coho1 and 10 drops of concentrated sulfuric acid are added and the sample is refluxed under a condenser for about 1 hour. It is transferred to a separatory funnel with 5 ml. of ether and 20 ml. of water. The separated upper layer is washed with water until free of acid. Anhydrous sodium sulfate is added to a few milliliters of the upper layer. Ten microliters are injected onto a chromatographic column, 8 feet in length, packed with 25% diisodecyl phthalate on 60-80 mesh chroniosorb P. h chromatogram is obtained isothermally a t a column temperature of 110’ C. Large peaks due to the ethyl ether and butyl alcohol are properly attenuated. Any acid salts present in the original sample are identified by retention time of the corresponding butyl ester established by calibration with known compounds.
baths and the results of analysis for alkali and acetate are shown in Table I. The analysis of a similar mixture a t five-fold increased alkali and acetate content was also made and included in Table I. Potassium salts were also substituted for sodium salts. I n all cases, no decrease in reproducibility or accuracy was observed. Sharp end points cannot be obtained without the addition of the ethyl ether. The volumetric determination of salts of organic acids other than acetic, will require insertion of appropriate conversion factors in the calculation. Some modification of the gas chromatographic procedure may be necessary for identification of salts of high molecular weight acids.
RESULTS AND DISCUSSION
To illustrate the potential accuracy and versatility of the volumetric procedure, known mixtures were prepared to resemble the stannate immersion
LITERATURE CITED
(1) Furman, N. H., ‘‘f3,tandard Methods of
Chemical Analysis, Vol. 11, 5th Ed.,
p. 2251, Van Nostrand, New York,
1939.
Table 1. Determination of Sodium Acetate and Sodium Hydroxide in Stannate Immersion Baths of Known Composition
Present“
% 1. Sodium acetate
1.03
Found % 1.02; 1.05
Sodium hydroxide 0.98 1.00; 0.99 2. Sodium acetate 5.0 5.00; 5.00 Sodium hydroxide 4.9 4.70; 4.70 Also contains 5% potassium stannate and 5% sodium pyrophosphate.
(2) Hillebrand, W. F., Lundell, G. E. F., Bright, H. A., Hoffman, J. I. “Applied Inorganic Analysis,” p. 291, Wiley, New
York, 1953. (3) Zbid, p. 698.
RECEIVEDfor review October 15, 1965. Accepted November 10, 1965.
Modified Karl Fischer Titration Method for Determination of Water in Presence of SiIanoI and Other Interfering Materials ROBERT C. SMITH and GENE E. KELLUM Dow Corning Corp., Midland, Mich.
b Determination of water in silicone materials containing silanol was accomplished by direct titration with Karl Fischer reagent (KFR). High molecular weight alcohols were used as sample diluents to minimize the alcoholsilanol interfering reaction in which water is produced. The method described has general application for water determination in other materials as ketones, aldehydes, organic acids, and vinyl ethers which react readily with methanol and release water. A modified biamperometric titration apparatus employing potentiometric strip chart recording of current changes was used. End points were determined by observations of rates of current decay which allowed differentiation of water originally present from water formed by relatively slow interfering reactions.
