethylthio)benzothiazole and methylene bisthiocyanate by reversed-phase high-performance liquid chromatography on Novapak CI8, with UV detection at 280 and 220 nm. The extraction step involved 3 h of Soxhlet extraction or heated ultrasonic agitation of solid blocks in 19 + 81 methanol - acetonitrile azeotrope, for recoveries of 90-loo%, dependent on the timber species and preservative carrier.In the absence of any sample clean-up, the RSD was 2-4% in two timber species. The method is suitable for processing large batches of sapwood samples treated with dilute emulsions, organic solvent solutions, or gel diffusion formulations. Keywords: 2-(Thiocyanometh y1thio)benzothiazole determination; meth ylene bisthiocyanate determination; high-performance liquid chromatography; preservative-treated timber 2-( Thiocyanome t hylthio) benzo thiazole (TCMTB) and methylene bisthiocyanate (MTC) are amongst many fungi- cidal compounds being tested as wood preservatives in an international attempt to find less environmentally damaging substitutes for chlorophenols. Tetra- and pentachlorophenol have been widely used both in oil solution for long-term timber preservation and in aqueous formulation (as sodium salts) for surface treatment of green timber against sapstain and mould fungi.Emulsions of TCMTB and MTC, either alone or in combination, have been tested for sapstain control with varying degrees of success in many c0untries.1-~ Attempts have been made4 to explain performance differences in terms of timber species, fungi and climate. Only recently has any work been undertaken to check, by chemical analysis, the distribution of the compounds in or on treated timber. An attempt to analyse treated timber for MTC by gas - liquid and high-performance liquid chromatography was reported5 to be unsuccessful owing to difficulties with extraction and separ- ation. An alternative bioassay technique was developed, and showed that poor performance of MTC against surface staining could be attributed to excessive diffusion into the timber, leaving sub-toxic surface concentrations of MTC.TCMTB is also being evaluated by this Department for use as a permanent wood preservative, where full impregnation with aqueous emulsions of TCMTB is being pursued using vacuum-pressure processes. Organic solvent solutions of TCMTB could also be used for permanent preservation of joinery. A third avenue under investigation6 is the use of TCMTB in fungi toxic gels and other concentrates in bandages applied to the ground line region of soft-rot affected CCA- treated eucalypt power transmission poles. Analytical methodology for TCMTB is absent from the literature. MTC has been determined in technical materials and aqueous samples by HPLC.7 These compounds are determined in formulations by one manufacturer8 by HPLC and GLC techniques, respectively, but no method is yet available for the analysis of timber.The development of a new timber preservative can involve thousands of analyses when establishing toxic thresholds, penetration mechanisms, distri- bution patterns and leaching or degradation characteristics of the compound in timber. Once the proposed preservative is adopted commercially, chemical analysis continues for quality control at the timber treatment plant and for routine checks by regulatory agencies. This Department, in administering the State legislation covering timber treatment, analyses many thousands of samples annually to check on compliance with retention and penetration requirements.It is therefore important that analytical techniques be simple, suitable for batch processing and reproducible. Generally, extensive clean-up or concentration steps are incompatible with these requirements. When timber that has been treated with organic solvent or oil-borne preservatives is ground, losses occur due to smearage of oils within the grinding chamber.9 An extraction technique suitable for whole pieces of wood is preferable to grinding. Further, the hygroscopicity of wood requires that preservative retention be specified at some reference moisture content. In Australasia there is a trend towards the adoption of “oven-dry moisture” as this standard, with retentions expressed as a percentage by mass of oven-dry wood (YO r n h ~ ) .~ O When whole pieces of wood are extracted and then oven-dried before weighing, the retention may be expressed in the desired form without the necessity of determining the moisture content on a “matched” sample, provided that it can be shown that the extracting solvent removes insignificant amounts of wood substance. This paper reports the development of an HPLC method that is suitable for the routine analysis of TCMTB and MTC in the sapwood of a softwood and a hardwood treated with dilute emulsions, organic solvent solutions, or diffusing bandage formulations. Experimental Instrumentation A Waters Associates high-performance liquid chromato- graphy system consisting of 6000A and M45 pumps, Model 660 solvent programmer, WISP 710B autoinjector and Model 450 variable-wavelength UV detector, was coupled to a Hewlett-Packard 3388A computing integrator with basic programmability and external events control.Apparatus Bransonic B-220 or B-52 ultrasonic baths were used in sample extraction. These are rated at 125 and 240 W, respectively. Reagents Re-crystallised TCMTB and MTC were supplied by the manufacturer, Buckman Laboratories, Memphis, TN, USA, at 99% purity. Methanol, acetonitrile and methylene chloride for extraction studies were of LC grade and supplied by702 ANALYST, JUNE 1986, VOL. 111 Waters Associates. Water was resin de-ionised before distil- lation and used within l d. Extraction of Sapwood Blocks, Laboratory Treated with Aqueous and Organic Solvent Solutions Small transverse section blocks, 25 mm x 20 mm x 4 mm, were cut from seasoned sapwood of slash pine (Pinus elliottii Engelm var.elliottii) and spotted gum (Eucalyptus maculata Hook). They were treated with either ( i ) 1% aqueous emulsion of Busan-1009 (Buckman Laboratories, Sydney) or (ii) a 3.3% solution of organic solvent formulation Busan-30 in Solvesso 150 (an aromatic solvent from Exxon Chemical Australia, of boiling range 191-212 "C). The concentration of active ingredients in these was (i) 0.1% TCMTB and 0.1% MTC and (ii) 1.0% TCMTB. The treatment schedule involved evacuation to -95 kPa for 20 min, flooding under vacuum, and maintaining at -96 kPa for a further 20 min. The system was then returned to atmospheric pressure and the blocks allowed to soak for 30 min. On removal, the blocks were weighed for solution uptake, stored wet (wrapped in poly- thene) for 3 d to allow any solute - subetrate interaction to occur, and then air-dried to their equilibrium moisture content (EMC) .Extraction solvents examined included 1,1, l-tri- chloroethane (BDH Chemicals, AnalaR), methylene chloride, acetonitrile and 19% methanol in acetonitrile (azeotrope). These were used at elevated temperatures in an ultrasonic bath, or using Soxhlet extraction with at least ten solvent changes per hour. Extraction of Sections from Poles Treated with Diffusion Bandages Sound Eucalyptus sp. transmission poles (copper - chrome - arsenate treated before installation) were bandaged with a product produced by Buckman Laboratories, Sydney, con- taining a diffusing gel formulation of TCMTB.After 18 months the sapwood was sampled with a core-boring bit and sectioned to provide blocks 4 mm thick in the direction of the grain. These blocks were extracted three times each with 19 + 81 methanol - acetonitrile azeotrope in an ultrasonic bath at 55 "C, and the combined extract analysed by the proposed HPLC procedure. The timber residue was exhaustively extracted, and the combined extracts concentrated and checked for the presence of remaining TCMTB using the HPLC procedure described below. Procedures Preparation Cut the sapwood analytical zone into transverse section blocks (maximum 4 mm along the grain) with a sharp saw, and secure samples of about 2 g for sapstain-treated wood or about 0.5 g for high-level treated wood. Extraction Extract the blocks by Soxhlet extraction or ultrasonication.Soxhlet extraction. Place the blocks in a 50-ml or 100-ml Soxhlet extraction chamber. Add the appropriate volume of 19 + 81 methanol - acetonitrile solution to the reservoir, and extract for 3 h at a minimum of ten cycles per hour. Evaporate the extract to less than 25 ml by redirecting the condensate to waste rather than returning it to the reservoir. Decant into a calibrated flask, and make up to the mark. Ultrasonication. Place the blocks in a 50-ml covered beaker with 10 ml of 19 + 81 methanol - acetonitrile solution, and place in an ultrasonic bath at 50-55 "C for 1 h. Decant the extract, add 7 ml of fresh solution and re-extract the blocks for a further hour. Repeat a third time, making the combined extract up to a final volume of 25 ml.Dry the extracted blocks at 103 "C overnight and weigh for oven-dry mass. Filter the extract through a 0.45-km PTFE filter, and submit 15 pl to HPLC under the following conditions. HPLC conditions Column. Waters Associates, Radial-PAK Cartridge 8NVC185 with 5-pm particles, CIS functionality and dimen- sions of 8 mm x 10 cm. Mobilephase. TCMTB, 2.5 ml min-1 of 60 + 40 acetonitrile - water; MTC, 2.9 ml min-1 of 6 + 94 acetonitrile - water for 7 min, 100% acetonitrile flush for 3 min and 6 + 94 acetonitrile - water for 8 min equilibration delay. Detection. TCMTB, 280 nm UV, 0.08 a.u.f.s.; MTC, 220 nm UV, 0.01 a.u.f.s. Quantitation. By peak area, against standards in aceto- nitrile. For sapstain treatment, use 0.0002% and 0.002°/~ mlV TCMTB and MTC.For high-level treatment, use 0.002% and 0.02% mlV TCMTB. MTC peak integration may require care in defining peak termination points because of base line swells (use BL MODE 1 on an HP 3388A integrator). Express the results as a percentage by mass of oven-dry wood. Performance Characteristics of the Method Recovery Small transverse section blocks were cut from seasoned sapwood of slash pine (30 mm x 20 mm X 4 mm) and spotted gum (30 mm x 16 mm x 4 mm) and placed in 50-ml beakers. The addition of 1.00 ml (slash pine) or 0.50 ml (spotted gum) of 1% Busan-1009 emulsion or 3.3% Busan-30 solution dropwise by pipette allowed permeation into the blocks, at realistic volumetric uptake rates of 420 1 m-3 and 260 1 m-3 for the softwood and hardwood, respectively. Drying of the spiked blocks was restricted for 2 d to allow any possible solute - substrate interaction to occur, and the blocks were then air-dried to EMC. They were extracted and analysed using the proposed procedure.Identical aliquots of the treatment solutions were added to empty beakers, evaporated to dryness under ambient conditions, and re-dissolved in 19 + 81 methanol - acetonitrile for comparison with the block extracts. The extracted blocks were then air-dried, ground through a 5-mm screen in a cross-beater mill and re-submitted to the proposed procedure, to check whether any additional recovery was available from the comminuted sample. Interferences Additional active ingredients, which could possibly be present in the preservative formulation include the insecticides aldrin, dieldrin, permethrin and deltamethrin, and the fungicides