for preparing the illustration, and K. T. Williams, E. B. Kester, and Yoshio Tomimatsu for technical advice. LITERATURE CITED
(1) Feinberg, B. G., Am. Chem. J . 49, 87 (1913).
(2) Hunter, I. R., Dimick, K. P., Corse, J. W.,Chem. & Znd. (London) 16, 294 (1956). ( 3 ) Hunter, I. R., Ferrel, R. E., Houston, D. F., J . A q r . Food Chem. 4, 874 (1956). (4) Parkinson, A. E., Wagner, E. C., ISD. EXG. CHEM., .4sa~. ED. 6 , 433 (1934).
(5) Tomada, J., J . SOC. Chem. Znd., Japan, 30, 747 (1927); J. SOC. Chem. Znd. 48, 79 (1929). (6) S’irtanen, A. I., Rautanen, N., Soumen Kenaistilehtz 19B, 56 (1946).
RECEIVEDfor review June 10, 1957. Accepted October 4, 1957.
Semimicro Hydrogenation with Electrically Generated Hydrogen JOHN W. MILLER’ and DONALD D. DeFORD Department of Chernisfry, Northwestern University, Evanston, 111.
b As hydrogenation apparatus previously described does not allow simple pressure regulation or easy measurement of reaction rates, a simple apparatus was designed. The conventional gas buret and leveling bulb were replaced b y a U-shaped electrolysis cell, which served as both a source of hydrogen and a pressureregulating device. The electrical current used to produce the hydrogen consumed in the hydrogenation was measured by an electronic coulometer, which also operated a stepping recorder. Curves were automatically plotted. A variety of organic compounds were hydrogenated with an and a standard accuracy to f 3 % If room temperadeviation of 2%. ture or atmospheric pressure does not change during hydrogenation, a hydrogen uptake of less than 2 ml. can be accurately measured.
H
has been used for the analysis of unsaturated materials for over 50 years. The “simplified type” of apparatus was first introduced in 1908 by Fokin (6) and Paal and Gerum (8). Many modifications of the original apparatus have been described (2, 7 , 9, l l , 12, Id), all based on volumetric measurement of hydrogen uptake. The lower limit of accurate measurements is dependent on the size of the gas buret. Most methods have a lower limit of 3 to 5 ml. of hydrogen consumption. Therefore, a n investigation was undertaken t o design and construct a hydrogenation apparatus which would not require volumetric gas measurements and would allow simple, automatic pressure regulation. With the restriction of measuring gas volumes eliminated, the method could be extended YDROGENATION
Present address, Research Division, Phillips Petroleum Co., Bartlesville, Okla.
to smaller volumes of hydrogen. The apparatus and method mere based on the use of electrically generated hydrogen as a source of gas and elimination of gas burets. Manegold and Peters (6) were the first t o use electrically generated hydrogen in place of the conventional gas buret. I t s use for the determination of relative rates of hydrogenation has been described by Farrington and Sawyer (4). The new apparatus is a modification of that of Manegold and Peters (6), which embodied a cell for electrically generating hydrogen, a gas coulometer t o measure the amount of electricity consumed, and a reaction vessel. This apparatus measured both the rate of reaction as indicated by the current intensity and the total amount of gas consumed as shown by the coulometer, but the current had t o be adjusted manually during a run, so that the rate of generation equaled the rate of disappearance of hydrogen. The apparatus was constructed t o measure large volumes of gas (liters) and was too complex for easy reproduction and maintenance. These disadvantages \yere overcome by automatic electrolysis, which regulated the pressure and varied the electrolysis current so that rates of consumption and generation of hydrogen were equal.
