Precision and Detection Limit. The precision of the method was evaluated by replicate analysis of samples ranging from 0.25 to 3.0 ppm nitrite. The relative standard deviation for ten replicates is 1.2% a t 0.25 ppm and decreases to 0.4% at 3.0 ppm nitrite. The detection limit of the automated analysis at 100% conversion of nitrate to nitrite is 17 ppb nitrate as determined by the method of Hubaux and Vos (28). This detection limit is determined by flow system generated noise, not by the photometer, and is primarily associated with high acid concentration in the sulfanilamide reagent. In a quiescent system, with the photometer noise the limiting factor, a detection limiting of 2 ppb nitrate is estimated. CONCLUSIONS The foregoing discussion indicates that the enzymatic procedure represents a significant improvement in the analytical methodology applicable to nitrate analysis. The detection limit with a continuous flow analytical system is improved by a factor of 3 from that obtained with the recommended EPA procedure (13).Adaptation of the analytical scheme to a nonflow system, where photometer noise is the limiting factor, can improve the detection limit an additional factor of 8. The ionic strength of the reaction media suggests that the method can be directly translated to sea water samples where the nitrate content is typically less than 50 ppb (29). In fact, early studies with a soluble enzyme preparation resulted in only a 20% linear decrease in enzyme activity as the ionic strength of the reaction media increased from 0.1 to 1.0 M (25). Studies with industrial effluent samples indicate excellent freedom from interferences. No evidence for reduction of nitrite was observed. This results directly from the fact that nitrate reductase induced by anaerobic growth of E. coli is a respiratory enzyme and is only involved in the terminal electron-acceptor step of respiration. Although the reported procedure utilized an immobilized enzyme, the method can be adopted to a soluble enzyme preparation. Because commercial preparations of nitrate reductase are rather expensive, it cannot be used in large amounts to provide complete conversion in a continuous flow analyzer with reasonable length reaction coils. With a nonflow analytical scheme lower enzyme concentrations and longer reaction times can achieve 100% conversion of the substrate to product. Thus, to attain the detection limit reported for the immobilized enzyme with soluble nitrate reductase a nonflow analytical scheme is the only practical approach.
ACKNOWLEDGMENT The use of facilities a t the University of Georgia's Fermentation Laboratory is gratefully acknowledged.
LITERATURE CITED (1) A. Nason, "The Enzymes", 2nd ed, Vol. 7, P. D. Bayer, H. Lardy, and K. Myrback, Ed., Academic Press, New York. N.Y., 1963, pp 587-607. (2) T. R. Camp, "Water and Its Impurities", Reinhold, New York. N.Y., 1963, Chapter 5. (3) D. R. Kenney, J. G. Konrad, and 0. Chester, J. Water Pollut. Control Fed.. 42. 411 11970). (4) F.-L.' Fisher, E. R. ibert, and H. F. Beckman, Anal. Chem., 30, 1972 (1958). (5) "Standard Methods For The Examination of Water and Wastewater", 12th ed. American Public Health Association, inc., New York, N.Y., 1965, pp 195-205. (6) C.A. Noli, lnd. Eng. Chem.,Anal.€d., 3, 311 (1931). (7) M. Toknoka, Collect. Czech. Chem. Commun., 4, 4444 (1932). (8) I. M. Kolthoff, W. E. Harris, and G. Matsuyma, J. Am. Chem. Soc., 66, 1782 (1944). (9) J. Romirez-Munoz, Anal. Chim. Acta, 72, 437 (1974). (10) J. B. Mullin and J. P. Riley, Anal. Chim. Acta, 12, 464 (1955). (11) J. D. H. Strickland and K. H. Austin, J. Cons., Cons. lnt. Explor. Mer., 24, 446 (1959). (12) C. R. Sawicki and F. P. Scaringelli, Microchem. J. 16, 657 (1971). (131 "Methods For Chemical Analysis of Water and Wastes", US. Environmental Protection Agency, Office of Technological Transfer, Washington, D. C., 1974, p 201. (14) G. B. Garner, J. S. Baumstark, M. E. Muhrer, and W. H. Pfander. Anal. Chem., 28, 1589 (1956). (15) R. H. Loweand J. L. Hamilton, J. Agric. FoodChem., 15, 359 (1967). (16) A. L. McNamara, G. R. Meeker, P. D. Shaw, and R. H. Hageman, J. Agric. Food Chem., 19, 229 (1971). (17) R. H. Lowe and M. C. Gillespie, J. Agric. FoodChem., 23, 783 (1975). (18) K. Sharp, Ph.D. Thesis. University of California, San Diego, San Diego, Calif.. 