pentachlorophenol, copper and zinc naphthenate, tributyl- tin oxide, dibutyltin oxide, azaconazole (1-{ [2-(2,4- dichlorop hen y 1) - 1 ,3-dioxolan-2-yl] me t h yl} - 1 H- 1 ,2,4- triazole), imazilil { 1-[2-(2,4-dichloropheny1)-2-(2- propeny1oxy)ethyll-lH-imidazole} and IPBC (l-iodo-2- propynylbutylcarbamate).Solutions containing equal amounts of these with TCMTB and MTC were prepared and examined for the absence of interference. Co-extractive wood components could also interfere, par- ticularly with MTC determination. Sapwood blocks of Pinus elliottii (slash pine), P. radiata (radiata pine), P. sylvestris (Scots pine), Fagus sylvatica (beech), Eucalyptus maculata (spotted gum), E.microcorys (tallowwood), E. regnans (mountain ash), E. obliqua (messmate), E. drepanophylla (grey ironbark) and Tristania suaveolans (swamp box) were subjected to the proposed extraction procedure, and extracts were checked for the absence of intefering peaks. To ensure that the extracting solvent did not remove wood substance, transverse-section blocks were cut adjacent to eachANALYST, JUNE 1986, VOL. 111 1 I I I I I I F 1 2 3 4 5 6 7' 1318 703 1 3 other along the grain of single pieces of uniformly seasoned sapwood of both timber species. One block was oven-dried at 103 "C overnight and the adjacent block was extracted by the proposed procedure before oven-drying. Triplicate sets were examined for differences in the mass losses produced. Repeatability and Reproducibility It is not possible to test the repeatability and reproducibility of the method on treated blocks because of the high natural variability in uptake, even in blocks treated in the same charge.This is true for both pure solutions and emulsions, because variations in density, the early-wood to late-wood ratio and permeability between blocks causes variations in retention, which add significantly to the observed RSD values, poorly reflecting the true precision of the method. To obtain a more homogeneous bulk for this work, small blocks of similar uptake were pooled and ground together to provide three samples for each test species. These samples were analysed in triplicate on three separate occasions, and a two-way analysis of variance of the data was used12 to obtain the precision estimates for the method.Results and Discussion Typical chromatograms are shown in Fig. 1. Run tim elm in Fig. 1. Chromatograms of extract from slash pine containing 0.021% mlm TCMTB and 0.027% mlm MTC: (a) 280 nm, 0.08 a.u.f.s.; (b), 220 nm, 0.01 a.u.f.s. 1, Wood extractives; 2, formulation components Busan-1009); 3, TCMTB (0.0017% mlV, this extract); 4, MTC f 0.0022% m/V, this extract); 5, acetonitrile flush Extraction Procedure Analysis of the Busan-1009 emulsion before and after treatment of blocks revealed marked changes in concentra- tion. This was attributed to stripping of the active ingredients from the emulsion into the timber blocks and precluded calculation of block retentions from mass uptakes. Hence extraction techniques could not be compared in terms of percentage recovery, but only on absolute amounts recovered per block. Extraction rates for TCMTB from blocks treated with 1% Busan-1009 are given in Fig.2. l,l,l-Trichloroethane and methylene chloride, although good solvents for the active ingredients, are poor extracting solvents for wood. More polar acetonitrile and the methanol - acetronitrile azeotrope have a greater ability to swell the wood cell wall and thus gain access to the compounds deposited within.9.11 On polarity consider- ations, methanol should be more efficient again, but was found to degrade TCMTB at higher concentrations in acetonitrile. Solutions containing 0.002% mlV TCMTB, after standing at 55 "C for 3 h, produced chromatograms in which the peak area of TCMTB decreased as the methanol content increased from 40 to 100%.There was a corresponding appearance and increase in area of another peak, which eluted before TCMTB , and was assumed to be a hydrolysis product. No degradation occurred when the azeotrope was used. The 19% methanol - acetonitrile azeotrope was selected as the most suitable, giving superior extraction rates after 3 h of Soxhlet extraction or ultrasonication at 55 "C with a solvent change each hour. Extraction of MTC by this procedure was also checked, and was found to be superior to the alternatives. Blocks treated with the organic solvent solutions were also subjected to this procedure, with negligible TCMTB found in the extracts after 3 h. Blocks from poles treated with gel - TCMTB bandages, having TCMTB retentions of 0.9 to 3% I- r P r , l , l l r / l - 0 I 2 3 4 5 6 7 0 1 2 Extraction time/h Fig.2. Solvent extraction of TCMTB from Busan-1009 treated blocks: (a) slash pine; ( b ) spotted gum. 1, 1,1,1-Trichloroethane, 55 "C ultrasonic bath; 2, methylene chloride, 35 "C ultrasonic bath; 3, methylene chloride, Soxhlet extraction; 4, acetonitrile, 55 "C ultrasonic bath; 5 , acetonitrile, Soxhlet extraction; 6, methanol - acetonitrile (19 + 81), 55 "C ultrasonic bath; and 7, methanol - acetonitrile (19 + 81), Soxhlet extraction Table 1. Recovery by the proposed method for two species Formulation Emulsion Organic solvent Recovery* Retention, Retention, Recovery* of Timber YO mlm MTC, TCMTB, % mlm TCMTB, species in wood Y O Yo in wood Y O Slashpine .. . . 0.08 99.6 100.0 0.8 99.4 Spottedgum . . 0.03 100.7 93.3 0.3 90.4 * Mean recovery of triplicate determinations.4 op Table 2. Precision estimates for the determination of TCMTB in treated wood Amount of TCMTB in wood, o/o m/m x lo4 Slash pine samples Spotted gum samples (i) Determination: 1 2 3 1 2 3 Dayl . . 1392 1562 1580 1155 1220 1206 1221 1228 1279 329 322 343 344 360 366 377 383 373 Day2 . . 1404 1462 1537 1210 1106 1163 1299 1319 1264 324 330 325 342 374 344 387 379 399 Day3 . . 1361 1412 1358 1190 1246 1159 1285 1238 1309 318 333 325 317 349 341 381 369 369 (ii) Analysis of variance: Degrees of Mean freedom square F-ratio Degrees of Mean freedom square F-ratio . . . . . . . . . . . . . . Days 2 2405.590 0.86 Days . . . . . . . . 2 360.7037 3.00 Samples .. . . . . . . . . . . 2 168255.8 - 2 6162.370 Days x samples . . . . . . . . . . 4 7537.260 2.71 Days x samples . . . . 4 1 18.8704 0.99 . . . . . . . . . . . . Residual . . . . . . 18 120.4074 Residual 18 278 1.930 - - Samples . . . . . . . . - Repeatability RSD12 = 4.05% Reproducibility RSD12 = 7.10% Repeatability RSD12 = 3.11 Yo Reproducibility RSD12 = 4.59% Table 3. Precision estimates for the determination of MTC in treated wood Amount of MTC in wood, % m/m x 104 ( i ) Determination: 1 Dayl . . 873 907 947 Day2 , . 871 894 905 Day3 . . 853 856 864 (ii) Anulysis of variance: Days . . . . . . . . Samples . . . . . . . . Days x samples . . . . Residual . . . . . . Repeatability RSD12 = 2.23% Reproducibility RSD12 = 6.04% Slash pine samples 2 3 850 870 851 902 899 929 925 769 793 885 906 899 811 813 812 858 853 887 Degrees of Mean freedom square F-ratio 2 5 107.440 13.73 2 13397.33 4 733.6 100 1.97 371.8900 - - 18 Spotted gum samples 1 2 3 110 111 118 114 119 119 121 123 125 116 113 114 119 116 116 120 126 131 105 114 106 113 111 113 118 118 118 ?i 2 2 Days x samples .. . . 4 2.277778 0.22 3 Degrees of Mean F F-ratio cl freedom square Days . . . . . . . . 2 94.11 1 11 9.14 247.0000 - Samples . . . . . . . . 2 P - . . . . . . \o cx o\ Residual 18 10.29629 Repeatability RSD12 = 2.75% Reproducibility RSD12 = 5.49% c 0 rANALYST, JUNE 1986, VOL. 111 705 mlm according to the above procedure, also released negli- gible TCMTB on further exhaustive extraction (less than an additional 2% relative). Recovery Recoveries at the retention levels tested, which should be typical of commercial production, are given in Table 1.Recovery is excellent for MTC from both species and for TCMTB from slash pine. Slightly poorer yet acceptable recoveries for TCMTB from spotted gum may reflect its extreme sapwood density of 950 kg m-3. Grinding the extracted sample gained an additional 1.9% from the emul- sion treated and 8.7% from the organic solution treated spotted gum, increasing recovery totals to 95.2 and 99.1%, respectively. No additional recovery was obtained from slash pine or for MTC. Interferences The extracts of untreated sapwoods and possible additional preservative components produced no significant peaks in the region of elution of TCMTB or MTC. As most preservative treatment specifications require only sapwood penetration, and most retentions refer to the sapwood zone, no heartwood should be included in an analytical sample.It was not considered necessary, therefore, to evaluate heartwood com- ponents for interference. Application of the method to the analysis of heartwood would require interference and recovery studies for the timber species concerned. The mean mass losses for slash pine were 11.32% by oven drying and 11.31% by the proposed procedure with oven drying. The corresponding values for the spotted gum were 12.58 and 12.69%. These differences are not significant: the procedure does not remove non-volatile wood substance from the sapwood of either test species. Repeatability and Reproducibility Tables 2 and 3 contain the replicated analyses for each sample on 3 d, and the analysis of variance used to calculate the precision estimates.With repeatability RSD12 in the region of 2-4%, and reproducibility RSD12 in the region of 4-7%, the precision is good for the material being analysed. The extreme variability in morphology and chemical composition, even within a single piece, and the resultant uneven preservative distribution makes the measurement of precision very diffi- cult, as any sample will contain particles from high and low retention zones within the piece. Precision estimates are rarely presented in published methods for the analysis of treated timber. Conclusions The method fulfils the requirements for a rapid, routine determination of TCMTB and MTC in treated timber, at a level of precision adequate for both the present developmen- tal phase of Busan preparations and future commercial use. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. References Butcher, J. A., Mat. Org., 1983, 8(1), 51-70. Eslyn, W. E., and Cassens, D. L., For. Prod. J . , 1983,33,65. Cserjesi, A. J., and Johnson, E. L., For. Prod. J., 1982,32,59. Dickinson, D. J., and Henningson, B., The International Research Group on Wood Preservation, Stockholm, 1984, Document No. IRG/WP/3245. Williams, G. R., Eaton, R. A., and Lewis, D. A., The International Research Group on Wood Preservation, Stock- holm, 1985, Document No. ERG/WP/3335. Chin, C. W., McEvoy, C., and Greaves, H., Int. J. Wood Preserv., 1982, 2, 55. Shustina, R., and Lesser, J. H., J. Liq. Chromatogr., 1984, 7 , 2653. Van Deren, J. M., Buckman Laboratories Inc., Memphis, Tennesee, personal communication, 1984. Kennedy, M. J., Int. J. Wood Preserv., 1986, submitted for publication. Leightley, L. E., and Kennedy, M. J., The International Research Group on Wood Preservation, Stockholm, 1983, Document No. IRG/WP/3227. Leutritz, J., Proc. A . W. P.A., 1971, 198. Youden, W. J., and Steinder, E. H., “Statistical Manual of the AOAC,” Association of Official Analytical Chemists, Wash- ington, DC, 1975, p. 80. Paper A51361 Received October 11 th, 198.5 Accepted January 22nd, 1986