A schematic diagram of the apparatus is shown in Figure 1. The sample, solvent, and catalyst are contained in the reaction flask, F . The flask is connected to the U-shaped electrolysis cell, Tz. by the gas manifold and drying tube, T I . The entire system is closed t o the atmosphere by stopcocks and by placing generator electrolyte in T z . The catalyst is prereduced with hydrogen; when reduction is complete, the increase in the hydrogen presfiire in the system causes the liquid to be pushed away from the electrode. The contact b e h e e n the electrode and electrolyte is broken, and the elec-
trolysis current is automatically shut off. The pressure in the system exceeds the atmospheric pressure by an amount which corresponds to the difference in liquid levels in the U-tube arms. When the sample is introduced into the solvent, reaction takes place and the hydrogen pressure in the system decreases, causing the liquid to make contact with the electrode. The rate of hydrogen generation is adjusted to equal the initial rate of hydrogen consumption. After this setting is made, the hydrogenation is automatic. Some degree of current control is achieved by employing a tapered electrode, so that its depth of immersion in the electrolyte governs the current flowing. As hydrogenation proceeds, the rate of hydrogen uptake decreases, and the liquid level falls, slowing down the rate of hydrogen generation. If the rate of generation exceeds the rate of consumption because of a slow hydrogenation reaction, the instrument will cycle on and off. During the off-cycle the pressure in the system slowly drops until the liquid makes contact with the electrode again. K h e n no hydrogen is generated for a given length of time, the reaction is assumed to be complete. A stepping recorder was placed in the coulometer circuit, so that the number of milliequivalents of hydrogen consumed was plotted against time. After the initial adjustments, the hydrogenation proceeds to completion automatically with no operator attention. The results are then calculated from the coulometer reading as registered on the recorder. APPARATUS
The reaction vessel, F , u-as constructed from a 25-ml. Erlenmeyer flask to which was added a standardtaper 14/20 joint, J,. A short length of 6-mm. tubing served as a side arm for introduction of sample. A female standard-taper 14/20 joint, sealed to a 3inch length of 1-mm. capillary tubing, connected the reaction vessel to the rest of the system. The ball and socket VOL. 30, NO. 2, FEBRUARY 1958
295
joint, Jz, was placed above the reaction flask to give flexibility to the system. The three-way 2-mm. capillary stopcock, C1, allowed the reaction flask to be evacuated before hydrogenation was begun. A water aspirator was satisfactory for evacuation. A hydrogen reservoir, R, a 500-ml. separatory funnel filled with mineral oil, and attached to the system by the threeway 2-mm. capillary stopcock, CB, served as a source of hydrogen for the flushing procedure. The absorption tube, TI,filled with 8-mesh calcium chloride, dried the hydrogen before it passed into the reacting system. The generator cell, T,, was a U-shaped drying tube closed on the left by a No. 0 rubber stopper through which extended the platinum generating cathode (1.0 X 1.4 cm.). The cathode was tapered so that the width at the bottom was one half that a t the top. A rod of reagent grade zinc served as the generator anode. The individual components were joined with 2-mm. capillary tubing with butt joints of Tygon tubing between components. Such construction allowed easy cleaning of the apparatus and made it flexible and less subject to breakage. The total volume of gas in the apparatus, including the electrolysis cell, drying tube, connecting tubing, and empty reaction flask, was 51.5 ml. To reduce errors arising from changes in temperature or pressure, this gas volume should be kept as small as possible. The side arm of the flask was closed by a rubber serum bottle stopper. Liquid samples were injected by a hypodermic syringe, which served as a weight buret. For solid samples a cup (shown in Figure 1) was constructed of l / P inch aluminum rod, inch in length. A hole 3/16 inch in diameter was drilled into the center of the rod to a depth of 3/8 inch. Two small holes were drilled opposite one another a t the top and a wire loop of Chrome1 A wire was attached. The cup was hung from a hook made from a needle bent a t the end and inserted through a serum bottle stopper. The cup was dropped into the catalyst-solvent mixture by a half turn of the stopper. The hydrogenation mixture was agitated by a 5/8-in~h magnetic stirring bar placed in the reaction flask. The current source was a SargentSlomin Electroanalyxer, chosen because low voltage current was desired to prevent arcing between the electrode and the liquid when contact was broken. Two direct-reading coulometers were used during the work. The first (5) was based on the principle of charging a condenser to a given voltage with the generation current and counting the number of times it was charged during a run. The second used an integrating motor (Model 120T-3, Summers Gyroscopy Co., Santa Monica, Calif.) to measure the amount of electricity passed during a run. The design and operation of these coulometers will be described in subsequent publications.
