4 RESEARCH AND DEVELOPMEWT TOPICS Proc. AnaZyt. Div. Chewz. SOC. Research and Development Topics in Analytical Chemistry The following are summaries of twelve of the papers presented at the Research and Develop- ment Topics in Analytical Chemistry meeting of the Analytical Division held on June 28th and 29th, 1978, at the University of Wales Institute of Science and Technology, Cardiff. Studies in Flow Injection Analysis B.Fields Department of Chernistvy, Univevsity College of Swawsea, Si9agleton Park, Swansea, SA2 8PP Unsegmented continuous-flow analysis or flow injection analysis (FIA) has been developed during the last 4 years as an automatic method of analysis that is simple, accurate and rapid, typical sampling rates being 120 per h o ~ r . l - ~ In all such systems, a stream of reagent or other carrier flows through a small-bore tube, the flow-rate being maintained by a constant-pressure or a constant-volume pump. At a point along the length of the tube an injection mechanism allows the sample to be injected into the stream and as the sample bolus passes down the tube it may react with reagent in the stream or undergo other reactions with the stream solution.Situated downstream from the injection point is a sensor that measures the extent of these reactions. As the sample is injected directly into the carrier stream it follows that there is an inter- facial region between the sample plug and the carrier and that during the course of mixing concentration gradients are established across the interface if the initial concentrations in the carrier and sample are different.Thus, if we have a system in which the sample solution is a mixture of metal ions at a given pH and the carrier is at a different pH and is a solution of a reagent that reacts with the metal ions, we would expect a well defined sequence of colour-forming reactions to take place across the interfacial region as the reagent and metal ions reacted and the pH ~ h a n g e d .~Jniiitnry, 1979 RESEARCH AND DEVELOPMENT TOPICS 5 To test the feasibility of this approach a crucial experiment is suggested by the pH - absorbance graphs of the lead and vanadium chelates of 4- (2-pyridylazo)resorcinol (PAR) obtained under static conditions. At pH 9, only the lead will react with PAR whereas at pH 2 only the vanadium forms a colour.If the carrier is a solution of PAR at pH 9 and the sample is a solution of lead(I1) and vanadium(V) a t pH 2, the pH across the bolus - carrier interface will vary between 9 and 2. Hence the peak obtained by measuring the absorbance downstream from the injection point will not have the Gaussian shape that is obtained when the sample solution contains only a single metal ion, and it might contain enough information for the two ions to be determined.Experimental The reagent stream, Typical peak profiles are shown in Fig. 1. M PAR solution, was buffered with ammonia at pH 9.9 and The outer peaks arise from the reaction of samples were injected as solutions of 2.5 x PAR with lead and the inner peak is due to vanadium. M hydrochloric acid.B i A C' Time Fig. 1. Peaks obtained from one sample containing a mixture of Pb(I1) and V(V) injected into a solution of PAR. (A) General shape of curve; (B) increasing concentration of V(V) while that of Pb(I1) is held constant; and (C) increasing concentration of Pb(I1) while that of V(V) is held constant. -4 straight-line calibration graph is obtained for lead in the presence of a constant or varying amount of vanadium, all of the peak heights being within the expected precision limits of a single sample (standard deviation 1-2%).I t is therefore simple to determine lead selectivity in the presence of vanadium. The resolution of the lead and vanadium peaks is greatest when the vanadium concentra- tion is reasonably high. This puts the range of concentration over which determinations of vanadium can be made close to that of the reagent concentration and hence the slope of the calibration graph is small.If the lead concentration is constant it is possible to determine vanadium with limited precision, but if the lead concentration is varied i t is difficult t o determine the base of the vanadium peak and, coupled with the small slope, this leads to calibration graphs that have not proved useful in the absence of a sophisticated method of peak analysis.lye believe, however, that these results demonstrate the feasibility of using unsegmented continuous-flow systems to obtain a range of conditions (pH gradient, masking gradient, etc.) over a single sample bolus so that resolution of absorbing species in multi-element determinations can be achieved.It seems certain from the results presented and difficulties highlighted that some computing power will be required for peak resolution and more precise determination of peak characteristics. The successful extension of the method to more complex systems depends on the ability to choose suitable combinations of reagents, pH, masking agents and metal ions, on the ability to process results rapidly and on the development of a satisfactory theory of FIA.6 RESEARCH AND DEVELOPMENT TOPICS Proc.Analyt. Div. Chern. SOC. Novel Phototransducer for Improved Peak Detection in FIA A significant development that enhances the sensitivity of the method and hence ultimately the precision of the determination of peak characteristics has been the introduction of a novel type of phototransducer designed specifically for use in FIA.5 Although simple, physically small and inexpensive, it allows sampling rates of up to 300 per hour and is capable of a lower detection limit of less than The light components are extremely cheap and robust and offer a long life, typically 20 000- 100 000 h.Interchangeable units permit most of the visible and near-infrared spectrum to be covered.g of ions in solution. With the associated electronics the total materials cost is about Q O . Design of Transducer Cell The cell is constructed from a Perspex block through which a central hole is drilled as the light path. The light components are glued into either end and two further holes are drilled, perpendicular to the light path and as close as possible to the light components, in order to connect the cell into the stream.A gallium phosphide light-emitting diode (LED) and a silicon phototransistor act as a light source and sensor, respectively. After assembly the transducer cell and flow tubes are lacquered black in order to exclude ambient light. In operation, the output current from the phototransducer is converted into voltage by a simple current to voltage converter circuit before connection to a chart recorder or oscilloscope.Spectral Considerations The spectral emission of a gallium phosphide LED centres on 565 nm with a band width of 30nm. The absorption maxima of transition metal chelates with PAR are in the region of 500-550 nm, but the overlap of the curves is shown in practice to give reasonable results.As the molar absorptivities of the various metal chelates a t 565 nm differ, it follows that the sensitivity of the transducer is different for each metal. Results and Discussion Calibration graphs have been obtained for the transducer response to injected samples of cobalt(I1) in the concentration ranges 0-100 p.p.b.(parts per lo9) and 0-0.5 p.p.m., plotting the mean peak height of two samples. The degree of linearity is good in view of the spectral range of the light source and the graphs are of equal quality and similar linearity (linear regression coefficients typically 0.998) to those obtained using a flow cell in a spectrophoto- meter as a detector, but the sensitivity and resolution are greatly increased.These increases are due to the high stability of the LED used as a light source. For cobalt concentrations of 100 p.p.b. and greater the relative standard deviation in the result is about 1%. Refractometry By virtue of the cell design and its optical configuration the transducer is sensitive under dynamic conditions to changes in the refractive index of colourless samples.The form of the peaks obtained is illustrated in Fig. 2 and the distance from the base line to the first peak is a measure of refractive index under conditions of medium or high flow-rate. Principle of Operation The principle of measurement of refractive index is based on the formation of a refractive index gradient between the sample and a standard stream solution, the magnitude of which is a measure of the original refractive index of the sample.The high sensitivity of the system is due to the physical shape of this gradient. If a solution of high salt content and hence high refractive index is injected upstream from the transducer into a stream of distilled water, when the sample passes through the transducer it will do so such that the lines of equal salt concentration (isohalines) are parabolic in shape due to wall drag, and according to the theory of flow in tubes under laminar conditions.Each isohaline will have a different refractive index to that adjacent to it. The system can be regarded as a series of liquid lenses that focus or diverge light in a direction and of a magnitude dependent on the dimensions of the parabola, the aspect of the parabola with respect to incident light, the direction of the refractive index gradient and the magnitude of the refractive index gradient.RESEARCH AND DEVELOPMEST TOPICS c D i E E E E Fig.2 . Typical peak profiles for refractive index responscs for sodium chloride solutions: (&I) 1 ; (B) 2 ; (C) 3 ; (D) 4; and (E) 5 g 1-I.Carrier stream, distilled water. The refractometric responses are temperature dependent and at low flow-rates diff usivity At medium to high flow-rates the relative standard deviation is about 1% and dependent. the detection limit about 0.010/, m/m of solute. References 1. 