Continuous Environmental Monitoring of Nickel ... - ACS Publications

---

Continuous Environmental Monitoring of Nickel Carbonyl by Fourier Transform Infrared Spectrometry and Plasma Chromatography William M. Watson’ Rohm and Haas Company, Spring House, Pa. 19477

Carl F. Kohler Rohm and Haas Texas, Inc., Deer Park, Tex. 77536

Nickel carbonyl was successfully monitored in a plant environment by Fourier transform infrared spectrometry (FTIR) and plasma chromatography (PC). Instrumentation based on both techniques proved capable of automatic, continuous monitoring of sub-one-part-per-billion (ppb) levels. T h e practical limits of detection were estimated a t about 0.2 to 0.3 ppb, the precision at f 0 . 2 ppb, and the accuracy a t f0.5 ppb a t the 1-ppb level in the environmental air of a process control room. Good correlation between the two instruments in side-by-side testing implied that the recommended wet chemical method gave both spuriously high and spuriously low monitoring data. Nickel carbonyl is an important but very toxic intermediate in the synthesis of acrylate monomers. For many years we monitored ambient plant air by wet chemical and instrumental (gas ionization cell) techniques; in addition, we conducted an extensive personnel monitoring program for excreted nickel. Evidence that nickel carbonyl, or a metabolite, might also be a source of cancer in rats prompted the establishment of an OSHA TLV of 1 ppb for nickel carbonyl ( I ) . This is the lowest value for an environmental air standard established t o date and posed a substantial monitoring challenge. Existing analytical methods for nickel carbonyl could not be made sensitive enough for application as rapid on-line monitors a t the 1-ppb level. The wet chemical method ( I ) requires the concentration of a t least 3 yg of Ni(C0)4 into an oxidizing solution. Over 400 L of air must therefore be sampled to reach even 1-ppb sensitivity, disqualifying the technique for rapid monitoring. The gas ionization instrument (“Billionaire”, Mine Safety Appliances, Pittsburgh, Pa.) is based on observing the change in ion conductivity across a sample cell that is irradiated with low-level cy radiation. Despite extensive efforts, the base line could not be stabilized below a response equivalent to about 50 ppb and no preconcentration technique was apparent. Fourier transform infrared absorption spectrometry, or, alternatively, plasma chromatography seemed practical routes to solving the continuous monitoring problem. The instruments envisioned would (a) monitor the nickel carbonyl concentration of ambient air in a process control room, (b) provide a permanent record of those measurements, and (c) trigger an alarm system if a predetermined excursion concentration were exceeded. They should (d) be sensitive to a fraction of a part per billion of nickel carbonyl, (e) be selective against known interferences a t that level, and (f) remain calibrated with a minimum of intervention by trained personnel. Both systems were found to meet these criteria.

Instrumentation Fourier Transform Infrared Spectrometry. Our instrument (2, 3 ) is a modified version of the Eocom System 7400, diagrammed in Figure 1. The infrared radiation is provided by a conventional globar source. The Michelson interferometer has 2-in. throughput optics. The 8-cm throw and feedback-stabilized motion of the movable mirror provide an 0013-936X/79/0913-1241$01.00/0

@ 1979 American Chemical Society

SAMPLE CELL I 4 2 m.1

.L I

FOUR1 ER TRANSFORM

+

POP I1 COMPUTER

~

L-J

I Figure 1. Functional diagram of FTIR monitor

FTIR resolution of 0.065 wavenumber. At this resolution the spectral detail of even the smallest molecules is limited by collision broadening a t atmospheric pressure. A single scan of the movable mirror takes approximately 40 s. The beamsplitter is CaF2. The folded path length of the White cell is 42 m. The cooled PbSe detector and the signal electronics are conventional. A minicomputer accomplishes the Fourier transform in core, without the benefit of a disk. For the sensitivity we desired, 8 scans were accumulated and transformed to provide one concentration read-out about every 6.6 min. Eight-hour averages were also calculated and recorded. An experimental sample with less than 1 ppb of nickel carbonyl has an extraordinarily small absorbance even though the molar absorptivity of the observed band is respectable. Hence, electronic stability and path length were a t a relative premium. Resolution, on the other hand, was not a problem once interferometry was adopted. An interferometrically produced absorption scan of the 5 pm region for nickel carbonyl in the presence of carbon monoxide is shown in Figure 2. Carbon monoxide a t 10 000 times higher concentration is present in the nickel carbonyl standard to stabilize it. The antisymmetric C=O stretching mode of Ni(CO)4 is a broad band centered a t 2058 cm-l(4). The sharp bands are the individual vibrational-rotational lines of CO. The 0.1-wavenumber resolution of the generated spectrum is sufficient to resolve parts of the broad nickel carbonyl band from the interspersed CO lines. One such 0.1-wavenumber channel, point B in the nickel carbonyl band, is compared to two background channels, A and C, where no interferences in the ambient air occurred. Some long-term drift in the three reference points does occur and was accounted for by rezeroing against tank nitrogen every 8 h. The absorptivity used in converting band intensity to actual concentration was determined on the instrument itself by introducing known concentrations of nickel carbonyl into the sample cell. Volume 13, Number 10, October 1979 1241

C

1

1

I . . . , l . , , , l . . . . l . . . , I , . . I. , , , , ! I * . , 2040

2010

cm-I

'2Ow

2070

Figure 2. FTIR absorption spectrum of 75 ppb of Ni(C0)4 in 750 ppm

of

I

co

15.00 ION DRIFT TIME

(m*)

Figure 4. Negative ion plasmagram of 2.2 ppb of Ni(C0)4 in air. Operating conditions: cell temp 52 O C ; cell voltage, -4000 V; inlet flow, 200 mllmin; drift gas flow, 500 mL/min air; ion gate, 0.2 ms

INTEGRATOR

e PRINT-OUT [PLPRM

CHART RECORDER

BOX-CAR INTEGRATOR

I30 min

I 1

1

Figure 3. Functional diagram of PC monitor 3 2 . 5 ppb

Plasma Chromatography. Our plasma chromatography monitor is a Franklin GNO (now PCP, Inc., West Palm Beach, Fla.) plasma chromatograph with an alarm system appended, as diagrammed in Figure 3. The principle of operation includes the ionization of the sample a t atmospheric pressure and the separation of the resulting ions by their respective masses (5, 6). The sample is introduced into an ionization chamber where the carrier gas is ionized by low-energy p radiation and subsequent collisions ionize the sample molecules. The stabilized ions are repetitively gated into an electrostatic field where ions of charge opposite to the far end of the field will traverse the field within a few milliseconds. Since each will gain the same total energy, the ions of different mass will pick up different velocities and will arrive a t the far end separated in time by mass. When the instrument is operated in the negative ion mode, nickel carbonyl shows up as a massive ion well separated from the other negative ions generated from environmental air (see Figure 4).This ion is believed to be the molecular parent ion, Ni(C0)4- (7). In the monitor the nickel carbonyl signal from each opening of the gate is signal averaged by a box-car integrator to obtain a sensitivity of 0.02 ppb in 12 s. Eight-hour averages are provided by an integrator. Calibration The accuracy of the calibration a t low concentration is obviously of prime importance, and several experiments were undertaken to prove the validity of our determinations. A typical experiment on the FTIR instrument is shown in Figure 5. A 32-ppb nickel carbonyl standard was produced by a dynamic dilution system and flushed continuously into the sample cell. The monitored concentration after equilibrating the cell is 32.5 ppb. A further dilution of 1O:l yielded an average of approximately 3.2 ppb, and a further 1:l dilution brought about a 1.8-ppb reading. Physical limitations in the dynamic dilution system prevented the production of lower 1242

Environmental Science & Technology

Figure 5. A calibration experiment on the FTIR instrument, 32.5 to 1.8

ppb of Ni(C0)4

....- pc

-F T I R

--- CHEMICAL

Figure 6. Comparative daily averages of Ni(COk concentration in a process control room as monitored by FTIR, PC, and wet chemical method

level gas standards. The plasma chromatograph was initially calibrated by a different dynamic dilution system consisting of a 5 cm i.d. tube, through which ambient air was driven a t 3300 L/min. A 9-ppm source of nickel carbonyl was introduced through a small tube in the side of the main dilution tube a t known flow rates. Downstream 200 mL/min of the air stream was removed via a second small tube and pulled through the PC by its internal pump. Baffles in the main tube assured adequate mixing of the spiked air stream. A calibration curve was established using concentrations as low as 1.1ppb. At this level the response remained within f0.12 ppb for 1-h intervals and f 0 . 2 ppb for a 24-h period. We were most interested in nickel carbonyl monitoring in the concentration range 0.1 to 1 ppb. Directly demonstrating

c) c)

In-

C

II

a m

-

I

FTIR

I

.I

.2

,. 3

A

MONTHLY

AVERAGE

( ppb)

Figure 7. Frequency distribution of the monthly averages of Ni(C0)d monitoring data obtained by FTlR and PC

the absolute accuracy of the instruments below 1 p’pb has proved to be quite a challenge because we have had difficulty creating reproducible nickel carbonyl concentrations at such dilutions. Both instruments, however, show a linear response down to this level. Because nickel carbonyl is notoriously unstable, the ambient air sampled is merely filtered to keep out particulate matter and no further sample preparation is attempted.

Continuous Monitoring Experience In plant process control rooms, both instruments proved quite capable of operating unattended for extended periods. The FTIR instrument required field service from the manufacturer on an average of once every 6 months. Most other malfunctions, requiring intervention by laboratory personnel, involved computer problems or the effects of power failures. We have realigned the optical paths periodically, although no great changes have been observed. In 3 years of monitoring, the PC required only one field trip by the manufacturer. However, it was determined that in actual plant use the PC must be baked out every few weeks to eliminate adsorbed contaminants. Interferences from specific compounds that might also be present in the control room air have been investigated. No problem situations have been encountered. Both instruments were initially placed in the same process control room. For 3 months we compared daily averages obtained by FTIR, PC, and a wet chemical method. The chemical method involves scrubbing ambient air through an oxidizing solution and determining the concentration of Ni2+ ion by a colorimetric technique (1,8).Figure 6 shows a comparison of the results for this period. The correlation of the daily averages produced by the two instrumental techniques is remarkably good for realistic monitoring at the 1-ppb level. The instrumental data also reinforce the notion that the traditional colorimetric method gives both spuriously high and spuriously low results for workday exposures. The wet chemical method was abandoned after several months of additional comparison without resolving this problem. More importantly, this comparison indicates that our monitors give credible 8-h averages for compliance records.

We estimate that the practical limits of detection of our instruments are about 0.2 to 0.3 ppb and that the precision is about f 0 . 2 ppb and the accuracy f 0 . 5 ppb at the 1-ppb level in actually monitoring the environmental air of a process control room. After the side-by-side evaluation of the three techniques, the two instruments were operated in the control rooms of two separate process areas. The monthly averages of each 8-h exposure interval for each instrument are displayed as a frequency distribution in Figure 7 . The grand ensemble average is 0.43 ppb ( s = 0.19, n =70), which is roughly equivalent to our estimate of the accuracy of the 8-h data. These instruments also act as “continuous” monitors, which give rapid warning of any potentially serious excursions. In this function they also provide a second line of defense against process accidents that might escape the notice of the less sensitive gas ionization instrument located within the process area itself. In summary, we believe that the effort to apply Fourier transform infrared spectrometric and plasma chromatographic techniques to the problem of continuous ambient air monitoring for a species at the 1-ppb level has been very successful. Our monitors for nickel carbonyl are quite capable of automatic, continuous analysis, producing sub-1-ppb determinations that are no more than a few minutes old.

Literature Cited (1) Arthur, J. L., “NIOSH Special Occupational Hazard Review and Control Recommendations for Nickel Carbonyl”, NIOSH Report No. 77-184. N.T.I.S. Publication PB-273 795. 1977. (2) Gillespie, R. E., paper presented at the F.A.C.S.S. Meeting, Atlantic City, Nov 22, 1974. (3) Kohler, C. F., Watson, W. M., paper presented a t the Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Cleveland. March 4. 1975. PaDer No. 181F. (4) McDowell, R. S., A m . Ind. H y g . Assoc. J., 32,621-4 (1971). (5) Cram, S. P., Chesler, S. N., J . Chromatogr. Sci., 11,391 (1973). (6) Karasek, F. W., A d . Chem., 46,710A (1974). (7) Wernlund, R. F., Cohen, M. J., paper presented at the Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Cleveland, March 3, 1975, Paper No. 105. ( 8 ) Brief, R. S., Venable, F. S., Ajemian, R. S., A m . lnd. Hyg. Assoc. J., 2 6 , 7 2 (1965). Received /or review December 8, 1978. Accepted J u n e 15,1979.

Volume 13, Number 10, October 1979

1243