D
of water in silicone materials using the conventional Karl Fischer reagent (KFR) method is complicated by the reaction of silanol with both methyl cellosolve in the reagent and with the methanol diluent. ETERMINATION
Gilman and Miller (7) reported that simple and unhindered silanols reacted quantitatively with KFR (6). d reaction similar to that proposed for orthoboric acid and KFR (15) was given:
+ + SOz + 2 CH30H + 2 H I + CHaHS04 (1)
R3SiOH IS RsSiOCH3
+
Grubb (8) stressed the fact that the silanol-methanol reaction R3SiOH
+ CH30H
+
R3SiOCH3
+ HzO (2)
occurred readily in the presence of K F R in preference to intermolecular condensation of silanol. The use of alcohols other than methanol, as reported in this paper, minimizes the silanol-alcohol reaction. Damm, Noll, and Krauss ( 5 , 19) employed K F R to determine water in siloxane polymers and resins containing silanol using differences in reaction rate of water and silanol with the KFR. Samples were dissolved in pyridine, given amounts of K F R were added, and the time for reaction to a biamperometric end point was measured a t
several intervals up to 30 minutes. A sharp break in the consumption us. time plot was obtained within a few seconds indicating rapid water reaction and slow silanol reaction. Extrapolation to time zero gave a measure of water content. A rapid, direct titration procedure was developed in our laboratory to determine water in the presence of silanol containing materials using commercially available KFR and a modified biamperometric apparatus for end point detection. The reaction of silanol with KFR (or, more specifically, the silanolalcohol reaction in which water is released) was minimized or eliminated without inhibiting the free water reaction appreciably by employing high molecular weight linear or branched alcohols as sample diluents. The method has general application to materials containing components which react readily with methanol and to systems where fading end points prevented accurate results previously. Ketones, aldehydes, organic acids, vinyl ethers, and other compounds are known to react with methanol to release water. Various solvents and solvent mixtures VOL. 38, NO. 1, JANUARY 1966
67
have been proposed as sample diluents to replace methanol to eliminate or minimize the side reactions or to improve solubility and homogeneity of samples. Typical diluents are pyridine, chloroform, dioxane, dimethylformamide, ethylene glycol, formamide, acetic acid, and C2-Cd alcohols. Solvent mixtures containing methanol, pyridine, or benzene with these materials are frequently used. Visual and electrometric (both potentiometric and amperometric) end points are in common usage for direct or back titrations with K F R (10, 12, 20). These titrations are frequently performed with potentiometric measurement between two metal electrodes polarized with constant current and have been classed as bipotentiometric titrations (21). Modified bipotentiometric techniques as that of Barbi and Pizzini (1) have been employed in an attempt to account for water produced by esterification of acetic acid with methanol. A small reproduced portion of K F R was added, and the time required for the small millivolt reading to return to 250 mv. m s recorded. K F R additions were continued each time 250 mv. was reached, and a plot of “transition time” us. total volume yielded a curve similar to a potentiometric titration break. The inflection point was used as the end point. The coulometric method of hfeyer and Boyd (11) applied a supplementary generating current adjusted to maintain an end point before sample addition. This approach would have general utility if the interfering side reaction remained a t a constant rate throughout the titration. In our titrations, a potentiometric measurement of current was employed similar to that reported by Potter and White (18). Afast responding potentiometric recorder was utilized for continuous current monitoring us. time. A similar system has been described (4) for optimizing and speeding the operation of an automatic titrator; however, it was not used as the end point detection device. Swensen and Keyworth (22) employed recorder end points in initial investigations for coulometric titration of water in benzene, but did not apply this detection method routinely. Murayama (16) has designed a n automatic recording amperometric titration unit with potentiometric recorder readout to titrate mercapto groups with AgN03. Titrant delivery was continuous, which is not suitable for K F R titrations. EXPERIMENTAL
Apparatus. The Sargent Model I11 manual polarograph was modified by replacement of the sensitive b u t sluggish responding galvanometer with one of two potentiometric recorders equipped with inputs for current measurement. One was the Bausch 68
ANALYTICAL CHEMISTRY
1-10 MILLIVOLT RECORDER JACKS
ATTENUATOR
ELECTRODE JPCKS
Z E R O ADJUST
U E
Figure 1 . Modified biamperometric circuit for potentiometric measurement of current 1.35-volt battery to give an applied potential of about 2 0 0 mv. 1.35-volt battery S.P.S.T. S.P.S.T. 270-ohm resistor 47-ohm resistor 5000-ohm, Daven and Co., 12-step attenuator, Type Spec. 7 5 4 5 - 1 1 000-ohm resistor 33,000-ohm resistor 1 00-ohm resistor 1 00-ohm resistor, 3- or 1 0-turn pot Duoplatinum electrode, Beckman 3 9 0 3 2
and Lomb V . 0 . X - 5 with 0- t o 0.01ma. span, 1000-ohm input impedance, 1/2-second response (5-inch scale), and 1 inch/minute chart speed. The other recorder was the Sargent M R S-72150 with 0- to 1.25-pa. span, 1000-ohm input impedance, 1-second response (25cm. scale), and 1 inch/ minute chart speed. A potential of about 200 mv. was applied to a Beckman (39032) platinum electrode pair for all titrations. For routine application, the simple biamperometric titration apparatus circuit shown in Figure 1 was employed. Here potentiometric recorders of 1-mv. or 10-mv. span were used. The burets were 2- or 10-ml. capacity and automatic zeroing type with attached, closed reservoir and Teflon stopcock (Kimax 17138 F, available from Arthur R. Thomas and Co., KO. 2469). A titration cell of about 100-ml. volume was constructed from a 55/50 T joint similar to a previously reported design (22). The outer joint was closed off and four openings added. The dual platinum electrode, buret tip, and a solvent introduction line were sealed in place with Silastic 731 RTV silicone rubber. The fourth opening was utilized for sample introduction and was closed with a rubber stopper. The lower portion of the cell contained a 6-mm. borosilicate glass drain. Purging with dry nitrogen was not necessary with the cell sealed in the above manner. Rapid mixing was provided with a magnetic stirrer. Reagents. Karl Fischer reagent (KFR), single stabilized solution,
Fisher Scientific Co. (SO-K-~), a modification of the formulation of Peters and Jungnickel (17 ) . Dodecanol, technical grade dodecyl alcohol (Lorol 5), Matheson Scientific, Inc., contains about 60% dodecanol and other alcohols in the Cs to CI8range. 2-Ethyl-1-hexanol, practical grade, Eastman, P 3608. Procedure. A modified biamperometric end point was employed as described below. The correct current multiplier setting on the manual polarograph, or the attenuator position of the apparatus in Figure 1, is found for any alcohol solvent in the following manner. Place 40 ml. of alcohol or mixed solvent in the cell and stir rapidly. While titrating the mater in the solvent with KFR, adjust the multiplier or attenuator setting to give deflections above base line of 25 to 30% of recorder span persisting for 10 to 15 seconds with additions of 0.25 ml. of KFR. Deflections of this magnitude occur after most of the water in the solvent has reacted, and a t this time the precipitation of pyridinium hydroiodide should be complete. The correct current span for the final end point is found when 0.05 ml. of K F R produces about 3 to 6% increase in current a t a stable current level of 50 to 70% of the recorder span chosen. The current should remain constant for several minutes. The solvent titration should require less than 4 ml. of KFR, or a maximum of about 0.06y0 water is allouable in the solvent. Using a capillary dropper, add a weighed amount of water to give about a 1-ml. titration and determine the K F R factor (grams of water per milliliter of KFR) for each solvent employed. Weigh liquid samples, or solids dissolved in a suitable solvent, to give approximately a 1.0-ml. titration. Inject the sample into the titration cell and titrate immediately to a constant current response of 1 minute near the current level of the above solvent titration step. If a constant current cannot be obtained, add 0.05-ml. increments of K F R until the same slope (current change us. time) appears on the recorder chart. Locate the end point where the slope is first reproduced with the given additions of KFR. Drain the cell and flush with dry pyridine or other solvent before refilling with alcohol. RESULTS AND DISCUSSION
Conventional “kick off” biamperometric end points in direct K F R titrations have been taken at the first sudden large increase in current which remained above the residual current for some time interval, as 10 seconds or 30 seconds. hlinimum millivolt readings steady for 10 to 15 seconds or 30 seconds are typically specified for potentiometric end points. These end point criteria are based on a rate of consumption of K F R from excess reagent to the equivalence point or the water side of the equilibrium.
