Determination of Ni(CO)4, Fe(CO)5, Mo(CO)6, and W(CO)6 in sewage gas by using cryotrapping gas chromatography inductively coupled plasma mass spectrometry Jo�rg Feldmann Department of Chemistry, University of Aberdeen, Meston Walk, Old Aberdeen, UK AB24 3UE. E-mail: j.feldmann@abdn.ac.uk; Fax: +44 1224 272921 Received 17th September 1998, Accepted 17th November 1998 Evidence for the occurrence of Ni(CO)4 in addition to Mo(CO)6 and W(CO)6 in fermentation gases from a municipal sewage treatment plant is presented for the first time.The gases were sampled at the top of the sewage sludge digester using Tedlar bags, and were analysed using cryotrapping followed by gas chromatography coupled with inductively coupled mass spectrometry (GC-ICP-MS). The use of an ICP-MS as an element-specific detector gives suYciently low detection limits for metals and was coupled to a packed column gas chromatograph.This method provides information about the speciation of volatile transition metals in contrast to previously used methods for the determination of Ni(CO)4 in gas samples. The element-specific detection of three diVerent isotopes (m/z 58, 60, 62) and the correspondence of the samples’ retention times with those of the standard provided convincing evidence that Ni(CO)4 is present in the fermentation gas.The concentrations found were in the sub-ppb level, which is at least one order of magnitude lower than the threshold level of 1 ppb (v/v). In addition, Mo(CO)6 and W(CO)6 were also measured in the sub-ppb range in contrast to the absence of Fe(CO)5. The stabilities of Ni(CO)4, Fe(CO)5, and Mo(CO)6 were tested in a carbon monoxide atmosphere.In the presence of distilled water, the following order of stability was found after 11 weeks: Fe(CO)5<Ni(CO)4<Mo(CO)6. In the presence of an aqueous solution containing nickel, molybdenum, tungsten and iron, however, only Fe(CO)5 was significantly decomposed (<0.3% recovery); Ni(CO)4 and Mo(CO)6 were stable after 11 weeks.No W(CO)6 was formed. The low stability of Fe(CO)5 in the presence of water could be the reason why no volatile iron compound was found in sewage gas. This study showed that GC-ICP-MS can be employed to identify species-specific traces of metal carbonyls in process gases such as sewage gas. The analytical method used should be able to measure very Aim of investigations small concentrations of volatile nickel compounds, should Nickel tetracarbonyl is used in the manufacture of catalysts, give information about the species, and should not be limited in nickel vapour plating, as an intermediate in nickel refining by other volatile organic or organometallic compounds (the Mond process), and in the manufacture of high purity present in sewage gas.DiVerent methods have been developed nickel powder. Nickel carbonyl is one of the most dangerous in the past for continuous or spot measurements, and chemicals known, because of its lipid solubility and volatility detection limits for the following analytical methods have (bp 43 °C). It can easily be inhaled and absorbed by traversing been reported for nickel tetracarbonyl: non-specific chemithe pulmonary alveolar membranes and the blood–brain bar- luminescence, 0.01 ppb (0.07 mg m-3);4 specific for Ni(CO)4, rier.1 It exhibits acute toxicity, as well as carcinogenicity and 1 ppb (7 mg m-3);5 and as a portable system, 0.13 ppb teratogenicity.Therefore very low threshold levels are set by (0.9 mg m-3);6 mass spectrometry, 10 ppb (70 mg m-3);7 occupational health and safety bodies in diVerent countries infrared spectrophotometry, 10 ppb (70 mg m-3).8 The (e.g., US limit value (time-weighted average, TWA), 0.12 mg absorption of volatile nickel compounds into a graphite Ni m-3 (50 ppb); French valeur moyenne d’exposition (VME) furnace with subsequent detection using atomic absorption limit, 0.12 mg m-3; and Swedish level limit, 0.007 mg m-3 spectrometry is not species-specific, but has excellent detection (1 ppb)). limits because the detection system is