Anal. Chem. 1909, 6 1 , 1325-1327
LITERATURE CITED (I) Mori, %ciao Anal. Chem. 1983, 55, 2414-2416. (2) Overton, J. R.; Rash, J.; Moor, L. D., Jr. Sixth InternationalGPC Seminar Proceedlngs, Miami Beach, FL, 1968,422-432. (3) Paschke, E. E.; Bidlingmeyer, 8. A.; Bergmann, J. G. J. folym. Sci., Porn. Chem. Ed. 1977, 15, 983-989. (4) U g k , c. V.; Alzicovici. A,; Mlhaescu, S. Eur. folym. J . 1985, 21, 677-679. (5) Berkowltz, Steven J. Appl. folym. Sci. 1984,29, 4353-4361. (6) Drott, E. E. Liquid Chromtography of Polymers and Related Materials;
Cams, Jack, Ed.; Chromatographic Science Series Vol. 8, Marcel Dekker: New York, 1976;pp 41-50.
1325
(7) Hibi, Kiyokatsu; Wada, Akio; Mori, Sadao Chromatographla 1886, 1 1 ,
--- - . ..
FSS-fiAl
(8) Provder, Theodore; Woodbrey, James C.; Clark, James H. Sep . Sci. 1971, 6 , 101-136. (9) Mori, Sadao Anal. Chem. 1981, 53, 1813-1818. (IO) Janssen, R.; Ruysschaert, H.; Vroorn, R. Makromol. Chem. 1984, 35,
153-158. (11) Russell, G.A.; Henrichs, P. M.; Hewitt, J. M.; Grashof, H. R.; Sandhu, M. A. Macromolecules 1981, 14, 1764-1770.
for review December 7, lg8&Accepted March 13, 1989.
Gas Chromatographic Determination of Water Using 2,2=Dimethoxypropane and a Solid Acid Catalyst Kevin D. Dix, Pamela A. Sakkinen, and James S. Fritz*
Ames Laboratory-US. Department of Energy and Department of Chemistry, Iowa State University, Ames, Iowa 50011
The amount of water in various organic and Inorganic substances Is determined by reaction with 2,2dknethoxypropane (DMP), folbwed by measurement of a product of the reactlon (acetone) by caplllary column gas chromatography. The reaction of water with DMP requires only 5 mln when Nafion Is used as a solid acld catalyst. Various experlmentai parameters are investigated to optlmlre the analytical procedure. The percentage of water In a variety of analytical samples was determined successfully.
for determining water in nitroglycerin-nitrocellulose pastes by GC. In the present method the sample is combined with a solution containing DMP and an internal standard. A small amount of Ndion (7) is added to catalyze the reaction of water with the DMP. Then an aliquot is injected into a gas chromatograph equipped with a capillary column and a flame ionization detector. The amount of water is calculated from the ratio of the peak areas (or peak heights) of the acetone and internal standard. The method is both sensitive and convenient; it has been applied successfully to a wide variety of analytical samples.
INTRODUCTION
EXPERIMENTAL SECTION
The determination of small amounts of water in various organic and inorganic compounds is of great practical importance. The Karl Fischer method (1) is perhaps the most widely used procedure for the determination of water. Although this method works well in many cases, the commercial reagents are rather costly, the visual titration end point is difficult to discern, and there are numerous interferences. Although water can be determined directly by gas chromatography using a thermal conductivity detector (TCD) (2), surprisingly few laboratories seem to use this approach. Furthermore, a packed column rather than a capillary GC column must be used because of the large cell volume of the traditional TCD. A few authors have used the acid-catalyzed hydrolysis of 2,2-dimethoxypropane (DMP), the dimethyl ketal of acetone, as a way to determine water.
Glassware and Apparatus. The reactions were carried out in 5-mL glass microreaction vessels fitted with Teflon-lined septa (Supelco Glass Co., Bellefonte, PA). Reagents and Chemicals. The 2,2-dimethoxypropane was obtained either from Aldrich Chemical (Milwaukee,WI) or from Eastman Kodak Chemical (Rochester, NY). Nafion 1100 EW resin (60-100 mesh) was purchased from C. G. Processing, Inc., (Rockland, NY) and was dried at 100 "C for 3 h under vacuum before use. Amberlyst-15 resin was obtained from Rohm and Haas (Philadelphia, PA) and dried as above. All other reagents were of reagent grade or better and were "blanked" before use, Distilled water was further purified with the Barnstead Nanopure I1 system before use. Gas Chromatography. A Hewlett-Packard 5790A gas chromatograph equipped with a flame ionization detector (FID) was used in the split mode. The split ratio was 80-100:l and was held constant during a series of experiments. The injection liner was packed with a small amount of 80-100-mesh silanized glass beads to prevent contamination of the column with nonvolatile components. The beads were changed periodically, and the injector was held at 150 "C. Two different carrier gas flow rates were used in this study. Initially, a flow of 2.5 mL/min of zero grade He was used with an oven profile of 5.2 min at 40 "C.Later, it was found that a flow rate of 5.0 mL/min and an oven profile of 2.2 min at 40 "C provided adequate resolution and decreased the analysis time. In each case, the oven temperature was stepped to 220 "C after the initial hold at 40 OC. This rapid increase in temperature served to remove any later-eluting peaks in less than 5 min. The column was a 30 m X 0.53 mm J+W DB-5 Megabore with a film thickness of 1.5 pm. The detector was an FID held at 250 "C. Reactant Solution. A reactant solution was prepared in a dry 100-mL volumetric flask from a 5-mL aliquot of DMP and
CH3C(OCH3)2CH3+ H 2 0 -%CH3COCH3+ 2CH30H Critchfield and Bishop (3) determined water by reaction with DMP in the presence of methanesulfonic acid and measured the acetone formed by infrared spectroscopy a t 5.75 bm. Hager and Baker ( 4 ) made a cursory investigation of the use of DMP for the indirect GC determination of water. Martin and Knevel(5) proposed a quantitative method for water by reaction with DMP and measurement of the change in height of the GC peaks of DMP and acetone. The method required accurate weighing of both DMP and acetone, as well as the sample itself, and the sensitivity of the method was somewhat limited. Blanco et al. (6) used a somewhat similar method 0003-2700/89/0361-1325$01.50/0
0 1989 American Chemical Society
1328
ANALYTICAL CHEMISTRY, VOL. 61, NO. 13, JULY 1, 1989
a 1-mL aliquot of 3-methylpentane, the internal standard, and diluted to the mark with pure solvent. This solution permitted a simple one-step addition of the necessary chemicals. Both ethyl acetate and dimethylformamide were used as the solvent with no problems. Standardization. Internal standardization was used (8). This method allows for small differences in the injected volume. First, the reactant solution was chromatographed to determine the amount of initial acetone present. Then 10 pL of acetone was added, and the solution was chromatographed again. By subtraction of the initial acetone, a response factor can be calculated. Since this factor varies only very slightly during a day, it is only necessary to do this periodically. Procedure. The required amount of Nafon resin was weighed into the reaction vessel, which was then capped. Then 1mL of the reactant solution was added to the vessel via syringe t o minimize uptake of water from the atmosphere. The vessel was shaken for a specific length of time and a 1-pL aliquot injected into the gas chromtograph. The area of the acetone peak represents the water blank of the system. Next, a specific amount of liquid sample was introduced via syringe, or solid was added to the reaction vessel. The mixture was shaken again for a specific length of time, and a 1-pL aliquot was introduced into the gas chromatograph. A ratio of peak areas of acetone to internal standard is calculated. This ratio, minus the blank, yields a relative response to acetone generated from the reaction of water in the sample. From the response factor previously determined for acetone, a value for the absolute amount of acetone is found. This value is easily converted to the amount of water by considering the stoichiometry of the reaction and the molecular weights of acetone and water: Ww = weight of water WA = weight of acetone
ww -- W~
1 mol of water 1 mol of acetone
X
MW of water MW of acetone
RESULTS Time of Reaction. The reaction vessel containing 1 mL of reactant solution, 2 mL of ethyl acetate, and 12 mg of Nafion was shaken, and aliquots were taken a t various time periods for gas chromtographic analysis. The acetone peak attains its maximum height after 5 min, and its height remains constant between 5 and 45 min of reaction time. From these results, 5 min was selected as the optimum reaction time for the determination of water. A similar experiment was performed using 12 mg of Amberlyst-15 cation-exchange resin instead of Nafion as the acid catalyst. The curve requires approximately 25 min to reach the plateau region. This experiment shows Nafion to be a superior catalyst. Amount of Nafion Resin. Using the reaction conditions above, we varied the amount of Nafion catalyst. I t was determined that at least 6 mg of Nafion was needed to reach the maximum height of the acetone peak within 5 min. For all remaining determinations of water at least 10 mg of Nafion was added as the catalyst. Calibration Curves. In the reaction system used it is necessary to know the range over which water can be determined and whether the formation of acetone is linear over this range. A nonlinear response indicates the possibility of a side reaction or the loss of acetone by other means. A 2-mL portion of dried ethyl acetate was added to 1 mL of the reactant solution and 12 mg of Nafion as before, and the amount of water in this blank was determined from the area of the acetone peak. Then a small amount of water was added via a 0.5-pL syringe, and the amount of water now present was determined chromatographically. This procedure was repeated until there was no further increase in the amount of acetone generated from the reaction. Figure 1 shows the relative response of acetone as a function of the amount of water added. The response is linear up to about 0.275% water
I4 W
9
12-
2 l0-
Somole
W
08a
E
06-
,
041 02
000
005
010
015 020 0 2 5 030 PERCENT WATER ADDED ( v / w l
035
040
Figure 1. Relative response of acetone for the addition of increasing amounts of water in a 2.0-mL sample of ethyl acetate. Other conditions are given in the text.