1974. (19) L. D. Bowers, L. M. Canning, Jr.. C. N. Sayers, and P. W. Carr. submitted for publication in Clin. Chem. ( Winston-Salem, N.C.). (20) H. L. Pardue and S. N. Deming, Anal. Chem., 41, 986 (1969). (21) D. R. Senn, P. W. Carr, and L. N. Klatt, submitted for publication in Chem. lnstrum. (22) S. Taniguchi and E. Itagaki, Biochem. Biophys. Acta, 44, 263 (1963). (23) H. Barnes and A. R. Folkard, Analyst (London), 76, 599 (1951). (24) J. Van7 Riet and R. J. Planta, FEBS. Lett., 5, 249 (1969). (25) D. R. Senn, PhD. Thesis, University of Georgia, Athens, Ga. 1975. (26) J. P. Danehy and C. W. Zubritsky 111, Anal. Chem., 46, 391 (1974). (27) D. R. Senn, P. W. Carr, and L. N. Kiatt, Submitted for publication in Anal. Biochem. (28) A. Hubaux and G. Vos. Anal. Chem., 42,849 (1970). (29) D. F. Martin, "Marine Chemistry", Marcel Dekker, New York, N.Y., 1968, p 146.
RECEIVEDfor review January 9, 1976. Accepted February 6, 1976. Taken in part from the Ph.D. dissertation of D. R. Senn submitted to the Graduate School, University of Georgia. Presented a t the 2nd Annual FACSS Meeting, Indianapolis, Ind., October 1975. This work was supported in part by Grant R-800858 of the Environmental Protection Agency. Oak Ridge National Laboratory is operated by the Union Carbide Corporation for the U.S. Energy Research and Development Administration.
Possible Formation of Bis(chloromethy1) Ether from the Reactions of Formaldehyde and Chloride Ion James C. TOU* and George J. Kallos Analytical Laboratories, Dow Chemical U.S.A.,Midland, Mich. 48640
Several reactions of formaldehyde and commonly used chloride salts have been investigated for the possible formation of bis(chloromethy1) ether (BCME). All experiments were carried out In a dynamic system which had been mathematlcally simulated. Both the original and presently Improved 958
ANALYTICAL CHEMISTRY, VOL. 48, NO. 7, JUNE 1976
derlvatiratlon-gas chromatographic methods were employed for the analysis of gaseous BCME suspected to be generated from the reactions. No evidence was found for the formation of BCME in the gas phase from the reactions studied at a detection limit of low parts per trillion.
Bis(chloromethy1) ether (BCME) has been shown to be a carcinogen (1,2). Extensive research on this subject has been conducted in many laboratories. Furthermore, BCME is known to be an impurity present in chloromethyl methyl ether (CMME), a common chloromethylating agent and possible industrial exposure to BCME has been of concern. Therefore, monitoring of BCME a t parts-per-billion levels in air is necessary for environmental control. Many analytical techniques have been developed for this purpose. They are high resolution mass spectrometry ( 3 ) , gas chromatography/mass spectrometry ( 4 ) , dual-column gas chromatographic system (51, gas chromatography/high resolution mass spectrometry, (6) and a gas chromatographic method employing derivatization (7). A technique (8,9)has also been developed for the direct analysis of BCME in a water-soluble anion-exchange resin where CMME is used as a chloromethylating agent. Evidence for the presence of BCME in the resin was not found a t the parts-per-billion level. Kinetic studies of the stability of BCME in both humid air (10) and aqueous solutions (11) show that BCME is very stable in humid air (t1/2 > 20 h) but hydrolyzes rapidly in aqueous solutions (t1/2 < 1min). Recent disclosure (12, 13) of the reported spontaneous formation of BCME from the reaction of hydrogen chloride and formaldehyde in the gaseous phase initiated a new concern about the presence of BCME in the environment. Both hydrogen chloride and formaldehyde are widely used in industry and both could possibly be present, a t low levels, in environmental air. However, experimental data have been inconsistant regarding the level of BCME produced from the reaction of hydrogen chloride and formaldehyde at low levels in air (12,13). Kallos and Solomon (12)found that BCME did not form at a detection limit of 0.1 ppb even a t concentrations of hydrogen chloride and formaldehyde in air higher than their threshold limit