element-specific [0.5 ng Nickel tetracarbonyl is produced by passing carbon monox- absolute mass, 10-3 ppb (0.007 mg m-3)],9 and an on-line ide over finely divided nickel.However, it is also known that preconcentration method for trapping nickel as nickel carit can be generated spontaneously in unexpected environments bonyl can detect 0.09 ng nickel as absolute mass.10 However, whenever carbon monoxide comes into contact with an active these methods are not able to separate and identify diVerent form of nickel.2 Nickel is concentrated in sewage, and values volatile nickel compounds. It was shown earlier11 that gas up to 100 mg nickel per kg of raw sewage sludge are normal.3 chromatography can be used to separate diVerent volatile After the fermentation process of this raw sludge, the nickel compounds, and therefore a combination of elementconcentration of heavy metals in the digested sludge is usually specific detection and chromatographic separation would give higher, but not for nickel (82 mg kg-1).2 Nickel occurs species information as well as excellent detection limits.mostly as organo-nickel complexes in the sewage and will be Previously described cryotrapping GC-ICP-MS was used to transformed into nickel sulfide during the fermentation identify volatile molybdenum and tungsten species as their process. However, other microbial transformations of nickel hexacarbonyls in landfill gas.Thus, the aim of this study was to identify volatile nickel metabolites, such as Ni(CO)4, in cannot be excluded. J. Environ. Monit., 1999, 1, 33–37 33sewage gases by applying this analytical technique for trace through a heated Teflon transfer line (0.3 mm od, 120 °C) to the torch of the ICP-MS (PQ2, VG Elemental, Winsford, amounts of volatile nickel compounds.UK). In addition, an aqueous solution containing 10 ng mL-1 Rh as a continuous internal standard was introduced as a wet Description of experimental procedures aerosol into the plasma using a Meinhard nebulizer. Rh can Reagents be used to monitor the stability of the ICP-MS, when gas samples are introduced into the plasma, so that ghost peaks Mo(CO)6 and W(CO)6 (both purchased from Pressure can easily be identified.Both gas flows were mixed together Chemicals, Pittsburgh, PA, USA) were used for the standard in a tee-piece (6 mm od) inserted between the spray chamber addition method to quantify the concentration of these comand the torch, replacing the quartz elbow usually in this pounds in sewage gas as well as in carbon monoxide.A 10 L position. The cryotrapping-cryofocusing GC-ICP-MS unit is gas cylinder of carbon monoxide containing Ni(CO)4, shown in Fig. 1. A thorough description of the method has Fe(CO)5, and Mo(CO)6 at unknown concentrations was used been published elsewhere.13 The operating parameters were as the diluted standard for identification purposes. NiBr2 the same as used for liquid samples (cool gas flow, 13 L min-1; (K&K Laboratories, Hollywood, CA, USA), FeSO4.7H2O auxiliary gas flow, 0.7 L min-1; nebulizer gas flow, (Fisher Scientific, NJ, USA), (NH4)6Mo7O24·4H2O, and 1.0 L min-1; power, 1350W).In a screening analysis, the WOCl4 (both Alfa Inorg. Ventron, Beverly, MA, USA) were mass range between m/z 48 to 200 was scanned using 320 ms used for the stability test.as the dwell time per channel. The following isotopes were detected in the peak hopping mode (20 ms dwell time per Sampling site channel ) for the qualitative determination of the isotopic Gas samples were collected from a municipal sewage treatment fingerprint of the gas samples: 50Cr, 52Cr, 54Cr/Fe, 55Mn, 57Fe, plant equipped with a mesophilic sewage sludge digester (Iona 58Ni, 59Co, 60Ni, 61Ni, 62Ni, 64Ni, 92Mo, 94Mo, 95Mo, 96Mo, Beach, Vancouver, Canada).Inside the pumping house, gas 97Mo, 98Mo, 100Mo, 103Rh, 180W, 182W, 183W, 184W, and 186W. pipes were connected to the top of the fermenters, from which For quantification purposes only 54Fe, 57Fe, 58Ni, 60Ni, 98Mo, gas samples containing methane and carbon dioxide were 