(v/v). At this point all of the DMP is consumed and the formation of acetone ceases. This point agrees with the point that can be calculated from the amount of DMP that was initially added. The correlation coefficient for the linear portion is 0.9995. Extrapolation of this line to zero relative response and subtracting the blank yield the amount of water in the 2-mL sample of ethyl acetate. This value, 0.047% (v/v), agrees closely with 0.045% found by using the internal standardization method. A similar series of experiments were performed with samples containing higher percentages of water. This is done by simply reducing the size of the sample. Only a 0.20-mL sample of ethyl acetate was used, and the water additions were similar to those before. A straight-line response is obtained for up to 2.75% water in the sample. The correlation coefficient for this line is 0.9991. Although the experiment was not performed, it seems that virtually any amount of water, even up to loo%,can be determined just be reducing the sample size. Since methanol is also produced in this reaction, its peak area can also be plotted. The correlation coefficient for such a plot from these data is 0.9963. Similarly, the decrease in the DMP peak was monitored with a correlation coefficient of 0.9973. Reproducibility. The reproducibility of the indirect GC method was estimated by independently determining the amount of water in six 2-mL samples of tetrahydrofuran. The mean of the six samples was 0.0523% water with a standard deviation of 0.0015%. This corresponds to a relative standard deviation of 2.8%. The relative standard deviation of the blank alone was 1.2% for the six runs. Limit of Detection. The limit of detection of water by the indirect GC procedure is apt to depend more on external factors than on the measurement method itself. Thus the detection limit depends to a major extent (a) on the ability to dry and keep dry Nafion, reactant solution, and glassware and (b) on avoiding adding water during transport or weighing. It is this "extra" water that makes up the blank. There are larger variances associated with a larger blank, and these inevitably result in a higher limit of detection. By careful handling and by drying the components of the reactant solution over calcium hydride and drying the Nafion a t 110 "C, we were able to determine as low as 0.001% water in various samples. Determination of Water i n Various Samples. Water in several organic liquids was determined both by the gas chromatographic method and by the well-known Karl Fischer (KF) method (see Table I). The KF solution was standardized just before use, and the titrations were performed in triplicate. The end point in the Karl Fischer titration is often difficult to determine visually, and in some cases precipitation
ANALYTICAL CHEMISTRY, VOL. 61, NO. 13, JULY 1, 1989
Table I. Comparison of Water Determination by Indirect GC and by Karl Fischer Titration % water (w/w) GC method KF titration
compound ethyl acetate tetrahydrofuran dimethylformamide methylene chloride dioxane
0.059 0.0355
0.056 0.0361
0.0075
0.0074
0.0121 0.489
0.0142
0.479
Table 11. Determination of Water in Liquid and Solid Samples
compound
sample size, mL
70 water (w/w)
water spikes, mg
added found
Liquid Samples methylene chloride toluene ethyl acetate 1,1,2-trichlorotrifluoro-
0.0121
0.50
0.0186
0.50
0.2309 0.0030
0.50 0.50
0.58
0.50 0.50
0.43
0.40 0.40
0.37 0.41
2.00
0.0064 0.4078 0.0029 0.0127 0.0800
0.50
0.50
0.05
3.0707
0.50
0.53
0.40
0.39
0.40
0.39
2.00 2.00 2.00 2.00
ethane "dry" ethyl acetate
methyl ethyl ketone carbon disulfide anisole tetrahydrofuran 1-propanol
2.00 2.00 2.00 2.00
0.51 0.49 0.48
0.58
Solid Samples oxalic acid dihydrate (28.6% H20) phenol
10
50
30.1 0.701
1327
Second, the amount of water in 20.0 mg of ascorbic acid was determined by the regular GC method. In both cases the amount of water in the ascorbic acid was 0.026% (w/w).
DISCUSSION It might be expected that samples containing high-boiling or nonvolatile components could not be analyzed for water by a GC method because of irreversible adsorption on the chromatographic column. However, our experiments indicate that this is not the case with the present method. Use of a capillary column means that a smaller sample is injected than in previous methods in which a packed GC column was employed. Insertion of glass wool and silanized glass beads seems to adsorb the nonvolatile sample components while allowing acetone to pass quantitatively onto the column. A rapid temperature increase a t the end of a chromatographic run serves to remove any high-boiling compounds that might be retained by the column. The absolute amount of water that can be determined in this procedure could obviously be increased by adding a larger amount of DMP to the sample. However, this could increase the blank and reduce the accuracy in samples containing a low percentage of water. It is also likely that other acetals or ketalrJ could be used in place of DMP for the determination of water. Registry No. DMP, 77-76-9; HzO, 7732-18-5; Nafion, 3946459-0; dimethylformamide, 68-12-2; dioxane, 123-91-1;methylene chloride, 75-09-2; toluene, 108-88-3; ethyl acetate, 141-78-6; 1,1,2-trichlorotrifluoroethane, 76-13-1;methyl ethyl ketone, 7893-3; carbon disulfide, 75-15-0; anisole, 100-66-3;tetrahydrofuran, 109-99-9;1-propanol, 71-23-8;oxalic acid dihydrate, 6153-56-6; phenol, 108-95-2. LITERATURE CITED
occurred, which made the end point even more difficult to locate. Despite these difficulties, good agreement was obtained between the percentages of water determined by the two methods. Several liquid and solid samples were analyzed for water by the GC method. Then a measured amount of additional water was added to each sample, and the total amount of water present was determined by the GC method. Table I1 gives results for the percentages of water in the original samples, the amounts of water added, and the amounts of added water found by the GC method. The excellent recoveries of added water demonstrate that the new GC method gives dependably accurate results for a wide variety of samples. Samples containing even a small amount of ascorbic acid cannot be titrated by the Karl Fischer method. However, no difficulty is encountered when the indirect GC method is used. This was demonstrated in two ways. First, a 2-mL sample of dimethylformamide was analyzed for water by the GC method before and after addition of 20.0 mg of ascorbic acid.
(1) Scholz, E. Karl Fischer Tihetion; Springer-Verlag: Berlin, 1984. (2) Mitchell, J., Jr.; Smith, D. M. Aquameby; Wiley-Interscience: New York, 1977. (3) Critchfield, F. E.; Bishop, E. T. Anal. Chem. 1961, 33, 1034. (4) Hager, M.; Baker, G. R o c . Mont. Acad. Sci. 1962. 22, 3. (5) Martin, J. H.; Knevel, A. M. J . h m . Sci. 1965 5 4 , 1464. (6) Blanco. J. A.; Rucci, A. 0.; Revuelto, S. C.; Dubini, A. A,; Gonzalez, R. A. Propellants Expos. 1979. 4 , 27. (7) Po&merlc Reagents and Cata&sts; Ford, W. T.. Ed.; ACS Symposium Series 308; American Chemical Society: Washington, DC, 1986; p 42. (8) Debbrecht, F. J. Modern Practice of Oas Chromatography, 2nd ed.; Grob, R. L., Ed.; Wiley: New York, 1985.
RECEIVED for review October 4,1988. Revised March 15,1989. Accepted March 28,1989. Ames Laboratory is operated for the U S . Department of Energy under Contract No. W7405-ENG-81. This work was supported by the Director of Energy Research, Office of Basic Energy Sciences, and by a research fellowship from Phillips Petroleum Co., Bartlesville, OK. Portions of this paper were presented at the 30th Rocky Mountain Conference, Denver, CO, August 1988.
