5
LITERATURE CITED
li
( I ) Bradley,
Figure 5. Vapor pressure curves of hydrocarbons in mixture of known composition
4
-I'
3
I
O 0
M
100
150
200 TEMPERATURE
250-
'sL O
DECOHPOSIllON
('Cl .
T
J
619 (1954). (3) International Critical Tables, T'ol. 3. D. 208. IllcGraw-Hill. Sew York.
1928. ( 4 ) Kotin, P., Falk, H. L., ?&der, P., Thomas, AI., A n i . M e d . dssoc Arch. Ind. Health 9, 153 (1951).
15) Mortimer, F. S., Murphy, R. V., Ind. Eng. Chent. 15, 1140 (1923). 16) Reid, A. F., U. P. Patent 2,628,892 (Feb. 17, 1953). ( 7 ) Tebbens, B D., Thomas, J. F , LIukai, ?VI., Am. M e d . dssoc. Arch. Ind. Health 13,567 (1956). (8) Zbid., 14, 413 (1956). (9) Tebbens. B. D.. Thomas. J. F.. San' born, E. Mukai, AI., .4h. I n d . Hyg. Sssoc. Quart. 18, 165 (1957). (10) Thomas, J. F., Tebbens, B. D., Miikai. LL, Sanbcrn, E. S . , ANAL. C H E M . ' 1835 ~ ~ , (1957 I .
s,,
IURUACE QlMlLNT
molecules from the surface of the solid, and the rate of return of molecules to the solid. Below these particular temperatures for each of the compounds, the equilibrium is shifted toward the solid phase. If the cold point of the temperature gradient in the furnace is adjusted to fall between the two equilibrium temperatures, preferably closer to the more volatile component. in this example pyiene, pyrene will be driven from the furnace and will condense as crystals in the cold finger.
R. S., Cleasby, T. G., J . Chem. Soc. 1953, p. 1690. (2) Hausmann, W., ASAL. CHEM. 26,
I n actual operation a vapor pressure curve is desirable for each component in a sample, but it is possible to determine empirically the optimuni conditions for fractionation in much the same manner as is done in any type of fraction collector. ACKNOWLEDGMENT
Conrad Kwasnicki, University of California, Berkeley, rendered significant assistance.
RECEIVED for review Sovrniber 4, 1957. .4CCEPTED Jul>-25, 1958. Division O f Industrial and Engineering chemistry, Air Pollution Symposiiim, 132nd Meeting, ACS, S e w York, S . T.,September 1957. Supported b r research grant RG-4281 of the Sational Institutes of Health, Public Health Service, cooperative effort within the University of California School of Pub!ic Health and Department of Engineering.
.Chemical Analysis by Measurement of Reaction Rate Determination of Acetylacetone W. J. BLAEDEL and D. L. PETITJEAN' Chemistry Department, University of Wisconsin, Madison, Wis.
b Work was undertaken to investigate the possible development of a more general, rapid, and accurate method of analysis based on kinetics than now exists. A preliminary study of the utility of the suggested approach was carried out with a well known reaction -the alkaline hydrolysis of ethyl acetaie. The developed method was then applied to acetylacetone determination by measurement of its reaction rate with hydroxylamine hydrochloride to demonstrate that it is equally applicable to integral-order or complex reactions. Simple systems free from interfering substances can be quickly analyzed with an ultimate accuracy of about o.3Y0. The method can probably b e extended to mixtures and samples containing chemically inert interferences with some loss in accuracy.
T
HE analytical utility of kinetics has been recognized for some time in inorganic analytical chemistry. and niicro-
1958
ANALYTICAL CHEMISTRY
chemical applications have been discussed in detail (6, 9, 94). Catalytic activity has found considerable qualitative application, allowing the detection of minute quantities of various elements. With increasing interest in the accurate determination of microgram quantities of materials, catalytic activity is being found extremely useful in quantitative applications as well. For example, the catalytic influence of iodide ion on the reaction between ceric ion and arsenious acid (19) permits the determination of microgram quantities of iodine and has been the subject of many papers for more than 20 years (11, 14). More recently, methods have been developed for determining submicrogram quantities of silver by its catalytic effect on the oxidation of manganous ion by persulfate ( 2 3 ) , and for determining comparable concentrations of copper by its catalysis of the autoxidation of resorcinol (IO). With the increasing importance of trace impurities, further quantitatire applications of kinetics can be expected in inorganic analytical chemistry.
The utility of kinetics in quantitative organic analysis, hon.ever, has not been as extensively recognized. Qualitative applications, such as the differentiation of sugars by time of osazone formation, and of organic halides by time of silver halide formation, are of frequent use in characterization Ivork. Quantitative applications, on the other hand, are rare. Perhaps the earliest and one of the best examples of the potential value of kinetics in quantitative organic analysis is found in a method for the resolution of mixtures of the normal butenes ( 5 ) . illthough modern infrared equipment now provides a simple solution to this problem, it n a s not available a t that time. Howeyer, the problem of resolving such mixtures was solved by detrrmination of the total butene concentration, the density of the mixture, and a pseudo first-order reaction rate constant