296
ANALYTICAL CHEMISTRY
REAGENTS
C.P. concentrated sulfuric acid was used to prepare the 6N acid. Platinum oxide catalyst and 10% palladium-on-charcoal were obtained from the American Platinum Works, Newark, N. J. The purity of several of the compounds used was unknown. The following materials were used as received from the manufacturer: cinnamic acid and acetophenone (Eastman Kodak Co.), and phenyl propyl ketone (Matheson, Coleman and Bell Division, Matheson Co., East Rutherford, N. J.). The low results obtained for these compounds were undoubtedly due to lack of purity. Reagent grade benzene was dried with sodium wire before use. The naphthalene had been purified by being crystallized four times from each of the following solvents: glacial acetic acid, acetone, and ethyl alcohol; it was further purified by refluxing for 39 hours in ethyl alcohol over Raney nickel. p-Nitrophenol was purified by recrystallization from 1 N sulfuric acid. I t s melting point was 112-13" C. PROCEDURE
By electrolyzing 6N sulfuric acid, hydrogen gas was produced a t the cathode. With all the stopcocks open to the atmosphere, hydrogen was allowed to flow through the system 20 minutes before any determinations were carried out. During the flushing and filling operations the current was adjusted to its maximum (1 ampere) by increasing the applied potential to full value. The reservoir was then filled with electrically generated hydrogen. While the apparatus was being flushed, the catalyst was weighed into the reaction flask and 5 ml. of solvent added. A weight of solid sample corresponding to a hydrogen uptake of between 2 and 22 ml. was placed in the aluminum cup. The cup was hung on the hook after the
solvent and catalyst had been added. High boiling liquids could also be weighed into the cup with no loss when the reaction vessel was evacuated. Liquid samples of high volatility were made up as standard solutions in the hydrogenation solvent, so that 1 ml. of the solution would take up the required amount of hydrogen. A 1-ml. tuberculin syringe (Yale No. 1YT, Becton, Dickinson, and Co., Rutherford, N. J.) was filled to the I-ml. mark with solution and weighed. After injection of exactly 1ml. of sample solution, the syringe was reweighed and the sample size was calculated from the loss in weight. The reaction flask containing catalyst, solvent, and magnetic stirring bar was attached to the apparatus. Stopcock Czwas turned so that only the hydrogen reservoir was connected to the reaction vessel, and the reaction flask was alternately evacuated and filled with hydrogen by proper adjustment of stopcock CI. After five evacuations, both stopcocks were turned so that the reaction vessel was connected directly to the generator cell. The magnetic stirring motor was set a t a speed which would allow rapid hydrogen transfer across the gas-liquid interface. The catalyst was reduced until the coulometer indicated no hydrogen uptake for 10 minutes. The recorder was turned on and the register on the coulometer was reset to zero. The sample was then introduced either by dropping the aluminum cup or by injecting 1 ml. of the standard solution into the solvent. The temperature and barometric pressure were noted. For the hydrogenation of aromatic nuclei, glacial acetic acid was used as the solvent and platinum oxide as the catalyst. Most of the other reductions were carried out in 95% ethyl alcohol; 10% palladium-on-charcoal was the catalyst. The catalyst usually weighed 10 to 20% as much as the sample. This amount was sufficient to permit the reaction to go to completion in less than an hour.
SAMPLE CUP
Figure 1.
Schematic diagram
of
apparatus
CALCULATION
OF
RESULTS
The number of counts on the coulometer was used to calculate the results. As the coulometers had been previously calibrated in terms of milliequivalents per count, the number of milliequivalents of hydrogen generated was the product of the number of counts and the calibration factor. This result was compared to the theoretical number of milliequivalents of hydrogen. The results were then calculated as the hydrogen number, defined in Equation 1 as n( 201.6) Hydrogen number = mol. wt. ~
=
No. of Detns.
Hydrogenation of Cinnamic Acid
Hydrogen Wt. Range ' No. Sample, Mg. Found
4 4 4 4 4
72.6-74.4 36.2-37.5 18.6-19.6 10.2-10.9 5.2-5.7
If the sample was added as a standard solution, the number of milliequivalents of hydrogen generated did not represent actual hydrogen uptake. When 1 ml. of solution was added, the liquid level in the generator cell was pushed a corresponding distance below the cathode. Even though hydrogen was consumed by the sample, no counts were registered until slightly more than 1 ml. of hydrogen had been used. For this reason it was necessary to add to the number of milliequivalents obtained fromlrthe coulometer the number of
%
Table II.
1.342" 1.339 1.330 1.324 1.343
HP Std. Max. Dev. Consumed, Dev., yo from Mean M1. 0.58 1.98 1.59 1.14 3.18
0.011b 0,034 0.029 0.019 0.052
11 5.5 2.7 1.6 0.84
Coulometric Hydrogenation of Unsaturated Compounds
KO.
of Wt. Range Compound Detns. Sample, Mg. Benzene 11 14.9-15.4 Naphthalene 6 20.6-23.3 p-Kitrophenol 8 47.2-54.3 Acetophenone 5 72.5-74.2 Phenyl propyl ketone 7 57.2-62.4 a I n terms of hydrogen number.