values, for example at 100 ppm each. On the other hand, Frankel, McCallum, and Collier (13) reported detection of 3 ppb of BCME generated from the above reaction. Concern for environmental control becomes even more pressing when the formation of BCME was reported from the reaction of formaldehyde and ionic chlorides in aqueous solutions (13). However, in another study (14) of the reaction of aqueous hydrogen chloride and formaldehyde a t concentrations up to 2000 ppm, BCME was not detected either in the aqueous phase or the gaseous phase above the reaction mixture with detection limits of 9 ppb and 1 ppb, respectively. Because of the possibility of the BCME formation and its impact on occupational health, the National Institute for Occupational Safety and Health ( 1 5 ) has been conducting an investigation of the possible presence of BCME in several industrial environments. Because of the above implication and its impact on our industrial environment, several reactions of commonly used chloride salts and formaldehyde a t different concentrations were studied in an attempt to investigate the possible formation of BCME. The experiments were carried out in a dynamic system which allows us to extend our detection limit to the parts-per-trillion level which has never been achieved before in such an analysis. This is in contrast to a commonly used static system, where only parts-per-billion sensitivity can be reached. A more specific analytical technique than the one previously reported (7) has also been developed for a more accurate and specific analysis of the reaction product.
RESULTS AND DISCUSSION Mathematical Simulation of t h e Reaction. A reactor with a volume Vt(1.) in liters is first charged with the reactants occupying volume Vr(l.). Therefore, the gas space volume in the reactor becomes V(1.) = Vt - V,. Both the gaseous phase and liquid phase in the reactor are stirred. This is shown in
To Vacuum
t
Stirrer
Active Charcoal Tower
Needle Valve
Flowmeter
D r y Ice Trap
Figure 1. The dynamic reaction assembly
Figure 1.At the beginning of a reaction, air is allowed to sweep over the gas space in the reactor a t a flow rate, f (l./min), and the gaseous components, including any BCME which may be formed, are swept into two reagent impingers. If BCME is present, it will react with the derivatizing reagents and the derivative formed is then extracted and analyzed as described in the Experimental section. Assuming the air flowing into the reactor was completely mixed with the gases in the reactor by stirring, the concentration C(pg/l.) of the generated BCME a t any instance in the gaseous phase is therefore uniform throughout the reactor. The amount of BCME carried out by the air into the impingers is then fC. It was found that no gross change occurred in the reactant concentrations throughout the experiment, This leads to the valid assumption t h a t BCME is generated at a uniform rate, r,(pg/min), from the reaction with release into the gaseous phase. It is known that BCME is stable in humid air (IO),but it is also known that BCME in the gaseous phase hydrolyzes with a rate, k(min-'), when it is in contact with an aqueous phase. Therefore, part of the BCME which is assumed to be generated from the reaction and released into the gas phase will be destroyed because of the hydrolysis. The amount being destroyed is kVC. The general equation for the material balance of BCME in the reactor can be written as: BCME into the reactor - BCME out from the reactor + BCME generated from the reaction BCME destroyed in hydrolysis = BCME accumulated in the reactor (1) Mathematically, the above material balance equation is rewritten in the following form: o - fC
+ rg - kVC = VdC/dt
(2)
Rearranging the variables and integrating, we have
(3) Then the following equation is obtained
This describes the concentration, C, of BCME in the reactor at any time t , The total amount, W in pg, of BCME trapped in the two impingers is therefore: W = L'fCdt ANALYTICAL CHEMISTRY, VOL. 48, NO. 7, JUNE 1976
(5) 959
.
Concentration Of BCME Fed Into The Reactor
Table I. Effect of Stirring Rates on the Rate of Hydrolysis of BCME in a Gaseous Phase Equilibrated with an Aqueous Phase
56
c
..........