1 Present address, -4lcoa Research Laboratories, Aluminum Co. of America, Sem- Kensington, Pa.
for the reaction of a large excess of the corresponding dibromides (formed from the butenes by bromination) n ith a solution of potassiuni iodide. The literature contains other applications of kinetics to quantitative organic analysis (13). Consideration of the variousapproaches led to the conclusion that accurate result> are obtainable only by expenditure of considerable time and effort. Furtherinore, none of these approaches can be easily applied to consecutive or reversible reactions. The present J\ ork n as undertaken to investigate the possibility of developing a more general method of analysis via hinetics, based upon measurement of the rate of a reaction a t several times during the early stages of the reaction. It was thought that a method utilizing modern instrumental techniques might rapidly give results of high precision in a vide variety of applications. Measurement of reaction rate during the early stages of a reaction might permit analytical application of reversible or consecutive reactions, n hich are generally considered to be of the least analytical utility. The recently developed high-frequency oscillator type of analytical instrument n-as used. Such an instrument is capable of high sensitivity, and its instantaneous response and ease of adaptation to recording of data appeared well suited to this type of problem. The nonlinear relation betiyeen response and concentration allon s only empirical rate measurements to be obtained. This feature restricts application of the highfrequency method to samples relatively free of interferences but does not prevent its application to complex reactions. The method 1%as del-eloped and tested with a well known reaction-the alkaline hydrolj sis of ethyl acetate-then applied to the determination of acetylacetone to illustrate its applicability to reactions that do not go to conipletion. METHOD OF ANALYSIS
-1known eight of sample is allowed to react nith a standard reagent solution, the reaction being one that results In a change in solution conductance. The progress of the reaction is followed v i t h the apparatus shonn in Figure 1. The reaction vessel is a part of the tuned circuit of a high-frequency oscillator. Changes in solution conductance as the reaction proceeds cause changes in the frequency of the loaded oscillator. The frequency changes are measured by heterodyning the signal from the loaded oscillator with a signal from a reference oscillator, the frequency of 17 hich is controlled manually. The difference (beat') frequency obtained from the mixer is measured with a frequency meter, the output of M hich is a direct current nearly directly proportional t o the h a t fre-
VESSEL
FREPUENGY METER
RECORDER
OSCILLATOR^ Figure 1.
Block diagram of apparatus
for measuring rate of reaction
quency. The changes in direct current are recorded, and the measured rate of change of recorder current gives a measure of the rate of the reaction. The amount of organic compound in the sample being analyzed is then determined by referring the measured rates of frequency change to a working curve plotted from data obtained with known amounts of the substance being determined. APPARATUS
Reaction Vessel Assembly. Structural details of t h e reaction vessel assembly found most satisfactory are shown in Figure 2 . The glass vessel itself, ,4,was derived from a 250-ml., extra high form electrolytic beaker by reducing its height and grinding the top edge and the base parallel. During- the grinding process the base thickness was reduced from about 0.09 to 0.04 inch to obtain a reaction vessel possessing fairly high sensitivity to concentration changes. Silver electrodes were formed on the side (continuous ring electrode) and bottom (disk electrode) of the vessel by applying a coat of D u Pont KO.4666 silver paint to the desired areas and fusing i t into the glass a t 250" C. A grounded shield of 4-inch diameter brass tubing, B , with upper and lower edges machined parallel was located concentrically around the vessel. Entrance and exit tubes (G and H, respectively) for circulation of thermostatically controlled water were provided by soldering short lengths of 5/16inch outside diameter brass tubing into holes drilled in the brass shield. Enclosure of the annular space around the vessel was completed by means of the npper and lower Lucite rings. The entire assembly was clamped together tightly by means of four vertical threaded hrass rods and nuts, I , J . .-1 n-atertight seal was obtained by applying a thin film of stopcock grease to neoprene gaskets, E , F , and fiber n-ashers. K . To connect the reaction vessel into the oqcillator circuit the brass shield, B , n-aq grounded to the oscillator chassis by nieans of a lug, L, soldered to the .hield. Soldered to the inside nall of the brass shield was a strip of spring brass. Ji, which pressed firmly against the silver electrode on the glass wall of the reaction 1-esqel. Connection of the ungrounded electrode into the circuit was made by a thin copper disk, S,
pressed firmly against the circular electrode on the base of the vessel by a spring, 0. The lower end of the spring was held in place on the upper constricted end of a machined brass connector, P , held in the loner Lucite ring, D. The loner end of P xias of a diameter suitable to fit snugly into a coaxial connector mounted on the oscillator chassis. The center hole in the upper Lucite ring, C, was bored on a 4-degree taper t o accommodate a similarly tapered Lucite stopper, &. The stopper was provided with a hole in the center, T , to accommodate a borosilicate glass propeller stirrer and holes, C, V, W , for addition of reagents, insertion of a thermometer, and insertion of rubber tubing through which gas could be delivered for maintenance of an inert atmosphere above the solution in the vessel, if necessary. Two additional holes were provided for pin jack R connections to a thermistor, S , the terminal leads of which were coated with Tygon paint (E.S. Stoneware Co., Akron, Ohio) to provide electrical insulation from the plectrolytic solutions in which it was immersed. Temperature Control and Measurement. The temperature of the reaction system n as controlled by pumping water from a thermostatically controlled water bath through t h e annular space surrounding t h e reaction vessel, using a 1/30-hp. ininiersion pump. The high-frequency field and ambient temperature each had a slight b u t noticeable effect on t h e temperature of the reaction system, which therefore mas usually slightly different from the bath temperature. -1 Western Electric Type V-519 thermistor and a Wheatstone bridge n ere used to measure the temperature of the reaction system with minimum time lag and maximum sensitivity. The bridge galvanometer, n i t h a sensitivity of 5.6 mm. per 0.01" C., permitted an accuracy of 0.001" C. in measuring temperature changes. Power absorption by the thermistor from the oscillator was negligible, the niaxinium effect being equivalent to a teniDerature rise of o n l f 0 . 