Interactive effects of temperature, salt concentration, and pH on head

Nov 1, 1979 - Interactive effects of temperature, salt concentration, and pH on head space analysis for isolating volatile trace organics in aqueous e...
0 downloads 6 Views 772KB Size
ANALYTICAL CHEMISTRY, VOL. 51, NO. 13, NOVEMBER 1979

2187

Interactive Effects of Temperature, Salt Concentration, and pH on Head Space Analysis for Isolating Volatile Trace Organics in Aqueous Environmental Samples Stephen L. Friant Academy of Natural Science, Philadelphia, Pennsylvania 19 104

Irwin H. Suffet” Department of Chemistry, Environmental Studies Institute, Drexel University, Philadelphia, Pennsylvania 19 104

A systematic approach is presented for the isolation and quantification of volatile trace organics from aqueous solutions by head space analysis. Fundamental information is obtained on the partition process for a multiple solute system consisting of model compounds under varying aqueous matrix condlons, Le., pH, temperature, and salt addition. Interactive effects between parameters are quantitatively shown by the use of the thermodynamic equilibrium partition coefficient. A general optimum head space isolation methodology is obtained from statistical evaluation of the effect of parameter variation on the partition coefficient. The optimum head space analysis sampling conditions of pH 7.1, 50 O C and 3.35 M sodium sulfate were determined from a statistical design. At the optimum conditions of this design, enrichment factors d the vapor phase of up to 66 times were achieved as compared to a reference state of pH 7.1, 30 OC, without electrolyte addition. Under optimum head space analysis conditions, river and drinking water can be routinely profiled for volatile trace organics.

Two different analytical approaches are utilized to determine organic compounds in water (1). The first consists of quantitative analysis of one pollutant such as bis(2chloroethyl) ether (2), a group of chemically related pollutants such as the trihalomethanes ( 3 ) ,or a pollutant and its related breakdown products such as an organophosphorus pesticide, fenthion, and its oxons, sulfoxides, and sulfones ( 4 ) . T h e second approach is a general “screening” procedure for organic compounds. This consists of a qualitative analysis with semiquantitative evaluation of the constituents found. An example of a general screening procedure would be the analysis of volatile organics responsible for taste and odor incidence in drinking water (5). T h e isolation method used to collect trace organic compounds from water is the primary basis of any analysis as it defines the type of compounds to be analyzed, the maximum recovery of a compound, the precision and accuracy of the method, and possible co-extractives that may interfere with a subsequent quantitative analytical step. Each isolation technique has a selective efficiency for specific compounds as reviewed by Suffet and Radziul (1). Thus the analytical problem is to define the parameters involved in an isolation method to enable optimization for a specific purpose. Vapor phase isolation of trace organics present in aqueous samples can be divided into two areas: dynamic volatile organic analysis and equilibrium head space analysis as discussed by Weurman (6). Two commonly used dynamic organic analyses are the gas stripping (purge and trap) procedure of Bellar and Lichtenberg (3)and the closed looped stripping procedure of Grob and Zurcher (7). The purge and trap method is used primarily for analysis of volatile organics of less than 2% solubility and boiling points below 150 “C (3). 0003-2700/79/0351-2167$01.OO/O

Head space analysis (head gas analysis) is the static sampling of the vapor phase in thermodynamic equilibrium with the aqueous phase. The initial liquid phase concentration is determined from the measurement of the equilibrated vapor phase concentration and the equilibrium partition coefficient. Trace organics which favor the vapor phase are easily determined a t the microgram per liter concentration level by head space analysis. Examples of these trace organics are hydrocarbons and chlorinated hydrocarbons (8). McAuliffe (8) found that vapor phase partitioning for the following chemical classes followed the order: alkanes > olefins > cycloalkanes > aromatics. It was also shown that within each chemical class an increased vapor partition was observed as the molecular weight increased. Head space analysis can selectively separate nonpolar volatiles from more water soluble alcohols and acids whereas purge and trap analysis can remove more of these polar compounds (7). Enhancement of vapor phase partitioning has been achieved by both the addition of an electrolyte and increased temperatures (9, 10). Quantification of the interactive effects of temperature and salt addition on the vapor phase partitioning process is of interest for both types of vapor phase trace organic isolation methods. This study was undertaken to investigate the quantitative effects of salt addition, temperature, and pH on the vapor phase partitioning of selected compounds utilizing an ANOVA analysis of an experimental design. The ult.imate goal was to understand how to incrdiase the vapor concentration of organics. Model compounds of high polarity and solubility were studied to investigate the limits of the ability of head space analysis to partition these materials into the vapor phase.

