The coal acids, a mixture obtained from the Dow Chemical Co., and the coal acids that had undergone the decarboxylation reaction, both containing known amounts of added m-toluic acid, were then silylated and chromatographed. Figures 2 and 3 show the type of chromatogram obtained. Using the response factors which were determined previously, the analysis of the coal acids was carried out. Table I1 summarizes the results. The starting acids contain fairly large amounts of phthalic, hemirnellitic, trimellitic, pyromellitic, and pentacarboxylic acids. The decarboxylated acids contain large amounts of benzoic, isophthalic, and terephthalic acids with only traces of higher acids. Thus, the decarboxylation reaction essentially converts all of the tri-, tetra-, and pentacarboxylic acids to iso- and terephthalic acids and the benzoic acid results primarily from decarboxylation of phthalic acid.
Table 11. Analysis of Coal Acids Coal acids. weight Z Before After Component decarboxylation decarboxylation 4 ... Benzoic Trace 3.3 Phthalic 10 Isophthalic }0.4 3 Terephthalic ... 2.8 Hemimellitic ... 5.5 Trimellitic >0.5 ... Trimesic Trace 10.1 Pyromellitic a ... Mellophanic4 ... Prehnitic" Trace 3.4 Pentacarboxylic Trace Trace Mellitic 17.5 25.5 Total a It was later shown that these acids could not be completely separated from pyromellitic acid so that the value given for pyromellitic represents the total tetracarboxylic acid concentration. It was assumed that the response factors were the same for all the tetracarboxylic acids.
The authors thank Joan Gordon for her assistance in the experimental portion of this work.
Once the relative response factors were determined, other mixtures of the acids containing different weights of each acid were prepared and analyzed using the appropriate response factors. Table I compares the known and found values of a sample mixture and shows that the analysis can be made satisfactorily.
RECEIVED for review March 24, 1967. Accepted April 17, 1967. Division of Fuel Chemistry, 152nd National Meeting, ACS, New York, N. Y., September 1966. Research supported in part by the Union Carbide Corp. Reference to a company or product name is made to facilitate understanding and does not imply endorsement by the U.S. Bureau of Mines.
0
ACKNOWLEDGMENT
Potentiometric Determination of Dimethylamine Borane and Hypophosphite A. F. Schmeckenbecher and J. A. Lindholm InternationaI Business Machines Corp., Components Division, East Fishkill, N. Y .
DIMETHYLAMINE BORANE (DMAB) and hypophosphite are now used as reducing agents to deposit metal fdms without electrolysis, The reducing agents are unstable under certain conditions. This report describes a new way of assessing the concentration of DMAB as a reducing agent in stock solutions. This will result in better control of the DMAB concentration in stock solutions, plating baths, etc. Several methods have been employed to oxidize the sample to be tested. Among the oxidizing agents used for hypophosphite determination are: potassium permanganate ( I , 2), bromine and iodine (3-5), mercury salts cerium(1V) (7, 8), molybdic acid (9), and ferric ions ( I O , 11). Other
(a,
L. Marino and A. Pellegrini, Gazz. Chem. Ifal., 43, I, 494 (1911). J. P. McCloskey, Plating, 51,689 (1964). E. F. