Anal. Chem. 1992, 6 4 , 346-349
346
(13) Kuo. J. E.; Milby, K. H.; Hinsberg, W. D.; Poole, P. R.; McGuffin, V. L.; &re. R. N. Clin. Chem. 1085. 31. 50-53. (14) Bailey, M. P.; Rocks, B. F.; Riley, C. I n Non-Isotopic Immunoassay; Ngo, T. T., Ed.; Plenum Press: New York, 1988; pp 187-197. (15) h e r , A.; Collasius, M.; Valet, G. Anal. Siochem. 1990, 191, 295-30 1. (16) Evangelista, R. A.; Pollak, A.; Gudgin-Templeton,E. F. Anal. Biochem. 1991, 197, 213-224.
(17) Chan, M. A.; Bellem, A. C.; Diamandis, E. P. Clin. Chem. 1987, 33, 2000-2003. (18) Christopoulos, T. K.; Lianldou, E. S.; Diamandis, E. P. Clin. Chem. 1990, 36, 1497-1502.
for review July 12, 1991. Accepted November 11, 1991.
Separation of Fluoride from Fluoroelastomers by Diffusion in Test Tubes Pothapragada Venkateswarlu* 3MlIndustrial Chemical Products Division and 3MISpecialty Adhesives and Chemicals Division, Building 236-2B-11, 3M Center, St. Paul, Minnesota 55144
The conventional procedure for separation of fluoride as trlmethylfluorosllane In Conway diffusion cells involves the use of grease for sealing the cells and also for closing the hole In the lid drilled for lntroductlon of hexamethyidlslioxane. We have developed a simpler procedure In which diffuslon Is carried out in 5-mL test tubes without the use of grease. Resuits of analysis of fluorlde following diffusion from water, urine, and bone samples are in excellent agreement with those obtained by other procedures not Involving diffusion. Separation of fluoride from partly and fully cured fiuoroelastomers is achieved by first grinding the samples In a liquid nitrogen mill and then using methyl ethyl ketone as an adjuvant to perchloric acid employed In the diffusion procedure.
INTRODUCTION Separation of fluoride as hydrogen fluoride from bone and enamel by diffusion in 50-mL polyethylene bottles was first introduced by Singer and Armstrong' and later adapted and modified by several investigators2-8 for analysis of fluoride in various materials. These procedures call for 24-48-21 diffusion at 50OC for quantitative recovery of fluoride. A significant development in this field was the rapid room temperature diffusion technique reported by Tavesg in which the fluoride is diffused as the more volatile and more hydrophobic trimethylfluorosilane (TMFS). These diffusions currently are carried out in Conway plastic diffusion cells or modifications thereof. These cells require use of grease for sealing to achieve quantitative fluoride recovery. In the procedure described in the present paper, the diffusion is carried out in 5-mL stoppered test tubes instead of in the more expensive Conway diffusion cells. The new technique is much simpler. The tedium of having to apply a continuous bead of grease around on the inside of the lids of a large set of cells is all avoided in the present procedure which does not need the use of any grease for sealing the test tubes. Diffusion of as many as 72 samples can be carried out in a test tube rack as small as 9 in. long, 4.5 in. wide, and 2.5 in. high. This amounts to considerable saving of bench space, using no more than the space occupied by 2 or 3 Conway diffusion cells. Stationary diffusion (diffusion with no shaking) is also possible.
This paper describes successful application of this technique in the determination of fluoride in water, urine, and bone samples. Adaptation of this new technique with certain mandaqry additional steps for the determination of fluoride generated in situ during the different stages of curing of fluoroelastomers is also described in this paper. Such information is needed in understanding the solid-phase cure chemistry of fluoroelastomers and has now been made available for the first time by the technique described in this paper.
EXPERIMENTAL SECTION Reagents. Reagent-grade chemicals and doubledistilled water were used throughout. Sodium fluoride stock solution, lo00 Mg F f L, contained 1.2105 g of sodium fluoride in 1 L of water and
was stored in a polyethylene bottle. This stock solution was diluted to obtain fluoride standard solutions of desired strength covering the range of fluoride in the samples. Perchloric acid (Mallinkrodt)used was 70%. This acid is corrosive and explosive; it should be handled with care. The acid was always stored ice-cold in a refrigerator. A 5% hexamethyldisiloxane (HMDS) reagent was prepared by adding 50 pL of HMDS (Dow Corning fluid 200, 0.65 centistokes) to 1 mL of ethyl alcohol. Materials. Materials analyzed included samples of water, urine, experimental rat bone ash, and partly and fully cured samples of a fluoroelastomer, Fluorel (3M), a copolymer of vinylidene fluoride and hexafluoropropylene. Curing of the polymer to different stages was carried out by blending 100 g of copolymer with 1.5 g of hexafluoroisopropylidenebis(4-hydroxybenzene)or bisphenol AF (cross-linkingagent), 0.475 g of benzyltriphenylphosphonium chloride (phase-transfer catalyst), 6 g of Ca(OH)2, and 3 g of MgO on a two-roller mill and, then, holding 15-g lots of the rubber compound at 300 O F from 2 to 16 min on a Monaanto oscillating disk rheometer. At selected time intervals the curing was interrupted and within seconds the sample was dropped in liquid nitrogen to freeze the reaction. Apparatus. Five-milliliter polystyrene stoppered test tubes (Falcon 2003, individually wrapped) and 5-mL polypropylene stoppered test tubes (Falcon 2063) served as diffusion cells. The stoppers of these test tubes have a concave recess which serves to retain enough sodium hydroxide solution for trapping HF or TMFS. The mechanical shaker (Eberbach) was modified to run under a d.c. motor speed and torque control system. A liquid nitrogen mill (Spex Freezer Mill, 6700) was used to grind the fluoroelastomer samples. Procedure A (Water, Urine, Bone, Dental Plaque, Enamel, Dentin, AshedfFused Biological, and Soil Samples). A few milligrams of a solid sample or up to 500 pL of a liquid sample
0003-2700/92/0364-0346$03.00/0 0 1992 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 64, NO. 4, FEBRUARY 15, 1992 -25pL
347
2 . 5 M NaOH
+
1 mm
L
cI
0
(A) STATIONARY (QUIESCENT)DIFFUSION
5
10
15
20
25
Hours of Diffusion
Diffusbn of W i as trlmethyifluorosilane in test tubes from bone under stationary (0)and nonstationary (0) conditions.
( 6 )NONSTATIONARYDIFFUSIONON SHAKER
Flgure 1. Diffusion of fluoride as trimethylfluorosilane in test tubes. (A) Stationary (quiescent) diffusion. (B) Nonstationary diffusion by
agitation on shaker. and 1.0 mL of ice-cold concentrated perchloric acid were placed in a 5-mL polystyrene test tube. In one hand the inverted cap containing 50 p L of 2.5 M NaOH was held as close to the mouth of the test tube as poasible and 50 & of HMDS reagent was added (squirted on the wall of the test tube) with an Eppendorf pipette held in the other hand. Immediately thereafter the test tube was capped tightly, by gently rotating the cap with slight pressure. (Until the additon of HMDS, the test tubes were held in test tube racks placed in shallow trays containing water and ice cubes.) The entire test tube rack full of test tubes was placed on a shaker with the side of the test tube rack inclined 80 the tat tubes were in a near-horizontal position to increase the surface area of the contents in the test tube. After carefully securing the test tube rack on the shaker, the speed and the torque of the oscillations of the shaker were gradually increased with the motor controls from zero to a preset point so there was no splashing of the contents onto the cap (Figure 1). Diffusion does take place even without the use of the shaker, but the test tube rack should be so positioned that the largest possible surface of the sample plus acid is maintained to facilitate the diffusion process. Alternatively,the test tubes can be placed on a flat surface (a glass sheet),with the rim of the stopper resting on a 0.5-1.0 mm thick strip of stainless steel (ruler) as shown in Figure 1. At the end of the diffusion period, which was usually overnight for convenience (see discussion later), the caps containing the NaOH with the trapped fluoride were transferred onto a fresh set of identical test tubes, containing 50 p L of 2.5 M perchloric acid and an appropriatevolume of a selected colorimetric reagent,'O fluorometric reagent," or a buffer for measurement with the fluoride e l e c t r ~ d e . ~The ~ J ~ contents were thoroughly mixed. Appropriate fluoride standards containing 50 p L each of 2.5 M NaOH and 2.5 M HCIOl were prepared. Spectrophotometric and fluorometric measurements were made in the same polystyrene tubes without any further transfer to other cuvettes. When aliquots of the sample diffusate8 were diluted and analyzed, the fluoride standards contained amounts of base (NaOH) and acid (HC104)equivalent to those in the diluted aliquota. It is advisable to run fluoride standards alongside the samples to ensure that the diffusion is proceeding as expected. Procedure B (Fluoroelastomer Samples). Using a nitrogen mill, the ektomer sample was ground (12 min) to a f i e powder, which at mom temperatwe aggregated itself into fine soft particles that further coalesced into small lumps, which had an excellent
surface area. Ten to twenty milligrams of these ground samples were weighed into the 5-mL polypropylene test tubes (not polystyrene test tubes), and 250 pL of methyl ethyl ketone (MEK) was added. The test tubes were stoppered to prevent evaporation of MEK and set aside for 15-30 min. The rest of the procedure from this point on is the same as in procedure A, starting with addition of 1mL of ice-cold perchloric acid. Diffusion of fluoride was allowed to take place over 48-60 h on the shaker. The exact time period for diffusion was determined by an earlier trial on a sample with the highest cross-link density for each experiment on curing. The fial diffusate solutions were analyzed for fluoride with the fluoride electr~de.'~J~
RESULTS AND DISCUSSION To facilitate diffusion of fluoride as H F using the SingerArmstrong procedures,7 a large surface area of samples plus acid (about 50 cm2) is maintained through the use of flat diffusion cells. In addition, the cells are placed in an oven at 50-60 "C. The Taves room temperature diffusion techniqueg requires, because of the danger of loss of fluoride as highly volatile TMFS, that the diffusion cell containing the fluoride trapping agent and the sample plus acid be first covered and sealed with grease-lined lid and then HMDS introduced through a 1-mm hole in the lid which is also sealed with grease immediately thereafter. In the present procedure no detectable loss of fluoride has been encountered even though HMDS was added before stoppering the test tube and no grease was employed for sealing. The relatively small surface area of the sample plus acid (about 1 cm2) inside the vertically held test tube and the immediate stoppering after addition of HMDS help prevent any detectable loss of fluoride. Then the tubes are arranged in a near horizontal position and gently agitated in a shaker (Figure 1). These manipulations and the relatively small head space volume (about 3 mL) within the test tube seem to contribute to the success of quantitative separation of fluoride by diffusion in test tubes without any loss of fluoride on adding HMDS. Figure 2 shows the rates of diffusion of fluoride from bone ash maintained under stationary (quiescent) and nonstationary (agitation on a shaker) conditions. The rates of diffusion of fluoride from water and urine were also similar. While the agitation did speed up the diffusion process as was demonstrated earlier by Taves: stationary conditions also yielded quantitative recoveries, provided the diffusion was allowed to proceed for longer periods (overnight). It is convenient to process a large number of samples overnight under these conditions. Because a shaker is not used, accidental splashing of sample plus acid onto the NaOH in the cap will not occur. For reliable results, it must be determined how long the diffusions should be carried out, with and without shaking, for each specific type of sample, sample size, and fluoride
348
ANALYTICAL CHEMISTRY, VOL. 64, NO. 4, FEBRUARY 15, 1992
Table I. Comparison of Results of Fluoride Analysis of Different Materials by the Present Diffusion Procedure and Other Procedures Not Involving Diffusion
sample
present diffusion method F- PPm mean f SE (n)
direct analysis (no diffusion) F- PPm mean f SE (n)
final mode of F analysis in both cases
3.05 f 0.038 (5) SH" 1.02 f 0.009 (4) ST" 1.01 f 0.003 (5) SH 1.21 f 0.019 (4) ST 1.22 f 0.008 (4) SH 4280 f 0 (3) ST 4227 f 22 (6) SH
2.99 f 0.015 (6) 1.09, 1.00
spectrophotometry1°
1.27, 1.17
p~tentiometry'~
4200, 4200
potentiometry16
water spiked with FSt. Paul city water human urine rat-bone ashb
potentiometry'*
"SH:Diffusion under agitation on the shaker. ST: Diffusion under stationary conditions. *Aliquotsof a bulk solution of bone ash were emdoved for the various analvses. Table 11. Effect of Methyl Ethyl Ketone (MEK) and Grinding of Samples in a Liquid Nitrogen Mill on Diffusion of Fluoride from Fluoroelastomers
35
30
I
0
sample finely shredded and snipped with scissors" no MEK MEK added Fppm Pppm 61
mean f SE
63 66 63 f 2.P
681 878 926 828 f 75
"Paricle size less than 2 mm
X 2
mm
I
2 25
sample ground in liquid
6
N2 mill no MEK
F-ppm
MEK added F-ppm
955 1215 1351 1174 f 116
2871 3019 3443 3111 f 171
X
0
ga 20
G
2". 0
content. For example, Taves et found that 48-h diffusion was required for quantitative separation of fluoride from soft tissues, although he9 had earlier demonstrated that fluoride could be separated from bone, urine, and serum supernatants (after protein precipitation) in 1-6 h by the very same diffusion procedure. As shown in Table I, results of analysis of water, urine, and bone samples by the present diffusion method and by other reference methods are in excellent agreement. Thus, it is possible to quantitatively separate fluoride as TMFS by diffusion in test tubes without the use of grease (procedure A). However, procedure A is not adequate for determining fluoride in fluoroelastomer samples. The fluoride generated in situ during curing is trapped within the heavily cross-linked polymer matrix which is insoluble in acids. The fluoride so generated is not easily accessible to acid and therefore not readily available for diffusion. This problem was now overcome by the use of methyl ethyl ketone (MEK) as an adjuvant to perchloric acid. MEK penetrates into the texture of the insoluble cross-linked mass and carries with it the acid and HMDS to the sites where fluoride is trapped. Thereby the trapped fluoride is protonated to H F and subsequently converted to TMFS which eventually diffuses out. This process is further enhanced by increasing the surface area of the sample prior to diffusion by grinding it in a liquid nitrogen mill to fiie powder. Whereas only about 60 ppm fluoride was detected in a fully cured fluoroelastomer sample by the standard diffusion procedure (procedure A), as much as 3000 ppm fluoride was found in the same sample when it was submitted to the same diffusion procedure after it was ground in the liquid nitrogen mill and wetted with MEK (Table 11). The rates of diffusion of fluoride as TMFS from a fully cross-linked fluoroelastomer sample under stationary and nonstationary conditions are shown in Figure 3. It is conceivable that, in spite of the additional steps described in procedure B, not all the fluoride trapped in the heavily cross-linked matrix is available for diffusion. Future
0
10
20
30 40 50 Hours of Diffusion
60
70
Flgure 3. Diffusion of fluoride as trh"eylflu0rosilane in test tubes from a fully cured fluoroelastomer sample under stationary (0)and nonstationary (0)conditions. Micromoledg BPAF
PCI
F-
I
a
I
200 180 160 140
10 5
' __-----
ODR
VI
E
-
z
60
120 100 80 60 40 20
40
20
f
$
a
0
0
5
10
15
20
Minutes (3W9F)
Flgure 4. Progressive changes in the levels of fluoride released (F-), free bisphenol AF (BPAF), the phosphonium ion (P'), and the cross-link density reflected by the torque (ODR) during the curing of VF,/HFP copolymer with BPAF.
innovations might reveal more diffusible fluoride. The present approach has yielded valuable information needed in understanding some aspects of solid-phase cure chemistry of fluor~elastomers.'~Some typical data are shown in Figure 4. The release of fluoride ion precedes and runs parallel to the rise in the cross-linkdensity determined with an oscillating disk rheometer (ODR). This reflects the creation, by dehydrofluorination, of unsaturated sites required for cross-linking. Also, it has been possible to determine which of the cure ingredients or combinations thereof could cause dehydrofluorination and to what extent. While Ca(OH)2and MgO have limited capacity to dehydrofluorinate the elastomer, and benzyltriphenylphosphonium chloride (PCl) has virtually no capacity for such dehydrofluorination, a combination of either of the bases with PCl is endowed with remarkably enhanced capacity for dehydrofluorination. Addition of bisphenol AF (BPAF) to Ca(OH), and PCI seems to further augment the
Anal. Chem. 1992, 6 4 , 349-353
Table 111. Release of Fluoride from VF,/HFP Copolymer at 335 OF by Selected Combinations of Compounding Ingredients
microname F released per gram of the copolymer
ingredients" added to VF2/HFP copolymer a b C
d e
f g
h i j
none
BPAF PC1 Ca(OH), Ca(OH)2,BPAF MgO MgO, BPAF MgO, PC1
Ca(OH)2,PCl Ca(OH)2,PCl, BPAF
55 46 61 347 269 807 677 1729 3399 4531
tions. Thanks are also due to D. L. Stanek and L. D. Winter for helpful discussions. NO. BPAF, 147861-1; F, i6w-488; H ~ O7732-185; , (FzC=CFCFS-H~CECF~), (copolymer), 9011-17-0.
REFERENCES (1) (2) (3) (4) (5) (6) (7) (8)
(9) (IO) (11) (12) (13) (14)
" Concentrations of ingredienta (parts per hundred) added to the elastomer were BPAF, 1.5;PC1, 0.475;Ca(OH),, 6.0;and MgO, 3.0. (15)
fluoride release (Table 111). These results demonstrate the usefulness of the present procedure.
ACKNOWLEDGMENT I wish to thank R. A. Prokop, R. A. Guenthner, and G. H. Millet for their interest and support during these investiga-
349
(16) (17)
Singer, L.; Armstrong, W. D. Anal. C h . 1954, 2 6 , 904-906. Stegm"a, H.: Jung, G. F. 2. Anal. Chem. 1959, 375, 222-227. Hall, R. J. AnaW~t1963, 88, 76-83. Frere, F. J. Anal. Chem. 1961, 33, 646-645. Wharton, H. W. Anal. Chem. 1962, 34, 12961298. Venkateswarlu, P.; Sita, P. Ana/. Chem. 1971, 49,758-780. Singer, L.; Armstrong, W. D. Anal. Bbchem. 1965, 10, 495-500. Baeumler, Von J.; Glinz, E. MM. Geb&te Lebensm. Wg7. 1964, 55, 250-264. Taves, D. R. Talenta 1868, 15, 969-974. Qreenhalgh, R.: Riley, J. P. Anal. Chlm. Acta 1961, 2 5 , 179-188. Taves, D. R. Talenta 1968, 75, 1015-1023. Frant, M. S.; Ross, J. W., Jr Science 1966, 154, 1553-1555. ORION RESEARCH Instructkm Menuel, Fluoride Elsctrodea, 1982. Taves, D. R.; Forbes, N.; Slbwman, D.; Hicks, D. In FluorMee: Effects on V q t a t k m , AnhnelsandHumns; Shupe, J. L., Peterson, H. B., Leone, N. C., Eds.; Paragon Press: Sat Lake City, 1983; pp 189-193. Singer, L.; Armstrong, W. D.; Vogel, J. J. J . Lab. Clln. M .1969, 74, 354-358. Singer, L.; Armstrong, W. D. Anal. Chem. 1968, 40. 613-614. Venkateswarlu, P.; Koib, R. E.; Guenthner, R. A. Po@w preprtnts 1990, 37.360-361.
RECEIVED for review July 25,1991. Accepted November 15, 1991.
Detection and Determination of Dilute, Low Molecular Weight Organic Compounds in Water by 500-MHz Proton Nuclear Magnetic Resonance Spectroscopy D. Bruce Fulton, Brian G. Sayer, and Alex D. Bain* Department of Chemistry, McMaster University, Hamilton, Ontario, Canada U S 4Ml Harold V. Malle Wastewater Technology Centre, 867 Lakeshore Road, P.O. Box 5068, Burlington, Ontario, Canada L7R 4L7 The feasiblltty of using hlgMleM proton NMR spectroscopy to analyze aqueous solutlons of organlc solutes at mlcromolar concentrations has been evaluated. N-Nltrosodhnethylamlne (NDMA) and benzene served as model compounds. The WATR (water attenuatlon by transverse relaxatlon) method of solvent suppresslon was optknlzed to permlt detectlon of the protons of analytes In the submicromolar concentratlon reglme. All data were collected In blocks to permlt quantltatlve estlmatlon of detectlon llmlts and experimental error. Externally referenced peaks heights, rather than peak Integrals, were used to measure solute concentration. The detection limn for NDMA, uslng a 500-MHz Instrument for 2 h, was 510 ng/mL and, for benzene, ushg a 4 4 acgulsltbn, was 35 ng/mL. The technlque was used to monltor the effect of UV lrradlatlon on a 10 pg/mL sample of NDMA. The NDMA was ellmlnated; the predomlnant photoproduct was dlmethylamine (DMA).
INTRODUCTION Concern about the contamination of freshwater systems by organic compounds is increasing, particularly where public water supplies may be affected. For example, the Ministry of the Environment in the Province of Ontario, Canada, has 0003-2700/92/0384-0349$03.00/0
recently lowered the allowable limit of N-nitrosodimethylamine (NDMA) from 14 parts per trillion to 8 ppt. The purpose of this work is to develop a method for the detection of low level concentrations of specific organic compounds directly in water without the requirement of separation or preconcentration. Proton nuclear magnetic resonance (NMR) spectroscopy is not usually considered as a method for analyzing aqueous solutions of very dilute (4Hg/mL) organic solutes, since it is perceived to suffer from certain disadvantagea. First, NMR spectroscopy is inherently less sensitive than other types of spectroscopy because it involves radio-frequency transitions. Second, proton NMR spectra of aqueous solutions are dominated by the water signal (effectively 110 M), which must be suppressed to bring micromolar solute signals within the dynamic range of detection. Third, quantitative measurement of NMR intensities requires very careful data acquisition and analysis.'P2 On the other hand, the rich information content of the spectra gives NMR certain advantages over other analytical techniques. In particular, the high resolution of chemical shifta allows for the analysis of a mixture of similar species without the need for chemical separation. Because NMR is nondestructive, the sensitivity problem can in principle be overcome by accumulating a sufficiently large number of transients. However, in practice one is limited 0 1992 American Chemical Soclety