The modified biamperometric end point in our new method was taken a t a slight excess of K F R where the rate of consumption of reagent was constant or zero as indicated by a constant decrease or no change in current. The current levels employed for end points were reproducible and stable in the absence of interfering side reactions. Bastin, Siegel, and Bullock (3) also titrated beyond the first current surge, and read current values with increasing additions of KFR to obtain an extrapolated equivalence point as would be normally used in amperometric titrations. However, each measurement must be made a t a stable current for accurate results. This type of end point would be difficult or impossible to obtain in the presence of an interference. Many factors influence the ease in which the new end points are observed in the presence of side reactions. Type of sample, rate of its interfering reaction, sample size, water content of the sample, and water content of the diluting solvent (or amount of KFR to predry the solvent) are considered to obtain reliable end points and results. If the interfering reaction is not too rapid or can be minimized, the KFR consumption due to the side reaction and that due to water present in the sample are readily differentiated using the recorder readout. The general shape and magnitude of the interference curves differ greatly, and the position a t which end points are chosen will vary slightly with operators and with the factors noted above. One advantage of this detection system is that these differences, sometimes quite subtle, are observed quickly and easily with continuous recording of current changes. The detection method is also versatile. The end point reported in the procedure generally describes one which worked best for our purposes; however, with other types of compounds and other investigators, a variation of this end point might be needed and could be established quickly. With high molecular weight alcohols as diluents, precipitation of the pyridinium hydroiodide occurred in the solvent titration with K F R because of the low solubility of the salt in this medium. The precipitation caused no difficulty in end point detection, provided the solubility equilibrium of the hydroiodide salt was established during the solvent titration step. Difficulty with false end points attributed to this type of precipitation has been reported ( I S ) . Coating of the platinum electrodes with excessive precipitation in 5 :4 pyridine-acetic acid solvent mixture gives obscured end points ( 2 ) . No detectable loss in sensitivity was observed in our titration method with precipitate present. Any deposits left on the electrodes were adequately washed off with pyridine between titrations.
CHART TRAVEL, 1 IN./MIN.
Figure 2. 1. 2.
3. 4.
Titration of water under varying conditions Titration of water in solvent Normal titration of water in a sample Titration of water in a sample which reduces sensitivity Titration of water in presence of interfering reaction
The potential applied to the platinum electrode pair was not critical, and 200 mv. was satisfactory for our titrations. Other potentials may be employed with a corresponding increase or decrease in sensitivity. The noise level increases somewhat with increasing applied potential, but a usable gain in sensitivity is possible. At 200 mv., the peak-to-peak noise at the end point was about 2 and 0.5y0of recorder span when using methanol and 4: 1 (2-ethyl-l-hexanol)-pyridine. Evidence of electrode reaction in methanol near 600-mv. applied potential was observed. A nearly linear current response with increasing potentials up to about 2 volts was found using the 4 : l (2-ethyl-l-hexanol)-pyridinemixture. The current response obtained with each addition of excess KFR was linear within limits with each solvent employed. For example, linear regions were found between 0.1- and 0.4-ml. excess for methanol, 0.1 and 0.8 ml. for 4 :1 dodecanol-pyridine, and 0.2 and 0.8 ml. for 4 : 1 (2-ethyl-l-hexanol)-pyridine. Generally, end points were taken to an excess of KFR giving current levels near the lower end of these linear regions. It was then possible to correlate current changes and reagent additions which allowed corrections for over-titrations when necessary. Titration to lower current levels (nearer the more conventional biamperometric end point) might be possible, but our experience has shown that in this region the response was generally not as stable because of incomplete reaction of xater with the smaller excess of KFR. Various linear, branched, cyclic, and aromatic alcohols in the Cz to CL2range were successfully employed as sample diluents. Better sensitivity and sharper end points are achieved when using 20
to 50 volume yo pyridine with the alcohols. The absolute current response a t the new end point using C4 and higher molecular weight alcohols and their pyridine mixtures ranged from about 90 pa. to 0.6 pa. in comparison to about 900 pa. with methanol. The relative current increases a t the end point were similar and were conveniently observed on the same recorder scale using the attenuated signal from either polarograph. Examples of typical recordings of titrations are shown in Figure 2. Curve 1 demonstrates the titration of water in the alcohol-pyridine solvent. The initial spikes in current with KFR additions, a, and the stable end point current level at 50 to 707, of span, 6, are shown. Curve 2 represents a titration of water in a sample which contributed no interference. A constant current level, c, near that of the solvent titration was obtained. Curve 3 shows the titration of water in a sample which caused deadening of current response as generally occurs with most silicones. Because the water reaction with KFR is somewhat retarded in the high alcohol medium, a slight hesitation, d , occurred before the water reaction commenced. Thus, a new end point current level was chosen before KFR addition which allowed for decrease of response. The end point was then taken to this current level, e . If the titration had been carried out to the original solvent titration current level, the results would have been high. The importance of fast current change response is evident, or this choice of new end point current level would not have been so simple. Curve 4 illustrates a titration of water in the presence of a side reaction such as a reactive silanol or acetone with alcohol. VOL. 38, NO. 1, JANUARY 1966
69
1.0-
-
0.9
2
i-
-
g 0.8 z
c 3 z v
0.7
s::
-
P
-
0.6
0.5
-
I
I
I
I
I
2
3
TIME, MIN
Figure 3.