-
Stirring rates (arbitrary units)
-
Gas phase
Calculated Experimental
6 (fast) 5 5 0 (no stirring)
2
0
4
6
8 10 Time, Min
12
14
16
Substituting C from Equation 4 into Equation 5 and integrating, we have = f
V
A
+ kV ( t +f+kV
e-(f+kV)t/V
- __ ) f+kV
(6)
The rate of generation of BCME, r g ,from the reaction can therefore be calculated from the experimentally determined quantity, W.
w(f+
V e-(f+kV)t/V - ( t + f kV f+kV The quantity, rg,is the fundamental quantity of the reaction. The following related quantities can then be calculated. 1) Under the flowing system, the concentration of BCME in the system and out of the system at any time can be calculated from Equation 4. When the system reaches equilibrium, the concentration becomes: Pg
=
f
+
c, = f +Pkg V
(8)
2) Assuming the air flowing over the reaction surface sweeps all the BCME out of the system immediately after it is formed, and leaves no chance for BCME to come in contact with the aqueous phase, the concentration of BCME in the air will be
This is similar to a situation of carrying out a reaction in a completely open reactor with an air drift rate, f , over the reactor. 3) If there is no air flowing into the reactor or f = 0, the situation is then the same as a static reactor which was used previously (13, 14). Then the concentration a t equilibrium conditions can be written as
4) Assuming that two molecules of formaldehyde are involved in the reaction with two chloride ions in the formation of one molecule of BCME, the apparent conversion of the formaldehyde is then (11)
where Mf and Mb are the molecular weights of formaldehyde and BCME respectively and Wf(g) is the amount of formaldehyde used in the reaction. Therefore, the apparent conversion, 9, is in the unit of pptlmin. 960
k (min-I)
8 (fast) 7 0 0 (no stirring)
0.53 0.41
0.24 0.16
18
Figure 2. Comparison of the calculated and experimental concentrations of BCME In the reactor
W
Liquid phase
ANALYTICAL CHEMISTRY, VOL. 48, NO. 7, JUNE 1976
The concentration unit is expressed in pg/l. in all the above equations. However, it is converted to conventionally used ppb or ppt volume-to-volume ratio in the final results. Validation of t h e Mathematical Simulation. In the mathematic formulations, the flow rate, the gas space volume and the time can be easily measured, and the rate of hydrolysis should be determined for every reaction system. The 2-1. reactor was first charged with 500 ~ mdistilled . ~ water. The rate of BCME hydrolysis was determined with a mass spectrometer equipped with a hollow fiber probe (8-10) and was found to be dependent on the stirring rates, both in the gaseous phase and the aqueous phase. The result is shown in Table I. It is understandable that the faster the stirring rate, the more rapid is the hydrolysis. Therefore, the stirring rates have to be kept constant in any experiment. In the validation of the mathematic simulations, a 10-1. BCME standard was prepared in air with concentration C , = 56 ppm and allowed to flow a t a rate off = 0.086 l./min. into the above reactor containing 0.5 1. of distilled water. This corresponds to rg = Cof = 23 Fglmin. The concentration of BCME in the reaction was monitored with the hollow-fiber probe connected to a mass spectrometer. The calculated and the experimental concentration of BCME in the system a t any time are shown in Figure 2. Excellent agreement between the calculated and experimental concentrations indicates the working hypotheses made are valid. The mathematical formulations successfully predict both the time (-10 min) and the concentration (-7 ppm) when the system reaches equilibrium. If the reaction is carried out for 2 h, all the exponential terms in the equations are negligible, which greatly simplifies the calculation. The above experiment was repeated with a 28.3-ppb BCME standard. The gas sweeping out of the reactor was passed through the two impingers containing the derivatizing solutions for 1.5 h with use of the apparatus as shown in Figure 1. In this case, the theoretical amount of BCME can be calculated from Equation 6 by substituting r g = C,f where C , = 28.3 ppb or 133 ng/l., Le.,
V e-(f+kV)t/V - W=- C0f2 f kV ( t + f + k V f+kV In this case, the theoretical amount of BCME was calculated to be 120 ng as compared to the experimental amount 150 ng. The major portion of the original amount (1.02 wg) of BCME into the reactor was destroyed by hydrolysis. The agreement is satisfactory, which further validates the mathematical formulations. The above experiment is termed as the recovery and was carried out for most of the reactions reported in this paper. Reactions a n d Analytical Results. Several commonly used chloride salts were studied. Among them, sodium chloride and calcium chloride are widely known to be present in industrial aqueous streams. Ferric chloride is used as a coagulant in waste water treatment. Magnesium chloride is one of the active components in cloth treatment formulations used in the textile industry. Instead of studying the reaction of
+
~~
~
Table 11. Reactions of Formaldehyde with Chlorides without Evidence of BCME Formationa Recovery (ng) Wcalcd,
tions k (min-') rg,pg/min, Eq. 7 A
0.41b
B
0.41b
C
0.41b
D
0.41b
E
1.58
F
0.12
N.D. (L.D. = 76) N.D. (L.D. = 67) N.D. (L.D. = 110) N.D. (L.D. = 62) N.D. (L.D. = 150) N.D. (L.D. = 19)
C, ppb, Eq. 9
N.D. (L.D. = 0.18) N.D. (L.D. = 0.16) N.D. (L.D. = 0.26) N.D. (L.D. = 0.12) N.D. (L.D. = 0.35) N.D. (L.D. = 0.05)
Ce,ppt, Eq. 4
C,, ppt, Eq. 10 8 , ppt/min, Eq. 11
N.D. (L.D. = 24) N.D. (L.D. = 21) N.D. (L.D. = 35) N.D. (L.D. = 20) N.D. (L.D. = 12) N.D. (L.D. = 20)
N.D. (L.D. = 21) N.D. (L.D. = 19) N.D. (L.D. = 31) N.D. (L.D. = 17) N.D. (L.D. = 12) N.D. (L.D. = 14)
N.D. (L.D. = 16) N.D. (L.D. = 14) N.D. (L.D. = 7.7) N.D. (L.D. = 13) N.D. (L.D. = 0.45) N.D. (L.D. = 0.06)
Welpti
Eq. 6
*..
...
160
170
...
...
...
...
43
48
270
420
a N.D., = Not detected. L.D., = Limit of detection. ng = nanograms, pg = picograms, ppb = parts per billion (volumeholume); ppt = parts per trillion (volume/volume). The rate of hydrolysis in the distilled water system was used in the calculations. A, 250 pH 0.9. B, 250 cm3 CaClZ,, (13%) + 250 cm3 HzCO,, (l%), pH 8.5. C, 250 cm3 NaC1. cm3 HCl,, (1.2%) 250 cm3 HzCO,, (l%), (15%) 250 cm3 HzCO,, (3%),pH 5.8. D, 250 cm3 CaClZ,, (13%) + 250 cm3 HzCO,, (1%)+ 4 cm3 concd HCl, pH 0.9. E, 450 cm9 Formalin (38% formaldehyde and 14% CH30H) 50 cm3 FeCls,, (0.56%), pH 1.9. F, 460 cm3 Formalin (37% formaldehyde and 1.5%CH30H) 90 g FeC13.6 H20, pH 0.9.
+
+
+
+
Table 111. Comparison of the Analytical Results from Two Techniques Employed in the Analysis of BCME Formations from the Reactions of Formaldehyde and Chlorides" rg?pg/min
Reack tions (min-l) Origb
G
0.25
Co, ppb
ImprC
Origb
0.20
0.07
ImprC
Origb
ImprC
Origb
ImprC
Recovery, ng Wexptl
Wcaicd
2.6
0.5 ppb
0.6 ppb
6.4
190
250
103 1.8 X 103
4.2
0.9 ppb
0.9 ppb
11
180
250
240
250
24
30
28
30
390
580
34
34
143
N.D. L.D. = 0.3 0.3
N.D. L.D. = 65 I
Origb
ppt/min
1.1X
N.D. L.D. = 110 H
ImprC
7,
cs, PPt
Ce, P P ~
190 250
72 N.D. L.D. = 0.13
0.45 0.6 N.D. L.D. = 59
N.D. L.D. = 60
N.D. L.D. = 60 93
N.D. L.D. = 33 200 250
N.D. L.D. = 0.15
0.4 N.D. L.D. = 42
360 466 N.D. L.D. = 60
N.D. L.D. = 0.6
N.D. L.D. =
N.D. L.D. = 0.18
110
N.D. = Not detected. L.D. = Limit of detection. Original technique, see Ref. 7 for details; analyzed as derivative A subject to interference. Improved technique, see text for details; analyzed as derivative B with more specificity. G, 460 cm3 Formalin (37% formaldehyde and 14% CH30H) 90 g FeC13, pH 0.6. H, 460 cm3 Formalin (37% formaldehyde and 14%CH3OH) 58 g NaC1, pH 3.2. I, Cloth treatment formulation (100 cm3 reactant containing MgClz 400 cm3 reactant containing low formaldehyde glyoxal), pH 4.5.