2 . 3. 4. 5 . RGiitka, J . , and Hanscn, E. H., --1na!j,tica Chim. Acta, 1975, 78, 146. RiiiiEka, J , and Stewart, J .IT. B., Analytica Chinz. ,?eta, 1975, 79, 79. Stewart, I<. I<., Beecher, G. R., and Hare, P. E., dnalyt. Biochem., 1976, 70, 167. Betteridge, D., and Fields, B., ,4?zalyt. Chem., 1978, 50, 654. Retteridge, D., Dagless, E. L., Fields, B., and Graves, N. F., ,4izalyst, 1978, 103, 897. Cyanoethylation Reactions in Catalytic Thermometric Titrimetry L. Dajer de Torrijos and E.J. Greenhow Departnteizt of Clz Pnzistvy, Chelsea College, L1!Tawresa Road, Loudmi, Sit-3 6LX Xcrylonitrile is used as an indicator reagent in the determination of very weak to strong acids by catalytic thermometric titrimetry.l The temperature rise marking the end-point of the titration is caused by exothermic indicator reactions catalysed by a small excess of the basic titrant.At the end-point, two reactions can occur, anionic polymerisation of the acrylonitrile (reaction 1) and cyanoethylation, if a suitable substrate is present (reaction 2) : nCH,=CHCN 1-+A(CH,CH,CN),CH,CHCN . . . . ( 2 ) ~ - I CH,=CHCN -f- A- -+ ACH,CHCN - ! KH l+ACH,CH,CN + R-; R- + CH,=CHCN -+ RCH,CHCN; RCH,CHCN + RH -+ RCH,CH,CN + R- (2) where A- is the catalyst and RH is the substrate.Recently, we have shown by gas - liquid chromatography that cyanoethylation products are major constituents of the final titration solution when primary or secondary alkanols are present as solvents in the titrant or titrand.* The titrant-catalyst is usually an alkoxide or a hydroxide ion.8 PYOC. A Ptalyt. Div. Clzewc. SOC. A4s cyanoethylation makes a significant contribution to the indicator process, it was decided to investigate its influence on the shape of the thermometric titration curves by evaluating the combined effect of using different substrates, in varying amounts, in the sample solvent with different titrant systems. The solvent mixtures alone, in addition to solutions of benzoic acid in them, have been titrated because previous experience has shown3 that the effects on the titration curves brought about by changes in the composition of the solvent system and titrant are often indicated more clearly in the “blank” titrations. The titration curves for acrylonitrile alone were much sharper when potassium hydroxide in methanol, potassium n-butoxide in butan-1-01 and potassium tert-butoxide in tert-butanol were the titrants than when secondary alkoxides in secondary alkanols were used.This suggests that indicator reactions involving secondary alkanols are much slower than those involving primary and tertiary alkanols. Primary alkanols are known to be cyanoethylated more readily than secondary alkanols and the shapes of the titration curves reflect this fact. However, tertiary alkanols do not undergo cyanoethylation to a significant extent at ambient temperatures, and it must be assumed, therefore, that polymerisation, but not cyanoethyla- tion, occurs with the tert-butoxide titrant.The titration curves for the solution of benzoic acid in acrylonitrile did not differ markedly in shape, but that obtained by using the n- butoxide reagent had the sharpest end-point inflection.When the tert-butoxide titrant was used for titrations of mixtures of acrylonitrile with different alcohols, the titration curves confirmed that the least sharp curve is obtained with a secondary alkanol, sec-butanol. Sharp end-point inflections were achieved by using, as the co-solvents with acrylonitrile, butan-1-01, benzyl alcohol and diphenylmethanol. The last compound is a secondary alcohol and it would seem that the aryl substituents influence its reactivity towards acrylonitrile.A poor titration curve, in terms of end-point sharpness, was obtained with triphenylmethanol as a co-solvent, but this tert-alkanol is an unusual alcohol in the ease with which it forms triphenylmethyl anions. When the content of sec- butanol in the solvent system was varied, the sharpness of the end-point inflection deteriorated and the rate of temperature rise decreased with the increase in sec-butanol content. In all of these titrations, sharper inflections were obtained by adding the dipolar aprotic solvent dimethylformamide to the mixture of acrylonitrile and the alcohols.Earlier studies3 have shown that the presence of this solvent causes the rate of the anionic polymerisation of acrylonitrile to increase because it facilitates the dissociation, and therefore the reactivity, of the alkoxide titrants.The addition of a small amount of butan-1-01 to a solvent mixture consisting of acrylonitrile and dimethylformamide improved the sharpness of the end-point inflection in titrations with the potassium tert-butoxide - tert-butanol reagent.This suggests that some cyanoethylation or chain-transfer polymerisation, involving the butan-1-01, can be beneficial in increasing the rate of the indicator reaction after the end-point. Solvents other than dimethylformamide have been examined ; none was more beneficial with respect to end-point sharpness, and the least satisfactory was toluene, which markedly reduced the rate of the polymerisation and/or cyanoethylation reactions.Vl’hen various potassium alkoxide and hydroxide reagents were used to titrate a mixture of acrylonitrile and sec-butanol, containing a high proportion (75%) of the latter, the titration curves were almost identical. This is to be expected because, when equilibrium is attained in the titrand, the titrant anion will exist mainly as sec-butoxide in all instances.The molarity of the titrant was found to affect the sharpness of the end-point when a solution of benzoic acid in acrylonitrile -pyridine (2 + 2) was titrated with potassium hydroxide in propan-2-01. Poor end-points were obtained with 1.0 and 0.5 M titrants, apparently because cyanoethylation occurred before the expected end-point , while a sharp inflection was achieved by using the 0.1 M titrant.In contrast, the use of a 0.5 M potassium tert-butoxide - tert-butanol titrant, which does not give rise to cyanoethylation, led to a satisfactory end-point inflection. An interesting phenomenon was observed when pyridine was used as a co-solvent and 0.025 M tetramethylammonium hydroxide was used as the titrant.The temperature rise at the end-point was only about one quarter of that observed with the other solvent - titrant systems, but if propan-2-01 was added to the solution after the end-point inflection had occurred the rise in temperature increased markedly. The usual temperature rise was obtained by including propan-2-01 in the original sample RESEARCH AND DEVELOPMENT TOPICSJ Q IZ iiary, 19 79 RESEARCH AND DEVELOPMEST TOPICS 9 solution.Apparently, inhibition of the indicator reaction by pyridine is prevented by the presence of an excess of an alkanol. When primary or secondary amines are included in the sample solution or solvent system cyanoethylation occurs readily at ambient temperatures, without requiring the excess of titrant, and results in a temperature rise before the end-point of the titration.The magni- tude of the temperature rise depends on the content of the amine undergoing cyanoethylation. However, provided that not all of the acrylonitrile has been consumed when the end-point is reached, a sharp end-point inflection is still observed and there is no objection to using these amines as co-solvents.In addition to the alkanol and amine solvents, the acid samples have labile