Effect of straight chain alco-
hol molecular weight upon silanolalcohol reaction 1.
2.
3. 4. 5. 6.
7.
4 :1 4 :1 4: 1 4 :1 4 :1
Dodecanol-pyridine Decanol-pyridine Octanol-pyridine Hexanol-pyridine Butanol-pyridine Ethanol Methanol
With high molecular weight alcohol diluents, the water reaction with KFR is much faster than most side reactions. I n this case the end point was chosen a t point f , where the first 0.05-ml. increment of KFR near the end point gave a reproducible slope. Once the water had reacted, the side reaction dominated the consumption of KFR, and the interference pattern was observed and allowed this choice of end point. Because of the presence of methyl cellosolve in KFR, it is important to keep the solvent and sample titration a t a minimum if an analysis of water in silanol-containing silicones and other materials reactive to alcohols is to be accomplished. If the solvent or mixture used as diluent contained more than 0.06% water, some drying technique was employed. Generally 5A Molecular Sieves were effective for drying of pyridine and alcohols. The magnitude of the sample interference determined the practical maximum sample size and titration volume. Only small titration errors were encountered if the interference was held to a maximum of 5% of current span during a 30-second period with a 0.05-ml. KFR addition. Generally, sample sizes were chosen to give a maximum of 1.0-ml. titration with KFR, thus allowing a rapid approach to the end point and minimum addition of methyl cellosolve. The KFR factor should be determined with proper addition of water to consume about the same 70
ANALYTICAL CHEMISTRY
volume of titrant as in the sample titration. Errors due to dilution or current response increase from the methyl cellosolve added during the titration are held to a minimum with these procedures. A useful qualitative comparison of effectiveness of the various solvent systems in inhibiting the interfering side reactions was quickly observed utilizing the following technique. A sample containing a small amount of water was introduced into about 50 ml. of pretitrated solvent which also contained 1 ml. of additional excess KFR. Then current decay us. time was recorded continuously using the same apparatus as described above. I n Figures 3, 4, and 5, plots of relative consumption of KFR os. time are presented for a silanolcontaining silicone fluid and other compounds which gave interfering side reactions with alcohols. I n each case, the sample size was reproduced so that the amount of water added to the various alcohols was constant. The amount of KFR consumed by the original water in the sample was the same when using each alcohol. The initial, rapid current decrease resulting from the KFR reaction with the original water was noted. Then a relative KFR consumption scale in milliliters proportional to the observed current changes was established on a normalized basis even though these current differences and levels varied by as much as a factor of 100. These curves were constructed for relative compari1.0-
0.9 -
0.8 D.
2
-
0.7 -
_I
-
0.6
-
L
k 3
0.5-
z 0.4
5 z P
-
0.3 -
Q P
0.2
-
0.1
-
Or
I
1
0
2
I 4
6
1 8
1 1
I
1 O
I
Z
TIME, MIN.
Figure 4. Effect of solvent mixtures upon acetone-alcohol reaction 1.
2. 3. 4. 5. 6,
4 :1 4 :1 5 :2 4 :1 1 :1
Dodecanol-pyridine Isopropanol-pyridine Acetic acid-pyridine Ethylene glycol-pyridine Methonol-pyridine Methanol
I 0
2
I
I
4
6
TIME, MIN.
Figure 5. Effect of solvent mixtures upon ethyl vinyl ether-alcohol reaction 1. 2. 3.
4. 5.