+
MgClz and H2CO alone, we have investigated the reaction of the formulated components containing MgCl2 and H2CO which are commonly used in the textile industry for cloth treatment. Also studied were the reactions of HCl,, and H2CO,, a t much higher concentrations than previously reported (13). The well characterized gas chromatographic technique utilizing derivatization of BCME ( 7 )was initially employed in the analysis of the reaction product. This is referred to as the original technique in the later discussion. The results are shown in Tables I1 and 111. No evidence was found for the formation of BCME in the reactions A-F as shown in Table 11. However, a gas chromatographic peak at the retention time of BCME derivative was observed in the reactions G-I and its intensity was used in the calculations of the levels of the
+
+
BCME formation. The results are shown in Table 111. Because of the above positive response with the original derivative technique, a second independent analytical confirmation of the formation of BCME from the reactions was highly desirable. The reported gas chromatography-mass spectrometric technique was first attempted (4).However, an excess amount of gases was trapped on the Chromosorb 101 (Johns-Manville) and subsequently caused excess high pressure in the mass spectrometer during monitoring. This made the above technique inapplicable for this type of analysis. The derivative of BCME formed and analyzed in the original derivatization-gas chromatographic technique ( 7 ) was well characterized and is believed to be generated from the following reaction ANALYTICAL CHEMISTRY, VOL. 48, NO. 7, JUNE 1976
961
A B
150 130
Column Temp -
Column Temp -
110°C
190°c
E
i
-
1
2
3
- Retention Time, Min -
I
Flgure 3. Gas chromatograms of the two BCME derivatives
0
l
l
40
l
80
l
l
l
120
l
160
l
l
200
l
240
Amount (ng) Of BCME Into The Derivative Solurion
Flgure 4. Gas chromatographic response curve of BCME as bis(trichlorophenoxy) derivative
/cl
A
Liquid Standard In Hexane
A
CleOCH20CH20CH3
C1CH,0CH2Cl
B:
+-
+ 2C1
+
/cl Column: 3-ft 0.1 % OV-101 on GLC 100. Flow rate: 70 cm3/min. 95:5 Ar/CH4. Sample size: 1.2 bl of the extracted derivatizing solution corresponding to 213 ng BCME/2 cm3 hexane extract
C
l
~
2
c'\
c' 1
0
C
'c1
H
,
0
~
+ 2C1C
Cf
B
c1 A I t is expected that the monitored derivative A, will not be specific for the indication of the presence of BCME if chloromethylal, ClCH20CHZOCH3 also exists in the system. Chloromethylal is expected to yield the same derivative as BCME, IC1
+
C1CH,0CH20CH3 C1q
O
-
-
A
+ C1-
c1
However, if the bis(trich1orophenoxy) derivative, B, is also formed in the derivatization, 962
ANALYTICAL CHEMISTRY, VOL. 48, NO. 7, JUNE 1976
monitoring B would be more specific in the determination of BCME. Early attempts in analyzing for B were not successful (16) with the then available gas chromatographic column technology. Recent continuing efforts in this direction now make such an analysis possible by using 0.1% OV-101 on a GLC 100 glass bead column at 190 "C. This is shown in Figure 3. Also shown in this figure is the analysis of derivative A using the same column but at 110 OC. Both gas chromatographic peaks are identified with the authentic derivatives prepared separately. We found no background peak at the same retention time. Fortunately the techniques developed for the determination of BCME with the use of two different derivatives demonstrate comparable sensitivity. No change in the derivatization procedure is necessary. The response of the BCME derivative a t different BCME levels were determined either by spiking the derivative solution with known amounts of BCME in hexane or by trapping known amounts of BCME at different concentrations in air. Linear relationships over the concentration range of interest are demonstrated in Figure 4. Employing the presently modified derivatization-gas chromatographic technique, the reactions giving positive response (for the formation of BCME) in the original technique were reinvestigated. No peak corresponding to the BCME derivative, B, was detected with the calculated limits of detection shown in Table I11 for each reaction.