(meq. Hz)( 100.8) wt. of sample (mg.)
Error,
-1.47 -1.69 -2.35 -2.79 -1.39 Theoretical hydrogen number is 1.362. In terms of hydrogen number.
(1)
where n is the number of double bonds. The experimental hydrogen number was calculated by substituting the proper values in Equation 2. Hydrogen number
Table 1.
Table 111.
No. of Detns.
Wt. Range Sample, Mg.
3 6 10
51 .5-53.9 25.7-27.2 12.0-13.0
%
hfax. Dev. from Means
7.754 7.865 4.348 1.678
7.578 7.671 4.273 1.638
-2.27 -2.47 -1.72 -2.38
3.26 1.44 1.74 1.04
0.152 0.173 0.127 0.024
1.361
1.330
-2.28
3.75
0.090
.Hydrogen S O . Theory Found
Error,
Std. Dev.,
%
Hydrogenation of Fumaric Acid"
Hydrogen KO. Theory Found
Error,
%
Std. Dev., %
1,737 1.737 1 . 737
+O. 23 2.18 1.741 1 'ill -1.50 1.72 +O. 23 4 82 1.741 a Practical rade material was recrystallized from 1N hydrochloric acid and vacuum dried at 50" before use.
6
milliequivalents of hydrogen equivalent to the volume of sample added. The volume was corrected to standard pres-
sure and temperature before it was used in any calculations. The total number of milliequivalents taken up by the sample was then the sum of those from the coulometer reading and those from the volume of sample added. RESULTS AND DISCUSSION
I
I
60
10
1 40
I 30
I
I
20
10
TIME IMINUTES)
Figure 2.
Typical rate curves
The accuracy and precision expected for the hydrogenation of a variety of compounds are shown in Tables I to 111. The standard deviation is less than 2% in all but four cases. The poor precision for benzene was probably due to partial loss of sample during introduction into the reaction flask. The average of three determinations of benzene on a hydrogenation apparatus identical to that described by Park, Planck and Dollear (9) was -2.227& This result agrees with that obtained on the new apparatus. The extrapolation procedure used to find the number of coulometer counts for the reduction of phenyl propyl ketone accounts for the high standard deviation of this compound. Typical rate curves are shown in Figure 2. Khen the nitro group had been completely reduced, the uptake of hydrogen continued a t a slow rate. The dffercnce in the rate of the main reduction and the slow reaction was large enough so that the "end point" could be determined by extrapolation of the two linear VOL. 30, NO. 2, FEBRUARY 1958
297
portions of the curves. The end points for the hydrogenation of the two aromatic ketones, acetophenone and phenyl propyl ketone, also were found by extrapolation. The curve after the break was linear only at an optimum catalyst sample ratio of 5%; higher ratios gave a rounded curve, which could not be used for extrapolation. I n such cases the optimum concentration of catalyst must be determined before unknowns are analyzed. Table I indicates the effect of sample size on the determination of cinnamic acid (average results). I n three of the five weight ranges a t least one of the four determinations gave a high result. It mas possible to hydrogenate microsamples with little loss in accuracy. The precision decreased slightly as the sample size decreased; this may be due to the effect of temperature on the liquid level in the cathode compartment. It was possible to reduce acetophenone rapidly to ethylbenzene, with no break in the rate curve corresponding t o the alcohol intermediate, by using a 3 to 1 sample-catalyst ratio. The consumption of hydrogen stopped abruptly upon completion of this reaction. Two such hydrogenations gave errors of -4.41 and -3.75%, respectively. Naphthalene might be expected to give low results because of its slow rate of hydrogenation. The lorn results on p-nitrophenol are unexplained. Three main factors affect the accuracy of the hydrogenations: temperature, pressure, and surface tension. With a dead volume of 46.5 ml. when the reaction Aask is filled with 5 ml. of solvent, a 1” change in temperature during a hydrogenation is equivalent to 0.16 ml. of gas. The final result may be high or lorn, depending on the direction of the temperature change. For relatively large samples that require on the order of 15 ml. of hydrogen. this error is only 1%, but, for every small sample, the error may be as large’as 20%. In this laboratory temperature fluctuations were small or negligible during a determination and the accuracy and precision were good. During the summer months the room temperature varied by as much as 10” C. during the day, but fluctuations during a run must have been small. I n many situations some form of temperature control n-ould have t o be maintained. The effect of temperature fluctuation could be lessened by decreasing the dead volume of the apparatus, by using a smaller reaction flask and no drying tube. The drying tube accounted for more than half of the dead volume. The