hydrogen and are, theoretically, capable of cyanoethylation. Fortunately, carboxylic acids and phenols do not undergo cyanoethylation under the titration conditions. Acidic thiol groups are readily cyanoethylated and alkyl and simple aryl thiols cannot be titrated.4 The thiol group of 2-mercaptobenzothiazole can be titrated although the group is readily cyanoethylated. \Ye found that both cyanoethylated 2-mercaptobenzothiazole and also O-cyanoethylated phenol are titrated as monofunctional acids by the catalytic thermometric method.Apparently, the cyanoethylation reaction is completely reversible for these compounds under the titration conditions. In earlier ~ t u d i e s , ~ the different titration values obtained when different solvents and different titrants were used in the determination of sulphanilamide were attributed to the latter being cyanoethylated to varying extents when the indicator reaction was initiated.This hypothesis has now been supported by the results obtained when samples of dicyanoethylated sulphanilamide are titrated ; the titration values were found to be dependent on the nature of the solvent system.The aim of this investigation has been to establish the optimum conditions for which the reactivity of the indicator reagents before the theoretical end-point is at a minimum and after the end-point is at a maximum. The information obtained has been used to devise suitable solvent and titrant systems for use with the acrylonitrile indicator reagent in the determination of weak acids.Two combinations are recommended. (i) An acrylonitrile - dimethylformamide (4 + 2) solvent system with 0.1 M potassium n-butoxide in butan-1-01 as the titrant. (ii) An acrylonitrile - dimethylformamide - butan-1-01 (3 + 2 + 1) solvent system with 0.5 M potassium tert-butoxide in tert-butanol as the titrant.Many of the effects of the titrants and sample solvents on the rates of polymerisation and cyanoethylation can be explained in terms of solvation of: (a) the alkoxide or hydroxide ions of the titrant; ( b ) the acrylonitrile; and (c) other solvents, e.g., dimethylformamide. In (a), solvation by alkanols reduces the reactivity of the alkoxide ions, and hence the rate of cyanoethylation ; in (b) , solvation of acrylonitrile by alkanols increases the electrophilicity and, consequently, the reactivity of the P-carbon of the acrylonitrile : s- ROH ...... 0- N = *r\ C.CHF CH,S+ thereby increasing the rate of cyanoethylation.These two solvation effects are clearly in opposition, and the outcome will depend on the nature and concentration of the solvating alkanols.In (c), solvation of dimethylformamide by the alkanols will reduce the effective concentration of the latter, thus increasing the reactivity of the titrant anion and reducing the solvation of the acrylonitrile; this will tend to promote the polymerisation reaction in preference t o cyanoethylation. Further, dimethyl- formamide will solvate the cation of the titrant and release the anion from the ion pair, thus increasing its catalytic powers.In the titrations of benzoic acid, catalysis before the end-point can occur only by the weakly basic benzoate ion, if the stirring during titration is efficient. References 1. Greenhow, E. J., and Spencer, L. E., Analyst, 1973, 98, 90.10 RESEARCH AND DEVELOPMENT TOPICS Proc. Annlyt. Div. Chem. SOC.2 . 3. 4. 5. Greenhow, E. J., Nadjafi, .I., and Dajer de Torrijos, L., Analyst, 1978, 103, 411. Greenhow, E. J., and Shafi, A. A., Analyst, 1976, 101, 421. Greenhow, E. J., and Loo, L. H., Analyst, 1974, 99, 360. Greenhow, E. J., and Spencer, L. E., Analyt. Clzem., 1975, 47, 1384. Application of Carbon-skeleton Gas Chromatography to the Analysis of Polychlorinated Compounds Aileen M.Prescott and Michael Cooke School of Ckemistvy, Uniuevsity of Bvistal, Bvistol, BS8 1TS Polychlorinated biphenyls (PCBs) have been prepared industrially since 1929 by chlorination of biphenyl with anhydrous chlorine using either iron filings or iron(II1) chloride as a catalyst. The product obtained is a complicated mixture of several PCBs. In the UK, PCRs are marketed as Arochlors by Monsanto.All Arochlors are characterkd by a four-digit number ; the first two digits represent the type of molecule (e.g., 12 represents biphenyl, 54 terphenyl and 25 and 44 are mixtures of biphenyl and terphenyl) ; the last two digits give the percentage by mass of chlorine, e.g., Arochlor 1260 is a 12-carbon system with 60% mjm of chlorine. Commercially, polychlorinated naphthalenes (PCNs) are manufactured as Halowases, Nibren waxes, Seekay waxes and Clonacire waxes.The three major types of Nibren waxes are D88, Dll6N and D130, and all contain between 50 and 60% mjm of chlorine. I t is thought that the amount of PCNs produced industrially is about 10% that of PCBs. PCBs together with $9'-dichlorophenyldichloroethylene (DDE) are now the most abundant of chlorinated aromatic pollutants in the ecosystem and have been identified in many species of British wildlife, rainwater and sewage sludge.They are present because they are very stable and fat soluble. The possible paths by which they could be dispersed are as wastes into rivers and lakes, as industrial smoke, and via direct contamination of foodstuffs. PCRs can enter the body directly through the skin, by inhalation as vapours, or by ingestion in food. In the past, the analysis of environmental samples containing residues of organochlorine pesticides (OCPs), PCBs and PCNs has proved difficult as these compounds are essentially very similar.Many different procedures have been tried, most of which have involved three stages, viz., extraction, clean-up and detection. Two main extraction techniques have been employed, Soxhlet extraction and washing, i.e., solvent partitioning with solvents such as hexane or acetone. For the clean-up stage adsorption chromatography, liquid - liquid partitioning, gel-permeation chromatography and volatilisation have all been used to separate the three classes of compounds.These procedures have been only partially successful as certain pesticides, aldrin, heptachlor and $$'-DDE in particular, have tended to remain in the PCB fraction and they then interfere in the detection and quantification stage.Normally gas - liquid chromatography with an electron-capture detector has been used for quantitation as it is very sensitive to the electronegative chlorine atoms of the PCBs and PCNs.As there are a larg? number of components in each YCB mixture, the gas - liquid chromatographic trace is very complicated. Normally quantification has been based on the areas of one or more selected peaks, total peak areas or heights of selected peaks. However, these methods tend to be inaccurate because not only do environmental samples contain complex mixtures of PCBs and PCNs but in the environment there is weathering and preferential biodegradation of certain PCB isomers. To overcome this, perchlorination of the PCB residue to decachloro- biphenyl has also been used.In 1972 Flotard and Veith used it in the analysis of sediments. Recoveries of about 85% for Arochlors were obtained by refluxing acidified sediments for 4 h. Separation is achieved as the vapour pressures of many pesticides and industrial chemicals are greater than those of water-soluble chemicals. The principle of carbon-skeleton gas chromatography is embodied in the work of Thompson, who showed that sulphur, nitrogen, oxygen and halogens in organic compounds can be replaced by hydrogen, and unsaturated bonds can be saturated by passing the compounds over a heated catalyst in a stream of hydrogen. In humans they cause cliloracne and liver disease.Steam distillation has commonly been used in flavour and drug analysis.Jnizicuy, 1979 RESEARCH AND DEVELOPMENT TOPICS 11 Method In this work either a neutral 37; palladium catalyst or a 5q4, platinum catalyst was used. Both were conditioned under a stream of hydrogen at 120 "C for 30 min, 210 "C for 30 min and then a t 310 "C for 6 h.A normal Pj7e 104 gas chromatograph fitted with dual flame- ionisation detectors was employed. The catalyst was packed into the part of the column that passed through the injection point heater and was thus maintained at the required temperature. Products were identified by a mass spectrometer coupled to the gas chromatograph.As the tempera- ture of the catalyst was increased from 140 to 305 "C there was a progressive decrease in the formation of bicyclohexyl and phenylcyclohexyl and an increase in the yield of biphenyl. At 305 "C biphenyl was the only product from the ,4rochlors used. I t is likely that at low temperatures loss