103Rh, and 184W were measured.collected. Evidence for some leakage in the pipes was recognized by the characteristic smellf the fermentation gas in the Stability test pumping house. For this study, 10 samples were collected. Two 400 mL Schlenk flasks closed oV with a silicone septum Sampling procedure and preconcentration were filled with carbon monoxide containing Ni(CO)4, Fe(CO)5, and Mo(CO)6.One flask was filled with 100 mL The Tedlar bags, equipped with a valve, were connected with deionized water. The other flask was filled with an aqueous polypropylene (PP) tubing to a small bypass valve in the metal solution (Me-sol ) of NiBr2 (107 mg Ni L-1), pipes. The pressure in the pipe was suYcient to fill the bags FeSO4·7H2O (54 mg Fe L-1), (NH4)6Mo7O24·4H2O (123 mg with a flow rate of about 5 L min-1.The gases were collected Mo L-1), and WOCl4 (125 mg W L-1). Both flasks were directly into 80 L Tedlar bags, which had been checked for purged and filled with CO again and stirred in the dark at blank levels. The temperature of the gas was approximately room temperature for 11 weeks using a magnetic stirrer. Only 33 °C. The bags were placed in a black plastic bag to prevent 1 mL of headspace gas was withdrawn from each flask after 4 UV light from entering and stored at 4 °C.The gas samples weeks and 11 weeks using a gas tight syringe. The gases were were cryogenically preconcentrated by trapping the gases on directly injected onto the cryogenic column through a septum the first U-shaped trap filled with Chromosorb (10% SP-2100 port.The 4 week samples were used to check for leakage by 60–80 mesh, Supelco) at -78 °C (dry ice/acetone slush). This the determination of nitrogen; no leaks were observed. relatively high temperature was chosen to avoid condensation of carbon dioxide and methane, the major components of sewage gas. In a cryofocusing step, the volatile species were Results and discussion volatilized by increasing the temperature of the trap from When carbon monoxide is stored in steel gas cylinders, iron -78 °C to 150 °C with a flow rate of 133 mL min-1 He.The pentacarbonyl can be formed. Steel contains many other transfer line was heated up to 120 °C by a nichrome wire. The metals, e.g., nickel and molybdenum, so that the formation of released gases were frozen ( liquid nitrogen) onto a second Ucarbonyl compounds other than that of iron can be assumed. shaped trap (6 mm od, 31 cm length), which was packed with CO from a steel gas cylinder was sampled in a Tedlar bag.Chromosorb (10% SP-2100 45–60 mesh, 6 mm od, 31 cm Only 1 mL of CO gas was injected onto the GC column, length, Supelco, Bellefonte, PA, USA) and cooled with liquid cooled at -78 °C.After separation, signals at m/z 58, 60, 61, nitrogen. The trapping eYciency was tested for non-polar organometallic substances (SnH4 bp: -52 °C, 0.1%; Me2SnH2 bp: 35 °C, 87%; n-BuSnH3 bp: 100 °C, 100%) which are easy to handle.12 A trapping eYciency of more than 90% can be assumed for non-polar compounds with a boiling point higher than 40 °C. Analytical procedure No clean-up procedure or derivatization was performed on the gas samples in order to avoid a change in the molecular structure of volatile metal compounds.The analytical procedure applied was a combination of thermodesorption of the cryotrapped sample and separation using a non-polar chromatographic column. The column was heated exponentially from -196 to 150 °C within 3 min and the gases separated using Fig. 1 The set-up of the cryotrapping-cryofocusing GC-ICP-MS unit. an He flow of 133 mL min-1. The reproducibility of this The transfer line, tubings and traps were electrically heated by a procedure is so good that the variation of the retention times nichrome wire. A, cryofocusing; B, measurements and simultaneous cryotrapping of a new sample. was less than 3 s.The separated sample was transported 34 J. Environ. Monit., 1999, 1, 33–3762, and 64 were recorded at a retention time of 60 s. The intensities of all the peaks can be fitted to the relative abundance of naturally occurring nickel. At 110 s, signals at m/z 54, 57, and 58 were recorded. 