ISSN:0003-2654    

DOI:10.1039/AN9861100701

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

ANALYST JUNE 1986 VOL. 111 707 Analytical Investigation of Some Fluorogenic Reactions of Indol-3-yl Acids with o-Phthalaldehyde Part 11.* Thin-layer Chromatographic Studiest Tereza C. M. PastoreS and Clausius G. de Lima$ Departamento de Quimica Universidade de Brasilia Brasl'lia D. F.-70.9 Brazil The fluorogenic reagent o-phthalaldehyde was examined in the presence of sulphuric acid as a possible spray reagent for the visual determination of some indol-3-yl acids including indol-3-ylacetic acid (IAA) after thin-layer chromatography. Concomitantly the effects of 2-mercaptoethanol or cysteine UV irradiation and gaseous electrical discharge were also examined jointly with the o-phthalaldehyde - sulphuric acid reagent. Comparisons between the systems studied were carried out and it was concluded that the best system for the detection of the compounds examined was the o-phthafaldehyde - sulphuric acid system principally if associated with an UV pre-irradiation treatment.Keywords Indol-3-ylacetic acid; o-phthalaldeh yde reagent; thin-la yer chromatography; ph ytohormones Several methods for the in situ determination of indol-3-yl acids through the use of fluorogenic spray reagents after thin-layer chromatography (TLC) have been proposed or used owing to the biological interest of some of these compounds which either are or behave as phytohormones. Some also increase their concentration in excreta or in cerebral or cerebrospinal fluid as a consequence of some illnesses principally owing to modifications in the human metabolism of tryptophan or serotonin.1-8 Formaldehyde,1-7 either in the vapour state or in solution (mostly in HC1 medium) or o-phthalaldehyde (OPA) - HCl - cysteineg are usually the fluorogenic reagents studied. Fluorogenic reactions in solution or on solid surfaces9 usually provide an extremely sensitive method for the determination of trace amounts of organic molecules. The utilisation of the formaldehyde - HCl reagent for the fluorimetric determination of several indole compounds associated with TLC has been described in reference 1. Using this reagent system indol-3-ylacetic acid (IAA) , 3-(indol-3-yl)propionic acid (IPA) 4-(indol-3-yl) butyric acid (IBA) and 5-hydroxyindol-3-ylacetic acid (5-HIAA) were detected (in silica gel) showing visual limits of detection of 5 , 10 10 and 50 ng respectively.Using paraformaldehyde vapours Cowles et aZ.2 detected 1 yg of IAA. Toneby,3 employing paraformaldehyde in solution detected IAA, 5-HIAA and 5-methoxyindol-3-ylacetic acid (5-MIAA) (25 , 6.25 and 12.5 ng per spot respectively). Formaldehyde in hydrochloric acid medium was utilised by Larsson et al.,4.5 when a limit of detection of 30 ng of IAA was found a much better result compared with the 10 pg previously found when paraformaldehyde vapours were used.6 Although 5-HIAA was detectable (3 pg) using the latter technique no signal was observed from 5-HIAA or IAA when OPA in glacial acetic acid was tested.6 Recently Kraus and Richter7 detected IAA IPA IBA, 3-indol-3-ylacetic acid (ILA) and other similar compounds, using the formaldehyde - HCl and TLC - pH-gradient tech-* For Part I of this series see reference 20.t Presented in part at SAC 83 the 6th International Conference on Analytical Chemistry Edinburgh UK 17-23 July 1983. $ Present address Laboratorio de Produtos Florestais IBDF-MA SG 12 Universidade de Brasilia Brasilia D.F.-70 910 Brazil. § To whom correspondence should be addressed. nique while studying the factors that could affect the TLC of such compounds. The detection of 5-HIAA after separation using TLC, was made possible as demonstrated by Korf and Valkenburgh-Sikkema,8 when the conditions employed were similar to those suggested by Maickel and Miller10 (OPA, 0.1% mlV in HCl - CH30H 1 + 1 VIV) modified by the previous atomisation of a cysteine solution (1% mlV).OPA - HCl was also employed in the presence of cysteine by Narasimhachari and Plutll for the identification of bufotenine (5-hydroxy-N,N-dimethyltryptamine) and 5-me t hox y-N,N-dime thy1 tryptamine . OPA (0.2 YO m/V) in concentrated sulphuric acid medium was proposed by Szabo and KaracsonyQ for the detection of some indole alkaloids (ergolenes and ergolines) after successful results. Using TLC and a appropriate instrumentation 1 ng of ergoline can be detected by this method. OPA has also been utilised as a spray reagent for the determination of several other compounds such as hist-amine,12-15 histidine,l4 hydrazines,16 peptides and amino acidsl7-18 and for compounds such as amphetamine mescaline and noradrenaline. 19 Turner and Wightman14 have reviewed the subject and the list of compounds that can be detected by the use of OPA continues to grow.In this work we describe the utilisation of OPA - H2S04 as a TLC visual detection reagent for the determination of some indol-3-yl acids based on previous work carried out in this laboratory.20 Previously it was observed that OPA reacts in sulphuric acid not only with 5-HIAA and 5-MIAA but also with IAA IPA IBA and indol-3-ylpyruvic acid (IPyA). The combination of this reagent with 2-mercaptoethanol (2-Me), previously proposed by Roth for the determination of amino acids,21 or with cysteine,8JOJl were also examined in an attempt to verify the effect of these spray reagents in the detection of the indol-3-yl acids. Also the effect of UV radiation (which in most instances showed a beneficial effect when the reactions of indol-3-yl acid with OPA - H2S04 were examined in solution2") and gaseous electrical discharge in the presence of low-pressure ammonia vapour,22 were investi-gated.Although the mechanism involved and the possible reaction products are not known this work describes a further and useful expansion of the utilisation of OPA as a TLC spray reagent for the detection and eventually the determination of indol-3-yl acids 708 Experimental Reagents The reagents and solvents were as described previously.20 Thin-layer glass plates (100 x 200 mm or 50 x 200 mm) or microscope slides (26 x 77 mm) used in the electrical discharge experiments were coated using silica gel without a fluorescent indicator (H Type 60 Merck Darmstadt FRG).One-dimensional (ascending) thin-layer chromatography was carried out the plates being washed previously with the solvent system employed to remove any impurities that might be present. The indole solutions (prepared in ethanol - water, 10 + 90 V/V) were applied to the plates (at 10 mm from the base) using a Hamilton lo-@ microsyringe (Hamilton Reno, NV USA). After the chromatography (the solvent front was 150-160 mm from the starting point) the plates were sprayed with the reagent system. The running times were 35-45 min. Apparatus A UV viewing box (Ultra Violet Products San Gabriel CA, USA) was used for viewing the plates. The irradiation of the plates (350 nm region) was carried out using the photochem-ical reactor described previously.20 Photographs were taken using a Polaroid assembly (Polaroid Cambridge MA USA, Model MP-4) and film (667 black and white Land Film 3000 ASA Panchromatic).In order to study the effect of a gaseous electrical discharge on the TLC of the indole compounds a discharge chamber was assembled similar to that described in the work of Davies and Pretorius.22 A high-frequency coil (4-5 MHz 2 kV, Speedvac Model T2 Edward High Vacuum Ltd. Crawley, Sussex UK) was employed in the experiment and a small mass of ammonium hydrogen carbonate was present in the chamber as recommended previously.22 Procedure The preparation of the plates the chromatographic process and the detection followed in general the usual procedure.1 The plates were air-dried (room temperature) after the pre-run spotting (2 or 4 pl) and the final run.After being chromatographed the plates were examined in the viewing box before spraying (so as to examine any possible native fluorescence) and again afterwards. The UV treatment of the plates was carried out in the reactor for 10 min when required. Results and Discussion OPA - H2S04 as a Spray Reagent In order to study the application of OPA - H2SO4 as a spray reagent some different solvent systems were examined. Initially a system consisting of CH30H - CHC13 (75 + 25 V/V) was tested but strong tails were observed. The addition of a ANALYST JUNE 1986 VOL. 111 small amount of glacial acetic acid drastically reduced or even eliminated the tails. Two similar systems were then examined: system I (CH30H - CHC13 - CH3COOH 75 + 25 + 0.4 VIV) and system I1 (CH30H - CHC13 - CH,COOH 75 + 25 + 0.2 V/V).With the exception of the RF values of ILA and 5-HIAA7 which were larger than the RF observed in system I1 (Table 1) there were no other major differences. Although neither of these two systems gave a perfect separation of most of the compounds system I was used to test the spray reagents proposed and examine the other conditions that were the objective of this work. Obviously the separation system suggested could only be used in circumstances where no more than one of the overlapping compounds is present. The best separation system is the one recently proposed by Kraus and Richter,7 in which a TLC - pH - gradient layer technique was successfully developed for the separation of several indol-3-yl acids including IAA IPA IBA and ILA.Experiments using a 0.15% m/V OPA solution (in 7 M H2SO4 with 10% V/V ethanol) showed positive results which can be seen in Table 1. The visually estimated limits of detection reveal that it is possible to detect 15 ng (equivalent to 72 pmol) of 5-MIAA per spot or 40 ng (230 pmol) of IAA. The worst result was obtained with IPyA (200 ng or 985 pmol). The spots were visible for only 24 h with the exception of 5-MIAA which was visible for one week. Fig. 1 shows a UV illuminated photograph of a typical TLC plate after being run and sprayed with the OPA - H2S04 reagent. Concentrations used for production of the photograph were usually high (200-400 ng per spot). Mild heating of the TLC plate (45-80 "C) in an open atmosphere (and room fluorescent light) after reagent spraying led to the appearance of a yellow fluorescent background in the sprayed region.Influence of UV Irradiation When the TLC plates were irradiated 10 min after the run, before the reagent spraying and then finally sprayed lower limits of detection were found. At the same time a colour Fig. 1. UV illuminated TLC of some indol-3-yl acids using OPA -sulphuric acid reagents. A IAA; B IPA; C IBA; D ILA; E IPyA; F 5-MIAA; and G 5-HIAA Table 1. Analytical characteristics of the indole compounds using the OPA - H2S04 spray reagent Limit of detection with pre- Spot colour RF Limit of detection/ irradiation/ ng (pmol) ng (pmol) Before After Compound System I System I1 per spot per spot irradiation irradiation IAA .. . . . . 1 .oo 1 .oo 40 (229) 20 (114) Blue - violet Greenish blue IPA . . . . . . 1.05 2 0.04 1.06 2 0.03 50 (264) 40 (212) Blue - violet Greenish blue IPyA . . . . . . 0.07 f 0.02 0.06 f 0.02 200 (985) 100 (493) Blue -violet Greenish blue ILA . . . . . . 0.28 k 0.07 0.21 f 0.02 40 (195) 10 (49) Blue Greenish blue IBA . . . . . . 1.11 _+ 0.02 1.07 k 0.02 50 (246) 30 (148) Blue - violet Greenish blue 5-HIAA . . . . 0.68f0.09 0.52 f 0.08 40 (208) 10 (52) Green Yellowish green 5-MIAA . . . . 1.01 k 0.05 1.04 f 0.03 15 (72) 5 (24) Bluish Blue (intense ANALYST JUNE 1986 VOL. 111 709 Table 2. Analytical characteristics of the indole compounds using OPA - 2-ME followed by 7 M H2S04 Limit of detectionhg (pmol) per spot Spot colour Immediate After irradiation Before Compound signal After 24 h (10 min) irradiation After irradiation - - - - - IAA .. . . . . IPA . . . . . . 400(2116) 200 (1058) 40 (212) Violet Whitish blue IPyA . . . . . . ILA . . . . . 400(1951) 40 (195) 40 (195) Blue Yellow IBA . . . . . . 400(1970) 20 (98) 40 (197) Blue Yellow 5-HIAA . . . 20(104) 2.0 (10) 2.0 (10) Orange Greenish yellow 5-MIAA . . . . 2.0(9.7) 2.0 (9.7) 5.0 (24) Light blue Greenish yellow - - - - -Table 3. Limits of detection of indole compounds using OPA -cysteine The simple spraying with OPA - 2-ME did not produce any visible spots. However after acid spraying IPA appeared initially followed by IBA and ILA. In spite of this the fluorophores obtained showed weak fluorescent signals lead-ing in consequence to high limits of detection (400 ng per Compound System I System I1 spot).Table 2 shows the results obtained where an exception can be observed with the 5-substituted indol-3-yl acids IAA . . . . . . (5-methoxy and 5-hydroxy) which showed the smallest of the IPA . . . . . . 200(1058) 200 (1058) IPyA . . . . . . - -- detection limits of the compounds examined. An examination ILA . . . . . . 200(975) IBA . . . . . . 200(985) 200 (985) of the plate after 24 h revealed that for most of the acids a 5-HIAA . . . . 40 (208) 40* (208) decrease was observed in the detection limits with the 5-MIAA . . . . 40 (194) 20* (97) exception of 5-MIAA where no change occurred. Limit of detectiodng (pmol) per spot - -* After 1 h. Effect of Irradiation on the OPA - 2-ME - H2SO4 System The exposure to UV irradiation (350 nm region) before spraying with reagent leads to the quenching of the fluorescent signals of most of the compounds except 5-MIAA.After 24 h, ng (pmol) however some compounds showed a positive signal. In Compound per spot Colour contrast applying UV irradiation 10 min after reagent IPA . . . . . . 40 (212) Yellow spraying improved the limits of detection of IPA ILA and ILA . . . . . . 10 (48)* Yellow IBA (40 ng per spot) ten-fold (Table 2). The limit of detection IBA . . . . . . 40 (197) Yellow of 5-HIAA did not change but a slight increase was detected with 5-MIAA. No signal was observed however in any of these attempts when IAA or IPyA were tested as analytes. Action of Cysteine Table 4.Limits of detection of IPA ILA and TBA using H2S04 as a spray reagent Limit of detection/ * The reading was carried out within 2 h of spraying. Table 5. Effect of plasma on TLC Exposed for 5 min Amount/ Exposedfor andsprayedwith Cysteine which has been used jointly with OPA as a TLC Compound ng per spot 5 min* OPA - H2S04* reagent spray in an acidic medium (HC1),8J0 was examined in IAA . . 400 + + two different media system I using an aqueous solution of IPA . . . . 400 - + OPA and cysteine (0.15% m/V of each) followed by spraying IPyA . . 200 + + with HzSO4; and system 11 in which a solution of cysteine ILA . . . . 400 + + (0.15% m/v) was first sprayed followed by the OPA - H2SO4 I B A . . . . 400 + reagent. Table 3 shows the results obtained. As in the previous utilisation of 2-ME no signal was observed with IAA or IPyA.5-HIAA . . 200 The visual limits of detection of the compounds lacking the 5-MIAA . . 200 substitution at the 5-positions were however higher com-pared with the limits found when the OPA - H2SO4 reagent was tested (Table l). One disadvantage found was that on using cysteine a strong fluorescent background signal appeared on the plate within 24 h. H2S04 as a 'pray Reagent H2S04 (7 M) was also examined as a possible fluorescent reagent. However only IPA ILA and IBA in small amounts showed any visible signal (Table 4). An immediate signal from ILA even with an amount of 200 ng per spot was not visible. However 10 ng per spot could be detected if the plate was re-examined 2 h after spraying.The intensity of the fluores-cent signal from IPA and IBA decreased with time; after 6 h no signals were present. Effect of Gaseous Electrical Discharge Following the same line of experiments carried out previously by Davies and Pretorius,22 TLC plates (microscope slides) were exposed to the action of a gaseous electrical discharge -- -- -* (+) Presence or (-) absence of fluorescence signal. change with a red shift was observed in most of the compounds examined (Table 1) resulting in colours and intensities that are more easily detected visually. Tests using irradiation after spraying showed a reasonable quenching in the signals. Only with amounts of 100 ng per spot was it possible to observe fluorescent signals and colour changes. Thus IAA and 5-MIAA showed weak white spots, IAA IBA and ILA changed to a light violet and 5-HIAA showed an intense green spot.Effect of 2-Mercaptoethanol The effect of the addition of 2-ME to the reagent (OPA) before acidification with H2SO4 was examined. This reagent system (OPA - 2-ME) has been used in a basic medium for the detection of several amino acids21 although there are appar-ently no references to its use in an acidic medium. The usual procedure was to spray the TLC plate after the run with the reagent (OPA - 2-ME) followed by acidification with 7 M H2S04 710 ANALYST JUNE 1986 VOL. 111 plasma. Table 5 shows the qualitative results obtained which proved to be unsatisfactory even with the large amounts tested. Only IAA IPyA and ILA showed positive results.It is interesting to observe that after exposure to the discharge and spraying with OPA - H2SO4 a quenching of the signals from the 5-substituted compounds was observed. Of the spray reagents systems tested the system OPA -H2SO4 would appear to be the best if all the compounds examined are to be detected. Further by coupling this system to a UV pre-irradiation treatment a decrease in the limits of detection is obtained. The association of 2-ME and UV irradiation seems however a possible choice when the 5-substituted derivatives are the only compounds to be detected. Presumably the differences observed between the systems could eventually be used for the characterisation of a compound especially if a thin-layer scanner coupled to a fluorimeter is employed.The authors thank E. F. Lima for technical assistance and S. G. de Lima for reviewing the manuscript. 1. 2. 3. 4. 5. 6. References Stahl E. “Thin-Layer Chromatography,” Academic Press, New York 1965. Cowles E. J. Christensen G. M. and Hilding A. C. J. Chromatogr. 1968 35 389. Toneby M. I. J. Chromatogr. 1974 97 47. Larsson L. I . Sundler F. and Hikanson R. J. Histochem. Cytochem. 1975,23 873. Larsson L. I. Sundler F. and Hikanson R. J . Chromatogr., 1976 117 355. Aures D. Fleming R. and Hikanson R. J . Chromatogr., 1968 33 480. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. Kraus Lj. and Richter R. Chromatographie GIT Suppl., 1982 63; Chem. Abstr. 1983 98 27056 h. Korf J . and Valkenburgh-Sikkema T.Clin. Chim. Acta, 1969 26 301. Hurtubise R. J. “Solid Surface Luminescence Analysis,’’ Marcel Dekker New York 1981. Maickel R. P. and Miller F. P. Anal. Chem. 1966,38 1937. Narasimhachari N. and Plaut J. J. Chromatogr. 1971 57, 433. Szab6 A. and Karacsony E. M. J. Chromatogr. 1980 193, 500. Shelley W. B. and Juhlin L. J. Chromatogr. 1966 22 130. Turner T. D. and Wightman S. L. J. Chromatogr. 1968,32, 315. Hikanson R . Juhlin L. Owman C. and Sporrong B. J . Histochem. Cytochem. 1970 18 93. Weeks R. W. Jr. Yasuda S. K. and Dean B. J. Anal. Chem. 1976,48 159. Lindberg E. G. G. J . Chromatogr. 1976 177 440. Schittz E. Schmackerz K. D. and Gracy R. W. Anal. Biochem. 1977 79 33. Gubitz G. Chromatographie 1979 12 779. Pastore T. C. M. Nicola E. M. de M. and de Lima C. G., Analyst 1984 109 243. Roth M. Anal. Chem. 1971 43 880. Davies R. D. and Pretorius V. J . Chromatogr. 1978 155, 229. Mendez E. and Gavilanes J. G. Anal. Biochem. 1976 72, 473. NOTE-Reference 20 is to Part I of this series. Paper A41298 Received August 24th 1984 Accepted January 8th 198

