ANALYST, FEBRUARY 1991, VOL. 116 149 Determination of Nickel Tetracarbonyl by Gas Chromatography Alexander Harper AEA Technology, Harwell Laboratory, Oxfordshire OX7 7 ORA, UK Capillary gas chromatography, with use of an electron-capture detector, has been assessed as a detection technique for nickel tetracarbonyl in gas samples. Detection limits as low as 1 part in 1011 can be achieved under the appropriate conditions. Keywords: Gas chroma tograph y; electron-capture detector; nickel tetracarbon yl Nickel tetracarbonyl is a volatile, thermally unstable material, which may be formed by the direct reaction of CO with nickel or nickel-containing alloys. Interest in the analysis of gases for Ni(C0)4 at low concentrations arises both from the toxicity of the material and from its role in the transport of catalytically active nickel in gas circuits. The carcinogenic behaviour associated with the chronic inhalation of Ni(C0)4 is reflected in the exposure guidelines recommended for this material, 100 vppb (parts per lo9 by volume) short term exposure limit (STEL) (10 min).Even more stringent controls have been applied in the past; the American Conference of Industrial Hygienists set a value of 1 vppb in 1959, although this was subsequently increased to 0.05 mg m-3 of nickel [equivalent to 20 vppb of Ni(C0)4].2 These low values clearly point to the need for a reliable method for the determination of Ni(C0)4 vapour at vppb levels for environmental monitoring. The transport of catalytic nickel in gas circuits involves the formation of Ni(C0)4 in relatively cool portions of the system and its subsequent thermal decomposition in regions of the circuit at higher temperatures.The catalytic nickel thereby formed can lead to the decomposition of hydrocarbon materials with the resulting formation of carbonaceous deposits in undesirable places. For example, in the primary cooling circuits of Advanced Gas-Cooled Reactors, where CO is maintained in the C 0 2 coolant to inhibit corrosion of the graphite moderator, the phenomenon could lead to the formation of layers of material of relatively low thermal conductivity on heat-transfer surfaces. In systems such as this, where the rate of gas flow is high, even trace concentrations of Ni(C0)4 in the gas phase can result in the transport of significant amounts of catalyst. Analysis of gases for Ni(C0)4 has been accomplished by two techniques: adsorption of the component of interest on to a substrate and subsequent assay of the substrate for nickel,3 and by exploitation of the chemiluminescent reaction between Ni(C0)4 and ozone .4-7 Adsorption techniques possess the advantage that they mag be calibrated without resource to gas standards containing Ni(C0)4. Determinations by such methods are, however, time consuming and require large samples of gas, especially at low concentrations.Further, the ultimate sensitivity is limited by the inevitable nickel contami- nation of suitable adsorbents. Chemiluminescence possesses the merit of sensitivity, but commercially available equipment is expensive, and the method can be subject to interference from the chemiluminescent reactions of other components of the sample matrix.6 Gas chromatography, coupled with a suitable detection system, offers the potential for good selectivity and sensitivity combined with a modest sample volume and analysis time.This paper describes the evaluation of gas chromatography, with electron-capture detection, for the determination of Ni(C0)4 in CO-C02 gas mixtures. Experimental Gas Chromatograph The gas chromatograph used in the present study was a Hewlett-Packard Model 5890, fitted with a six-port gas- sampling valve and a constant-current electron-capture detec- tor (ECD). A facility was available-to allow operation of the oven at sub-ambient temperatures. Data acquisition, storage and analysis were effected by using a Hewlett-Packard Model 59970 workstation.All the experiments described were performed with the use of a 10 m x 0.53 mm capillary column, coated with a 2.65 pm film of 5% diphenyl-95% dimethyl polysiloxane gum (Hew- lett-Packard). The use of a capillary column restricts carrier gas flow-rates to values too low to allow operation of the ECD on carrier gas alone. A subsidiary gas feed (make-up gas) was therefore supplied to the detector. This has the advantage that carrier gas composition is not limited to those gases suitable for ECD operation. The analytical conditions finally adopted for the assessment of sensitivity are summarized in Table 1. Materials Nitrogen (high purity, oxygen free) was obtained from Air Products. Carbon monoxide (99.5%) was obtained from the same source and passed over active charcoal to remove traces of Ni(C0)4 and Fe(CO)5 before use. Gaseous C 0 2 was obtained from converters charged with solid carbon dioxide (Distillers).Nickel tetracarbonyl (>97%) was obtained in liquid form from Pfaltz and Bauer, and was used without further purification. Table 1 Summary of chromatographic conditions Non-isothermal 10 m x 0.53 mm coated 10 m x 0.53 mm coated with 2.65 ym film of 5% with 2.65 p n film of 5% diphenyl-95% dimethyl diphenyl-95% dimethyl polysiloxane gum polysiloxane gum Parameter Isothermal Sample