of chlorine is followed by or coupled with hydrogenation of the aromatic rings.At higher temperatures a secondary reaction involving the dehydrogenation of the cj-clic system is also present. At catalyst temperatures less than 280 "C hydrogenation of the rings was less pronounced although gas chromatographic - mass spectrometric studies indicated that s o n e tetrahydronaphthalene was present. Using this technique, polychlorinated terphenyls (PCTs) were converted into a mixture of 0- wz- and p-terphenyl at 305 "C.At 205 OC naphthalene gave two compounds, which, from mass-spxtrometric studies, were suggested to be tetra- and decahydronaphthalene. As the temperature was increased the peak heights decreased and at 305 "C no peaks remained; presumably the naphthalene skeleton was completelj? destroyed at this temperature. The results for PCNs were similar to those for naphthalene.At 280 "C small amounts of bicyclo- hex\-1, phenylcyclohexyl and biphenyl were eluted. As the temperature was decreased the amount of bicyclohexyl increased. At 180 "C conversion into bicyclohexyl was quantitative, and PCNs were converted only into decahydronaphthalene. Initially a 5y0 SE-30 column was used to separate the biphenyl and naphthalene after catalysis.Later gas - solid chromatography with a rubidium chloride column was employed, as inorganic salts have been found to give excellent separations of hydrocarbons and also very good reproducibility over a long period of time. Hydrogen was used as the carrier gas. At low temperatures hydrogenation of the aromatic rings tended to occur. Dechlorination of PCXs was much easier.At 305 "C PCNs were quantitatively converted into naphthalene. LTsing the 5% platinum catalyst, convzrsion of PCNs was poor. PCBs were also completely destroyed at 305 "C. Analysis of Environmental Samples Samples were taken from three areas of the Severn estuary-Aust, Sharpness and Arlingham. From the first two sites, sediment from an area of 0.09 m2 and a depth of about 5 CM was taken.At Arlingham there were two very different samples-a sandy sediment and coal particles. These samples were separated bj- washing with doubly distilled water. The coal appeared as an upper layer and was removed for separate analysis, as carbon has strong absorbing qualities and it was likely that any contaminants would be partitioned unequally.The samples were extracted by blending with doubly distilled water and then steam distilling for 2 h. After steam distillation the sample was removed and dried to a constant mass. The unit was washed with hexane and the combined extract and washings were concentrated to 2 cm3. For carbon-skeleton gas chromatography a 376 palladium catalyst was used together with a 2% rubidium chloride column.Temperature programming was used to obtain good separation. Xmong the compounds identified were a range of polynuclear aromatic hydrocarbons including naphthalene , met hylnap h t halene, biphenyl, dimet h ylnapht halene, phenant hrene / anthracene, methylphenanthrene, fluoroanthene, pyrene and chrysene. Only trace amounts of diphenylet hane, formed by dechlorination of DDT and related compounds, were found. This suggests that only trace amounts of these particular compounds were present. Biphenyl and naphthalene were found in the samples before catalysis and so biphenyl and naphthalene levels before and after catalysis were measured.Additional biphenyl or naphthalene was assumed to arise from declilorination of PCBs or PCXs. The contaminants were extracted into 20 cm3 of hexane.12 RESEARCH AND DEVELOPMENT TOPICS Proc.Analyt. Div. Chew. SOC. The possibility that other species such as hydroxybiphenyls may contribute has not been precluded, but as yet we have been unable to catalyse hydroxybiphenyl to biphenyl. The amounts of PCBs and PCNs found in the samples are given in Table I. TABLE I PCBS AND PCNS FOUND I N SEDIMENTS Values are given in parts per billion ( lop9 g g - l ) dry mass.Total Satural Total li’atural -1roc;hlor Site biphenyl biphenyl naphthalene naphthalene 1260 Xust . . .. . . . . 791 81.4 177 163 1689 Sharpness . . .. . . 589 118 269 266 1 120 (low carbon content) . . 46.0 10.1 ND* XD” 71 (high carbon content) . . 1120 144 1143 1021 2 322 Xrlingham ,Wingham * ND = not detected.These preliminary results suggest that extraction by steam distillation followed by carbon- skeleton gas Chromatography with either a flame-ionisation or mass-spectrometric detector is a practical alternative method for the determination of organochlorines in the environ- ment. Some Aspects of Quantitative Thin-layer Chromatography Samuel J. Lyle and M. Saber Tehrani The Chemical Laboratories, The University of Kent at Cantevbwvy, Kent, C T 2 7NH Thin-layer chromatography (TLC) often provides the means whereby complex mixtures can be separated into their constituent components without resort to expensive or elaborate equipment.However, quantitative determination1+’ of the amount of a separated com- ponent is not as readily accomplished as, for example, in gas - liquid or liquid - liquid chromatography.In general, for TLC in which conventional plates are used, quantitative methods are either based on in situ measurements or determinations in solution following elution of the resolved components from the chromatogram. The in situ determination is commonly performed by a scanning method making use of some physical property such as the reflectance or transmission of light, fluorescent emission or radioactivity. A component eluted from the chromatogram can be determined by any one of several conventional solution techniques.The component to be determined, together with a fixed mass of thin-layer adsorbent, is removed from the plate, mixed thoroughly and a single optical reflectance measurement made on it relative to pure adsorbent preferably also taken from the dried plate.A method based on this technique was suggested3 about 15 years ago but does not appear to have acquired much popularity. Results are compared with those obtained by application of (1) the elution procedure with a spectrophotometric finish and (2) a conventional light-scanning (densito- metric) method. A Pye Unicam SP500 spectrophotometer with reflectance attachment was used for reff ectance measurements. For densitometric determinations the Chromoscan (Joyce, Loebl & Co.), a one-dimensional scanner, was used.Preliminary studies were carried out on the 2,4-dinitrophenylhydrazones of formaldehyde and n-butyraldehyde ; later work related to some red water-soluble dyes. TLC was performed on silica gel G layers supported on glass.For reflectance measurements the variable depth depressions in the Pye Unicani sample holder required excessively large samples for present purposes even when used at minimum In the work described here a different type of technique is examined.J m i i a y , 1979 RESEARCH ASD DEVELOPMEK f TOPICS 13 depth (2 mm) settings. An aluminium insert was therefore constructed, reducing the sample diameter from 24 to 14 mm and the depth to 1 mm; this has the effect of reducing the amount of diluent to less than one fifth, making it possible to load each depression with about 100 mg of silica gel G adsorbent.In calibrations and determinations, known amounts of component applied to thin-layer plates were removed and “diluted” with pure adsorbent to a constant mass, thoroughly mixed and reflectance readings taken.Some tests were made by direct weighing of adsorbent (200 mg was chosen), but to reduce the time required for a measurement a simple apparatus was constructed to enable a constant area of adsorbent containing and surrounding the spot to be removed for mixing and measurement. This layer-removal apparatus con- sisted of a hollow metal thin-walled tube (i.d.3.9 cm), which can be pushed through the adsorbent layer to make contact with the glass backing. A sharp blade attached to a shaft passes down the inside of the tube. The blade is fixed at right angles to the shaft and is of a width that just allows clearance from the walls of the tube. IVhen the shaft is rotated the blade detaches the confined layer from the glass backing.Inversion of the plate over a watch-glass transfers the required mass of sample, provided that the layer is of constant thickness and composition. The uniformity of thickness of the layer is partly dependent on the variability in thickness of the glass backing. I t was found that for the usual layer thickness of 0.25 mm, window glass was an unsatisfactory backing as its thickness could \Tary by as much as the supported adsorbent layer.Glass supplied by Gallenkamp & Co. was found to be sufficiently uniform for the purpose. A comparison of masses of adsorbent removed by the layer-removal apparatus gave the following results : using home-made silica gel G plates (20 x 20 cm plates; 88 CR glass from Gallenkamp) the mean mass (25 measure- ments) was 129.1 mg (standard deviation 10.0 mg), and using Xerck pre-coated silica gel 60 plates the mean mass was 146.7 mg with a standard deviation of 3.2 mg.The larger standard deviation with the home-made plates is thought to be partly due to inhomogeneity within the thin layer and partly to variation in the thickness of the glass.An approximate maximum loading of the thin layer with the 2,4-dinitrophenylhdrazones was 40 pg per spot. Typical calibration graphs for the reflectance and densitometric scanning methods are given in Fig. 1. The quantity (1 - R)2/2R, where R is the ratio of the intensity 150 100 50 0 20 40 Mass of component/pg Fig. 1. Calibration graphs for ( a ) reflectance and ( b ) densito- metric methods.The 2,4-dinitrophenylhydrazone of formalde- hyde on silica gel G is represented. The wavelength of measurement in (a) was 348 nin and in ( b ) a Violet 405 filter was used in the Chromoscan. of light scattered from the sample surface to that from the reference substance, is a function of the concentration of light-absorbing species in the system under consideration. For low concentrations (less than 10 pg of component per sample in this instance) it is insensitive to concentration changes, and only approximates to the direct proportionality with concentra- tion expected from theoretical considerations* at the upper end of the range and beyond in Fig.1. In the densitometric method the instrument response becomes relatively insensitive to concentration change a t hydrazone loadings above 25 pg.Hence, the two methods are complementary with regard to amount of component to be determined in the sample. Benzene and acetone were used in determinations by elution and spectrophotometry. Only14 RESEARCH ASD DEVELOPMEKT TOPICS Proc. Analyt. Dic. Cbzcm. SOC. the latter solvent eluted the hydrazone derivatives quantitatively in 10 ml of solvent, although the former eluted a constant fraction and could be used.Absorbances were in the range 0.05-0.35 for 5 4 0 pg per 10-ml sample. Table I summarises results obtained by the different methods for the formaldehyde and n-butyraldehyde derivatives chromatographed in admixture on silica gel G plates. When compared with the component masses taken, TABLE I SOME RESULTS FOR THE 2,4-I)Ir\'ITROPHEN~LHY-DRAZONES OF FORMALDEHYDE AKD n-BUTYRALDEHYDE SEPARATED ON SILICA GEL G PLATES DEVELOPED GSISG BENZENE - LIGROIN (B.P.60-80 "C) (3 + 1 VjV) AT 25 "C Formaldehyde clcri\ati\-c -7 Mass of sample Mean mass Method taken/pg found/pg Elution method . . . . . . 10.5 10.9 20.0 20.9 Reflectance method (mass adjustment) . . . . . . 10.5 10.2 removal apparatus) .. . . 9.5 10.0 Scanning method .. . . 9.0 10.4 21.0 21.6 Reflectance method (using layer- 20.5 21.2 17.0 18.0 Rutyraldehyde derivati\-c 7- i - 7 No. of lllass of ineasiiremen ts sample Mean mass per mean mass t ken / pg f ou 17 d ,us recordccl 10.5 11.1 7 20.0 I9.G 10.5 11.2 12 21.0 20.2 9.5 9.8 12 9.0 8.6 10 20.5 20.4 17.0 17.9 good agreement is obtained with the means of several determinations by each of the methods.The spread of values about the mean within each group of results was about &2-30;6 for elution, 5% for reflectance and 5-10yo for the densitometric measurements. The agreement between values expected and found when using the scanning method is poorer in comparison with the other two methods; this can be attributed largely to variation in (1) layer thickness and composition and (2) spot size and shape.lq4 Three red water-soluble dyes were separated by TLC and determined by the three methods.Relevant properties and results of the determinations are recorded in Tables I1 and 111. TABLE I1 SOME PHI'SICAL PROPERTIES A S D K , VALUES FOR THREE W.4TER-SOLUBLE DYES (9 + 2: + 1 v/v) AS DEVELOPING SOLVENT SYSTEM SEPARATED ON SILICA GEL G AT 25 "C USING ETHAXOI, - BLJTAN-~-OL - WATER AIolar extinction Colour coefficient * / Index No. 1 mol-l c n r l \\'avclength~/nm Amaranth (F, D and C Red No.2) ~ . IS 186 20 100 522 0.10 Erythrosinc B (F, D and C Red S o . 3) . . 45 430 68 000 522 0.95 Rhodainine B . . . . . . . . 45170 85 800 552 0.45 * In SOYo V / V ethanol - water. t Of maximum absorption in SOO,/, V / V ethanol - water.It can be seen that good agreement is obtained between expected and ohserved amounts, particularly for the reflectance and scanning methods. The scanning method was more successful here than in the measurements of the hydrazones, probably because the spot sizes and shapes were more reproducible. The elution method gave low results, which can be attributed to incomplete removal of the dyes from the silica gel by the eluting solvent.From the results, it can be concluded that the reflectance method examined is able to complement a scanning method such as that used here, regarding component mass range. To obtain a mean value in a determination within about 5% of the "true" value it was deduced that at least five replicate measurements by the elution method and ten by each of the others were required.I t was estimated that, starting with the dried developed plates,Jmz itavy, 1979 RESEARCH ,4KD DEVELOPMEXT TOPICS 15 TABLE I11 SEPARATION OF THREE RED DYES ON SILICA GEL G USING ETHANOL - BUTAN-~-OL - WATER (9 + 2 + 1 VjV) AS DEVELOPING SOLVENT AT 25 “C AND THEIR QUANTITATIVE DETERMINATIONS Mass of each -1mount of dye recoveredlpg s o .of J. dye in I separations and Method admixture/pg _\maranth Erythrosine Rhodamine measurements Elution* , . . . 9.7 Keflectance . . 9.6 Scanning . . . . 9.1 8.9 9.3 9.0 9.2 9.0 10 9.6 9.2 10 s.9 9.4 10 * IZ‘ith 80’6 T7/17 ethanol - water and absorbance measurement a t the wavelengths listed in Table 11. the times required were 45 min (elution), 40 min (reflectance using the layer-removal apparatus) and 20 min (scanning). Reliable determinations by the elution and reflectance methods therefore take similar times.Silica gel and alumina are satisfactory but Kieselguhr less so for reflectance determinations. References 1 . 2. 3 . 4. Shellard, E. J ., Editov, “Quantitative Paper and Thin-Layer Chromatography,” Academic Press, Perry, S.G., *\mas, R., and Brewer, P. I., “Practical Liquid Chromatography,” Plenuni Press, Ne\!- Frodyma, 31. M., Frei, R. \Y., and \Villiams, D. J., J . Clzvomat., 1964, 13, 61. Goodall, R. R., J . Clavomat., 1976, 123, 5. London, 1968. York, 1972. Some Analytical Problems in the Determination of Mercury in Biological Materials by a Cold Vapour Technique P.J. Barlow and D. R. Crump Depavtment of Coiastvztction and En-civonme?ztal Health, Univevsity of -4 stoqz iY1 Bivmiwgham, Bivminghain, 134 7ET A. K. Khera and D. G. Wibberley Depavtnzeitt of Pharnzacjr, Cnivessity of A stoiz i n Biwninghanz, Bivnzinglzam, B4 7E T Most mercury occurs in nature as a red crystalline sulphide called cinnabar. Until fairly recently, mercury has been used widely in medicine, but the toxic properties of mercury and its compounds have been recognised since ancient times.The mechanism of mercury toxicity is related to the strong affinity between mercury compounds and sulphydryl (SH) groups in biological materials, particularly in enzymes.lY2 Industrial use of mercury, estimated3 at about 10000 tonnes per annum, may cause pollution problems because it creates high levels of mercury in very limited locations.Therefore, it is most important to reduce the emission of mercury to the environment and it is the task of the analytical chemist to measure low levels of mercury accurately, so that even small changes in mercury concentrations can be detected. There are a large number of analytical techniques that can be used for the determination of mercury, including mass spectrometry, gas - liquid chromatography, atomic-absorption spectrometry with electrothermal atomisation, neutron activation analysis, colorimetric analysis and cold vapour atomic-a bsorption spectrometry.Cold Vapour Technique The method utilises the fact that mercury is the only element (other than inert gases) that has an appreciable vapour pressure at room temperature and of which the vapour is almost wholly monoatomic.Mercury has a low affinity for oxygen; a relatively high con-16 RESEARCH AND DEVELOPMEKT TOPICS Proc. A nalyt. Dkl. Chenz. Soc. centration of mercury atomic vapour can be maintained in air at room temperature. The relatively high vapour pressure also means that no thermal energy is required for vaporisa- tion and atomisation of elemental mercury.A cold vapour mercury atomic-absorption system consists basically of a light source emitting mercury resonance lines, an absorption cell and a detector system. In the work reported in this paper a Perkin-Elmer mercury analysis system fitted to a Model 360 atomic-absorption spectrophotometer was used.Determination of mercury in biological materials by the cold vapour technique involves two distinct stages : destruction of organic matter and sample preparation, and measure- ment of the mercury in the absorption cell. Once a sample has been digested, the procedure for mercury analysis is basically the same regardless of the biological material to be tested. The digested sample is diluted to 100 ml with distilled water and treated with nitric and sulphuric acids in the presence of potassium permanganate in order to oxidise all of the mercury present to the mercury(I1) form (Hg2+).The excess of permanganate is reduced with hydroxylammonium chloride and the mercury is reduced to metallic mercury with tin(I1) chloride. An aerator is placed in the sample solution and a circulation pump moves the air trapped in the system through the solution, thus evaporating the mercury and carrying the vapour through the absorption cell.Mercury vapour in atomic form absorbs the 253.7-nm radiation emitted from the light source. The change in energy is then detected and read out in the usual way on the atomic-absorption spectrophotometer. Standard solutions were prepared on the day of analysis from a 1000 pg ml-1 mercury stock solution. The standards were prepared in the range 0.1-1.0 pg of mercury and a rectilinear graph was obtained.Sample Preparations Human Hair Samples London). for 60 min at 140 "C. Hair samples were prepared using Digby decomposition vessels (Digby Chemical Services, A 100-mg amount of hair was digested with 3 ml of concentrated nitric acid The solutions were then diluted to 100 ml with water.Soil Samples acid for 1Q h. filtrate diluted to 100 ml with water.) Soil samples were prepared by digesting 1 g of dry soil with 20 ml of concentrated nitric The samples were gently heated on an electric hot-plate, filtered and the Plant Samples These samples were prepared in the same way as soil samples.Milk Samples concentrated nitric acid at 110 "C for 60 min in the Digby decomposition vessels. The milk samples were prepared by digesting 2 ml of pasteurised milk with -5 ml of Placental Samples were weighed after removing excess of blood. added and the samples were placed in a water-bath at 55 "C for 2 4 11. almost dissolved after this time. approximately 100 "C for 5 min.Deep-frozen placental samples were defrosted and approximately 4.0 g of the material Concentrated nitric acid (20 ml) was then The samples were To complete the digestion the samples were heated to These solutions were then diluted to 100 ml with water. Roman Bone and Soil Samples A considerable amount of Roman bones and soils have been collected from a Roman cemetery in the South of England; in order to estimate and compare the levels of a number of heavy metals with those in present-day samples.To date only a small number have been examined. The bone samples were prepared by crushing approximately 1.5 g of bone and placing it in 20 ml of concentrated nitric acid. The mixture was heated in a water-bath at 55 "C forJnrlltnvy, 1979 RESEARCH AND DEVELOPMENT TOPICS 17 2-4 h, which resulted in complete digestion, and the solutions were then diluted to 100 ml with water.Roman soil samples were prepared in the same way as described for soil samples. Teeth Samples There is a possibility that some of this mercury could be released into the digestive system, particularly when foods of low pH are consumed. To examine this possibility, teeth samples with and with- out mercury fillings were kept for 24 h in solutions of pH 4 and 7 and the resulting mercury released was then determined.Mercury is commonly used in the amalgam used in teeth fillings. Results and Discussion The results of the determination of mercury in soils, hair, placentae and bones aregiven in Tables I and 11. TABLE I MERCURY LEVELS IN SOILS Samples No.of samples Rangelng 8-l RIean/ng g-l .&gricultural soils (Cheshire) . . 9 33-1 3 1 51 Rural soils (Birmingham) . . 4 19-30 26 Roman soils . . .. .. 6 29-88 50 Yrban soils (Birmingham) . . 9 30-180 99 TABLE I1 MERCURY LEVELS IN HUMAN HAIR, PLACENTAE AND BONES Hair .. .. .. 10 330-5 200 110 Placentae . . . . . . 4 6.80-14.70 9.80 Bones (Roman) . . . . 6 33.6-151.0 81.9 Sample No.of samples Rangelng g-l Meaning 8-l The mercury levels given in Tables I and I1 are in close agreement with those reported b!~ other workers. Mercury was also determined in nine grass samples, and was found to range between 37 and 88 ng g-l dry mass with a mean of 64 ng 8-l. Mercury levels in milk were below the detection level of 0.01 pgml-l. When a known amount of mercury in solution was added to placental, soil and hair samples before wet digestion, the recoveries were 86, 90 and 95% respectively; therefore, only a small amount of the mercury present in solution was lost during the digestion and preparation of samples.Mercury in Teeth values are given in Table 111. teeth with fillings than without fillings. Mercury levels extracted from teeth samples with and without fillings at various pH The results indicate that mercury levels are higher from TABLE I11 MERCURY LEVELS EXTRACTED FROM TEETH Filling present pH Mercury level/ng No 7 20 No 4 40 Yes 7 44 Yes 4 162 Yes 7 29 Yes 4 191 Problems with the Analytical Method and the laboratory environment. The most significant analytical problems were contamination from glassware, reagents18 RESEARCH AND DEVELOPMENT TOPICS Proc.APzaZyt. Div. Cheriz. SOC. The precautions taken to Drevent contamination were as follows : 1. 2. 3. 4. 5. 6. The glassware : (i) (ii) (iii) (iv) (v) (vi) X reagent blank must be run in order to verify the purity of the reagents. When complex sample preparation is necessary, recovery studies should be made and standards should be carried through the same digestion procedure.Actual operation of the system should be preceded by completion of all sample preparation require- ments. All precautions should be taken to avoid mercury contamination from the laboratory. A large volume (up to 100mI) of distilled water is used and therefore it is necessary always to test the water before carrying out any analysis, as even slight contamination can have a large effect.Also, an acid wash of the B.O.D. bottle between each deter- mination was found to be essential. During wet digestion all samples should be completely digested. It is most important to exclude water vapour from the absorption cell by the use of a desiccant tube. all glass equipment was soaked in 2% Decon 90 concentrate overnight; rinsed twice with hot tap water; washed with concentrated nitric acid ; again rinsed twice with hot tap water; rinsed twice with distilled water and allowed to drain; all glassware were then covered during storage.Conclusion The cold vapour technique for the determination of low levels of mercury in biological materials gives results that compare well with literature values; good recoveries of mercury are obtained and it is a fairly rapid method for use as a routine analytical procedure.References 1 . 2. 