6 0 ~ oC. Oscillators and Mixer Unit. The oscillators and mixer unit were essentially t h e same as those described in the literature for a 30-1Ic. instrument (2) although modifications nere made to improve stability, to obviate the use of cement in mounting circuit components, and to improve control of the reference oscillator frequency. ?;em oscillator coils were wound on partially hollowed-out Teflon rods of 1-inch diameter, with a tapped hole in the solid end, permitting them to be mounted on the chassis with machine Scren s. Zero-temperature coefficient capacitors were substituted throughout both oscillator circuits, and circuit components n ere rearranged to make their placement in the individual oscillator circuits more nearly identical. T o reduce the effect of drafts, metal boxes, rented a t the top to allow dissipation of heat from the tubes, n-ere Tecurely mounted on the oscillator chassis in such a way that they comVOL. 30, NO. 12, DECEMBER 1958
1959
4
ww
U
V
Q
Figure 2. assembly
Reaction vessel
Omitted are lugs on brass shield used for grounding the capadtors connected to thermistor leads
b Figure 3. Frequency changes recorded during reaction of 107.5 mg. of ethyl acetate with standard base solution
pletely enclosed the oscillator tubes, forming a dead air space around them. The trimmer capacitor previously used on the reference oscillator was replaced by a micrometer capacitor similar in design to one recently described by Axtmann ( 1 ) and had a usable range of about 0 to 1 ppf. A study of oscillator stability revealed that the maximum drift rate was about 200 C.P.S. per minute. Frequency Meter. The beat frequency output of the mixer was measured by means of a direct-reading frequency meter similar in principle to others used in this laboratory (9). However, incorporation of a multivibrator gave an output independent of input signal amplitude for any signal strong enough to activate the multivibrator (16). The frequency meter output was recorded on a 0- to 1-ma. Esterline-Angus recorder a t a chart speed of 6 inches per minute. Time measurements were obtained by measuring distances on the chart with a 50-inch stainless steel strip graduated in tenths of an inch and with the first 14 inches further subdivided into hundredths of an inch. The frequency meter was calibrated before and after each reaction rate analysis with a signal generator, using a 100-kc. crystal-controlled oscillator as a frequency standard. The stability of the frequency meter was such that the beat frequency could be measured as precisely as the recorder chart could be read (0.1 to 0.2% of full scale). Power Supply. The regulated power supply used for the oscillators and for the output stage of the frequency meter has been described (7). Although the power supply exhibited very good stability under large variations in load, its output voltage was excessively sensitive to changes in line voltage. The effect of line voltage
1960
ANALYTICAL CHEMISTRY
variations was reduced by using a Raytheon line voltage regulator, to the extent that a 10-volt change in line voltage produced less than a 0.1% change in the frequency meter output. DEVELOPMENT
OF METHOD
The alkaline hydrolysis of ethyl acetate was used to develop and test the method. This hydrolysis, thoroughly studied by many investigators, is a straightforward bimolecular reaction (8,20) : 0
I
CHaC-ocsH6
+ OH-
+
l
+
CHa -0C2H6OH
The rate constant for this reaction is such that it was conveniently measurable with the 30-Mc. instrument used Table 1. Data for Reaction of 107.5 Mg. of Ethyl Acetate with Standard Base Solution
T.A.M.,. G.S.R.,b T.A.M.,. G.S.R.,b Sec. Mm. Sec. Mm. 20 67.3 240 39.4 40 63.0 270 37.7 60 58.8 300 36.6 80 55.0 360 35.9 100 52.0 390 35.3 120 49.0 420 34.7 150 45.7 450 33.9 180 43.0 480 33.7 210 41.3 Recorder reading for 33.3 kc. Before run, 0.838 After run, 0.838 Calibration of G.S.R.,b reading of 30 mm. corresponds t o 30.000" C. a Time after mixing. Galvanometer scale reading.
'
in this work. The observed frequency change is caused by the change in conductance of the solution ( l 7 ) , brought about by conversion of highly conducting hydroxyl ion to less conducting acetate ion as the hydrolysis proceeds. The rate of frequency change may therefore be taken as a measure of the reaction rate. Procedure. The procedure used is best understood by reference t o Table I and Figure 3, data obtained for the reaction of 80 ml. of 0.0075N sodium hydroxide solution with 10 ml. of a solution containing 107.5 mg. of ethyl acetate. An 80-ml. aliquot of standard base solution, contained in a polyethylene bottle stored in the water bath, was transferred by pipet to the reaction vessel and stirring was begun. An atmosphere of oxygen was maintained over the solution in the storage bottle during the transfer and in the reaction vessel throughout the analysis. When the solution in the reaction vessel reached thermal equilibrium, the 0- to 40-kc. scale on the frequency meter was calibrated a t 33.33 kc. and the calibration reading recorded as in Table I. The reference oscillator was then adjusted to a frequency about 70 kc. above that of the loaded oscillator, recording the beat frequency on the 0to 80-kc. scale. Next, a 10-ml. aliquot of ethyl acetate solution was pipetted from a container stored in the water bath to the reaction vessel, using a pipet that had been partially immersed in the water bath with the immersed end sealed by a rubber policeman. The time of addition of ethyl acetate solution was automatically recorded by the decrease in beat frequency caused by dilution of the sodium hydroxide solution. At the same moment, timing of the reaction with a stop watch was started manually. After adding the ethyl acetate solution, the reference oscillator was again adjusted to a frequency above that of the loaded oscillator. The frequency meter was switched to the 0- to 40-kc. scale and a final adjustment made to a beat frequency of about 40 kc. As the reaction proceeded, the beat frequency decreased to zero beat, then increased as the frequency of the loaded oscillator became higher than that of the reference oscillator. After the beat frequency increased to about 35 kc,, the reference oscillator was readjusted to begin another frequency interval. Recording of beat frequency changes was terminated after three or four such intervals and the frequency meter calibrated as before. Under the conditions used in this study, one fourth to one half of the sodium hydroxide was consumed during the time required to measure three such intervals. To measure temperature, deflections on the bridge galvanometer were read a t 20- to 30-second intervals, beginning immediately after the start of the first frequency interval. After the last frequency interval, a thermometer was
Table II.