56Fe was not measured, because of the high background from 40Ar16O+ in the plasma.However, the isotopic fingerprint fits the isotope pattern of iron. At 170 s, all masses from molybdenum recorded signals according to the natural isotopic fingerprint. The metal carbonyls were separated according to their extrapolated boiling points [Ni(CO)4, log p=7.690-1519/T , bp 42.7 °C; Fe(CO)5, log p=8.3098 – 2050.7/T , bp 104.7 °C; Mo(CO)6, log p= 11.174 – 3561.3/T , bp 156.3 °C; W(CO)6, log p=11.523 – 3872/T , bp 174.9 °C; p is the partial pressure in mmHg and T is the temperature in °C].The identification of the metal species is based on the chromatographic separation and on the element-specific detection (ICP-MS). The chromatographic separation showed a linear relationship (R2=0.9832) between the boiling points or the extrapolated boiling points of volatile standards (Ni(CO)4, Fe(CO)5,Mo(CO)6, W(CO)6) and their retention times [bp=1.048·r.t.-22.19; bp, boiling point (°C); r.t., retention time (s)].The separation for the neutral metal carbonyls shows the same separation characteristics as for neutral methylated metal and metalloid species, such as SnH4, Me2 SnH2, Me4 Sn, AsH3, Me3 As, etc.). This relationship can be used to identify unknown peaks according to their element-specific detection and their correlated boiling point.Mo(CO)6 standard gave the same retention time as the unknown Mo peak (170 s). Using standard addition methodology, 0.14–0.19 ppb (v/v) (0.6–0.8 mg Mom-3) was determined to be the concentration of Mo(CO)6 in the CO. Nickel and iron carbonyls were quantified by a semi-quantitative Fig. 2 The detection of 300 ng Fe(CO)5, 0.14 ng Ni(CO)4, and 0.6 pg Mo(CO)6 in carbon monoxide stored in a gas cylinder. method using aqueous nickel and iron standards. It was shown earlier14 that this method can produce an uncertainty of about ±30% for volatile metal( loid) compounds. The amounts of isotopes matched the isotope ratios of the naturally occurring Ni(CO)4 and Fe(CO)5 were determined to be 49–54 ppb molybdenum and tungsten used as the standards, as shown (0.13–0.14 mg Ni m-3) and 80–148 ppm (200–370 mg earlier for landfill gas.16 The peak at the retention time of 60 s Fe m-3).The chromatograms measured simultaneously are showed matching at m/z 58, 60, 61, and 62. The occurrence shown in Fig. 2. The detection limit for a volatile molybdenum of a volatile sulfur compound with a boiling point of about compound was below 0.1 pg when 3s of the baseline was 40–45 °C could cause the high intensity at m/z 64 by forming considered.Nickel and especially iron were very much aVected 32S32S+ and/or 32S16O2+ (Fig. 4). The amount of Ni(CO)4 in by molecular interferences in the plasma of the ICP-MS. the sewage gas could only be assumed using the semi- According to the semi-quantification method, the detection quantitative method to be between concentrations of 0.5 and limits can be assumed to be two orders of magnitude higher 1.0 mg Ni(CO)4 m-3 (0.07–0.14 ppb).The amount of than the detection limits of Mo. Conservative detection limits Mo(CO)6 was slightly higher (3.0–3.6 mg Mo(CO)6 m-3; of 10 pg Ni absolute could be assumed for volatile nickel and 0.25–0.30 ppb) and W(CO)6 was determined in the 100 pg Fe for volatile iron compounds.If only 1 L of gas is 0.01–0.015 mg W(CO)6 m-3 range (0.0006–0.001 ppb). How- sampled, a detection limit of 0.01 mg m-3 (approximately ever, no volatile species of iron, chromium, cobalt, and manga- 0.004 ppb) can be calculated. Since detection limits depend on nese could be detected. the absolute mass of volatile nickel, the lowest concentration While the tungsten concentration found was similar to the in a gas sample can easily be reduced to 0.0004 ppb when the concentration in landfill gas, the concentration of volatile gas volume has been increased to 10 L.These detection limits molybdenum in the sewage gas was