3. Westermark, T., and Lunggren, K., “Mercury Contamination of Man and His Environment,” Hughes, ?V. L., Anit. N . Y . Acad. Sci., 1957, 65, 454. Passow, H., Rothstein, A., and Clarkson, T. W., Pharm. Rev., 1961, 13, 185. Proceedings of IAEA Symposium, International Atomic Energy Agency, Vienna, 1972.Application of lon-selective Electrodes to Environmental Pollution Problems in the Steel Industry D. S. Macintyre, B. G. Cooksey and J. M. Ottaway Depavtment of Ptcvc and Applied ChemistJy, Uwivevsit-v of Strathclyde, Cathedval Street, Glasgow, GI 1 X I - Strict legislation governing the discharge of industrial effluents into the environment gives rise to a need for continuous monitoring of a variety of different substances.Ion-selective electrodes are ideally suited to continuous monitoring but, like most analytical techniques, they suffer from interference effects.l The steel industry, like many other heavy industries, produces large amounts of some- times potentially toxic waste and it is often useful to know whether substances such as fluoride or cyanide, for example, are being discharged into the environment in complexed or ionised forms, as the toxicities of the two forms frequently differ.2 The lanthanum fluoride ion-selective electrode was used to measure fluoride in a variety of steelworks effluents and natural water samples.Tests showed that the electrode was not susceptible to any direct interference effects other than from hydroxyl ions, which restricted the operating pH to not greater than 7.0 at a fluoride concentration of 0.1 p.p.m.Com- plexation of fluoride with hydrogen ions also restricted the operating pH to not less than 4. Aluminium and iron gave low fluoride recoveries when present in fluoride standard solutions owing to the formation of soluble metal - fluoride complexes.Atomic-absorption spectro- metry showed that Ravenscraig Steelworks effluents contained only very low concentrationsJanuary, 1979 RESEARCH AND DEVELOPMEKT TOPICS 19 of aluminium (less than 2.0 p.p.m.), but other British Steel Corporation effluents are known to contain more than 25 p.p.m. of aluminium. Tests were carried out to establish whether fluoride recoveries in the presence of aluminium could be improved by the addition of various decomplexing agents.EDTA solutions at pH 6.0 gave poor fluoride recoveries (of the order of loo,/, at 50 p.p.m. of aluminium). This is presumably because metal ion - EDTA4 com- plexes are strongly pH dependent. The best decomplexing buffer was found to be a mixture of 0.1 M sodium acetate and 0.1 M sodium citrate at pH 6.0.However, only 70% fluoride recoveries were obtained in the presence of 50 p.p.m. of aluminium. Fluoride recoveries of greater than 90% could be obtained when 5 g of sodium citrate were added to 100-ml samples, which were boiled and cooled before the addition of the previously mentioned citrate - acetate buffer. However, this type of system is not acceptable for continuous on-line analysis and so attention was turned to automated techniques.-4 Technicon AutoAnalyzer TI fitted with an automatic distillation module was used with an Orion combination fluoride electrode fitted with a continuous flow cell to measure auto- matically total fluoride concentrations. An EIL 7050 expanded-scale pH meter and a Honeywell 195 chart recorder were used to measure and record the electrode potentials.Fig. 1 shows the apparatus. ,Analysis of 40 effluent and natural water samples for fluoride by this technique showed good agreement with similar analyses carried out colorimetrically. A regression line of y = 0.01 + 0.99% was obtained with 95% confidence limits on b of 0.95 and 1.03. Analysis of the same samples by the fluoride electrode without distillation also compared favourably, giving a regression line of y = 0.03 + 1.01% with 95% confidence limits on b of 0.98 and 1.04.This illustrated that a good measure of total fluoride concentrations in the samples could be obtained simply by mixing solutions with citrate - acetate buffer and analysing them directly by the fluoride ion-selective electrode. Effluents containing high concentrations of aluminium would require the automatic distillation procedure, which ga1.e fluoride recoveries of greater than 90% in the presence of 50 p.p.m.of aluminium. The determination of cyanide presents more problems than the determination of fluoride. T'ery low concentrations of free cyanide are extremely toxic to fish.Thiocyanates and complex cyanides are considerably less toxic, but variations in temperature, pH and the amount of sunlight may give rise to variable amounts of cyanide pollution downstream of any discharge point. It is therefore useful to be able to measure free cyanide, thiocyanate and complexed cyanide concentrations independently. However, most techniques available do not distinguish between cyanide and thiocyanate and delicate separation techniques are often required in order to measure complexed cyanide concentrations.Ion-selective electrodes for the direct determination of both thiocyanate and free cyanide are aL7ailable commercially, but both types of electrode suffer from problems of insensitivity and unacceptable interference effects.Free cyanide, however, can be determined indirectly by using a silver ion-selective electrode and a potassium silver cyanide 'KAg(CN),] indicator solution. Frant et aL3 reported a detection limit of less than 0.02 p.p.m. of cyanide for this technique, which should not suffer from serious thiocyanate interference because the solubility product is not exceeded. One major interference in this technique is sulphide.Tests showed that as little as 0.08 p.p.m. of sulphide gave readings of 0.1 p.p.m. of cyanide when no cyanide was in fact present. The interference results from destruction of the potassium silver cyanide indicator solution. Calibration graphs were plotted for cyanide in the presence of 0.05 M lead nitrate with and without sulphide present. These results showed that 100% removal of sulphide interference was not possible by this method, and conse- quently sulphide was found to interfere in cyanide determinations in the range 0.01-0.1 p.p.m.of cyanide. The finite amount of sulphide expected to remain in solution owing to tlie partial solubility of lead sulphide was negligible and cannot explain this interference. The colorimetric technique for the determination of cyanide showed considerably less sulphide interference, but sample pre-treatment stages such as distillation or dialysis were essential in the analysis of steelworks effluents, owing to highly coloured and turbid samples.Further problems arise from the fact that reactions such as the pyridine bipyrazalone colori- metric reaction are also sensitive to thiocyanate, and when solutions containing thiocyanate are distilled an apparent recovery for cyanide in excess of 25% is recorded.This effect can, of course, be corrected for by independently measuring the thiocyanate concentrations, for example by using the iron(II1) nitrate colour reaction. However, considerable errors can20 RESEARCH AND DEVELOPMENT TOPICS R o c . Analyt.Dia. Chew. SOC. arise when this technique is applied to steelworks effluents, as thiocyanate concentrations are frequently higher than cyanide concentrations by a factor of 10. An apparatus similar to that shown in Fig. 1 was used to distil cyanide-containing samples and to determine the cyanide content with the silver ion-selective electrode. Cyanide recoveries of only 2-3% were obtained when thiocyanate solutions in the range 0-3 p.p.m.were analysed by this technique. This indicates that about 22% of the thiocyanate distils over. /c--7\ 31 Sampler I V Heating bath with distillation coil Distillation solution Waste -1 Buffer Orion pH meter combination and recorder fluoride electrode Fig. 1 . Simplified diagram of apparatus used for the determination of total fluoride concentrations.It was also found that recoveries of free cyanide of about 98% were obtained when iron- complexed cyanides were distilled and analysed by this technique. Atomic-absorption analyses showed that the concentration of metal ions that form very stable metal - cyanide complexes, for example, cobalt and nickel, were negligible in steelworks effluents at Ravens- Craig Works. Samples that were exposed to ultraviolet radiation before distillation gave a I I I 0.1 1 .o 10 100 200 0.01 Cyanide concentration, p.p.m.Fig. 2. Cyanide calibration graphs obtained by automatic distillation and analysis with a silver sulphide electrode : (A) potassium cyanide ; (B) iron-complexed cyanide : (D) thiocyanate; and (C) thiocyanate after automatic treatment with ultra- violet radiation, before distillation.Jar z 21 m y , 19 79 RESEARCH AND DEVELOPMENT TOPICS 21 constant recovery of about 45% of cyanide from thiocyanate. This was carried out using a Technicon automatic ultraviolet digestor and the results are shown in Fig.2. By using this technique, it is possible to obtain measurements of free cyanide, complexed iron cyanide and thiocyanate using automated ion-selective electrodes.Free cyanide measurements are made by analysing samples mixed with potassium silver cyanide, free cyanide plus complexed iron cyanides are measured with an ion-selective electrode after auto- matic distillation (an approximately 30-fold excess of thiocyanate can be tolerated) and free cyanide plus complexed cyanide and thiocyanate can be measured by treatment with ultraviolet radiation followed by distillation.This work illustrates the ever increasing importance of ion-selective electrodes in analvtical chemistry. The authors thank the Ravenscraig Laboratory of the British Steel Corporation for providing facilities to carry out this work. References 1. 2 . 