Corrections Applied to Data Shown in Table I and Figure 3
Uncorr. Time Time % Corr. t o Be Applied t o Time after Length of Length of Interval Mixing, Interval, Frequency Temp. Dev. from Sec. SecSa meter change 30' C. -0.55 -0.27 55.8 -0.25 66.5 -0.15 -0.36 66.8 -0.26 149.0 -0.09 -0.25 -0.24 245.0 82.6 -0.10 -0.04 120.7 -0.25 425.5 To nearest 0.1 second. * To nearest O.lyo.
Corr.
Total
Time Length of Interval 55.2 66.3 82.1 120.2
%
-1.1 -0.8
-0.6 -0.4
Q
inserted in the solution and simultaneous thermometer and galvanometer readings were obtained, permitting translation of galvanometer readings to temperature in degrees Centigrade.
Table 111. Values of Af/At Obtained from Data of Tables I and II and Figure 3
Time after Mixing, See.
of Data. Using the 0.84-ma. ordinate as a reference, the time length of each frequency interval Treatment
and the time after mixing a t the center of each interval were measured and recorded as in Figure 3 and Table 11. Corrections were then applied for: the deviation of the reference ordinate from the reading for calibration frequency, the change in solution conductance with temperature, and the temperature dependence of the reaction rate constant. Although the 0.84-ma. ordinate was used as a reference in measuring frequency intervals, the time required for a 66.7-kc. frequency change was desired, necessitating a correction for the deviation of the reading for 33.33 kc. from 0.84 ma. For the data of Table I, the measured time length of each interval r a s in error by 100 (0.840-0.838)/0.838 = +0.25%, so a negative correction of 0.25% was applied. The correction for temperature changes between intervals was made on the basis of the thermistor-bridge sensitivity and the temperature dependence of frequency, derived from the slope of the response curve (frequency os. conductance) and the temperature coefficient of conductance. Assuming sodium hydroxide to be the predominant electrolyte during the early stages of the reaction, a 0.052% correction was applied for each millimeter change in galvanometer deflection. Bridge voltage polarity %-assuch that the corrections were of the same sign as the change in galvanometer deflection. Corrections for variation of the reaction rate constant with temperature Kere applied by determining the temperature of the reaction system a t the center of each interval and correcting to 30.00" C. The temperature a t the center of each interval was calculated from the bridge sensitivity and the simultaneous thermometer and galvanometer readings obtained a t the end of each run. The temperature coefficient reported in the literature, 5.5% a t 30" C., was used (20). This was justifiable because it was
0
100 200 300
1.5
n
2 0 0
w
-
Time Length, of 66.7-Kc. Intervals, Af/W Sec. C.P.S./Sec. 47.8 1395 59.4 1122 72.2 923 92.5 721
SECONDS
-
-
v)
\
t n -
' 100 ' I 120 140 * MG. OF ETHYL ACETATE ADDED
055p 80
I
Figure 4. Working curves for determination of ethyl acetate by its rate of. reaction with sodium hydroxide
found that, over the range of solution conductance involved in this work, the slope of the instrument response curvei.e., plot of frequency us. conductancedid not vary significantly \\-ith temperature. The corrected time length of each 66.7-kc. frequency interval was then plotted against the corresponding time after mixing. From the best smooth curve (very slight curvature) drawn through these points, time lengths of frequency intervals were obtained a t 0, 100, 200, and 300 seconds after mixing. For plotting the final working curves, these interpolated and extrapolated data were converted to values of Af/At (Table 111). VOL. 30, NO. 12, DECEMBER 1958
1961
Table IV.
Calibration Data for Determination of Ethyl Acetate and Acetylacetone
hIg. Time Added in after Compd. 10-Ml. Uixing, Detd. Aliquot Sec. Ethyl acetate 71 2 0 100 200 300 89.6 0 100 200 300 107.5 0 100 200 300 143 1 0 100 200 300 Acetylacetone 23 98 50 100 150 31 17 50 100 150 38 37 50 100 150 47 96 50 100 150
Lf / At, C.P.S./See., for Samples A B L. 920 925 926 766 762 763 635 638 634 532 533 531 1153 1168 1144 940 935 942 773 772 768 639 636 1372 1401 1395 1122 1122 Ill7 923 922 923 720 722 72 1 1812 1852 1431 1440 1093 1095 838 843 983 980 980 846 844 843 732 i29 731 1244 1253 1048 1050 889 889 1508
-
1237
1024 1817 1443 1161
Discussion of Results. Duplicate or triplicate runs were made with various amounts of ethyl acetate t o obtain the results summarized in Table I V and Figure 4, in which t h e average of several values obtained a t each concentration and time after mixing is plotted as a single point. ( I t should be pointed out t h a t t h e wavy form of the curve for Aj/At a t 200 seconds after mixing is real and can be explained on the basis of the instrument response curve.) Aqueous solutions of ethyl acetate can be analyzed by referring values of Af/At for such solutions, measured as above, t o the calibration curve of Figure 4. The interpolated values of AflAt in Table IV indicate a precision of about 0.3% in determining 4 f l A t at the times after mixing selected for the calibration curves of Figure 4, while a poorer precision resulted from the uncertain extrapolation to zero time. The degree of reproducibility is well within the estimated limits of experimental error and indicates adequate detection and control of the variables involved. The error in determining concentration from Figure 4 increases with increasing time after mixing, due to the decreasing slopes of these curves. The error in determining concentration from a value of 4 f l A t at zero time would obviously be equal to the error of Af/4t a t zero tinic. and at 100 seconds after niixing the equality still holds approximately. However, the lesser slope of 1962
ANALYTICAL CHEMISTRY
1831 1449 1164
ilv . Value of AflAt
924 764
636 532 1155 939 771 638 1389 1120 923 721 1832 1436 1094 841 981 844 731 1249 1049 889 1508 1237 1024 1824 1446 1163
% Uev. from Av. Value of s j / ~ t ,4v.