4 : 1 (2-Ethyl-1 -hexanol)-pyridine 4 : 1 Dodecanol-pyridine Isopropanol 5 : 2 Acetic acid-pyridine Methanol
son only and were not intended to give accurate quantitative KFR consumption values. We preferred to use the direct titration for quantitative determinations and did not exploit the latter technique for this use. Figure 3 clearly demonstrates the effect of the high molecular weight alcohol diluents in minimizing the silanolalcohol reaction. Also, the slight reduction of the water reaction rate is evident. Similar data indicate that branched alcohols are somewhat more effective than their linear isomers. Qualitative decisions on the merits of solvents or solvent mixtures in reducing side reactions with acetone were made from the data shown in Figure 4. Branched and high molecular weight linear alcohols in pyridine minimized the ketal reaction appreciably. Curve 3 is unusual compared with the others and probably reflects the additional effects of a second side reaction of carbinol (methyl cellosolve from the KFR) and acetic acid. I n Figure 5, the abnormally slow reaction of KFR with water in a vinyl ether diluted with high molecular weight alcohol and pyridine is apparent. This reduced water reaction rate causes the end points to be poor (see application section below). These curves suggest that the 5 : 2 acetic acid-pyridine mixture as recommended by Barnes and Pawlak (9)certainly is not very effective in inhibiting the interfering vinyl ether-alcohol reaction. Cooling of the titration solvent and sample has been employed to reduce the
O’t
i
0
!
0 0
1
1 2
l 3
l 4 TIME
l 5
l 6
l 7
~ 8
9
MI”
Figure 6. Effect of temperature upon silanol interference from a polysiloxane-diol fluid using 3 : 1 (2-ethyl1 -hexanol)-pyridine solvent 1. 2. 3.
28’ C.
4.
-19’
methyl and phenyl, trifluoropropyl and methyl, etc. The silanol content of these materials ranged from 1 to 4%. Various silicone resins containing about 1 to 7% total hydroxyl were analyzed for water content also. Representative precision of results are presented in Table I. Varying amounts of water were determined in the presence of several types and amounts of silanol with relatively consistent precision. Determination of water in alkyl subI stituent monomer silanols generally was not possible because their reaction with any alcohol tried was too rapid. Cooling of the sample and solvent may help
11OC. - 4 O
Table 1.
c.
Precision of Titration of Water in Silicone Fluids and Resins
C.
rate of interfering side reactions with K F R (3, 14, 17). This technique may be used to some advantage with our titration method also. Figure 6 shows the effect of lower titration temperatures upon a silanol-alcohol interfering reaction of moderate rate. Here the sample was added to pretitrated solvent, and a n end point established with K F R additions at various time intervals.
Sample HO [(CH+SiO],H Dow Corning 802 resin Dow Corning 840 resin Dow Corning 2105A resin Dow Corning 2-6018 intermediate Sample 1 Sample 2
Table II. APPLICATIONS
The analysis of water in many types of silicon fluids and resins which contained silanol was possible using the new direct titration method. Water was determined in various fluids of the type
Sample HO [( CH&SiO],H
I I
HO(-SiO-).H,
Approximate total hydroxyl,
70 HzO
Std. dev.
Rel. std. dev., 70
1
N 5 8 10 11
0.612 0,0770 0.351 0.0287
0.0161 0.00289 0.00506 0.00114
2.63 3.76 1.44 3.98
7 5
5 6
0.403 0.180
0.0102 0.00893
2.52 4.95
5% 4 1
3
Orig. water content,
Water added,
mg./g.
0.65 0.77
Dow Corning 840 resin
3.30
Dow Corninn 2-6018 intermediGe
2.07
where R was methyl,
21
Titration of Added Water in Silicone Materials
Dow Corning 802 resin
R
in some cases as indicated above. Generally, phenyl-substituted materials of this type reacted slowly and precise water determinations were possible, N M R resolution of absorption due to water hydroxyl and various silanol hydroxyl types (9) should allow some quantitative measure of water in monomers. Sensitivity of such a method would probably be limited to about 0.1% water. I n Table 11, data from studies in which water was added to these silicone materials demonstrate good recovery of the “spiked” water and indicate that good accuracy is possible. No general
mg./g.
Total water calcd., mg./g.
Total water found, mg./g.
Difference, mg./g.
1.02 6.16 1.08 3.41 1.06 7.97 0.96 5.17
1.67 6.81 1.85 4.18 4.36 11.27 3.03 7.24
1.62 6.77 1.85 4.33 4.39 11.3 3.04 7.06
-0.05 -0.04 0.00 +0.15 +0.03 0.00 +0.01 -0.18
R
Table 111.