volume 0.25 ml 3.0 ml Column Injector temperature 30 "C 30 "C Column temperature 30 "C -30 "C for 1 min then ramp at 40 "C min- I to 10 "C Make-up gas Nitrogen Nitrogen Make-up gas flow-rate 60 ml min- 1 Carrier gas flow-rate 2 ml min-1 2 ml min-1 Detector temperature 60 "C 60 "C Carrier gas co co 60 ml min- 1150 ANALYST, FEBRUARY 1991, VOL.116 Standard Gases Stock mixtures of Ni(C0)4 in CO-C02 were prepared by transferring between 0.8 and 1.0 kPa of Ni(C0)4 vapour into an evacuated aluminium cylinder. The cylinder was then pressurized, with as little delay as possible, to 4 MPa with a gas mixture of 2% v/v CO in C02. Carbon monoxide was maintained in the mixture to suppress the decomposition of Ni(C0)4 to metallic nickel and CO. Although this mixture was not analysed directly, dilutions made from it suggest an Ni(C0)4 concentration in the range 150-200 parts per million by volume (vppm). Standards in the range 1-5 vppm were prepared by transferring about 60 kPa of the stock mixture into an evacuated cylinder and pressurizing to 4 MPa as before.Subsequent dilutions of this standard provided mixtures with nominal Ni(C0)4 concentrations as low as 5 vppb. Cylinders with nominal Ni(C0)4 concentPations greater than 0.5 vppm were standardized by passing a known volume of gas over active charcoal and assaying the charcoal for nickel. The method has been described in detail elsewhere.3 This technique allows the standardization of gas mixtures independently of any Ni(C0)4 source. Gas mixtures contain- ing 4.5 and 1.2 vppm of Ni(C0)4 were analysed again after 3700 and 2850 h, respectively, at laboratory temperature (20 k 2 "C). No significant change in Ni(C0)4 concentration was observed, indicating the stability of these gas mixtures.Gas mixtures analysed in this way were, therefore, regarded as primary standards. Analysis of mixtures with nominal concentrations below 0.5 vppm was effected by gas chromatography. A calibration graph was prepared by dynamic dilution of a primary standard with C02 by using commercially available electronic mass flow meters (Brooks Instruments Model 5850). This graph was used to standardize mixtures with concentrations down to 50 vppb. Dilution of this mixture allowed the standardization of more dilute mixtures in the same manner. Dynamic dilution and a series of standards allowed the production of a continuous range of concentrations, covering several orders of magnitude, for the assessment of instrument performance. Results Column Performance Experiments involving the use of a 0.25 ml gas sample with a helium carrier gas flow-rate of 2 ml min-1 showed that, on the column described above at 30 "C, the major components of the mixture (CO and C02) were essentially not retained.Nickel tetracarbonyl was clearly separated from these materials, and was eluted as a clean, Gaussian peak. Experiments with carbon monoxide samples, which were known to contain both Ni(C0)4 and Fe(CO)S, showed Fe(CO)S to be eluted much later than Ni(C0)4, as might be expected from their relative boiling points [Ni(CO), 43 "C; Fe(CO)S 102.8 " C ] . ~ ~ ~ The column, therefore, provides adequate resolution of the two carbonyl compounds formed by the direct reaction of CO with a metallic substrate. Similar results were obtained with a CO carrier.Although the studies described in this paper were per- formed with CO-C02 as the sample matrix, the same chromatographic conditions should provide satisfactory resol- ution for the investigation of gaseous environmental samples, as the major components of air are not retained on the type of column used here. Carrier Gas Composition The significance of this parameter lies in the possibility of premature decomposition of Ni(C0)4 on the column or in the detector. This phenomenon could be reduced by the addition of CO to the carrier gas. The response of the detector to a 10 vppb gas standard was measured by using pure helium and CO as carrier gases. With the exception of the detector temperature, which was 105 "C, the other conditions were those shown in Table 1 for the isothermal example.With helium as the carrier gas the peak area obtained was 90 arbitrary units, whereas with carbon monoxide the peak area was 890 arbitrary units. These results indicate that detector response can be substantially enhanced by using CO as the carrier gas. Detector Temperature The sensitivity of the ECD is known to be a function of temperature. 1" The significance of detector temperature in these analyses was assessed from the response of the detector to a 0.25 ml sample of a 19 vppb Ni(C0)4 standard. The carrier gas was pure carbon monoxide and the make-up gas pure nitrogen. Peak area is shown as a function of detector temperature in Fig. 1. Detector response decreases steadily with tempera- ture. It is therefore clear that, for optimum response, the detector should be operated at a relatively low temperature, but that the temperature should be constant to ensure reproducibility. The latter criterion requires operation at temperatures significantly above ambient; a value of 60 "C was adopted.Oven Temperature The most straightforward mode of operation is to maintain a constant column temperature throughout the analysis. In