Xlax.
0 25 0 22 0 26 0 13
0 43 0 26 0 31 0 18 1 13 0 43 0 39 0 31 1 22 0 27 0 11 0 14 1 09 0 35 0 09 0 36 0 20 0 24 0 27 0 40 0 10 0 00
0 75
0 0 0 0 0
29 26 24
81
21
0 03
0 09 1 09 0 31 0 09 0 30 0 14 0 12 0 14 0 36 0 10 0 00
0 38 0 21 0 13
0 38 0 21
0 1;
the 300-second curve of Figure 4 causes a threefold greater error in concentration than in the value of AflAt referred to the curve. Superficially, this would seem to indicate that only Af/At values obtained near zero time would be of value. It will be shown later, however, from data obtained with acetylacetone, that values of A f l 4 t a t various times after mixing are useful in detecting interferences that affect instrument response. A precise measurement of initial rates would make application of the method to resolution of mixtures an attractive possibility and would also provide a convenient means of correcting for interferences. Several attempts were therefore made to determine A f l U a t zero time with greater precision than sho1v-n in Table IV, but all were unsuccessful. Briefly, the unsuccessful approaches were: derivation of a theoretical espression to guide the extrapolation, enipirical methods of straight-line extrapolation, and measurement of instantaneous rates nearer zero time with a differential type of frequency meter (3). Conclusions Regarding Applicability of Method. The results obtained with ethyl acetate indicate t h a t the method yields highly precise determinations 01 single constituents in samples free of interferences. The question of interferences will be discussed in following sections of this paper. T h e method is potentially applicable t o any reaction in which t h e
constituent nhich is sought reacts to cause a change in conductance. Resolution of binary mixtures can probably be performed with high precision by using the same total reactive group concentration in each reaction rate analysis. Optimum precision can be expected only for niivtures free of interferences. A sacrifice in precision must be made for samples containing appreciable interferences or in cases where the same total reactive group concentration is not always used. This is because such cases are best handled by use of values of Af,'Lt a t zero time, obtained by extrapoiation with a resultant loss in prccision. This situation might be miproved upon by better instrumentation. Consideration of the above factors indicates that application of the method is best made to reactions that do not go to completion, an example of which folloa q. DETERMINATION OF ACETYLACETONE
The determination of many organic compounds cannot be carried out in a simple, straightforward manner because of the incompleteness of reactions n hich they are known to undergo. Bcetylacetone was until recently in this category, owing to the incompleteness of its reaction n i t h the reagent used niost conimonly for determination of ketones (4). The classical procedure for determination of ketones depends on reaction lyith hydroyylamine hydrochloride : Ri
)C
= 0
+ HOKH3C1
R2
Ri C'
=
XOH
+ HC1 + H20
R2'
where R1and Rzcan be hydrogen. Pyridine is used to displace the equilibrium in the above reaction by combining n i t h the hydrochloric acid liberated. i i t the end of the reaction, pyridine hydrochloride formed is titrated lvith standard base to determine the amount of ketone involved in the reaction. For many carbonyl compounds this reaction goes to completion in 30 minutes a t room temperature, while others may require 2 hoursornioreofheatingat % " t o 100°C. for complete reaction. In some cases heating merely increases the rate of approach to a less favorable equilibrium a t higher temperatures, and long cooling times are then necessary to establish a condition of approxiniate completion a t room temperature. For acetylacetone and hydroxylamine hydrochloride, Siggia reports that the reaction is over 95.5yo complete after 2 days' reaction at room temperature ( 2 1 ) . Another disadvantage of the above method is that the end point in the titration of pyridine hydrochloride is not accurately defined. Several methods of overcoming this difficulty have been developed, but the best are semiempirical
in nature or give errors of 2% or more (18, 82). It was felt that application of reaction rate measurements to the reaction of hydroxylamine hydrochloride with acetylacetone would eliminate the problems mentioned above. Also, results obtained with ethyl acetate indicated that, the reaction rate method might yield results of greater accuracy in a shorter time than the above methods. These beliefs are supported by data. Choice of Conditions. T o niinimize t h e effect, of interferences and t o avoid t,he wavy forin of working curve obtained with ethyl acetate. conditions \yere chosen so as t o confine measurements to the linear portion of t h e response curve for t h e instrument. The response curve was found t o be nearly perfectly l i n a r over the specific conductance range of 6.5 X 10-4 to 10.5 x lo-' ohm-' em-', and the reaction of hydroxylamine hydrochloride with acetylacetone resulted in a n increase of conductance due to liberation of hydrochloric acid. Therefore, the hydroxylamine reagent used n-as of such a concentrat'ion that the specific conductance of thcL reaction system a t time of mixing was w r y near the lower limit of this range. The rate of oxinie formation is st'rongly dc,pentlrnt upon pH. -4study of the reaction of acetone lvith hydroxylamine to form acetoxime revealed a niaximum rate of formation around p H 4.5, with the rate dropping off rapidly a t higher and lower pH values (16). -4siniilar dependence of tlie rate of oxime formation may be expected of other ketones and aldehydes and is of particular importance in the case of acetylacrt'one. Reactions carried out with solutions of hydroxplaniine hydrochloride indicated that the rapid change in pH during the early stages of the reaction n-as very undesirable. I n order to obtain tlie desired solution conductance a t zero time, the required concent'ration of the salt resulted in a pH of about 4.2 a t the start of the reaction. As the reaction proceeded and the pH decreased, the reaction rate constant for the reaction decreased in a nonlinear manner. This caused the plot, of time lengths of frequency intrrvals L ' S . times after mixing to be very irregular in shape and decidedly useless for the intended purpose, Buffering the syst'em n-as not a practical solution to t'his difficulty, as it R-ould have nullified the increase in conductance caused by liberation of hydrogen ion during the reaction. Therefore, a reagcnt solut~ionwas used which contained an excess of hydrochloric acid and a lon-er concentration of hydroxylamine hj-drocliloride to obt'ain the desired solution conductance. By adjusting the p H of the reagent solution to a value such that the reaction rate constant varied nearly linearlj- and only slightly during
1.8
SECONDS AFTER MIXING
,,f
Table V. Corrected Values of Af/At Obtained with Unknown Samples of Acetylacetone
K t . of 4icetylacetone Reacted, Mg. 26.86 28.77 33.57 36.69 43.16
Af/ At, C.P.S./Set. Seconds after Mixing 50 100 150 1088 930 798 985 840 1159 1336 1117 938 1191 989 1446 1347 1098 1663
Table VI. Results Obtained on Analysis of Unknown Acetylacetone Samples
11-t. of Scetylacetone Reacted, Ng.