Sample Acetone” Methyl ethyl ketone Acetaldehydec
Titration of Added Water in Interfering Organic Compounds
Solvent 4:1 dodecanol-pyridine 4:1 (2-ethyl-l-hexanol)-pyridine 4:1 dodecanol- yridine 4:1 (2-ethyl-l-Eexanol)-pyridine 4:1 dodecanol-pyridine 4:1 (2-ethyl-1-hexano1)-pyridine 4:1 dodecanol-pyridine 4:1 (2-ethyl-1-hexano1)-pyridine 4:1 dodecanol-pyridine 4:1 (2-ethyl-1-hexano1)-pyridine
Orig. water, mg./g.
Added water, mg./g.
2.8 2.7 1.6 1.7 3.9 4.1 1.2 1.4 0.39 0.53
1.6 1.6 2.2 2.2 1.8 1.8 1.2 1.2
Total water calcd., mg./g.
Total water found, mg./g.
Difference, mg./g.
4.4 4.3 3.8 3.9 5.7 5.9 2.4 2.6 1.5 1.6
4.3 4.3 3.9 3.9 5.7 6.0 2.4 2.6 1.4 1.5
-0.1 0.0 +O.l 0.0 0.0 +0.1 0.0 0.0 -0.1 +o. 1
Ethyl vinyl etherd Isobutyl vinyl 1.1 ethers 1.1 a End points easily obtained even though current levels somewhat unstable. Stable current at end point observed. Unstable current at end points, but still possible to obtain accurate results. Water reaction abnormally slow in presence of this vinyl ether. Good end points difficult to obtain, although results in this case were good. Precision would probably be poor. e Water reaction was slower than normal, but end points were not difficult to obtain.
VOL. 38, NO. 1, JANUARY 1966
71
bias toward low or high results was observed and a maximum relative error of 3.6% was obtained. These titrations were carried out using 4 : l dodecanolpyridine diluent. As shown in Table 111,the use of two high molecular weight alcohol-pyridine mixtures had general application to systems where methanol is known to cause serious interfering side reactions. I n many cases, the previously reported solvents and solvent mixtures are not as effective in eliminating or minimizing the side reactions. A maximum relative error of 7y0 was found. I n all these examples, except with vinyl ethers, the water reaction rate was normal and end points were easily obtained. With vinyl ethers, the rate of the water-KFR reaction was inhibited greatly. End points were difficult to observed with the rapid titration method. The data in Figire 5 suggest that a small excess of K F R should be taken in addition to the normal end point excess, the sample added, and about 3 to 4 minutes allowed for the water reaction. Then the titration may be completed to the initial current level of the excess reagent. Advantages using the new titration method were also found in determination of water in noninterfering samples. h comparison was made on precision and speed of titrations with a sample of water in methanol using the new method and
other end point detection devices commonly employed, including a n automatic titration unit. Increased speed of titrations (2 to 3 times faster) is possible with 2 to 6 times better precision using the new method. Titrations using the modified apparatus were carried out by a continuous addition of K F R at a rate of about 1 m1./5 seconds until levels were approximately matched. Small additions of K F R were then made to complete titration of the least traces of mater to a n end point of closely matched current levels. The method and apparatus described have been employed routinely in our laboratories by hourly technicians for about a year. They have proved to be valuable for rapid and precise determinations of water in the presence of interfering compounds. We also use the same end point method for KFR titrations of water in noninterfering materials dissolved in methanol. The apparatus required was simple to assemble, and the only major expense was the potentiometric recorder. LITERATURE CITED
(1) Barbi, G. B., Pizzini, S., ‘AKAL. CHEIII.35. CHEW 35, 309 11963). (1963). (2) Barnes, ’L., (2yB;rnes, L., Pawlakj Pawlak, 11. S., Zbid., 31,
1875 (1959). (3) Bastin, E. L., Siegel, H., Bullock, A. B., Zbid., p. 467.
(4) Burns, E. A., Muraca, R. F., Zbid., 34, 848 (1962). (5) Damm, K., Noll, W., Kolloid 2. 158, 97 11958). (6) Fischeg, K., Angew. Chem. 48, 394 f 1 9.?.5\ \ - - - - I .