one order of magnitude would be comparable to those reported earlier14 for other higher than that reported in landfill gas.16 The nickel concen- volatile metals (Sn, Hg, Pb, Bi) and metalloids (As, Se, Sb, trations were one order of magnitude lower than the threshold Te) as published by Donard and co-workers.15 These detection level of 7 mg m-3 [1 ppb (v/v)]. However, the concentrations limits for the speciation of nickel are among the lowest ever measured (0.5–1.0 mg m-3) are three orders of magnitude reported and could be used to determine volatile nickel, iron, higher than the reported concentrations of volatile nickel molybdenum, chromium, and tungsten compounds in unexpeccompounds, assumed to be Ni(CO)4, in urban air ted environments such as sewage treatment plants.(0.00014–0.0048 mg m-3).17 Since these gases were leaking into The sewage gas samples were preconcentrated and analysed workplace air, it would be necessary to monitor workplace air within 4 h after sampling at the sewage treatment plant.Up for these compounds. However, these data show that the to 1 L of sewage gas was preconcentrated onto the column, concentration measured in the sewage gas will be negligible which was cooled at -78 °C.The identification of the nickel when it undergoes fast dilution in workplace air. Stability tests carbonyls was rather diYcult because the sewage gas contains for these volatile carbonyl complexes were performed in order a lot of other organic and organometallic compounds, which to shed some light on their stability in the presence of distilled can form molecular clusters in the plasma of the ICP-MS.water and a metal containing solution (Me-sol ) under oxygen- However, signals for tungsten, molybdenum, and nickel were free and dark conditions. The headspaces of the flasks, which recorded. The traces for m/z 58, 60, and 62 are shown in contained the metal carbonyls, were sampled after 11 weeks Fig. 3. Iron pentacarbonyl could not be detected in the gas. The fingerprint for the diVerent molybdenum and tungsten and analysed for volatile Ni, Fe, Mo, and W compounds. The J. Environ. Monit., 1999, 1, 33–37 35with the metal solution, and no Fe(CO)5 was detectable in the presence of pure water. The Ni(CO)4 concentration did not decrease in the presence of the metal solution; in fact, a slightly but not significantly higher concentration was determined for Ni(CO)4.In contrast, only 0.7% of the Ni(CO)4 could be recovered in the headspace after 11 weeks in distilled water. The amount of Mo(CO)6 remaining in both flasks was the same as the initial concentration in respect of the error of ±30% during the experiment. No W(CO)6 was formed in these experiments. The results of these stability tests indicate that iron pentacarbonyl is not stable in the presence of water, which might be the reason why no volatile iron compound was determined in the sewage gas.It seems that Mo(CO)6 was more stable under these conditions in comparison to the nickel and iron compounds. Nickel carbonyl seemed to be stable in the presence of a nickel solution in a CO atmosphere, and less stable if no nickel was present in the aqueous phase. It is well known that Ni(CO)4 can be synthesized if nickel and sulfide are present in aqueous solution.18 As an intermediate, the sulfide will be oxidized to elemental sulfur and immediately reduced by CO to sulfide again, and thus sulfide acts as a catalyst.Under 1 bar CO, NiS in water has been reported to be quantitatively transformed to Ni(CO)4.18 So far nothing is known about the transformation of iron and molybdenum into their volatile neutral carbonyls in aqueous solution.As the conditions in the fermentation tank of a sewage plant are reducing and free sulfide is available, a purely chemical reaction could be responsible for the generation of Ni(CO)4 at low temperatures (35 °C). Nickel as well as molybdenum and tungsten can be reduced by a suitable reducing agent (e.g., hydrogen produced by the micro-organisms in the fermentation