26.86 MG. OF AGETYLACETONE ADDED
Figure 5. Working curves for determination of acetylacetone by its rate of reaction with hydroxylamine hydrochloride
28.77 33.57
Seconds after Mixing 50
100 150 50
100 150 50
100 150
R t . of ilcetylacetone % Error Found, of Mg. Detn. 26.86 0 0 26.97 $0 4 26.92 +0 2 28.77 0 0 28.91 +0 5 28.83 $0.2 33.56 0 0 33.68 $0.3 33.64 $0.2 36 61 -0 2 36 52 -0 5 36 38 -0 8 42 93 -0 5 43 12 -0 1 43 02 -0 3
36 69 50 the reaction, satisfactory results mere 100 obtained. 150 Necessary Corrections. The types 43 16 50 of corrections necessary for this re100 150 action were essentially t h e same in principle as those applied in t h e ethyl acetate study. T h e method of correcting for the frequency meter calitylacetone n as dried over anhydrous bration was exactly t h e same as above, sodium sulfate and distilled through a whereas t h e correction for temperature 15 x 1 inch column packed n i t h glass beads. The first portion was dischanges n-as only 0.025% per mm. carded and small fractions were change in galvanometer scale reading. separately until t h e disTo correct measured values of ~ f l ~ collected t tillation was nearly complete. The to a common temperature, a knowledge first 20 ml. of these fractions showed a of the temperature dependence of this gradual increase in refractive index from quantity was necessary. This was deabout 1.446 to 1.4492. The following termined experimentally from ~ j / ~ 20 t ml. of small fractions had an essenvalues a t zero time, for a given contially constant value of 1.4492; these centration of acetylacetone, obtained a t were combined and used in preparing temperatures of 28", 30°, and 32.5" C. solutions for reaction-rate runs. To permit use of a single set of workAs the p H varied during a reaction, the ing curves, it was necessary that the temperature dependence was determined hydroxylamine reagent be of the same a t p H values of 2.8 and 3.2, the approxiconcentration for all reaction rate runs. mate limits of the pH range over which As hydroxylamine hydrochloride is not A j f l l t measurements were made during a a primary standard, the following prorun. At 30" C., the temperature used cedure was used in preparing each fresh in the procedure, temperature coeffisolution of this reagent, which was cients n ere found to be 7.47, a t a pH of 0.003300M in hydroxylamine hydro3.2, and 5.7% a t a pH of 2.8. As the chloride and 0.00100OJ4 in hydrochloric acid. A 0.463-gram portion of h4erck greatest precision in determining conreagent grade hydroxylamine hydrocentration may be expected from measchloride was weighed into a 2-liter voluurements nearest zero time, and the pH metric flask, 20.00 ml. of O.lOOOA* a t zero time n a s about 3.2 for the rehydrochloric acid was added, and action rate analyses, a temperature cothe contents were diluted to the mark efficient of 7.0% was used in correcting with distilled water, giving a solution all l j ' & measurements to a common slightly more concentrated in hydroxyltemperature of 30" C. Temperature coramine hydrochloride than desired. A rections never exceeded 0.1" C.. so this portion of this solution was then reapproximate method of correction probacted with an acetylacetone solution of known concentration, and value4 of ably introduced errors smaller than 0.17, A j ' A t nere obtained a t various times in the corrected values of l j f / A t , after mixing. By referring the observed Reagents. A 50-ml. volume of values of Af, At to a calibration curve Eastman Kodak practical grade aceobtained by reacting thc same acetylVOL. 30, NO. 12, DECEMBER 1958
1963
acetone solution with standard hydroxylamine solutions of known concentrations, the actual concentration of the fresh solution was calculated. The solution was then diluted with the required amount of 0.001000N hydrochloric acid to give the desired concention.