the present situation, adequate temperature control could be maintained at an oven temperature of 30 "C. The use of an isothermal column, however, limits the maximum sample size that can usefully be employed. The need to maintain satisfactory resolution limits carrier gas flow-rate, and, hence, the rate at which the sample loop can be purged. An excessive sample volume results in malformed peaks and poor resolu- tion.Under isothermal conditions, a 0.25 ml sample was the largest that could be used reliably. One strategy for increasing sample size is to operate the column at low temperature during the flushing of the sample loop. Under these conditions, Ni(C0)4 progresses extremely slowly through the column and is concentrated in a narrow band at the start of the column. Raising the temperature then allows elution of this material as a sharp peak. Fig. 2 shows the effect of this strategy on a 1.0 ml gas sample containing 19 vppb of Ni(C0)4. In this example, the column temperature was kept at -20 "C for 1 min, then ramped at 40 "C min-1 to a loo0 1 900 m m Y m a $' 700 600 0 0 0 3 500 50 100 150 200 Detector temperature/"C Fig.1 Effect of detector temperatureANALYST, FEBRUARY 1991, VOL. 116 h v) C c .- = 1000 2 F 2 & 100. c .- - m Y m a 10 15 1 - . 175 loooo - IJY c .- 5150 - 2 F $125 - + .- Y a, c ;loo - ? 8 75 n L c 0 - + 1.0 1.5 2.0 2.5 3.0 3.5 tlm i n Fig. 2 Effect of temperature programme on peak shape. A, Temperature programmed; and B, not temperature programmed Table 2 Calibration data at low concentration Isothermal Non-isothermal Slope 87.72 2046 Standard error of slope 0.90 56 In terce p t 5.5 17 Correlation coefficient 1 .OO 0.99 Number of data 9 20 Standard error of intercept 3.1 12 Detection limit (vppb) 0.08 0.01 constant temperature of 10 "C. A maximum sample size of 3 ml could be accommodated by using the temperature programme given in Table 1.Sensitivity Calibration graphs prepared under both isothermal and non-isothermal conditions, as defined in Table 2, are shown in Fig. 3. Over the range of concentrations considered the graphs are significantly non-linear. The relationships between peak area (A) and concentration ( c ) in vppb are given by: A = 105c0.87, for the isothermal example; and A = 1158c'JXl, for the non-isothermal example. In order to establish the detection limit, only data with peak areas of less than about 300 arbitrary units were considered. Over this restricted range the linearity was excellent in both instances. Unweighted linear regression by least squares was used to determine the slope and intercept of the best line through the data. These values, together with the associated standard errors and correlation coefficients, are given in Table 2.Also shown in Table 2 are the values for the detection limits. These have been defined as those concentrations at which the lower 95% confidence limit becomes equal to the calculated y-intercept. 1 1 The results of this exercise are summarized in Table 2. The sensitivity of the method is clear from the detection limits quoted. That the difference in detection limit between isothermal and non-isothermal exam- ples is not pro ruta with sample size reflects the greater degree of scatter in the non-isothermal data. 0.001 0.01 0.1 1 10 100 Ni(C0h (vppb) Calibration graphs. A, 0.25 ml loop size (isothermal); and B , Fig. 3 3 ml loop size (non-isothermal) Conclusions Gas chromatography with an ECD has been shown to be an effective technique for the determination of Ni(CO)4 in CO-C02 gas mixtures.Appropriate choice of carrier gas and chromatographic conditions allows detection limits as low as 0.01 vppb to be obtained. Although the method has been devised for the analysis of CO-C02 gas mixtures, the technique should also be appropriate for the analysis of air samples. In the latter instance, however, consideration would have to be given to the stability of the gas sample if analysis and sampling took place at different locations. This work was jointly funded by the United Kingdom Atomic Energy Authority and Nuclear Electric plc under the Thermal Reactor Agreement. The author is grateful to these bodies for permission to publish this paper. 1 2 3 4 5 6 7 8 9 10 11 References Occupational Exposure Limits, HSE Document 40/90, HM Stationery Office, London, 1986. American Conference of Governmental Industrial Hygienists List, 1989-90. Eller. P. M., Appl. tnd. Hyg., 1986, 1, 115. Stedman, D. H.. Tammaro, D. A., Branch, D. K . , and Pearson, R., Anal. Chem., 1979, 51, 2340. Stedman, D. H., and Tammaro, D. A., Anal. Lett., 1976,9,81. Houpt, P. M . , van der Waal. A., and Langeweg, F., Anal. Chim. Acta. 1982, 136, 421. Hikade, D. A., Stedman, D. H., and Walega, J. G., Anal. Chem.. 1984, 56, 1629. Handbook of Chemistry and Physics, ed. Weast. R. C., CRC Press, Boca Raton, FL, 60th edn., 1980, B-101. Handbook of Chemistry and Physics. ed. Weast, R. C., CRC Press. Boca Raton, FL, 60th edn., 1980, B-86. Grob, R. L.. Modern Practice of Gus Chromatography, Wiley-Interscience, New York, 1985, p. 262. Sharaf, M. A.. Illman, D. L., and Kowalski, B. R., Chemo- metrics, Wiley-Interscience, New York. 1986, p. 128. Paper 0/02273J Received May 22nd, I990 Accepted September 27th, I990