298
ANALYTICAL CHEMISTRY
use of 85% sulfuric acid permitted elimination of the drying tube, as the vapor pressure of water is negligible over it. However, as the low conductivity of 85% acid limited the current to a few milliamperes, only microsamples could be analyzed. Because one arm of the generator cell is open to the atmosphere, changes in atmospheric pressure also produce errors. A pressure change of 1 mm. of mercury is equivalent to approximately 0.06 ml. of hydrogen. It is possible that a pressure increase could be compensated by a temperature increase, so that no error would be introduced. It is unlikely that pressure would change by 1 mm. of mercury during a half-hour determination; thus, the error from this source would, in general, be small. Because of surface tension effects the liquid level in the cathode arm of the generator cell a t the moment when contact with the cathode is broken is about 1 mm. lower than the level a t which contact is re-established. With the cell used this difference in height corresponds to a volume of about 0.13 ml. For this reason there is a dead zone corresponding to this volume of hydrogen. The dead zone corresponds to a possible error of 1% when approximately 15 ml. of hydrogen are consumed. The percentage error increases as the volume of hydrogen uptake decreases. If desired, the dead zone could be reduced to a negligible value by use of a cathode compartment of smaller cross section or more elaborate manostats. The apparatus should be useful for rate studies, as the rate curves are automatically plotted. Extremely small changes in rates are easily noted from the recorded curves. Saphthalene shows a small break at 40% reaction; this is due to the difference in reaction rates of naphthalene and Tetralin (1, 10). The “end point” of a hydrogenation can be determined by the sharp break in the recorded curve. The apparatus can be used t o study rates only when the rate of reaction is slower than the rate of transfer of hydrogen across the gas-liquid interface. The rate of transfer has been shown to be dependent on stirring (12, 13). For rate studies the reaction vessel described by Vandenheurel (22) should be the most satisfactory. Because the hydrogenation rate curves n ere used only to locate the end points, the reliability of these curves was not studied. The curves for a compound such as naphthalene showed differences from determination t o determination, undoubtedly due to the effect of agitation.
The low results are what might be expected for the compounds used and no known systematic error exists in the system. A thermostated apparatus would eliminate one large source of error. SUMMARY
A simple hydrogenation apparatus, in which electrically generated hydrogen replaces the conventional gas buret, is useful for determination of a wide variety of unsaturated functional groups. As microsamples can be determined with the same precision and accuracy as larger samples, the method is attractive for trace analysis. Although this method requires relatively complex electrical components, ease of pressure regulation and completely automatic operation make this method one of the simplest for the determination of hydrogen numbers. ACKNOWLEDGMENT
The authors acknowledge the helpful comments of R. H. Baker, who supplied the purified naphthalene sample. One of the authors (J. R. I f . ) wishes to thank the United States Rubber Co. for partial financial assistance in the form of a fellowship. LITERATURE CITED
(1) Baker, R. H., Schuetz, R. D., J . Am. Chem. SOC.69, 1250 (1947). (2) Castille, A., Bull. SOC. chzm. Belg. 46, 5 (1937). (3) DeFord, D. D., Toren, C. E., North-
\yestern University, Evamton, Ill., unpublished results, 1955. (4) Farrington, P. S., Sawyer, D. T., J . Am. Chem. SOC.78, 5536 (1956). (5) Fokin, S., J . Russ. Phys. Chem. SOC. 40, ‘700 (1908). (6) Manegold, E., Peters, F., Kolloid-2. 8 5 , 310 (1938). ( 7 ) Oemler, -4.S., Obold, H. K., Hawkins, W., Mitchell, J., Delan-are Chemical Symposium, Feb. 19, 1955. (8) Paal, C., Gerum, J., Ber. 41A, 805 (1908). (9) Park, F. C., Planck, R. W., Dollear, F. G.. J . Am. O d Chemists’ SOC.29. 227 (i952). (10) Schuetz, R. D., Ph. D. thesis, Northwestern University, Evanston, Ill., 1847
(11) TajLlor; H. S., Strother, C. O., J . Am. Chem. SOC.56, 586 (1934). (12) Vandenheuvel, F. A., A 4 ~CHEM. ~ ~ . 1:
RECEIVED for review December 17, 1956. Accepted September 26, 1957.