l
From the above investigation, no evidence was found for the formation of BCME in the gas phase from the reactions studied at a detection limit of low parts per trillion. In the case of the FeCl3 reaction which has also been studied previously (13), our negative observation of BCME is in contrast to that reported, where 210 ppb of BCME were detected in the vapor above the solution. The discrepancy may have been caused by the lack of specificity of the high resolution mass spectrometry employed in the analysis. Two types of possible interferences can be encountered: 1)Common ion interference-The ion, C2H40C1+, can also be generated from compounds, other than BCME, containing a ClCH20CHp group, such as chloromethylal (ClCH20CH20CH3), dichloromethylal ( C ~ C H Z O C H ~ C H etc. ~ C ~2)) ,Different ion interference-Under the mass spectrometric resolution ( R = 3500) employed (13),the ion, C2H40Cl+ (mle = 78.9950), still suffers interferences from those ions with comparable masses such as C4H3Si+ (mle = 79.0004), C2HOF2+ ( m / e = 78.9995). Therefore, it is possible that the results reported in the previous study (13)represent only maximum values, but they do not represent the true concentrations of BCME generated.
EXPERIMENTAL Reagents. Bis(chloromethy1) ether was obtained from K&K Laboratories. Methanol and hexane, distilled in glass, were obtained from Burdick & Jackson Laboratories, Muskegon, Mich. Formaldehyde solution, sodium hydroxide, sodium methoxide, ferric chloride hexahydrate, and sodium chloride were Baker analyzed reagents and obtained from the J. T. Baker Chemical Co., Phillipsburg, N.J. 2,4,6-Trichlorophenol was synthesized a t The Dow Chemical Co. and was recrystallized with hexane. Calcium chloride was Mallinckrodt analytical reagent. Formaldehyde solution (37% HzCO and 1.5% CH30H) was made by refluxing paraformaldehyde obtained from the Celanese Corp. in water and adding 1.5%CH30H. Nuclear magnetic resonance analysis showed the distribution of chain length in this formalin solution to be very similar to the distribution of formalin containing 37% HzCO and 14% CH30H. The cloth treatment formulation commonly used in the textile industry consists of two components, one containing MgC12 and the other, low formaldehyde glyoxal. Apparatus a n d Procedure. The reaction apparatus used in this study is shown in Figure 1.The reactor was a 2-1. tubular glass vessel with a Teflon stopcock a t each end, a standard ground glass joint connection in the center for the magnetic stirrer and two 18/9 ball joints provided with rubber septa. The rubber septa were covered on the inside of the reactor with Teflon sheet to avoid any possible interaction with reactant or product. The joint parts were used for injecting BCME gas standards and connecting the fifteen-head silicone fiber probe to a cycloidal mass spectrometer. The impingers containing the derivatizing reagent were connected in series to the reactor and the dry ice trap on the other side. The dry ice trap collected any methanol vapor or other organics so that correct flow of air could be adjusted through the reaction system with a needle valve and flowmeter. The activated carbon trap was used as an additional safety device. The reaction apparatus was thoroughly cleaned, rinsed with water and acetone, and well dried before every reaction. In a typical reaction the 2-1. flask was charged with the appropriate amount of formaldehyde and chloride as shown in Table 11. The reaction mixture was stirred magnetically a t setting 7 (Sargent Magnetic stirrer) and the head space being stirred a t constant setting 5 (GT-21 stirrer, G. K. Heller Corp.). A 10-1. Saran bag, which had been previously rinsed twice with dry air and then filled with dry air, was connected on the other side of the reactor to supply the gas through the reactor into the impinger reagent solutions a t a constant flow of 0.086 l./min. The two impingers were provided with 15 ml each of the optimized reagent ( 7 ) .All reactions were run for 2 h. For the recovery runs, either 0.133 or 1.33 kg BCME in hexane were injected into the