3. Orion Research, “_lnalytical Methods Guide,” Orion Research Inc., Cambridge, Mass., 1975.Erichson Jones, J . R., “Fish and River Pollution,” Butterworths, London, 1964. Frant, 31. S., Ross, J . \T., Jr., and Riseman, J. H., .4iznZyt. Chein., 1972, 44, 2227. Some Studies in Inductively Coupled Plasma Emission Spectroscopy J. F. Alder, R. M. Bombelka and G. F. Kirkbright The high-frequency inductively-coupled argon plasma (ICP) has become established during the past decade as a versatile tool for emission spectrochemical analysis, offering the analytical chemist a stable, high-temperature source and high sensitivity for the determination of a wide range of e1ements.l Analytical growth graphs are commonly found to be linear over a concentration range of 4 or 5 orders of magnitude from the detection limit to the onset of self-absorption, and inter-element effects are small or absent owing to the high gas and electron temperatures encountered in the analyte emission zone.2 In order to characterise this zone, we have measured excitation and ionisation temperatures and studied the role of water vapour on various parameters in the ICP.The criterion for the existence of local thermal equilibrium (LTE) in a plasma is that bound states should have at least a 10-fold higher probability of de-excitation by collision than by emission of radiation ; when this occurs at sufficiently high electron densities (about 2 x 1022m-3 for argon) the gas temperature, T,, electron temperature, Te, excitation temperature, Tex, and ionisation temperature, Tion, are found to be ~ i m i l a r .~ Deviations from LTE will result in a disparity between these temperatures with Te rising and T , falling below the equilibrium value, and consequent over- and under-populations of various energy levels. Fe I was chosen for the determination of the electronic excitation temperature in the ICP and Ca, Ra, Zn, Cd, Fe, Mg and Ti for the determination of the ionisation tempera- ture via the Saha equation.The electron density was determined from the Stark half-width of the Stark broadened line Hp (486.1 nm) after deconvolution of the Doppler component and the instrumental function. The instrumental system employed in this work consisted of a 2-kW, crystal-controlled, radiofrequency generator operating at 27 MHz (International Plasma Corp., Model 120-27) with a manual matching network, a three-tube demountable silica torch, a concentric glass nebuliser (Meinhard Associates, Model T-230-A2) and a l-m scanning monochromator (Rank Hilger, Monospek 1000) fitted with an EN1 6256 B photomultiplier tube and a plane diffrac- tion grating blazed at 330 nm.The radiofrequency forward power was set at 1 200 13’ and the argon coolant gas flow-rate was 12.8 1 min-l; no plasma gas was employed.Measure- ments were performed at two injector gas flow-rates, 0.9 and 1.3 1 min-1, and the internal22 RESEARCH AND DEVELOPMENT TOPICS Proc. Analyt. Div. Chem. SOC. diameter of the injector tube, an important parameter that determines the axial velocity of the sample, was 1.60 mm. Fe I was chosen for the determination of the excitation temperature, and 20 emission lines were selected with upper energy levels, E,, in the range 26 875-55 754 cm-l.Boltzmann plots of Ix3/gf against E, were constructed at three different viewing heights, 10, 20 and 30 mm, in the tail flame above the top of the load coil. The excitation temperature in each instance calculated from low-energy lines was found to be substantially lower (about 1500 K) than that calculated from high-energy Fe I lines.In addition, the ionisation temperatures obtained from atom and ion line intensities of the seven elements studied were found to be similar in magnitude and to correspond more closely to the values of the excitation tempera- ture obtained from the high-energy Fe I lines.Radial intensities calculated via the Abel inversion of lateral data also showed a non-linearity of the Boltzmann plot, and hence the non-LTE properties of this excitation source were demonstrated. The non-linearity of the Fe I Boltzmann plot can be explained in terms of an over-population of the lower energy levels with respect to higher energy levels owing to radiative de-excitation from the upper levels that is not balanced by the inversz absorption process.Hence, care should be taken in equating the electron temperature to the excitation temperature derived from low-energy lines for this two-temperature non-LTE plasma. On pneumatic nebulisation of aqueous solutions into the ICP, a substantial amount of water vapour enters the source. At 6 000 K, water is almost entirely present as H and 0 atoms, with a small percentage of other species (H+, Of, 02+, e, OH, H,O, H, and 0,).5 The presence of water might be expected to modify the conditions in the analyte zone.A small U-tube containing dry coarse silica gel was mounted between the nebuliser chamber and the plasma torch. The silica gel served to remove water from the aerosol while still permitting a small amount of sample particles to enter the plasma, without affecting the argon flow-rate.The amount of water removed was determined from the increase in mass of the U-tube. It was found that both the electron density and cadmium ionisation tempera- ture decreased as water was progressively removed. The continuum emission intensity, being a sensitive function of the electron density, was also found to decrease. To show this effect conclusively, electron densities, cadmium ionisation temperatures and the Fe I excita- tion temperature were measured at different viewing h3ights in the plasma both with water and with most of the water removed. Again, a significant reduction in each parameter was observed when the aerosol was dried substantially. It can therefore be concluded that the presence of small amounts of water vapour in the ICP increases the excitation parameters. An increase in the electron density would be expected with water present, as under the actual conditions the degrees of ionisation of both hydrogen and oxygen are expected to be about 5 times higher than that of argon. When ultrasonic nebulisation rather than pneu- matic nebulisation is employed, far higher rates of water transport to the plasma are obtained. In this instance lower temperatures might result if a large fraction of the energy available was used to dissociate the substantial amount of water present. Sample excitation and ionisation in the ICP is most likely to be the result of direct electron collisions in a stepwise fashion, with excitation of high levels occurring by excitation from intermediate energy levels, the probability of such a process increasing as the physical size of an excited atom increases. Although Penning ionisation of analyte atoms with excited argon atoms may play a part in excitation and ionisation (collisional lifetimes of all four 4s Ar I levels are extremely short in the ICP and thus the distinction between metastable and other excited argon atoms should not be made in the ICP), a two-temperature plasma with an electron temperature 1 000-2 000 K higher than the gas temperature, where elec- trons alone are principally responsible for analyte excitation and ionisation, is to be favoured. ,4 series of studies to measure spatially resolved gas temperatures in the ICP from Doppler half-widths using a piezoelectric scanning Fabry - Yerot interferometer has been initiated in this department. Using this technique, we hope to obtain in the near future reasonably accurate spatial values of Tg, which would then serve as absolute minimum values of Te. Knowledge of these parameters, together with n e , is essential to an understanding of the excitation mechanism in the ICP, and to explain why ion lines are more intense than commonly expected. Whilst Mermet and Trassy6 and Roumans and De Boers favour a As the transition probabilities of Fe I are known to be accurate to aboutJa?iuary, I979 RESEARCH AND DEVELOPMENT TOPICS 23 mechanism involving Penning ionisation to explain these observations, we feel that a moderately high electron temperature, closer to the value obtained from the high-energy Fe I lines, can explain this effect equally well. References 1. 2. 3. 1. 5. 6. 7. Fassel, V. .1., and Kniseley, R. N., Analyt. Chem., 1974, 46, 1110*1 and 1155.1. Kornblum, G. R., and De Galan, L., Spectrochim. .4cta, 1977, 32B, 455. Bacri, J., Gomes, A\. >I., and Benzaid, S., ,I. Phjvs. D , .4ppl. Ph?is., 1976, 9, 1743. Bridges, J . &I., and Kornblith, R. L., A s t v o p h ~ ~ s . J . , 1974, 192, 793. Rurhorn, F., and Wienecke, R., Z. Phys. Chem., 1960, 215, 285. Mermet, J. M., and Trassy, C., Rm. Phj)s. AppZ., 1977, 12, 1219. Roumans, 1’. W. J . M., and de Boer, F. J . , Spectrochim. Actn, 1977, 32B, 365.