(7) Gilman, H., Miller, L. S., J. A m , Chem. SOC.73, 2367 (1951). ( 8 ) Grubb. W. T., Zbid.. 76. 3408 (1954). (9) Hampton, J. ’F.,Lacefield, G. W:, Hyde, J. F., Znorg. Chem. 4, J659 (1965). (10) Kolthoff, I. AI., Elving, P. J., “Treatise on Analytical Chemistry,” Part TI/ Vol. 1 (Mitchell, J.), pp. 84-87, Interscience, New York, 1961. (11) Meyer, A. S., Boyd, C. XI., Zbid., 31, 215 (1959). (12) Mitchell, J., Smith, D. M., “Aquametry,” pp. 71-102, Interscience, New York, 1948. (13) Zbid., p. 104. (14) Zbid., pp. 155-156. (IS) Zbzd., p. 256. (16) AIurayama, M.,U. S.Patent 2,834,654 iMav 13. 1958). (17) Peter;, E. D.,‘ Jungnickel, J. L., ANAL.CHEM.27, 450 (1955). (18) Potter, E. C., White, J. F., J . Appl. Chem. 7, 309 (1957). (19) Noll, W., Damm, K., Krauss, W., Farbe Lack 65, 17 (1959). (20) Stock, J. T., “Amperometric Titrations,” pp. 166-174, Interscience, New Ynrk 1965. -----I
(21) Zbid., pp. 54-55. (22) Swensen, R. F., Keyworth, D. A., ANAL.CHEX 35, 863 (1963). RECEIVEDfor review July 26, 1965. Accepted October 22, 1965. 13th Anachem Conference, Wayne State University, Detroit, Mich., October 1965.
Pyro lysis f o r St ruc t ure Dete rmina t io n of Cyclopropene and Cyclopropene Fatty Acids JOANNE L. GELLERMAN and HERMANN SCHLENK University of Minnesota, The Hormel Institute, Austin, Minn. Pyrolysis of cyclopropane fatty esters at 350” C. with Si02 as catalyst yields numerous olefinic isomers. The structure of the original ester can be concluded from the products of ozonization-hydrogenation of the olefinic mixture. The same method has been applied to sterculic ester. The cyclopropene ring requires 160” C. with Si02 for cleavage. Dienoic esters result and the products of oxidative fragmentation are again characteristic for the original structure. Hydrogenation with Pd-Pb catalyst in methyl acetate reduces sterculic ester selectively to dihydrosterculic ester while common unsaturated fatty esters remain unchanged. The latter are reduced selectively by PtOz in methanol without concurrent hydrogenolysis of the cyclopropane ring. Proper combination of hydrogenations with pyrolysis enables determining the structure of a cyclic acid without isolating it in purity.
72
ANALYTICAL CHEMISTRY
L
FATTY ACIDS containing a cyclopropane or cyclopropene ring are of interest as constituents of certain bacterial and plant lipids ( 1 , IS). Methods for identification of their structure-i.e., for locating the position of the ring-methylene group in the chain-involve chemical reactions which lead eventually to branched and/or normal chain esters. These are then subjected to any of the conventional oxidation methods for identifying the position of a side methyl group, double bond, or keto group in fatty acids (6, 7 , ONG-CHAIN
Id).
The procedures reported here are based on pyrolytic cleavage of the ring followed by ozonization-hydrogenation of the resulting mixture of unsaturated compounds. The methods may hold some advantage above others in regard to time, simplicity, and amount needed. I n gas-liquid chromatography (GLC) of cyclopropane fatty acid methyl esters, certain phase supports give rise
to several peaks which indicate decomposition of the ring ( 8 ) . Catalysts, temperatures, and periods of heating were studied to develop a controlled method for ring fission which is independent of GLC. Silicic acid as used for adsorption chromatography catalyzes the reaction in a sealed tube a t 350” C. very satisfactorily. A mixture of mainly monoenoic compounds results and only minimal amounts of polymers are formed. In pyrolysis of a cyclopropene ester-Le. , methyl sterculate-heating with silicic acid for 15 minutes a t 160’ C. is sufficient and the recoveries are as satisfactory as \Tith cyclopropanes. The dienoic products from sterculate resemble those which Kircher and coworkers (20) obtained under quite different conditions from the corresponding hydrocarbon, sterculene. Cyclopropane and cyclopropene fatty esters must not necessarily be isolated in purity to apply the pyrolysis method.