Saran bag containing 10 1. of dry air. Then this standard was run similarly over the reaction mixture a t 0.086 l./min into the impinger solutions for a certain time to determine the recovery efficiency. All reactions and recovery runs were carried out a t room temperature. BCME Rates of Hydrolysis. To determine the rates of hydrolysis of BCME in the headspace of each reaction mixture, a high concentration of BCME vapor standard (2800 ppm) was made with dry air in a Saran bag and 10 ml of this standard was injected into the headspace of the reaction. Both liquid and gaseous phase were stirred with arbitrarily fixed rates as described previously. The silicone rubber hollow fiber probe-mass spectrometric technique ( 4 9 )was used to follow the depletion of BCME by monitoring the intense ion peak a t m/e 79. Chromatographic Conditions. The gas chromatographic work was carried out on a Hewlett-Packard Model 5710 gas chromatograph equipped with a 63Nielectron capture detector. In the initial analysis of BCME as derivative A, the column used was 5-ft X %-in. glass tubing (2-mm i.d.1 packed with 120/140 mesh textured GLC 100 glass beads coated with 0.1% OV-275. The column temperature was 130 O C and the flow rate 40 ml/min, using Ar/CH4 (95:5). The injection and detector temperatures were 150 and 250 O C , respectively. For the bis(trich1orophenoxy) derivative, B, the separation was carried out on a 3-ft X Yd-in. (2-mm id.) glass column packed with 0.1% OV-101 on 120/140 textured glass beads (GLC 100) a t 190 "C. This column can also be used for analyzing derivative A, when operated a t 110 "C as shown in Figure 3. The temperature of the injection port was 200 OC and that of the detector was 300 OC. The Ar/CH4 (955) flow was set a t 70 ml/min. Validation of Bis(trich1orophenoxy) Derivative. The validation of the bis(trich1orophenoxy) derivative B was carried out as follows. Two gl of pure BCME was added to 100 ml of hexane. Fifteen milliliters of the optimized derivatizing reagent (7) were pipetted into six 12-dram Kimble screw-cap vials. To the vials was added, respectively, 8, 6, 4, 2, 1, and 0 11 of the concentrated standard. The resulting standards were equivalent to 212.8, 159.6, 106.4, 53.2,26.6, and 0 ng of BCME. Similarly, the same injections were made in six different Saran bags containing 10 1. of dry air. The standards from the bag were pulled through two impinger derivatizing solutions at 0.086 l./min as carried out in the reactions. Further steps in processing these standards are described in previous work (7). The linearity of responses for the standards is shown in Figure 4.
ACKNOWLEDGMENT The authors thank W. Blaser for the discussions in the column technology, L. B. Westover and H. Gill for their valuable consultations and encouragement, and J. Heeschen for his nuclear magnetic resonance analysis. LITERATURE CITED B. L. Van Duuren, B. M. Goldschmidt. C. Katz, L. Langseth, G. Mercado, and A. Sivak, Arch. Environ. Health, 16, 472 (1968). R. T. Drew, S.Larkin. M. Kuschner, and N. Nelson, Arch. Envlron. Health. 30,61 ( 1 9 7 9 , and the references therein. L. Collier, Environ. Sci. Techno/.,6, 930 (1972). L. A. Shadoff.G. J. Kallos, and J. S.Woods, Anal. Chem., 45, 2341 (1973). R. L. Wilklns and L. S.Frankel, U S . Patent 3,807,217 (1974). R. P. Evans, A. Mathias, N. Mellor, R. Silvester,and A. E. Williams, Anal. Chem., 47, 821 (1975). R. A. Solomon and G. J. Kallos, Anal. Chem., 47,955 (1975). L. B. Westover. J. C. Tou, and J. H. Mark, Anal. Chem., 46, 568 (1974). J. C. Tou, L. B. Westover, and L. F. Sonnabend, Am. lnd. Hyg. ASSOC.J., 36, 374 (1974). J. C. Tou and G. J. Kallos, Anal. Chem., 46, 1866 (1974). J. C. Tou, L. B. Westover, and L. F. Sonnabend. J. fhys. Chem.. 78, 1096 (1974). G. J. Kallos and R. A. Solomon, Am. lnd. Hyg. Assoc. J.. 34, 469 (1973). L. S.Frankel, V. S.McCallum,and L. Collier, Envlron. Scl. Techno/.,8,356 (1974). J. C. Tou and G. J. Kalios, Am. lnd. Hyg. Assoc. J., 35,419 (1974). T. Marceleno, P. E. Philip, and J. Bierbaum. The American Industrial Hygiene Association Conference,Miami, Fla., May 12-17, 1974.
Private communication with R. Solomon, Hydro Science inc., Knoxville, Tenn. (1975).
RECEIVEDfor review November 20, 1975. Accepted March 8, 1976.
ANALYTICAL CHEMISTRY, VOL. 48, NO. 7 , JUNE 1976
963