process) and the resulting active metal can directly react with carbon monoxide. However, it cannot be excluded that microbial transformations of insoluble or soluble nickel compounds may be responsible Fig. 3 Detection of Ni(CO)4 in 100 mL sewage gas measured at m/z for the generation of the volatile nickel carbonyl in the 58, 60, and 62 using GC-ICP-MS.The first peak at m/z 58 might not fermentation process. Furthermore, nickel and molybdenum be a nickel signal if the isotope ratio is considered. are essential elements for methanogens, e.g., the nickel tetrapyrrole prosthetic group of methyl CoM reductase is an essential enzyme of methanogenesis.In fact, one of the Ni ions in CO dehydrogenase appears to bind CO and a methyl group. If these enzymes break down, a CO transfer to the released metal is not unlikely. This study showed that the GC-ICP-MS method can be employed to identify the occurrence of volatile nickel compounds in sewage gas. The stability tests shed some light on why no iron pentacarbonyl can be detected in landfill and sewage gas.However, how nickel tetracarbonyl is formed, i.e., in a purely chemical reaction or by the action of microbial Fig. 4 Isotopic fingerprint of the signal at 60 s retention time by transformation, cannot yet be answered. determination of Ni(CO)4 in sewage gas using GC-ICP-MS (standard: Ni(CO)4 in CO gas cylinder).Acknowledgements quantified concentrations in Table 1 were compared to the initial concentrations of the carbonyls in CO. The recovery I gratefully thank Prof. F. W. Sunderman Jr. for useful discussions. I would also like to address my special thanks to rate for Fe(CO)5 was determined to be only 0.3% of the concentration of Fe(CO)5 in CO before the flask was filled Prof.W. R. Cullen, who gave me the opportunity to be part Table 1 Metal carbonyls in carbon monoxide gas stored dry in metal gas cylinders, and their concentration after 11 weeks in contact with distilled water and a metal containing solution (Me-sol ) in glass flasks. The concentrations measured in sewage gas and landfill gas are also shown. Concentrations are given in ppb (v/v) Medium Fe(CO)5 Ni(CO)4 Mo(CO)6 W(CO)6 CO (cylinder) 80000–148000 49–54 0.14–0.19 n.d.CO (Me-sol )a 340 60 0.3 n.d. CO (dist. water)a n.d. 0.35 0.2 n.d. Sewage gas n.d. 0.07–0.14 0.25–0.30 0.0006–0.001 Landfill gas16 n.d. n.d. 0.047–0.071 0.0006–0.0012 aAfter 11 weeks; n.d., not detected. 36 J. Environ. Monit., 1999, 1, 33–377 J. H. Campana and T. H. Risby, Anal. Chem., 1980, 52, 468.of his research group and for his support and interest in this 8 R. S. McDowell, Am. Ind. Hyg. Assoc. J., 1971, 32, 621. work. Finally, I thank the Iona Beach Sewage Treatment Plant 9 L. Filkova and J. Ja�ger, J. Chem. Listy, 1986, 80, 1207. for the opportunity to sample the sewage gas. 10 D. Erber and K. Cammann, Analyst, 1995, 120, 2699. 11 F. W. Sunderman, Jr., Arch. Environ. Health, 1968, 16, 836. 12 J. Feldmann, PhD Thesis, Cuvillier, Go� ttingen, Germany, 1995. References 13 J. Feldmann and W. R. Cullen, Environ. Sci. Technol., 1997, 31, 2125. 1 F. W. Sunderman, Jr. and A. Oskarsson, in Metals and Their 14 J. Feldmann, J. Anal. At. Spectrom., 1997, 12, 1069. Compounds in the Environment, ed. E. Merian, VCH, Weinheim, 15 C. Pe�cheyran, C. R. Quetel, F. M. Matin Lecuyer and O. F. X. 1991, pp. 1101–1126. Donard, Anal. Chem., 1998, 70, 2639. 2 S. Hetland, I. Martinsen, B. Radziuk and Y. Thomassen, Anal. 16 J. Feldmann and W. R. Cullen, Environ. Sci. Technol., 1997, 31, Sci., 1991, 7, 1029. 2125. 3 R. C. Kistler, F. Widmer and P. H. Brunner, Environ. Sci. 17 L. Filkova and J. Ja�ger, C. Hygenia, 1986, 31, 255. Technol., 1987, 21, 704. 18 H. Behrens and E. Eisenmann, Z. Anorg. Allg. Chem., 1955, 278, 4 D. H. Stedman and D. A. Tammaro, Anal. Letters, 1976, 9, 81. 155. 5 P. M. Houpt, A. Van Der Waal and F. Langeweg, Anal. Chimica Acta, 1982, 136, 421. 6 D. A. Hikade, D. H. Stedman and J. G. Walega, Anal. Chem., 1984, 56, 1629. Paper 8/07277I J. Environ. Monit., 1999, 1, 33&nda