Model Compound Selection for Characterization of the Air:Water Partitioning Process. Criteria for the selection of the four model compounds (Table I) were based on several factors. These included (1) ability for simultaneous GC analysis, (2) aqueous solubility, and (3) volatility. In addition, the compounds should represent industrial process chemicals and should be representative of several major chemical classes. The choice of model compounds that represent major chemical classes was determined using the concept of the solubility parameter (6T). The solubility parameter is a measure of a compound’s polarity in a pure state; the square root of the cohesive energy density (11). It was felt that the selection of compounds based on this concept would yield results that could be generalized since the solubility parameter is not solely dependent on functional groups but is a measure of the compound’s total polarity. The major applications of the solubility parameter theory have been in chromatography, reverse osmosis membrane rejection predictions, and polymer chemistry (11). The solubility parameter approach applied to the chromatographic 0 1979 American Chemical Society

2168

ANALYTICAL CHEMISTRY, VOL. 51, NO. 13, NOVEMBER 1979

Table I. Model Compounds and Experimental Conditions

polarity (11 ) dipole orientation dipole orientation proton donor proton acceptor

6

initial aqueous phase concn, (14) compound mgl L 9.5 methyl ethyl ketone 5.64 11.0

nitroethane

20.90

12.0

n-butanol

40.49

10.1

p-dioxane

93.30

Table 11. Statistical Design ( 3 x 3 x 2 ) for t h e Investigation of Salt, pH, and Temperature Effects on the Partitioning Process run no. 1

9 10 11

12 13

1100-1200 mL 1000-1040 mL 5.0 mL 7.1 0.00-3.35 M 0.00-10.02 0.20 M

EXPERIMENTAL Apparatus. All chromatography was conducted on a Tracor MT-550 gas chromatograph equipped with dual flame ionization detectors. The chromatographic column was 8 feet X l / * inch i.d. stainless steel packed with 20% SE-30 on 80/100 mesh Gas Chrom Q. The column temperature was maintained isothermally a t 130 "C. Inlet, outlet, and detector temperatures were 180,200, and 200 "C, respectively. Glassware in contact with both water and vapor phases was silanized to minimize surface absorption. First, the glassware was washed with detergent. This was followed by rinsing with distilled water and air drying. The dry surface was treated with Glas-treet (Alltech Associates) and rinsed with anhydrous methanol. The sampling bottle used for laboratory test conditions was a modified 2-L Pyrex reagent bottle. The neck was reformed by using $30 glass O-ring joints and the top was rounded and sealed. The two joints were sealed by a Buna rubber O-ring and a compression clamp. To allow syringe sampling of the vapor two (l/,-inch diameter) glass sidearms were attached. A stainless steel (1/4 to l/g inch) Swagelok reducing union was attached to the glass sidearms by '/(-inch rubber O-rings and a Swagelok nut. The sampling port was sealed by a chromatographic silicon septum (15). The gas syringe was a Precision Scientific (Baton Rouge, La.) pressurizable Series A-2 10.0-mL gas syringe equipped with sideport needle and stop/go valve. The valve permitted the sampling of larger vapor volumes by allowing compression of the sample prior to injection, thereby reducing peak broadening. The syringe also permits easier injections since the pressure in the syringe can be equal to the GLC column head pressure.

1.41 1.41 1.41 1.41 1.41 1.41 3.35 3.35 3.35 3.35 3.35 3.35

8

5h

partitioning process is described by functional probes (methyl ethyl ketone, n-butanol, nitroethane, and dioxane) (11-14. T h e polarity probes are representative of ketones, alcohols, nitro groups, and ethers. These functional groups are representative of the intramolecular forces of dispersion, dipole orientation, and proton donor and acceptor capabilities ( 11 ) . Keller (12) emphasized that nitromethane is an excellent probe for classical polarity (dipole orientation) with little acid-base function. The same appears to hold true for nitroethane (6T = 11.0) except at alkaline p H where a n aci-nitrogen effect becomes important. Methyl ethyl ketone (JT = 9.5) was considered t o be more of a combined polarity by Hartkopf e t al. (13)than of specifically dipole orientation as originally described. Dioxane (6T = 10.1) and butanol (6, = 12.0) have predominant proton acceptor and donor forces, respectively. T h e total solubility parameter values of these compounds ranged from 9.4 t o 11.6 on a scale of 6 for nonpolar alkanes t o 23.5 for very polar water (14).

0.00 0.00 0.00 0.00 0.00 0.00

2 3 4 5 6 7

General Experimental Conditions equilibration time vapor volume liquid volume volume of vapor samples PH sodium sulfate concentration ionic strength orthophosphate buffer (15)

salt concnM

14

15 16 17 18

temperature, pH

"C

4.5 7.1 9.1 4.5

30 30 30 50 50 50 30 30 30

7.1

9.1 4.5 7.1 9.1 4.5 7.1 9.1 4.5 7.1

9.1 4.5 7.1 9.1

111

121 131 112 122 132 211 221 231 212 222 232

50

50 50 30 30 30 50

31 1

321 331 31 2 3 22 332

50

50

30 pH 4.5 7.1 9.1

code

50

0.00

salt concentration, M 1.41 3.35 0.00 1.41

4.19 3.90 4.56

21.3 20.0 18.7

118

109 105

21.3 20.0 18.7

39.8 37.6 35.0

3.35 234 260 229

All experiments were run isothermally in a constant temperature air bath controlled to k0.5 "C. All samples were stirred on a magnetic stirrer. Reagents and Chemicals. All chemicals used were reagent grade or better. Solvents were pesticidal quality. Procedures. Table I shows the group of four model compounds studied, aqueous phase concentrations, and experimental conditions. The ionic strength and pH of all experiments were adjusted first to 0.2 M with orthophosphate buffers (26). Ionic strength was subsequently adjusted with Na2S04as desired. All sample and standard solutions were prepared based on the density of the pure solute. Sample solutions were prepared by individually pipeting with a microsyringe a known volume of solute into a 1-L volumetric flask filled with the appropriate buffer solution. Standards were prepared (by density) in carbon disulfide. The vapor phase concentration of the solutes was determined by comparison of the area of the vapor phase injections to standard curve areas of the solutes in carbon disulfide. Experiments requiring the addition of an electrolyte were completed by placing the salt (anhydrous sodium sulfate) in a clean sample bottle at the required experimental temperature 16 h prior to the study period to reduce the time required to dissolve the salt. Samples and standard solutions were prepared daily. A saturated solution of Na2S04was used that contains 475 g/L (3.35 M) when dissolved in 0.2 M orthophosphate buffer. A solution of Na2S04containing 200 g/L (1.4M) when dissolved in 0.2 M orthophosphate buffer was also used. A maximum sampling temperature of 50 "C was chosen to minimize water vapor condensation in the syringe during pressurization and enable ease of sample handling. A minimum temperature of 30 "C was chosen to maintain the temperature of equilibration of the head space bottle above ambient room temperature. Vapor samples were withdrawn through the sidearm sampling port of the sample bottle with a gas syringe. The syringe was flushed with the vapor phase prior to withdrawing the sample. After sampling the vapor, the syringe valve was closed and the

ANALYTICAL CHEMISTRY, VOL. 51, NO. 13, NOVEMBER 1979

Table IV. ANOVA Table for Methyl Ethyl Ketone SOURCE ss DF A B AB C AC BC ABC ERROR TOTAL

0.0 248535

1.0000000

0.1930203 0.0271621 0.0002594 0.0001816 0.0003457 0.0004975 0.0017965 0.248 116 6

2.0000000 2.0000000 2.0000000 2.0000000 4.0000000 4.0000000 18.0000000

MS

F CALC

PRCiB O F F

0.0248535 0.0965102 0.0135810 0.0001297 0.0000908 0.0000864 0.0001 244 0.0000998

249.0214 966.9896 136.0759 1.2997 0.9096 0.8659 1.2462

1.0000 1.0000 1.0000

2169

1 2

3 4 5 6

0.7029 0.5770 0,4953 0.6729

I

35.0000000 A. Temperature B. Salt Concentration

GROUP MEANS GRAND MEAN IS

0.072781

CORRECTION FACTOR

0.19069

1. 0.046506 2. 0.099056

C. pII

___-

1. 0.011533

1. 0.073391

2. 0.031092 3. 0.17512

2. 0.015711 3. 0.069228

Table V. Two-Factor Interaction of Significances for hfethyl Ethyl Ketonea K

temperature, "C C, ( 3 0 ° C ) C, (50 "C) mean

A, (0.00 M ) 0.421 X 2.00 x 1.21 x

salt concentration ( A ) A, ( 1 . 4 1 M ) 2.47 x 3.75 x 3.11 x

-

A, (3.35 M ) 11.1 x lo->

24.6 x 11.9 x

mean 4.66 x

10.1x 10.: '1.39 X 10.'

a Notes: Means of partition coefficients a t levels specified are averaged over pH 4.5, 7.1, and 9.1. The 90% confidence interval for A,C, is: 23.8-25.4 x Conclusions: Statistically significant maximum yield two-factor interaction occurs for A,C, (3.35 M salt concentration a t 50 "C).

volume reduced to l/lo sample volume followed by direct injection into the chromatographic column. Calculations. The basis for the evaluation of the magnitude of parameter effects was the partition coefficient (K):

where [A], and [A], are the equilibrium vapor and water phase concentrations of a solute A, respectively. The terms 7 , and y, are the corresponding activity coefficients. At equilibrium the activity terms are equal and in dilute solutions they approach unity. Equation 1 can then be rewritten in terms of solute equilibrium weights and volumes:

where W,, W, and V,, V, are the equilibrium weights and volumes for the solute in the vapor and water phase, respectively. If WT is the weight of the total solute in the system, W , = WT- W,. Substituting for W , in Equation 2 and defining the equilibrium volume ratio V,/ Vwas a yields: (3)

In a head gas experiment, if the initial volume of water and vapor are changed owing to the addition of an electrolyte, the initial a value must also be changed as the water volume increased the vapor volume decreases. The correction factor (CF) can be used: CF = V,/ Vvt where Vv,is the initial vapor volume. Therefore the initial a is multiplied by CF to obtain the new experimental a value. If no electrolyte is added, the CF term is unity while the addition of salt decreases CF to less than 1. Sampling Parameters. From preliminary work, the parameters shown to influence the partitioning process most significantly were pH, temperature, and the addition of an electrolyte (16). A quantitative investigation of these parameters for the vapor-water partitioning process was completed by a statistical design using an analysis of variance (ANOVA) to evaluate interactive results. The experimental design investigated the effects

of parameter variation on the partitioning process by a 3 X 3 X 2 analysis of variance, Table 11. The 18 experiments and replicates were completed using a random order ( 1 7 ) . Results of the design were reported as the equilibrium partition coefficient,K, and were used as input data to an IBM 370/168 computer using available APL statistical package ANOVA (18).

RESULTS AND DISCUSSION Statistical Design. Table I11 is a three-dimensional presentation of the results obtained from the 3 x 3 X 2 statistical experiments. Methyl ethyl ketone data are shown as a n example. The computer results of the analysis of variance summarizing main effects are shown in Table IV where A, B, and C represent temperature, salt concentration, and pH, respectively. As indicated in T h l e IV by the large probability of F , a significant two-factor interaction exists between salt concentration and temperature. Table V shows a two-factor interaction of significance existed a t both 1.41 and 3.35 M sodium sulfate and 50 "C. T h e results drawn from the statiistical design are: (a) p H had no effect on methyl ethyl ketone, butanol, and dioxane. The optimum p H for nitroethane was either 4.5 or 7.1; both were found to be equivalent. (b) The optimum salt concentration for all compounds was 3.35 M sodium sulfate. (c) The optimum sampling temperature for all compounds was 50 "C. (d) For all two-factor interactions, the maximum K occurred for either p H 4.5, or 7.1, 50 "C and 3.35 M sodium sulfate. From the above statistical approach the general head space analysis conditions for the optimum isolation of volatile trace organics was p H 7.1, 50 "C, and 3.:35 M sodium sulfate. T h e partition coefficients are shown in Table VI for all compounds investigated a t the optimum and reference state conditions. The order of increasing partition coefficients a t all experimental conditions is dioxane < butanol < nitroethane < methyl ethyl ketone. A plot of the partition coefficients is shown in Figure 1 for methyl ethyl ketone under the various conditions of salt and temperature.

2170

ANALYTICAL CHEMISTRY, VOL. 51, NO. 13, NOVEMBER 1979

Table VI. Partition Coefficient of Model Compounds

at the Optimum and Reference States

K x 10-3

compound methyl ethyl ketone nitroethane butanol dioxane

optimum reference conditions, condition, pH 7.1, 50 "C pH 7.1, 30 'C, 3.35 Ma no salta 2 60

72.5

"I4

50'

:I

/

3.90

44.3

2.89 0.746

13.7

0.278

0.2 M orthophosphate buffer.

Analysis of the Effects of Salt and Temperature on the Partitioning Process. An advantage of employing the statistical design approach for the selection of optimum head space analysis conditions is that interactions or main effects of interest can be factored from the design to enable a more comprehensive data evaluation. Salt and temperature effects expressed as enrichment factors are shown in Table VI1 for all compounds. Enrichment factors are defined as the ratios of the partition coefficients at the condition of interest to the reference conditions of p H 7.1, 30 "C, without the addition of salt. From Table VI1 the dominant force responsible for the partitioning process is salt addition a t the 3.35 M level, ranging from 2.7 to 5.2 times larger than the temperature effect. The effect of salt addition at the 1.41 M level yielded enrichment factors approximately equal to the effect of temperature although the salt-temperature interactions were significantly greater. The total enrichment factors shown represent the interactive effects of salt (3.35 M) and temperature (50 "C) over the reference conditions. Total enrichment factors ranged from 25 to 67 for the model compounds. The order of increased total enrichment factors is nitroethane < dioxane < butanol < methyl ethyl ketone. It is worth noting that the magnitude of enrichment is not the same for all compounds. This is in agreement with the work of Kepner et al. (19) and Nawar and Fagerson (20). They found that enrichment increased the vapor phase concentration but the relative proportion of each compound recovered was considerably altered. Table VI1 shows that the increased partition coefficients and enrichment factors are not in agreement. The order of increased partition coefficients also does not agree with the vapor pressuretemperature relationships of the pure solute (15). This qualitative discrepancy between pure vapor pressure trends and enrichment factors supports Dravnieks and O'Donnell (21) who indicated that vapor phase partitioning of anything but the pure solute is not only a function of vapor pressure but also of activity coefficients, presence of salt, complexing ability of the solute, and pH. Analysis of the Effect of pH on the Partitioning Process. The results of the analyses of variance showed that methyl ethyl ketone, dioxane, and butanol were not affected by pH. They apparently do not exhibit significant acid or base intramolecular bonding forces in the aqueous environment. Nitroethane showed maximum and equivalent vapor phase concentration at pH 4.5 and 7.1. This was expected since nitro substituted compounds in aqueous solution can form acinitrogen complexes; hence, solute-solvent interaction would be a t a minimum in acid or neutral solutions. The pH effect exhibited by nitroethane is discussed by Gould (22). An addition of base to a n aliphatic nitro compounds-e.g., nitroethane-consumed an equimolar quantity of base. This indicates a neutralization reaction. The reaction of the nitro group and base is due to the ionization of the C-H bond and

30'

r

1

L

I

I

I

I

0

05

io

15

20

25

30

SALT CONCENTRATION

35

d!

Figure 1. Partition coefficients for methyl ethyl ketone vs. salt con-

centration for varying temperatures causes a delocalization of the negative charge in the resulting anion. Henry's Law of Dilute Solution. The effect of varying the initial solute concentration on the partition coefficient was determined for the model compounds. The water phase concentrations were simultaneously varied over 100-fold from low concentrations of methyl ethyl ketone (0.051 mg/L), nitroethane (0.187 mg/L), n-butanol (0.363 mg/L), and dioxane (0.834 mg/L). The study was completed a t the optimum analysis conditions of pH 7.1,50 "C, and 3.35 M sodium sulfate. The partition coefficients were found to be independent within experimental error of initial water phase concentrations over a 100-fold change of the initial concentration. This was found to hold for all compounds except nitroethane where a large variance was noted at the lowest concentration investigated while at the two higher concentrations excellent agreement was obtained. This adherence to Henry's law for a multiple solute system implies that all compounds were sufficiently dilute to minimize intra-solute-solute interaction. According to Henry's law, the initial water phase solute concentration can be determined for a constant volume system from a knowledge of the equilibrium partition coefficient and an experimental determination of the equilibrium vapor phase concentration (Equation 3). This is particularly useful for the isolation and quantification of a specific solute. Detection Limits of Head Space Analysis. T h e calculated theoretical detection limits for the four model compounds are shown plotted against the reciprocal of the partition coefficient (1/K) in Figure 2. The theoretical limits were determined by assuming a nominal flame ionization detector limit of 50 ng absolute amount injected and calculating the initial water phase concentration by Equation 3. For head space analysis this is the amount in 5.0 mL of injected vapor. The detection limits ranged from 50 hg/L for methyl ethyl ketone to 740 kg/L for dioxane. The smaller 1 / K is, the lower the detection limit. Figure 2 can be used to determine the theoretical detection limits for any solute from the measured partition coefficient under comparable conditions. The head space analysis detection limits can be further enhanced by decreasing the equilibrium vapor to liquid volume ratio. Equations 3 and 4 can be used to determine the effect of varying the vapor to liquid volume ratio, a , on the percent solute concentration in the vapor ((A)J/(A)J. Figure 3 shows the curves obtained by solving Equation 3 for W , and varying the 01 ratios and partition coefficients. The total solute weight ( WT) is held constant and at a concentration of 100 hg/L for varying partition coefficients of 0.2 to 1.0 and varying a ratios of 0.19, 0.65, and 1.00. The total volume of the system, VT,

ANALYTICAL CHEMISTRY, VOL. 51, NO. 13, NOVEMBER 1979

2171

Table VII. Individual and Total Effects of Salt Concentration and Temperature on the Partition Coefficient enrichment factors-ratio of K values B A - total 0.00 t o 3.35 M temperature, salt, AIB plus 30 to 50 “ C compound 0.00 to 3.35 M 30 to 50 “ C methyl ethyl ketone nitroethane butanol dioxane

5.1 4.0 5.4 4.7

28.0 10.8 18.4 19.3

5.2

66.7 25.1 56.1 49.3

2.7

5.4 4.1

loor

’I 80

0 z BOW

U z

8

40-

DETECTION LIMIT ( N g / l )

Figure 2. Theoretical detection limits for model compounds plotted vs. the reciprocal of the partition coefficients. The nominal FID detection limit is 50 ng and the amount in 5.0 mL of injected vapor is calculated was 2.0 L. In all cases, for a decreasing a ratio the solute percent concentration in the vapor phase increased. This is equivalent to an increase in sensitivity of head space analysis by injecting into the GC a larger percentage of the solute. The change in percent solute concentration in the vapor is greatest for larger partition coefficients. An a ratio change from 1.00 to 0.19 will increase solute concentration by 2.7-fold for a partition coefficient of 1.00. For a partition coefficient of 0.2, the increase is 1.8-fold. The effect of a on the vapor phase enhancement of compounds with partition coefficients less the 0.2 is minimal. This analysis assumes that the total vapor volume is sufficiently large to maintain the head gas equilibrium upon sampling.

Comparison of Experimental Partition Coefficients to Literature Values. Table VI11 compares the partition coefficients obtained in this study vs. literature values. The literature values for n-butanol and dioxane were found to be within the 95% confidence limits of the experimental values found in this study. Other experimental values were greater than literature values. This could be due to the fact that experimental values were run under a constant ionic strength, p H , and temperature of 30 “C. Literature values were completed under a constant temperature of 25 “C while pH and ionic strength were not controlled. The second value shown for butanol was in a saturated solution of sodium sulfate (475 g/L) at 30 “C, while the partition coefficients obtained by Nelson and Hoff (9) were 600 g/L sodium sulfate solution a t 28 “C. The excellent agreement between partition coefficients, although salt concentrations differ significantly, was probably due to the fact that once the salt saturation concentration was exceeded no further salting out effect would be observed. An underlying assumption of this work is that each solute molecule in a mixture acts independently of all other solute present in dilute aqueous solution (Henry’s law). This assumption is demonstrated when one compares partition coefficients obtained by this research to the available literature values. All literature values were obtained with only one solute in solution. In this study, all partition coefficients were

02

04

06

08

10

K-RATIO

Figure 3. Calculated percent concentration ratio vs. the vapor to liquid volume ratio ( a )for theoretical partition coefficients. The initial aqueous phase concentration in the system is set to 100 p g / L and the total volume of the head gas container is 2.0 L obtained in a multisolute solution. If solutesolute interaction was present, it would be expected that the partition coefficient would decrease as a function of the most volatile solute present. This is not the case, as all values obtained are comparable literature values obtained in a single solute solution. Applications of Head Space Analysis. The head space approach to quantitatively determine organic compounds has been demonstrated in this paper. Specific conditions can be set for the specific analytical purpose. The standard deviations of the partition coefficients that were determined are reported in Table VIII. Analysis can be improved by the addition of an internal standard (19). T o calculate the initial water concentrations of volatile solutes, id1 that is needed is the solute K and a measure of the equilibrium vapor concentration. This work suggests that salt addition is more effective in increasing vapor phase partitioning of moderately polar compounds than just temperature although the interactive effects are far superior to salt addition and temperature individually. This also applies to purge and trap analysis which can utilize the same approach to increase the efficiency of analysis. The general “screening” procedure for qualitative analysis with semiquantitative evaluation has been utilized for taste and odor profiling in the food industry. In this case the head gas composition is much more meaningful that purge and trap analysis as has been clearly described in a recent ACS symposium book, “Analysis of Food and Beverages, Headspace Technique” (25). A primary reason for the investigation of head space analysis was to subsequently develop a method that would isolate volatile organics causing taste and odor in drinking water. Isolation of possible taste and odor organics under the enhanced conditions allows determination of the initial water phase concentrations. These concentrations can then be used to determine the vapor phase concentrations of the organic compounds presented to the consumer of drinking water. All that is needed is the partition coefficients of the

2172

ANALYTICAL CHEMISTRY, VOL. 51, NO. 13,NOVEMBER 1979

Table VIII. Comparison of Experimental and Literature Partition Coefficients compound methyl ethyl ketone

K experimentaln

3.90 t 0.01 x 10-3

nitroethane butanol butanol' (sodium sulfate saturation) dioxaned

2.89 i 0.13 x 7.46 i: 0.57 x 13.7 0.1 x 1 0 ' ~ +_

2.78

i

0.54 x

K literatureb

ref.

2.13 x 10-3 3.57 0.03 x 1 0 - 3 N.F.~ 3.6 2 0.4 x 13.3 x 1 0 - 3 d 3 e

(53)

(24) (23) (9)

1.85 x

(23)

Experimental conditions: pH 7 . 1 , 0.00 M salt concentration at 30 "C. Experimental conditions: not reported other than temperature of 25 "C. Experimental condition this study: pH 7 . 1 , 3.35 M salt concentration at 30 "C. Experimental condition this study: 4.23 M salt concentration at 28 'C, pH not reported. e Literature values within 95% confidence limits of experimental data. The remaining values d o not fall within the 95% confidence limits. f N.F. = No value found. organic compound a t the enhanced and drinking water conditions, respectively. Application of the method of head space analysis as described was utilized for GC/MS of a drinking water sample in Philadelphia. The results of mass spectral identification of the compounds found in the drinking water showed the presence of toluene, two C-2 benzene isomers, CHC13, CHC1Br2, CHC12Br, and 1,1,2,2-tetrachloroethane. T h e quantitative effect of salt and temperature was also studied. The optimized head space conditions of 50 "C with a saturated salt solution was compared to 24 "C without salt addition. An increase in peak response was observed of 8- and 22-fold for chloroform and bromodichloromethane, respectively. The analysis was completed in a 125-mL bottle containing 100 mL of tap water. A 50-yL gas sample was injected onto the GC using a 63Nielectron capture detector. Thus the methodology should be capable for analysis of microgram/liter quantities of organics in drinking water.

LITERATURE CITED (1) I. H. Suffetand J. V. Radziul, J. Am. Water Works A m . , 68, 520 (1976); Erratum, 69, 174 (1977). (2) I. H. Suffet and P. R. Cairo, J . Environ. Sci. Health. A13, 117 (1978). (3) T. A. Bellar and J. J. Lichtenberg, J . Am. Water Works Assoc., 66, 739 (1974). (4) I. H. Suffet, G. Dozsa, and S. D. Faust, Water Res., 5, 473 (1971). (5) I. H. Suffet and S.Segall, J . Am. Wafer Works Assoc., 63, 605 (1971). (6) C. Weurman, J . Agr. Food Chem., 17, 370 (1969).

(7) K. Grob and F. Zurcher, J . Chromatogr., 117, 285 (1976) (8) C. D. McAuliie, "Mark Pollution Monitoring (Petrdeum)",Natl. &K. Stand. (U.S.), Spec. Pub/.. 409, Dec. 1974. (9) P. E. Nelson and J. E. Hoff, J . FocdSci., 33, 479 (1968). (10) W. G. Jennings, J . Food Sci., 27, 366 (1972). 1111 A. F. M. Barton. Chem. Rev.. 75., 731 11975). ~, (12j R. A. Kelier, J.'CbrOmatogr. 'Sci., 11. 49 (1973). (13) A. Hartkopf, S. Grunfeld, and R . Delumyea, J . Chromatogr. Sci., 12, 119 (1974). (14) B. L. Karger, L. R. Snyder, and C. Eon, J. Chromatogr., 125, 71 (1976). (15) G. D. Christian and W. C. Purdy, J . flectroanal. Chem., 3, 363 (1962). (16) S. L. Friant. Ph.D. Thesis, Drexel University, Philadelphia, Pa.. 1977. (17) 0. L. Davies, "Design and Analysis of Industrial Experiments", 2nd ed., Hafner. New York. 1971. (18) J. Pr&, "Statisticai Program Package in APL", 5th ed.,State U. College of New York. 1973. (19) R . E. Kepner, H. Maarse, and J. Strating, Anal. Chem., 38, 77 (1964). (20) W. W. Nawar and I. S.Fagerson, Food Techno/., 16(11), 107 (1962). (21) A. Dravnieks and A. O'Donnell, J . Agr. FocdChem., 19, 1049 (1971). (22) E. S.Gould, "Mechanism and Structure in Organic chemistry",M , Rlnehart & Winston, New York, 1959. (23) A. G. Vitenberg, B. V. Toffe, 2. St. Dimitrova, and I. L. Butaeva, J . Chromatogr., 112, 319 (1975). (24) L. Rohrschneider, Anal. Chem., 45, 1241 (1973). (25) "Analysis of Food and Beverages, Headspace Technique", G. Charalambous, Ed., Academic Press, New York, 1978. \~

RECEIVED for review November 15, 1978. Accepted July 30, 1979. Presented before the Division of Analytical Chemistry, 176th National Meeting of the American Chemical Society, Miami Beach, Fla., Sept. 15, 1978. This research was supported by the Philadelphia, Pa., Water Department under the leadership of Water Commissioner Carmen F. Guarino.

Determination of Alkoxy1 Substitution in Cellulose Ethers by Zeisel-Gas Chromatography K. L. Hodges," W. The

Dow

E. Kester, D. L.

Wiederrich, and J. A. Grover

Chemical Company, Midland, Michigan 48640

An improved Zeisei gas chromatographic technique has been developed for the determination of molar substitution in cellulose ether derivatives. The method utilizes adipic acid to catalyze the hydriodic acid cleavage of the substituted alkoxy1groups quantliatlvety to thelr corresponding alkyllodides. An in-sku xylene extraction of the alkyliodides in a sealed vial allows for the determination of methoxyi, ethoxyl, hydroxyethoxy, or hydroxypropoxy substitution in mixed or homogeneous cellulosic ethers.

Cellulose ethers are used extensively as thickeners, binders, lubricants, emulsifiers, and film formers. Their capability to perform this large variety of tasks depends on the number of 0003-2700/79/035 1-2172$01 .OO/O

moles of ether (alkoxyl) substituted per anhydroglucose unit, molar substitution; and on the number of hydroxy groups substituted, degree of substitution. The ability to quantitatively determine the molar substitution is therefore important in adjusting the solubility, thermal gelation point, viscosity, and other physical properties associated with solutions of these cellulose ethers. A wide variety of analytical methods have been developed over the years for the determination of molar substitution, largely owing to the variety of ethers being marketed. T h e classical Zeisel distillation method (I) has been combined with gas chromatography (2-10) to obtain the selectivity needed for the analysis of mixed cellulose ethers. 0 1979 American Chemical Society