As noted by Marasco ( l a ) ,solutions of hydroxylamine hydrochloride are unstable and should be prepared fresh every 2 or 3 days. Reaction rate analyses indicated a rate of decrease of hydroxylamine hydrochloride concentration of about 1% per week. Procedure. The procedure is essentially identical to t h a t employed in the ethyl acetate study discussed above, except for a minor difference in the method of adjusting the reference oscillator in recording frequency intervals. As the present reaction results in an increase in solution conductance and a decrease in frequency, the reference oscillator is adjusted to a frequency below that of the loaded oscillator a t the start of each frequency interval. Working Curves. Data for working curves were obtained by reacting 80.0 ml. of hydroxylamine reagent with 10.00-ml. aliquots of aqueous solutions containing 24 to 48 mg. of acetylacetone. The final corrected results are shown in Table IV and Figure 5, A single point on Figure 5 represents the average value obtained from two or three runs, with the exception of the point corresponding to 38.37 mg. of acetylacetone, for which only one run m-as made. As in the case of the ethyl acetate study, the average deriation of Af/At measurements is of the order of 0.3%, well within the estimated limits of experimental error. The working curves in Figure 5 are consistently regular in form, with no reversals in curvature as were obtained with ethyl acetate. Determination of Concentration by Reference to Working Curves, Several solutions containing known amounts of acetylacetone were run as unknown samples. A fresh bath of hydroxylamine solution was prepared for these runs, and the concentration adjusted as described above before using it to analyze these samples. Reaction-rate runs were made, and the corrected values of Af/At referred to enlarged plots of the working curves shown in Figure 5. The results are summarized in Tables V and T’I. The average error in the analysis of these five samples was about 0.3%. I n only one of the determinations was an error greater than 0.57, obtained, and this was the result from Af/At a t 150 seconds after mixing for the sample containing 36.69 mg. of acetylacetone per 10 ml. of solution. Although the error of the Af/At measurement m-as only 1964
ANALYTICAL CHEMISTRY
Table VI!. Interferences in Determination of Acetylacetone
Present in 10-M1. Alirta Acetylacetone Soh. NaCl. me. 2.464.92 9.84 17.2 EtOH, ml. 0.5
Acetylacetone Found, hlg., from ~ f l ~ t Seconds after Mixing 50 100 150 31.17 31.07 30.50 28.67
30.98 30.59 29.46 27.71
30.66 30.09 28.63 26.42
31.24 31 6 -~ I_. 31.24 31.16 30.98 30.95 Each aliquot contained 31.17 mg. of
1.o 2.0
31.12 31.24 30.53
acetylacetone.
0.5%, the slight slope of the working curve resulted in a larger error in the final analytical result. Interferences. Chemical interferences consist of substances t h a t react with the hydroxylamine hydrochloride a t a n appreciable rate or t h a t significantly change the p H of the system. Other interferences are those t h a t do not affect the reaction but have a n effect on the instrument response. The latter are of two types-one type alters the shape of the response curve by causing the dielectric constant of the reaction solution to be different from that obtained with only the hydroxylamine reagent and acetylacetone in solution; the other changes the conductance of the system sufficiently to alter the sensitivity of the instrument to changes in solution conductance. Chemical interferences Lvere not investigated because they would depend upon specific applications of the method; each case would probably require its own special treatment. Nonchemical or chemically inert interferences can be investigated in a more general manner, however, as they are dependent upon the physical properties of the substances involved rather than their chemical identity and behavior. To determine the extent of nonchemical interferences, runs were made with solutions containing 31.17 mg. of acetylacetone per 10-ml. aliquot. Also present in these solutions were varying amounts of sodium chloride, t o alter the conductance of the system, or of ethyl alcohol, to change its dielectric constant. The results are summarized in Table VII. The effect of inert electrolyte increases rapidly with its concentration and with increasing time after mixing. This was predictable on the basis of the response curve for the instrument, which shows a decrease in sensitivity with increasing conductance a t conductance values above the range of nearly linear response. The effect of inert solvents,
as represented by the addition of ethyl alcohol, is less pronounced. It was attempted t o extend the interference study to still higher concentrations of ethyl alcohol. However, the heat of solution of ethyl alcohol produced such large temperature changes a t the beginning of the reaction that the effect of large amounts could not be studied because of the difficulty involved in trying to measure the temperature changes. The effect of chemically inert interferences can be reduced by employing smaller sample sizes and measuring Af/At a t higher instrument sensitivity. A lower precision of results would be expected as the effect of oscillator instability increases, but the loss of precision might be compensated for by the ability to handle larger amounts of interferences. CONCLUSIONS
It may be concluded that measurement of reaction rate during the early stages of a reaction is capable of yielding highly accurate results in a short time, Using the equipment and procedure described, data obtained during the first 5 to 10 minutes of a reaction are capable of yielding results accurate to 0.3% for simple systems. The method is applicable to integral-order or complex reactions and can probably be extended to mixtures and samples containing chemically inert substances with some loss in accuracy. Although a high-frequency instrument was used in the present investigation, this approach could probably be extended to other types of instruments. SF‘ith some instruments, such as a spectrophotometer, simplification and broader application might result because of a more linear instrument response and/or a lower temperature coefficient of the variable measured. ACKNOWLEDGMENT
This work n’as supported in part by the Wisconsin Alumni Foundation. During 1952-1953, D. L. Petitjean held the Merck Fellowship in Analytical Chemistry. LITERATURE CITED
(1) Axtmann, R. C., ANAL. CHEM.24, 783 (1952). (2) Blaedel, W. J., Malmstadt, B. V., Ibid., 22, 734 (1950). (3) Ibid., 24, 450 (1952). (4) Bryant, W. M. D., Smith, D. M., J. Am. Chem. SOC. 57,57 (1935). (5) Dillon, R. F., Young, W. G., Lucas, H. J., Ibid., 52, 1953 (1930). (6) Feigle, F., “Specific, Selective and
Sensitive Reactions,” Academic Press, New York, 1959. (7) Hoadley, J. C., Radio and Television News 43, 47 (August 1950). (8) Jensen, F. W., Watson, G. M., Beckman, J. B., AXAL. CHEM. 23, 1770 (1951).