ISSN:0003-2654    

DOI:10.1039/AN9861100707

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

ANALYST JUNE 1986 VOL. 111 711 Application of Computer-based Pattern Recognition Procedures in the Study of Biological Samples. Comparison of the Cuticular Hydrocarbon Profiles of Different Colonies of the Black Imported Fire Ant Jeffrey H. Brill," Tom Mar Howard T. Mayfield and Wolfgang Bertsch Department of Chemistry University of Alabama P.O. Box H University AL 35486 USA Computer-based pattern recognition procedures have been applied to gas chromatographic data obtained from the cuticular hydrocarbons of different colonies of the black imported fire ant. These included both supervised and unsupervised learning methods based on the ARTHUR chemometrics system package. The results obtained from ten samples indicate significant differences in the hydrocarbon profiles of different colonies of the black imported fire ant.The relative effectiveness of the different procedures in making correct classifications was approximately 90%. Keywords Computer-based pattern recognition; imported fire ants; cuticular hydrocarbons The primary function of the cuticular hydrocarbons is to prevent excessive water loss from the insect.1 However it has been shown that the cuticular hydrocarbons are also involved in chemical communication in insects.2-5 The cuticular hydro-carbons of the imported fire ants have been examined and characterised. They have been shown to be complex mixtures containing homologous series of n-alkanes and also internally branched monomethyl and dimethyl alkanes.677 It has been suggested that the cuticular hydrocarbons of the imported fire ants are involved in species and caste re~ognition.~?5 Another function of the cuticular hydrocarbons might be in colony recognition.Worker ants are sensitive to alien colony odours. Alien intruders in a colony appear to be detected by contact chemoreception,498.9 which would indicate that the cuticle plays a role in recognition. Thus it might be expected that there would be differences in the cuticular hydrocarbon profiles from different con-specific colonies. The cuticular hydrocarbon profiles from a number of different colonies of the black imported fire ant Solenopsis richteri were investigated using a variety of computer-based chemometric data analysis methods in order to determine if there were statistically significant differences between the different colonies.Theory Pattern recognition is a process whereby a hidden property of a collection of objects (in this instance colony profiles) can be detected and/or predicted by using indirect measurements on the individual objects. 10 In many instances a single discrimi-nating measurement cannot be found. Only a combination of measurements provides sufficient information. The human observer is the best pattern recogniser when dealing with two or three measurements per object. However when the number of measurements and objects greatly exceeds three or four the problem can only be handled successfully by using computer-based pattern recognition procedures. Pattern recognition procedures have proved useful in a number of different applications inside and outside of chemistry.These range from weather forecasting and speech patterns to bacterial taxonomy pyrolysis of high polymers, insect semiochemicals correlation of aromas in flavours, forensic methods and clinical chemistry. The main areas in which pattern recognition procedures have been applied are * Present address Department of Chemistry University of Missouri Columbia MO 65211 USA. spectroscopy in particular the deconvolution of unresolved spectra and in electrochemistry.11 Pattern recognition can be divided into two main classes-supervised and unsupervised learning. In supervised learning, the categories to which the data vectors belong are known to the computer by means of the training set. A mathematical model or discriminant function is generated for each category in the training set.This then allows unknown objects to be classified.llJ2 In unsupervised learning there is no training step as the classes to which the objects belong are not known to the computer (although the investigator may know). The computer then attempts to determine any natural clusterings in the data set which would indicate an underlying structure in the data. In supervised learning it is advantageous to subject the data set to one or more pre-treatment procedures in order to facilitate the extraction of useful information from the data. The goal of pre-processing is (a) enhanced separation of the different categories to make classification easier and (b) reduction of the dimensionality of the data to make classifica-tion more efficient .11-13 Pre-processing usually consists of three steps scaling feature weighting and feature selection.In these studies we have used autoscaling which scales the raw data so that the resulting new features all have a mean of zero and a variance of 1.0. In this way any inadvertent bias that may have occurred as a result of differences in the magnitudes of the features is removed.12J3 Weighting tech-niques assign weights to individual features based upon the ability of the features to classify the objects in the data set. Large feature weights are assigned to those features that enhance a given classification whereas features that detract from the classification are assigned smaller weights. 13 We have used the Fisher weighting method in these studies.13 It is advantageous to favour the features that carry the largest amount of information.If the additional features are used, noise or information unrelated to the problem of interest is introduced into the system. This increases the difficulty of the analysis and may detract from its reliability. The process of choosing which features to use is known as feature selec-tion. 11-13 It is practically impossible for a human to visualise the data when it is displayed in n-dimensional space (where n > 3). Therefore computers are used to project an approximation of the points from n-space into two-dimensional space to permit visual inspection of the data. This can be carried out using a procedure known as non-linear mapping (NLM) whereby non-linear combinations of the n-coordinates of the n data vectors are taken.14 In NLM the multi-dimensional data ar 712 ANALYST JUNE 1986 VOL.111 represented in two dimensions such that the interpoint distances are preserved. Another popular display method is the Karhunen - Loeve (K - L) projection.12-14 This procedure is based on principal components analysis and consists of the extraction of the eigenvalues and corresponding eigenvectors of the data covariance matrix. Because the information in the data set is preserved in as few features as possible the K - L transformation is often used for data reducti0n.12~13 The ARTHUR package provides the four classification methods used in this application. These methods are the K nearest neighbour (KNN) method the linear learning machine (LLM) method the multi-category classification (MULTI) method and the SIMCA method (named for statistical isolinear multiple components analysis).For the KNN procedure classification is based upon the distance of an object to its K nearest neighbours. The objects are considered as points in an n-dimensional hyperspace, where n is equal to the number of measurements made on each object. This procedure is based on the assumption that nearness in space between two points is a good measure of similarity between the corresponding objects.10 The LLM procedure compares the categories in a pairwise fashion and determines a hyperplane using a feedback procedure in such a way that the two classes fall on opposite sides of the hyperplanes.15 The MULTI procedure is a hyperplane discriminant algorithm similar to LLM.However it is used for data with more than two classes. The procedure attempts to achieve a simultaneous separation of all the classses. In the training step rn hyperplanes are generated (where m is the number of classes) such that the kth hyperplane describes the separation of the kth category from the rest of the data.13 The SIMCA method is related to methods of factor analysis and principal components analysis. The method generates a principal components model for the feature space occupied by each category. 12,13 The two methods of unsupervised learning applied to the data in this study were hierarchical clustering (HIER) and the minimal spanning tree (TREE). HIER is based upon the relative similarity of a set of data vectors.16 Initially each data vector is considered to be a separate cluster.A similarity matrix is then generated. The data vectors are then clustered in a stepwise fashion beginning with the two objects that have the greatest similarity until only a single cluster remains. The TREE method generates a minimal spanning tree over the data vectors. The tree is then pruned to determine the natural clustering of the data.16 Experimental Samples Samples of approximately lOO(b2000 workers were each taken from 10 different colonies of the black imported fire ant. The specimens were collected from nests in west-central Alabama and east-central Mississippi. Some of the colonies sampled were adjacent to each other.The samples were kept at -25 "C until used. The cuticular hydrocarbon profiles were establi-shed using a recently described procedure. This method is based on dynamic headspace analysis. Details are described elsewhere. 17 Instrumental The samples were examined by capillary gas chromatography using a Hewlett-Packard 5830A gas chromatograph fitted with a Perkin-Elmer 900 series injection port to accept a pyroprobe insert (Chemical Data Systems Oxford PA). A 16 m x 0.25 mm glass capillary coated with a 0.25 pm film of immobilised OV-1 (prepared according to the method of Grob and Grobls) was used. The column was temperature programmed from 80 to 300 "C at 8 "C min-1. Helium was used as the carrier gas at a flow-rate of approximately 1 ml min-1. The splitting ratio was set to approximately 100 1.Each sample was treated with an internal standard containing 250 ng each of tricosane and dotriacontane. Gas chromatography - mass spectrometry (GC - MS) was also performed on a number of the samples on a Hewlett-Packard 5985A instrument operated in the electron impact (EI) mode. A 15 m X 0.32 mm fused silica capillary coated with a 0.10 pm film of DB-5 phase (J & W Scientific Rancho Cordova CA) was used. Other chromatographic conditions were the same as above. Data Analysis From each of the ten colonies approximately 16 workers of roughly the same size were randomly selected. These were analysed individually by pyroprobe dynamic headspace analy-sis.17 The retention time and absolute area for each peak in the hydrocarbon region of each chromatogram were encoded on to computer cards and the data together with the marker peaks for the SETUP procedure,lg were transferred into a UNIVAC 1100/61 mainframe computer.The internal stan-dard was included in the marker peaks. SETUP is an algorithm for organising gas chromatographic data into data sets for chemometric analysis. The program generates a reference list of all unique peaks in the chromatographic data base. Each chromatogram is given a set of marker peaks, which serve as internal retention time standards. The marker peaks are used by the SETUP algorithm to standardise the retention times in the data base.19 The resulting data set consisted of 160 data vectors comprising 52 features. Each category was represented by an average of 16 data vectors.The raw data set was submitted to the ARTHUR program for chemometric analysis. A number of pre-processing and feature selection procedures were examined to enhance the discriminating ability of the data in the raw data set. A new data set was obtained by first autoscaling the raw data then weighting all the data vectors using the Fisher weighting method.13 Finally a third data set was obtained from the weighted data set by determining the number of weighted features that gave the greatest separation (discrimination) between the categories. After visual inspection of the non-linear maps five features were determined to be the most significant. The four classification methods discussed above were then tested on the final data set.The data set was divided into approximately eight training set - validation set combina-tion for testing the classification procedures. The computer was given the proper classifications in the training sets. It was then given random validation patterns from samples in the data set known only to the investigator and asked for a classification. The leave-two-out method was used in selecting the data vectors to be included in each training set. This refers to a method in which two data vectors from each category are assigned to the validation set and the rest are assigned to the training set. This is repeated until all data vectors from the data set have been used as validation set patterns at least once. In addition the autoscaled unweighted data set was examined by means of the two unsupervised learning pro-cedures described above.Results and Discussion Fig. 1 shows the non-linear map (NLM) for the raw data set, before Fisher weighting. There is a very poor separation among the ten categories in the data set. Following Fisher weighting and feature selection the separation between the categories is markedly improved. Fig. 2 shows the NLM using the five features that gave the greatest separation between the categories. Not all the categories are completely separated ANALYST JUNE 1986 VOL. 111 713 h 0 ++++ 0 f 1 . + + A + Fig. 1. Non-linear map (NLM) of the hydrocarbon profile data set for the ten colonies. Raw data set before re-processing. Colonies: 0 1; A 2; + 3; 0,4; . 5 ; 0 6; A 7; 8 8; 0 9; and .10 Fig. 2. Non-linear map (NLM) of the hydrocarbon profile data set for the ten colonies. Best five features by Fisher weight. Colony symbols are the same as in Fig. 1. Nearest neighbour colonies 1 and 2; 3 and 4; 5 and 6 ; and 8 and 9 This could in part be due to the relatively large number of categories present in the data space. Moreover the ten profiles are from the same species and are therefore not greatly dissimilar. Despite this it is significant to observe that the pairs of nearest neighbour colonies show the greatest separation from each other (Fig. 2). A few outliers were observed in the data set (Fig. 2). The two most noticeable were in category 6. The individual chromatograms for these outliers were examined closely. It appeared that a few pea& in these chromatograms had not been properly integrated causing problems in the treatment of data.For easy perception the data in Figs. 1 and 2 are transformed and displayed in two dimensions. Non-linear mapping is a method whereby a multi-dimensional data space is reduced to a two-dimensional space for display purposes. This reduction can be accomplished only approximately. Every data vector in the multi-dimensional data space (where n >> 3) has a certain distance to every other data vector. In non-linear mapping these distances are calculated and then considered to be constants. Upon reduction to two dimen-sions non-linear mapping attempts to preserve the interpoint distances by an error minimisation method.1°J4 It must be remembered that NLM is a method for displaying multi-A A A 0" Fig.3. Comparison of non-linear mapping (NLM) and Karhunen -Loeve (K - L) projection of the hydrocarbon profile data set for the ten colonies. Best five features by Fisher weight. (a) Karhunen -Loeve rojection first vs. second prmcipal component; (6) non-linear map. Eblony symbols are the same as in Figs. 1 and 2 IS I 20 25 30 20 25 30 trim i n Fig. 4. Sample chromatograms of the hydrocarbon profiles of four of the ten colonies. IS1 and IS2 internal standards. The arrows indicate the five best features by Fisher weight. H and I are nearest neighbour colonie 714 ANALYST JUNE 1986 VOL. 111 Table 1. Results of the K nearest neighbour classification tests on the ten colony profile data set Training set Test set No.of Y O No. of YO K value Category misses Correct misses Correct 1 1 39 65.2 5 68.8 2 7 93.3 2 87.5 3 13 88.4 2 87.5 4 1 99.1 0 100 5 7 92.7 1 92.9 6 16 84.6 2 86.7 7 26 79.7 3 83.8 8 14 87.5 2 87.5 9 6 94.2 2 86.7 1 45 59.8 6 62.5 2 9 91.3 1 93.8 3 6 94.6 1 93.8 4 2 98.2 0 100 5 1 99.0 0 100 6 24 76.9 4 73.3 7 46 64.1 5 72.2 8 0 100 0 100 9 13 87.5 2 86.7 10 0 100 0 100 10 0 100 0 100 3 Overall YO correct 65.6 92.5 88.3 99.2 92.7 84.9 80.1 87.5 93.3 60.2 91.7 94.5 98.4 99.1 76.5 65.1 87.4 100 100 100 Table 2. Results of the linear learning machine (LLM) classification test on the ten colony profile data set Category 1 2 3 4 5 6 7 8 9 10 Training set Test set Overall No.of misses 20 0 5 7 0 4 21 21 0 0 YO Correct 99.0 99.7 99.6 99.8 99.0 98.9 100 100 100 100 No. of misses 23 19 8 17 18 20 24 15 11 13 YO Correct 91.9 93.2 97.2 94.0 93.7 92.8 92.1 94.7 96.0 95.3 YO correct 98.1 99.1 99.4 98.9 99.2 98.9 98.1 98.4 99.5 99.4 Table 3. Results of the multi-category (MULTI) classification test on the ten colony profile data set Training set Category 1 2 3 4 5 6 7 8 9 10 Overall No. of misses 36 3 0 5 0 6 35 0 2 0 87 YO Correct 67.9 97.1 95.5 94.3 72.2 98.1 92.0 100 100 100 100 No. of misses 7 2 1 5 2 3 6 2 1 0 29 YO Correct 56.3 86.7 93.8 68.8 85.7 80.0 66.7 87.5 93.3 81.4 100 Test set Overall % correct 66.4 95.8 99.2 92.2 98.2 92.5 71.5 98.4 97.5 90.7 100 dimensional data in a form that is easily perceived by humans.If classes overlap in the display they may still be separated in multi-dimensional space by any one of a number of classifi-cation methods. Fig. 3 compares the non-linear mapping (NLM) and the Karhunen - Loeve (K - L) methods of displaying the data using the five best features by Fisher weight. It can be seen that the NLM method is superior to the K - L method and gives a better separation of the categories in two-dimensional space. Table 4. Results of SIMCA classification test on ten colony profile data set Training set Test set Overall Category misses Correct misses Correct correct No.of YO No. of YO YO 1 2 3 4 5 6 7 8 9 10 50 9 0 14 15 50 49 1 22 0 55.4 91.3 87.5 84.4 51.9 61.7 99.1 78.8 100 100 7 1 0 2 4 8 9 0 3 0 56.3 93.8 87.5 71.4 46.7 50.0 80.0 100 100 100 55.5 91.7 87.5 82.7 51.3 60.3 99.2 79.0 100 100 This was also found to be the case when the NLMs and K - L projections of the best n features (3 d n G 8) from the same data set were compared.20 Fig. 4 shows the hydrocarbon profiles of four of the ten colonies. The five most discriminating features are indicated by the arrows. It can be seen that there are differences between the different profiles.Although other non-hydrocarbon substances are also removed during sampling, they do not interfere with the hydrocarbon profile.17 GC - MS analysis showed that the components eluting between the two internal standards were hydrocarbons.21 The results for the KNN method ( K = 1 and K = 3) are shown in Table 1. For most categories (colonies) the results are satisfactory considering that a relatively large number of classes were being compared simultaneously. Category 1 yielded the poorest performance. It can be seen in Fig. 2 that this category shows overlap into a number of other categories which might account for its modest performance. The LLM method gave excellent results (see Table 2). However this method compares the categories in a pair-wise fashion.Nonetheless it is significant that a very good discrimination is obtained when the 10 categories are com-pared in a binary fashion. A more appropriate hyperplane discriminant method is the MULTI procedure. Table 3 shows the results from th ANALYST JUNE 1986 VOL. 111 71 5 .$ 0.5 E ClJ - .-.-v) 1 .o 8 7 4 1 6 3 5 2 10 9 Fig. 5. Hierarchical dendrogram of the ten colony profile data set. (Note the numbers along the horizontal axis indicate the colonies) Fig. 6. Minimal spanning tree plot of the ten colony profile data set. The numbers indicate the colonies. The shaded objects represent outliers MULTI method. The results are excellent for most categories. Categories 1 and 7 gave only moderate results. This is not unexpected as these categories tend to overlap a number of other categories in the two-dimensional data space (see Fig.2). The SIMCA method gave the poorest results of the classification methods examined (see Table 4). In the NLM, categories 1 6 and 7 show the greatest scatter and/or overlap (see Fig. 2). This is manifested in the results of SIMCA which is a method based on principal components analysis and attempts to generate a principal components model for the feature space occupied by each category. The results from the HIER method as applied to the autoscaled unweighted data are shown in Fig. 5. From the dendrogram it can be observed that the various categories tend to be separated from one another. This is especially the case for the pairs of nearest neighbour colonies.However, some overlapping does occur especially by data vectors from categories 1 6 and 7. This can be compared with the overlapping in the NLM (Fig. 2). In addition there are also three outliers at the extreme left of the dendrogram (Fig. 4). These outliers correspond to the outliers in categories 6 and 7 in the NLM (Fig. 2). Fig. 6 shows a plot of the TREE generated from the autoscaled unweighted data. The plot indicates how the data vectors link together. The numbers identify to which colony the particular objects belong (0 represents colony 10). It must be emphasised that the TREE algorithm was not informed as to which category the individual data vectors belonged. This was also the case for HIER. The investigator however knew the category identities for each data vector.This information was added to the TREE plot after the procedure had been completed and the plot generated. It can be seen that with the exception of a few outliers the data vectors from the same category link (cluster) together very well. The major branch points link different categories. Conclusion These multi-variate studies indicate that there are statistically significant differences between the hydrocarbon profiles of different colonies of the black imported fire ant. The differences appear to be most pronounced between colonies which are adjacent to each other. Earlier studies in which three different colonies of the black imported fire ant were compared yielded similar results.21,22 Workers from neigh-bouring colonies are much more likely to encounter one another while foraging than workers from more distant nests.Therefore if the cuticular hydrocarbons are involved in colony recognition it might be expected that the greatest differences in the hydrocarbon profiles would be observed between neighbouring colonies. The relative efficiency of the different classification methods used in these studies was satisfactory considering that the data set consisted of a relatively large number of closely related categories (i.e. colony profiles from the same species). The best of the procedures used was the LLM method which gave an average efficiency for all categories of approximately 99%. However as the LLM method is based on a pairwise comparison of the categories the results are exaggerated and should be interpreted with some caution.The next best procedure was the MULTI method which had an average efficiency of approximately 91%. The results of MULTI gave a truer measure of the classification ability as this method examines all the categories simultaneously. The KNN method had an average effectiveness of approximately 88%. The SIMCA method had the poorest efficiency of the methods examined with an average effectiveness of approxi-mately 81%. Although these classification procedures are based on different mathematical principles it is remarkable that classification efficiencies of better than 80% were obtained 716 ANALYST JUNE 1986 VOL. 111 The use of pattern recognition procedures in data treatment requires a substantial degree of human involvement because of the iterative nature of multi-variate data analysis.The resolution of a problem should not rely only on a single method. Instead a number of different methods should be applied in order to confirm the interpretation of the results. In these studies we have used six different pattern recognition procedures-two based on unsupervised learning and four based on supervised learning. The results of each of these procedures consistently indicate that there are statistically significant differences in the cuticular hydrocarbon profiles of different colonies of the black imported fire ant. The graphical display methods also show a perceptible clustering of the data vectors by category (colony). Finally this work provides an example of the use of pattern recognition procedures in the analysis of multi-variate data derived from biological samples.Such techniques provide the analytical chemist or biologist with a powerful means of evaluating and interpreting multi-variate data sets. This work was supported by a BRS grant from the University of Alabama. We thank Dr. John Brand Dr. Dave Fletcher and Dr. Earle Cross for their helpful comments and discussion during these studies. 1. 2. 3. 4. 5. References Chapman R. F. “The Insects Structure and Function,” The English Universities Press Ltd. London 1969 p. 435. Nelson D. R. Dillwith J. W. and Blomquist G. L. Insect Biochem. 1981 11 187. Howard R. W. McDaniel C. A. Nelson D. R. Blomquist, G. L. Gelbaum L.T. and Zalkow L. H. J. Chem. Ecol., 1982 8 1227. Howard R. W. and Blomquist G. L. Annu. Rev. Entomol., 1982,27 149. Vander Meer R. K. and Wojcik D. P. Science 1982 218, 806. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. Lok J. B. Cupp E. W. and Blomquist G. L. Insect Biochem. 1975 5 821. Nelson D. R. Fatland C. L. Howard R. W. McDaniel, C. A. and Blomquist G. L. Insect. Biochem. 1980 10 409. Bradshaw J. W. S. and Howse P. E. in Bell W. J. and Carde R. T. Editors “Chemical Ecology of Insects,” Sinauer Associates Inc. Sunderland MA 1984 p. 443. Wilson E. O. “The Insect Societies,’’ Harvard University Press Cambridge MA 1971 p. 272. Kowalski B. R. and Bender C. F. J. Am. Chem. SOC. 1972, 94 5632. Jurs P. C. and Isenhour T. L. “Chemical Applications of Pattern Recognition,” Wiley New York 1975. Duewar D. L. Koskinen J. R. and Kowalski B. R., Documentation for ARTHUR version 1-8-75 Chemometrics Society Report No. 2 Laboratory for Chemometrics Depart-ment of Chemistry University of Washington Seattle WA, 1975. Kowalski B. R. Editor “Chemometrics Theory and Appli-cation,” ACS Symposium Series 52 American Chemical Society Washington DC 1977. Kowalski B. R. and Bender C. F. J. Am. Chem. SOC. 1973, 95 686. Nilsson N. B. “Learning Machines,” McGraw-Hill New York 1965. Harper A. M. Duewar D. L. Kowalski B. R. andFasching, J. L. in Kowalski B. R. Editor “Chemometrics Theory and Application,” ACS Symposium Series 52 American Chemical Society Washington DC 1977 p. 14. Brill J. H. and Bertsch W. Insect Biochem. 1985 15 49. Grob K. and Grob G. J. Chromatogr. 1981 213,211. Mayfield H. T. andBertsch W. Comput. Appl. Lab. 1983, 2 130. Brill J. H. PhD Dissertation University of Alabama 1985. Brill J. H. Mar T. Mayfield H. T. and Bertsch W. J. Chromatogr. 1985 349 39. Brill J. H. Mayfield H. T. Mar T. and Bertsch W. J. Chromatogr. 1985 349 31. Paper A51424 Received November 18th 1985 Accepted December 30th 198