~
(9) Kolthoff, I. M,, Livingston, R. S., IND.ENG.CHEM., ANAL. ED. 7, 209 (1935). (10) Lambert, R. H., ANAL. CHEM. 24, 868 (1952). (11) Lein, A,, Schwarte, K.,Ibzd., 23, 1507 (1951). (12) Marasco, M., I n d . Eng. Chem. 18, 701 (1926). (13) Mitchell, John, Jr., “Organic Analysis,” Vol. 11, pp. 237-52, Interscience, New York, 1954.
(14) Morau, J. J., ANAL, CHEM,24, 378 (1952). (15) Olander, 2. Physik. Chem. 129, 1 (16 Reich, H. J., Ungvary, R. L., Rev. !c:2:?str. 19, 43 (1948). (17) Reilley, C. K.,McCurdy, W. H., Jr., ANAL.CHEM.25, 86 (1953). (18) Roe, H. R., Mitchell, J., Jr., Ibid., 23, 1758 (1951). (19) Sandell, E. B., Kolthoff, I. M., J . Am. Chem. SOC.56,1426 (1934).
(20) Shrivastava, H., J. Indian Chem. SOC.17,387 (1940). (21) Siggia, S., “Quantitative Organic Analysis via Functional Groups,” p. 18, Wiley, New York, 1949. (22) Smith, D. M., Mitchell, J., Jr., ANAL.CHEM.22, 750 (1950). (23) Underwood, A. L., Burrill, A. hf., Rogers, L. B., Zbid., 24, 1897 (1982). (24) West, P. W., Zbid., 23, 176 (1951). RECEIVED for review June 10, 1957. Accepted June 17, 1958.
Ammonium Molybdate as Spraying Agent for Paper Chromatograms of Reducing Sugars H. EL KHADEM and S. HANESSIAN Chemistry Department, Faculty o f Science, Alexandria University, and Starch Products Co., lid., Alexandria, Egypt
b Paper chromatograms of reducing sugars sprayed wth ammonium molybdate revealed spots which did not fade with time, while the background remained colorless if not exposed to strong light.
A
VERNON( I ) have suggested t h e use of 10% aqueous ammonium molybdate, concentrated hydrochloric acid and ammonium chloride as a spraying agent for paper chromatograms of reducing sugars and easily hydrolyzable nonreducing disaccharides such as sucrose. It was found t h a t ammonium molybdate alone readily revealed reducing sugars and offered several advantages over the above reagent. The background of the chromatograms remained RONOFF AND
colorless instead of turning blue, the chromatograms were less sensitive t o heat, so that they could be rolled and put in a drying oven without acquiring heat marks, the paper chromatograms did not deteriorate on keeping, and the spraying reagent could be stored unchanged for long periods. Paper chromatograms having spots containing 25 y of each of the sugarsarabinose, xylose, rhamnose, glucose, mannose, galactose, fructose, sorbose, glucosamine hydrochloride, maltose, and lactose-were developed with the upper layer of a mixture of butanolethanol-water-ammonia (40: 10:49:1) (2). The d r y chromatograms were sprayed with 10% ammonium molybdate and heated to 100” for 10 minutes. The spots revealed were first yellow, and after about 6 hours turned blue-gray on
a colorless background. It has been reported (3) that they got decidedly darker and clearer on standing, which is a big advantage over most of t h e available spray reagents that give spot F which fade with time. The spots could be observed to greater advantage in daylight or under ultraviolet light. The background of the chromatograms remained colorless for several months, provided they were not exposed to strong light. LITERATURE CITED
(1) Aronoff, S., Vernon, L. P., Arch. Biochem. 28,424 (1950). (2) Hirst, E. L., Hough, L., Jones, J. K. N., J . Chem. SOC.1949,928. (3) Lamblou, M. G., Blouin, F., private
communication. RECEIVEDfor review April 10, 1958. Accepted August 7, 1958.
Solvent Extraction of Chromium with Acetylacetone JAMES P. McKAVENEY’ and HENRY FREISER2 Department o f Chemistry, University o f Pittsburgh, Pittsburgh, Pa. ,The fact that the hydrated chromium(lll) ion is inert to chelate formation and solvent extraction under normal conditions has been used t o effect the separation of chromium from aluminum, iron, vanadium, molybdenum, and titanium. The latter metals are separated from an aqueous solution a t pH 2.0 by extraction with a 1 to 1 mixture of acetylacetone (2,4-pentanedione) and chloroform. After extraction of the interfering ions, the aqueous raffinate is refluxed with acetylacetone to convert the hydrated chromium(lll) ion to the chelate form.
The chromium acetylacetonate is then extracted with a 1 to 1 mixture o f acetylacetone and chloroform. The red-violet extract has been used for the determination of chromium above 0.20% in ferrous materials. Lower chromium contents are determined by the diphenylcarbazide procedure, following isolation with acetylacetone.
T
is a continuation of the study on the solvent extraction of metal ions encountered in ferrous analysis. Although chromiuni acetylacetonate has been known since 1899 HIS PAPER
(6) the peculiar behavior of chromium toward solvent extraction with acetylacetone has only been recently noted (1). Direct shaking of a chromiuni(111) solution with a 1 t o 1 mixture of acetylacetone and chloroform did not result in any extraction of chromium in contrast to the effect on such metals RS aluminum(111), iron(II1) t itaniuni(1V) , molybdenum(VI), and vanadium (III), (IV), and (V) (2) n-hich rrere easily ~
Present address, Crucible Steel Research Laboratory, Pittsburgh 13, Pa. * Present addresa, University of Arizona, Tucson 25, Ariz. VOL. 30, NO. 12, DECEMBER 1958
1965