ISSN:0003-2654    

DOI:10.1039/AN9861100711

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

ANALYST, JUNE 1986, VOL. 111 717 SHORT PAPERS Use of Open-circuit Pre-concentration of Sulphide Ion Voltammetry at the Parts per Billion Level of Sample Sambamoorthy Jaya, Talasila Prasada Rao and Gollakota Prabhakara Rao Cen tra I Elec tro c h em ica i Resea rch ins titu te, Ka ra iku di-623006, India in Stripping A procedure based on open-circuit pre-concentration using a silver disc electrode was developed. This procedure allows sulphide to be pre-concentrated at concentrations ranging from 0 to 40 p.p.b., with subsequent stripping through a linear cathodic potential sweep. The procedure is precise and reliable, with a similar sensitivity to and a better selectivity than those of conventional cathodic-stripping voltammetry. The procedure is useful for monitoring parts per billion levels of sulphide in tap water samples with a precision of 2%. Keywords; Open-circuit pre-concentration; sulphide determination; water analysis; trace analysis; stripping voltammetry The determination of wlphide is most often carried out by using either ion-selective electrodesl>2 or by cathodic-stripping voltammetry (CSV).s9 The latter procedure utilises either a hanging or a static mercury drop electrode for anodic pre-concentration of sulphide as its mercury salt.s.8 Shimizu and Osteryoung9 developed a procedure for the determina- tion of sulphip pre-deposited electrolytically on to a silver electrode and employing either a linear sweep or a differential pulse for the cathodic stripping Chemical pre-concentration procedures, viz., ion exchange, solvent extraction, coprecipitation, distillation, etc., are widely used in conjunction with various analytical techniques to enhance the sensitivity.These procedures, however, have limitations such as tediousness, the risk of losing the analyte of interest during the pre-concentration process and the stringent demands on the purity of the reagents used.10 It is far more advantageous to carry out the pre-concentration directly on the system whose analysis is desired. Electrochemical pre-concentration methods have therefore been preferred and widely used over the years for the determinations of various species.lc14 This paper reports the open-circuit pre-concentration of sulphide on to a silver electrode, leading to a silver sulphide- coated surface , without the aid of any external polarisation.This coated surface is subjected to cathodic stripping, based on which the determination of sulphide can be carried out. Experimental Reagents Sodium sulphide solution, 0.5 M . Dissolve 8.406 g of Na2S.5H20 (Riedel, Hannover, FRG) in 1 M sodium hydroxide solution and dilute to 100 ml with 1 M sodium hydroxide solution. Standardise the solution iodimetrically. l5 Sodium hydroxide solution, 2 M . Dissolve 8 g of NaOH (BDH Chemicals, Poole, UK) in conductivity water and dilute to 100 ml. Apparatus A three-electrode glass cell of 100-ml capacity with a centrally placed silver disc working electrode (7 mm diameter), a platinum foil counter electrode and a standard calomel reference electrode (SCE) was used. For stirring, a REMI magnetic stirrer was used.The working electrode surface was polished using emery papers of increasing fineness starting with grade 1/0 to 4/0. A Wenking Model LB 75 M potentiostat was used in conjunction with a Wenking Model VSG 72 scan generator for potential control and sweep, and the i - E curves were recorded on a Digilog XY-2000 recorder. High-purity nitrogen was used to de-aerate the solutions. Experiments were carried out at room temperature (30 k 1 "C). Procedure Pre-condition the silver electrode at a constant potential of -1.2 V as described elsewhere9 to obtain a constant current ( d o PA). Set the potential to the start potential, viz., -0.6 V, of the cathodic sweep. Place the selector knob of the potentiostat in the open-circuit mode (Er position). Stir the sample solution, containing 0-40 p.p.b. of sulphide in 50 ml of 0.2 M sodium hydroxide solution, for 10 min. Stop the stirring and turn the potentiostat control knob to the current measurement mode ( I mode) without a rest period, then record the cathodic-stripping voltammogram and measure the peak currents or the charge passed. Establish the concentra- tion of sulphide by reference to a calibration graph prepared for 0.5-40 p.p.b. of sulphide. Alternatively, any potentiostat can be used by switching off the potentiostat during the pre-concentration period. Results and Discussion Cyclic Voltammetric Studies A typical cyclic voltammogram in 0.2 M sodium hydroxide solution at the silver disc working electrode in a quiescent solution is shown in Fig. 1, A.Its characteristic features include two peaks that occur in the anodic scan at potentials of -0.97 and -0.8 V and another peak at -1.08 V in the cathodic scan. These peaks are due to monolayer formation of AgOH by adsorption and subsequent reduction of the surface-adsorbed AgOH to silver.16>17 On addition of lob6 M of sulphide to 0.2 M sodium hydroxide solution, a new peak appears at -0.7 V (Fig. 1, B) that is due to the oxidation of Ag(0) to silver sulphide. However, the corresponding reduction peak is not discernible during the cathodic scan of the cyclic voltammogram. Cathodic-stripping Voltammetric Studies Fig. 1, C and D, show the cathodic stripping voltammograms obtained after pre-concentration either in the open-circuit mode ( i . e . , by keeping the potentiostat in the E, position where the counter electrode remains disconnected) or with the potentiostat switched off, for times of 2 and 5 min, respectively.The open-circuit potential during the718 ANALYST, JUNE 1986, VOL. 111 100 0 a - 2- -100 -200 - 300 D 1 I I -1.2 -0.8 -0.4 E N Fig. 1. Cyclic (A and B) and cathodic-stripping voltammograms (C and D) obtained usin the silver disc electrode. A, Cyclic voltammo- gram obtained in 0.f M NaOH solution; B, cyclic voltammogram obtained in 0.2 M NaOH + 10-6 M Na2S solution; and C and D, cathodic stripping voltammograms obtained in 0.2 M NaOH + 10-6 M Na,S solution with pre-concentration for 2 and 5 min, respectively pre-concentration was found to be -0.71 V, which corresponds to oxidation in the presence of sulphide.When the cathodic-sweep voltammogram was drawn after the above pre-concentration, a cathodic reduction peak was obtained with a peak potential at -0.94 V, as can be seen from Fig. 1, C and D. Under otherwise identical conditions, when pre-concentration is carried out potentiostatically at -0.4 V for 2 and 5 min, identical cathodic-stripping peaks are obtained. This indicates that both electrochemical pre-concentration and in situ chemical modification results in the formation of the same species. Effect of Start Potential of the Cathodic Scan In order to study the effect of the start potential during the cathodic scan, the initial potential was adjusted to different values from -0.4 to -0.9 V vs. SCE in steps of 0.1 V. During these studies, sulphide was pre-concentrated from 30 p.p.b. solutions on to the silver electrode by in situ chemical modification as described earlier. After a known pre- concentration period of 2 min, the linear-sweep cathodic- stripping voltammogram was recorded from the selected start potential. As can be seen from Fig. 2, A, the cathodic- stripping peak is not affected up to a start potential of -0.7 V. Hence an optimum start potential of -0.6 V was utilised in the subsequent experiments. Effect of Sweep Rate The effect of varying the sweep rate from 6 to 400 mV s-1 in the potential range -0.6 to -1.2 V on the cathodic-stripping peak of silver sulphide is shown in Fig. 2, B. The cathodic- stripping voltammetric signal reaches saturation at sweep rates above 180 mV s-1. Hence a sweep rate of 220 mV s-1 was chosen for subsequent experiments.The stripping signal was found to be reproducible in repetitive determinations. Calibration Graph and Precision The calibration graph obtained by the recommended proce- dure was linear over the concentration range 0-40 p.p.b. of sulphide with a detection limit of 0.5 p.p.b. in a total volume of VImV s - 1 0 200 400 600 400 1 I I A -0.3 Fig. 2. Effect of start potential (A) and sweep rate (B) on the cathodic-stripping voltammetric signal of 30 p.p.b. of sulphide using the silver disc electrode; 0.2 M sodium hydroxide solution; deposition time, 2 min Table 1. Analysis of tap water with 10 min of open-circuit pre-concentration Volume Sample taken/ No. ml 1 45 2 45 3 45 4 40 5 40 Amount of sulphide added, Amount of sulphide Recovery, 0 0 100 1.0 1.00,0.98,1.0, 1.02 100 2.0 1.96,1.98,1.98,2.00 99 10.0 9.90,9.85,9.90,9.95 99 20.0 20.0,20.5,20.0,19.5 100 p.p.b. recovered, p.p.b. Yo 50 ml with an open-circuit pre-concentration period of 10 min. The coefficient of variation [CV = (standard deviation x 100)/mean] for the determination of 15 p.p.b. of sulphide was found to be 1.6% for five replicate determinations. Interference Studies The interference of various anions in the determination of sulphide was not investigated by Shimizu and Osteryoung.9 We therefore carried out interference studies with the addition of 3000 pg dm-3 of nitrate, nitrite, perchlorate, chloride, bromide, iodide, thiocyanate, sulphate, sulphite and carbonate during the determination of 30 pg dm-3 of sulphide by both electrolytic and open-circuit pre-concentration and subsequent cathodic-stripping voltammetry.Thiocyanate, iodide, nitrite and bromide interfered in the determination when electrolytic pre-concentration was utilised. On the other hand, only nitrite interfered in the determination of sulphide by utilising open-circuit pre-concentration. Further, mixtures of the above anions did not affect the determination of sulphide. Analysis of Tap Water Samples The determination of sulphide in tap waters was carried out by the recommended procedure after the addition of 2 g of EDTA (to complex heavy metals if present) and 0.8 g of ascorbic acid to 50 ml of the solution. The recoveries obtained on addition of known amounts of sulphide to tap water are also given in Table 1.The results indicate the utility of the developed procedure for pollution monitoring of sulphide in natural waters. Conclusion The introduction of in situ chemical modification in place of electrolytic pre-concentration results in a precise, reliable andANALYST, JUNE 1986, VOL. 111 719 selective procedure for the determination of sulphide with comparable sensitivity and detection limits. The authors are grateful to Prof. K. I. Vasu, Director, CECRI, Karaikudi, for his keen interest in this work and kind permission to publish these results. 1. 2. 3. 4. 5. 6. References “Instruction Manual, Sulphide Ion Electrode, Silver Ion Electrode, Model 94-16,” Orion Research, Cambridge, MA, 1970. Glaister, M. G., Moody, G. J., and Thomas, J. D. R., Analyst, 1985, 110, 113. Berge, H., and Geroschewski, P., Fresenius Z . Anal. Chem., 1965,207, 110. Miwa, T., Fujii, Y., and Mizuike, A., Anal. Chim. Acta, 1972, 60, 475. Youssefi, M., and Birke, R. L., Anal. Chem., 1977, 49, 1380. Florence, T. M., J . Electroanal. Chem., 1979, 97, 219. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. Florence, T. M., Anal. Lett., 1978, 11, 913. Florence, T. M., J. Electroanal. Chem., 1979,97, 237. Shimizu, K., and Osteryoung, R. A., Anal. Chem., 1981, 53, 584. Vydra, F., Stulik, K., and Julakova, E., “Electrochemical Stripping Analysis,” Wiley, New York, 1976. Copeland, T. R., and Skogerboe, R. K., Anal. Chem., 1974, 46,1275A. Jagner, D., Analyst, 1982, 107, 593. Florence, T. M., J . Electroanal. Chem., 1979, 97,219. Christie, J. H., and Osteryoung, R. A., Anal. Chem., 1976,48, 869. Brainina, K. Z., Talanta, 1971, 18, 513. Jaya, S . , Prasada Rao, T., and Prabhakara Rao, G., in preparation. Burstein, G. T., and Newman, R. C., Electrochim. Acta, 1980, 25, 1009. Paper A51341 Received September 23rd, 1985 Accepted January 17th, 1986

ISSN:0003-2654    

DOI:10.1039/AN9861100717

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

ANALYST, JUNE 1986, VOL. 111 72 1 Direct Determination of Nickel in Human Plasma by Zeeman-corrected Atomic Absorption Spectrometry Jan Rud Andersen," Bente Gammelgaard and Susanne Reimert Royal Danish School of Pharmacy, Department of Chemistry AD, 2 Universitetsparken, DK-2 100 Copen hagen, Denmark A procedure is described for the direct determination of nickel in human plasma in which the samples are diluted before quantitation by means of Zeeman-corrected atomic absorption spectrometry. The limit of detection is 0.09 pg 1-1 and the limit of quantitation 0.31 pg I-'. For healthy, unexposed subjects, the mean value of nickel in plasma was found to be 0.65 k 0.35 pg 1-1 ( n = 6) (range 1.28 k 0.02 to 0.32 k 0.05 pg 1-1). The method fails to give the recommended value of 1.8 k 0.6 pg I-' for nickel in the NBS reference material RM 8419, bovine serum. Instead the concentration was found to be 0.48 2 0.04 pg 1-1 ( n = 3).It is suggested that the recommended value may be in error. Keywords: Nickel determination; human plasma; Zeeman graphite furnace; atomic absorption spectrometry; NBS reference material 84 19, bovine serum Nickel is used for a wide variety of purposes in modern industry and, because it is a constituent of stainless steel, virtually every individual in the industrialised part of the world is in daily contact with the metal. Nickel is an allergenic and in some of its compounds a carcinogenic element,' hence the need for reliable analytical procedures for monitoring nickel in human tissues and body fluids has been emphasised during recent years.2 The concentration of nickel in plasma is often used as a measure of the body burden of the element, but the quantitation of nickel in human plasma has proved to be difficult, and no direct methods giving acceptable results have been disclosed so far.Until recently the majority of reported results were obtained either by the IUPAC recommended procedure relying on solvent extraction and electrothermal atomic absorption spectrometry,2 or by the very sensitive electrochemical method known as adsorption ~oltammetry.~ Both methods require exhaustive mineralisation of the samples prior to analysis, which make them time consuming and prone to contamination. In 1984 a more direct method was introduced by Sunderman et aL4 that only requires nitric acid deproteinisation of serum or whole blood before quantitation by means of Zeeman- corrected atomic absorption spectrometry.The values obtained by this method suggest that a substantial lowering of the reference value for nickel concentrations in unexposed individuals is needed. Earlier methods indicated that the value for nickel in human serum was approximately 2 pg l-1,5 whereas a mean of 0.46 -t 0.26 1.18 1-1 (n = 39) has now been found. In this paper an even more direct method is reported, in which only the simple dilution of human plasma is needed before quantitation by Zeeman-corrected atomic absorption spectrometry. Using this method results comparable to those reported by Sunderman et al.4 were achieved. However, the method fails to give the recommended concentration of nickel in the NBS reference material RM 8419, bovine serum.6 was used. The atomisation signals were displayed on a Perkin-Elmer R 100A recorder, and their peak heights printed out on a Perkin-Elmer PRS 10 printer.The hollow-cathode lamp current was 25 mA, and the 232.0-nm line with a spectral band pass of 0.2 nm was used. The graphite furnace temperature programme is given in Table 1. The internal argon gas flow was stopped and the Zeeman correction was on during atomisation. Pyrolytically coated graphite tubes were used. As temperatures may vary slightly between instruments for a given setting, they should be regarded as approximate values only. Reagents Nitric acid was of Suprapur quality and Triton X-100 was of scintillation grade, both purchased from Merck, Darrn- stadt, FRG.A certified 1 g 1-1 nickel reference solution (Titrisol, Merck, Darmstadt, FRG) was used, and aliquots of this were appropriately diluted to yield working standards. Only Milli-Q water was used, and all laboratory ware (sample containers, sample cups and pipette tips) was rinsed in 4 M nitric acid before use. Procedure Plasma samples were stored at -20 "C prior to analysis. The NBS bovine serum was also distributed and stored in a frozen condition. After thawing, the samples were diluted 1 + 1 with a solution 10-3 M in nitric acid and 0.1% in Triton X-100. Injections of 50 pl were used, and the concentrations evaluated either by comparison with a calibration graph obtained on human plasma or, alternatively, by the method of standard additions.The two approaches give identical results. Experimental Instrumentation A Perkin-Elmer Zeeman 5000 atomic absorption spec- trometer, equipped with a Perkin-Elmer AS-40 autosampler, * To whom correspondence should be addressed. Table 1. Graphite furnace programme Step TemperaturePC Ramp/s Dry I 100 40 Char 1250 30 Atomise . . . . . . . . 2700 O* Clean 2800 1 Cool 20 2 . . . . . . . . . . Dry I1 . . . . . . . . 200 60 . . . . . . . . . . . . . . . . . . . . . . . . . . . . * Maximum power heating. Hold/s 20 10 20 6 2 5722 ANALYST, JUNE 1986, VOL. 111 Results and Discussion The proposed method has several advantages; it is fast, precise, not prone to contamination because of its simplicity and has an adequate sensitivity. The limit of detection, defined as the blank value plus three times the standard deviation of the blank value, is 0.09 pg 1-1; and the limit of quantitation, defined as the blank value plus ten times the standard deviation of the blank value, is 0.31 pg 1-1.7 These values were calculated from 20 determinations of the blank, which was Milli-Q water treated as described above for the plasma.The results obtained by analysing plasma from six healthy volunteers are shown in Table 2. The mean value of 0.65 k 0.35 pg 1-l is in good agreement with the results reported by Sunderman et al.4 However, our attempts to validate the accuracy of the method by analysing the NBS reference material RM 8419, bovine serum,6 have so far been in vain. It has a recommended nickel concentration of 1.8 k 0.6 pg 1-l, but we found a concentration of 0.48 k 0.04 pg 1-1 (n = 3).The reason for this discrepancy is not clear. Obviously, a systematic error on our part could be the explanation but we have not been able to identify any errors, and we do not believe any are present for the following reasons. Firstly, as we are analysing a reference material, sampling errors are absent. Secondly, there is no sample pre-treatment, hence no errors are introduced prior to quantitation. Thirdly, the characteristic mass obtained by the method was found to be 13 pg per 0.0044 A s, in exact agreement with the value found by Slavin and Carnrick.8 This ensures that matrix interference is minimal, as is also indicated by the discovery that quantitation by a calibration graph or by standard additions gives identical results.In fact, an aqueous calibration graph can be used. It further guarantees that nickel is not lost during the charring Table 2. Nickel in plasma from unexposed, healthy individuals [Ni]/pg 1-1 (mean k s.d.; n = 3) 1 0.59 2 0.04 2 0.42 2 0.04 3 0.32 k 0.05 4 1.28 2 0.02 5 0.82 k 0.03 6 0.47 2 0.02 Mean . . . . 0.65 20.35 Sample step. Finally, nitric acid deproteinisation of the plasma samples prior to analysis as described by Sunderman et aL4 give results identical with those obtained by the direct method described here. The alternative solution to the accuracy problem is that the recommended value of the reference material is in error. The values on this pool of bovine serum are recommended values and are as such “subject to scrutiny and revision.”6 For the reasons given in this paper we feel that the value for nickel may be in error, but clearly the experience of others with respect to the nickel content of the material is needed. In the meantime we feel confident in the proposed method.In conclusion, we have presented a method for the determination of nickel in human plasma that by most standards is ideal, as it is fast, simple, precise, sensitive and not prone to contamination, etc. On the other hand, it fails to give the expected result for a reference material. The reason for this is not clear and warrants further work. Note Added in Proof After the acceptance of this manuscript we analysed the reference material Seronorm Trace Elements (Nyegaard Diagnostics, Oslo, Norway). It is a freeze-dried serum and it has a preliminary information value for nickel of 3.1 pg 1-1; we found 2.93 k 0.34 pg 1-1 (n = 9). References 1. 2. 3 . 4. 5. 6. 7. 8. Nriagu, J. O., Editor, “Nickel in the Environment,” Wiley, New York, 1980. Brown, S. S., Nomoto, S . , Stoeppler, M., and Sunderman, F. W., Pure Appl. Chem., 1981, 53, 773. Pihlar, B., Valenta, P., and Niirnberg, H. W., Fresenius 2. Anal. Chem., 1981, 307, 337. Sunderman, F. W., Chrisostomo, M. C., Reid, M. C., Hopfer, S. M., and Nomoto, S., Ann. Clin. Lab. Sci., 1984, 14, 232. Versieck, J., and Cornelis, R., Anal. Chim. Acta, 1980, 116, 217. Veillon, C., Lewis, S. A., Patterson, K. Y., Wolf, W. R., Harnly, J. M., Versieck, J., Vanballenberghe, L., Cornelis, R., and O’Haver, T. C., Anal. Chem., 1985, 57, 2106. American Chemical Society Committee on Environmental Improvement, Anal. Chem., 1980, 52, 2242. Slavin, W., and Carnrick, G. R., Spectrochim. Acta, Part B, 1984, 39, 271. Paper A51 45 7 Received December 20th, I985 Accepted January 20th, 1986

ISSN:0003-2654    

DOI:10.1039/AN9861100721

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

ANALYST, JUNE 1986, VOL. 111 723 Study of the Nitroprusside - Sulphydryl Test for Aromatic Thiols James P. Danehy Department of Chemistry, University of Notre Dame, Notre Dame, IN 46556, USA The Gmelin test for sulphide ions and organic thiolate anions, in which sodium nitroprusside [disodium nitrosopentacyanoferrate(l1) dihydrate, Na2FeNO(CN)5.2H20] gives a reddish purple colour in ammonia solution, has been found to be positive for non-aromatic thiols but negative with aromatic thiols. Keywords: Nitroprusside; thiols; mercaptans Since its introduction more than a century ago the nitroprus- side test (Gmelin test) has been widely used for demonstrating the presence of the sulphydryl group in aqueous solutions. No exception has yet been reported to the generalisation that an intense reddish purple colour appears instantaneously on the addition of one drop of sodium nitroprusside [disodium nitrosopentacyanoferrate(I1) dihydrate] solution to an ammo- niacal solution of a thiol (10-4-10-5 M RSH).It was a great surprise, therefore, when it was recently observed that a routine application of the test to a good sample of 4-mercapto- benzoic acid gave a completely negative test. A quick check with cysteine and mercaptoacetic acid exonerated the reagent but further investigation, prompted by this puzzling obser- vation, has failed to reveal a single aromatic thiol that gives a positive test. Specifically, each of the following compounds gives a negative test over a wide range of concentrations: thiophenol; o-thiocresol; rn-thiocresol; p-thiocresol; 2-mer- captobenzoic acid; 3-mercaptobenzoic acid; 4-mercapto- benzoic acid; 2-mercaptophenol; 4-mercaptophenol; 4- chlorothiophenol; 2-aminothiophenol; 4-aminothiophenol; 1-thionaphthol; 2-thionaphthol; 2-mercaptobenzothiazole; and 4-mercaptobenzenesulphonic acid.The thiolate anions corresponding to the three nitrothiophenols have deep red colours. The addition of nitroprusside to an appropriately diluted solution does not produce any observable change in the hue or intensity. No effort has been made in this investigation to survey aliphatic thiols as the literature records that an indefinitely large number of them react positively with nitroprusside. The author does not know of any aliphatic thiol that does not react in this way. On surveying the pertinent literature from 1907 through to mid-July, 1985, it can be seen that there has been a sustained interest in nitroprusside, particularly in modifying the compo- sitions to suppress the sensitivity to carbonyl compounds that can also give rise to colour.Vachekl reported the spectro- photometric determination of five substituted mercapto- purines, thiourea, five substituted thioureas and five substi- tuted mercaptopyrimidines in a medium of acetic acid - methanol - water (4 + 1 + 1) after reaction with sodium nitroprusside in 0.1 M NaOH. Recently,ZJ thiols have been detected by the nitroprusside test on anion-exchange resin beads and specific cation effects4 have been studied. However, no-one has yet reported the anomaly of which this paper is the subject: that the Gmelin test is positive with non-aromatic thiols and negative with aromatic thiols.There is no concrete evidence on which to base a rationali- sation for this dichotomous behaviour. It can scarcely be due to bulk or steric factors, as tert-butyl mercaptan, penicillamine and 2-mercapto-2-methylpropanoic acid react positively. Nor can it be due to the pK, value because at the pH value established using an excess of ammonia solution, all of the thiols are predominantly in the thiolate anion form. pK, values for aromatic thiols extend over a relatively narrow range from 4.5 to 8.2, and are thus encompassed by the aliphatic thiols, whose pK, values range from 3.5 to 11.4.5 References 1. 2. 3. 4. 5. Vachek, J., Pharmazie, 1960, 15, 707. Grdinic, V., and Jaksevac-Miksa, M., Acta Pharm. Jugosl., 1982, 32, 197. Saidul, Z . Q., Izzatullah, and Ahmal, N., Microchem. J., 1983, 28, 33. Leeuwenkamp, 0. R., Vermaat, C. H., Plug, C. M., andBult, A. A., Pharm. Weekbl., Sci. Ed., 1984,6, 195. Danehy, J. P., and Parameswaran, K. N., J . Chem. Eng. Data, 1968, 13, 386. Paper A51383 Received October 28th, 1985 Accepted January 22nd, 1986

ISSN:0003-2654    

DOI:10.1039/AN9861100723

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

ANALYST, JUNE 1986, VOL. 111 COMMUNICATION 725 Material for publication as a communication must be on an urgent matter and be of obvious scientific importance. Rapidity of publication is enhanced if diagrams are omitted, but tables and formulae can be included. Communications should not be simple claims for priority: this facility for rapid publication is intended for brief descriptions of work that has progressed to a stage at which it is likely to be valuable to workers faced with similar problems. A fuller paper may be offered subsequently, if justified by later work. Manuscripts are usually examined by one referee and inclusion of a Communication is at the Editor's discretion Extent of the Heating of Microwave Cavities at Large Modulation Amplitudes and Its Effect on Quantitative Analysis by ESR Spectrometry D.Thorburn Burns, Barry G. Dalgarno and Brian D. Flockhart" Department of Pure and Applied Chemistry, The Queen's University of Belfast, Belfast BT9 5AG, Northern Ireland Keywords: Magnetic field modulation; microwave cavities; quantitative ESR measurements; transition metal ions One of the parameters that markedly affects the signal to noise ratio of an ESR spectrometer is the amplitude of the magnetic field modulation (H,). If H, is much less than the line width, the output signal is approximately proportional to the slope of the absorption graph, and the peak to peak amplitude (A) of the derivative graph increases linearly with an increase in H,. As H, approaches and then exceeds the line width, this linear relationship ceases, the graph reaches a maximum, and then subsequently decreases slowly.Bryson et al. 1-3 have examined the effect of H, on A of the first derivative ESR signal from various transition metal ions in aqueous solution. The variation of A with H, for Mn2+, VO2+, Cu2f and Fe3+ was found to be linear up to a value of H , of ca. 10 G; for Cr3+ the value was ca. 8 G. At higher values of H,, all the A vs. H, graphs showed a marked deviation from linearity; in the papers2.3 devoted to Fe3+ and Cr3+, this deviation was attributed to the effect of the thermal instability of the Varian E-231 cavity. Heating of a resonant cavity is known4 to occur at modulation amplitudes larger than ca. 10 G. What has not been previously realised is the extent of this heating, and therefore the experimental procedure that is required to deal with this effect in any quantitative ESR study of aqueous solutions of transition metal ions.Experimental ESR measurements were made with a Varian E-109 spec- trometer operated at ca. 9.5 GHz with a magnetic field modulation of 100 kHz. The flow system used has been described elsewhere .5 Results and Discussion The variation in cavity temperature with H , for the Varian E-231 cavity is shown in Fig. 1; with H,,, = 40 G, the (empty) * To whom correspondence should be addressed. cavity reached a temperature of 61 "C (ambient temperature, 25 "C). Under similar conditions, the Varian E-231/E-232 dual sample assembly attained a temperature of 36 "C. When H , was increased from zero to 40 G, an aqueous solution 60 50 Y 40 H,IG 16 12 v) c 3 c .- 4 Fig.1. A, Cavity temperature as a function of modulation ampli- tude, H,. Peak to peak amplitude, A , of the first derivative curve as a function of modulation amplitude for a lo-* M solution of chromium(II1) in water. B , Solution static in the sample cell. C , Solution circulating through the cell.726 ANALYST, JUNE 1986, VOL. 111 contained in a Varian E-248 aqueous solution sample cell (“flat” cell) and mounted in the E-231 cavity reached a maximum temperature of 62.5 “C in 14 min. When the experiment was repeated with the solution contained in a cylindrical tube of 5 mm o.d., a temperature of 35 “C was recorded. The plot of A against H,,, for a 10-2 M solution of chromium(II1) in water contained in a flat cell is shown in Fig.1. This graph, with its marked curvature beginning at ca. 10 G, is closely similar to that published by Bryson et aZ.3 Fig. 1 also displays the plot of A against H,,, for a chromium(II1) solution of the same concentration that was circulated through the flat cell at a rate of 50 cm3 min-l by means of a peristaltic pump. This graph is strictly linear, as the temperature of the flowing solution was constant (21 “C in this experiment). Clearly, for transition metal ions such as Cr3+, the deviation from linearity in the A vs. H,,, graph as the value of H,,, is increased results from cavity heating of the sample solution which is static in the flat cell. Modulation broadening of the spectrum for Cr3+ is negligible at H,,, = 40 G. Conclusion Contrary to previous recommendations ,I-3 quantitative ESR measurements on aqueous solutions of transition metal ions require neither H,,, to be kept to values <10 G nor a delay of 5-15 min after adjusting H,,,.These restrictions are circum- vented by the use of a simple flow system whereby a steady temperature can be maintained in the ESR cell. Modulation amplitudes up to 40 G (the highest amplitude obtainable with commerically available spectrometers) can therefore be used, if required, to optimise the signal to noise ratio of the spectrometer. 1. 2. 3. 4. 5 . References Bryson, W. G., Hubbard, D. P., Peake, B. M., and Simpson, J., Anal. Chim. Acta, 1975, 77, 107. Bryson, W. G., Hubbard, D. P., Peake, B. M., and Simpson, J., Anal. Chim. Acta, 1978, 96, 99. Bryson, W. G., Hubbard, D. P., Peake, B. M., and Simpson, J., Anal. Chim. Acta, 1980, 116, 353. Goldberg, I. B., and Bard, A. J., in Elving, P. J., Bursey, M. M., and Kolthoff, I. M., Editors, “Treatise on Analytical Chemistry,” 2nd Edition, Part I, Volume 10, Wiley, New York, 1983, p. 253. Burns, D. T., Dalgarno, B. G., and Flockhart, B. D., Anal. Proc., 1985, 22, 379. Paper A61102 Received April lst, 1986

ISSN:0003-2654    

DOI:10.1039/AN9861100725

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

727 ANALYST, JUNE 1986, VOL. 111 BOOK REVIEWS Analysis of Hazardous Substances in Biological Materials. Volume 1 Edited by J. Angerer and K. H. Schaller. DFG German Science Foundation. Pp. x + 222. VCH Verlagsgesell- schaft. 1985. Price DM90. ISBN 3 527 26095 1 (VCH Verlagsgesellschaft); 0 89573 075 8 (VCH Publishers). This is a compilation of methods produced by the West German Working Group “Analytical Chemistry” of the Commission for the Investigation of Health Hazards of Chemical Compounds in the Work Area. Each method is developed by one member of the group and then thoroughly tested by another member. The publication in English of this wealth of experience is to be welcomed. An introductory section in the book deals with common problems such as collection and timing of specimens and evaluation of the reliability of methods through the criteria of sensitivity, imprecision, inaccuracy, detection limit and sensi- tivity and through the use of quality control.This is followed by a description of the methods. Aromatic amines and benzene derivatives in urine are determined by high- performance liquid chromatographic methods; chloroben- zenes , chlorophenols and hippuric acids in urine and carboxy- haemoglobin in blood by gas - liquid chromatography; beryllium, cobalt and nickel in urine and cadmium in blood by electrothermal atomic absorption spectrometry; lead in blood by solvent extraction - flame atomic absorption spectrometry; bromide in urine using an ion-selective electrode; thallium in urine by inverse voltammetry; and phenol and trichloroacetic acid by photometric procedures.For each method a full list of equipment and reagents is given together with details of specimen collection and sample treatment. Performance with respect to precision, accuracy and detection limit is thoroughly assessed and sources of errors are discussed. The text is illustrated with examples of typical chromatograms and calibration graphs. A useful section for each analyte is a brief chemical description and an assessment of its health hazard with guidelines for reference values and action limits. The objective of standardised methods is not one that is likely to find universal acceptance. It is this reviewer’s opinion that the development of varied approaches to analytical determination should be encouraged and the quality of performance of laboratories assessed through participation in quality control schemes or through analysis of reference materials.However, in not all instances is this yet possible and it is useful to have access to methods that have been thoroughly evaluated, which can be used when development time is limited or can be used for comparison purposes. This book is therefore recommended reading for all involved in the analysis of specimens for monitoring occupational exposure. D. J . Halls Modern Practice of Gas Chromatography. Second Ed it ion Edited by Robert L. Grob. Wiley-lnterscience. 1985. Price f75.20. ISBN 0 471 87157 5. This book has the same general layout as the First Edition. It has the immediate attraction of a more readable typeface, although some of the figures are still of poor quality.Several sections of the book have been substantially expanded. In Part 1 (Theory and Basics) there are separate chapters on packed columns and capillary columns and a new chapter on the optimisation of separations. Qualitative and quantitative analyses now, more logically, follow chapters on detectors and other instrumentation in Part 2. Part 3 of the book, covering some applications of gas chromatography, now includes the environmental, petrochemical and polymer fields, in addition to food, physico-chemical measurements, drugs and clinical applications. There are two appendices, the first of which contains examples relating attenuation and chart speed changes to peak areas and would be better summarised in Chapter 8.The second appendix is a useful list of abbrevia- tions and symbols, although some are pointless or so obscure that they are of no practical value. Each chapter includes a list of references, to which additions have been made since the First Edition. However, it is disappointing to find that in some of the new chapters references later than 1980 are not very common. The book covers a wide range of topics relating to gas chromatography, with little overlap between the contributions of the various authors. There are, however, some surprising omissions. For example, in the chapter on detectors, there is only a brief discussion of the thermal energy analyser with one reference, 10 years old. There have been many useful papers since that time and chemiluminscent detection has been applied to several types of compound in addition to nitro- samines.Modified flame-ionisation detectors are also men- tioned, but again there is only superficial coverage, with an inadequate number of references. The chapter on drugs concentrates on pharmaceutical compounds, although gas chromatography plays a very important role in the examin- ation of illicit drugs. Despite the title, there are several instances where space is devoted to outdated techniques and equipment, for example, in the chapters discussing area measurements and detectors. The book contains a wealth of information, but in the reviewer’s opinion is unbalanced. It contains too much detail on some aspects of chromatography and minimal coverage on equally important areas.Terry Gough Physico-chemical Behaviour of Atmospheric Pollutants. Proceedings of the Third European Symposium, Varese, Italy, 10-12 April, 1984 Edited by 6. Versino and G. Angeletti. Pp. xiv + 666. Riedel. 1985. Price f59.75; $84. ISBN 90 277 1873 3. This volume is the Proceedings of a Symposium organised by the Commission of the European Communities in order to review progress and results from the third R & D Programme on the Environment. It is therefore an up-to-date review and includes topical problems such as acid deposition. A total of 68 papers covering different areas related to the physical chemistry of air pollutants are included within the sub- divisions of identification and analysis of air pollutants, chemical and photochemical reactions, aerosol characteris- ation, pollutant cycles and transport and modelling.The section on the identification and analysis of pollutants gives particular attention to problems of chemical artefacts, which may be minimised by the use of diffusion tubes for sampling. Further sampling problems with atmospheric aero- sols are also identified, as is the need for standardisation of collection devices as a basis for intercomparison work. A wide range of analytical problems are also included in the section on aerosols, with several studies being concerned with the difficult problem of measurements “on the ground.” Analytical work contributes significantly to studies in other sections of the book and any chemist involved with air728 ANALYST, JUNE 1986, VOL. 111 pollutant studies is bound to find papers of interest.The volume contributes substantially to the understanding of air quality problems and identifies many opportunities for further research programmes in this extensive field of study. R. S. Barratt Structure Determination by X-Ray Crystallography. Second Edition M. F. C. Ladd and R. A. Palmer. Pp. xxiv + 502. Plenum Press. 1985. Price $39.50. ISBN 0 306 41878 9. The fact that a second edition of this rather specialised book has appeared after 8 years is undoubtedly an unsolicited testimonial to the value of the original version and to its standing among practising and neophyte crystallographers. It is safe to predict that this new, extended version will enjoy equal success, because of its didactic qualities and the fact that the sections covering methods in X-ray structure analysis have been expanded appreciably to cover recent advances in technique and computational methods.Further, by current standards, it is competitively priced. The first third of the book provides the necessary back- ground information, covering the various aspects of crystal geometry and the preliminary examination of crystals by optical and X-ray methods. Such material is available in a number of other publications but the merit of the present text lies in the clarity of the presentation. A linking chapter on the intensity of scattering of X-rays by crystals leads on to the four major chapters covering X-ray structure analysis. The induc- tive approach employed is very effective, both for its lucidity and the skilful use of examples, beginning with sodium chloride and culminating in a chapter covering in depth the large molecules 2-bromobenzo[b]indeno[ 1,2-e]pyran and potassium 2-hydroxy-3,4-dioxocyclobut-l-en-l-olate mono- hydrate. Those who conscientiously study these chapters will be well placed to begin work in this field. As with the First Edition, much material is consigned to a series of appendices filling 55 pages. Some readers will doubtless find this layout inconvenient. It is particularly surprising that experimental methods as a whole are so treated; arguably, they could form a chapter in the main text. However, this is a minor criticism of an excellent book, one that will be widely used. W. F. Maddams

ISSN:0003-2654    

DOI:10.1039/AN9861100727

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

728 ERRATA ANALYST, JUNE 1986, VOL. 111 Simultaneous Multi-element Analysis by Continuum Source Atomic-absorption Spectrometry with Graphite Probe Electrothermal Atomisation John Carroll, Nancy J. Miller-lhli, James M. Harnly, David Littlejohn, John M. Ottaway and Thomas C. O’Haver Analyst, 1985, 110, 1 153-1 159 Page 1156, Table 4, “Probe” column: the entry “29” against “V” should read “39.” Page 1158, left-hand column, line 19: “volatile” should read “involatile.” Determination of Hydroquinone in Skin-toning Creams Using Hig h-performance Liquid Chromatography Jane Firth and Ian Rix Analyst, 1986, 1 I1 , 129-1 37 Page 131, left-hand column, lines 27 and 28: ‘‘a” should read “fl 0.” A New Way of Organising Spectral Line Intensity Ratio Fluctuations - Bo Thelin Analyst, 1986, 11 1 , 41 9-422 Page 420, equation (5) and the following paragraph should read: R(I&,/&Jn’) = D(E)R( T)/kT + R(c&n/Pmfnf) ( 5 ) The choice of terms on the right-hand side of this equation maximises the R value of the intensity ratio.For this we have assumed that the R value of the C factor ratio, R(Ckn/P,fnr), is independent of R( T), the maximum relative temperature fluctuation. Using equation (5) it is possible to obtain a straight line with a direction coefficient of (1/kT)R(T) when plotting R(Ihn/&rnf) versus D(E) for all studied line pairs of the elements a and b because equation (5) is a linear differential equation. In this equation the R value of the C factor ratios is assumed to be constant. This assumption does not mean that the relative fluctuations of the factors in C are necessarily small. Large fluctuations arising from “bulk” effects such as changes in the instrument parameters or the density of the sample constituents will not significantly affect the relative deviation of the ratio of the C factors between two simultaneously measured spectral lines. Page 421, right-hand column, lines 21-23 should read: However a linear graph is always obtained with D(E) = Ihv,, - hv,,,~l but not with D(E) = IE, - Em,]. Page 421, caption to Fig. 2, the equation on the second line should read:

ISSN:0003-2654    

DOI:10.1039/AN9861100728

出版商:RSC    

年代:1986

数据来源: RSC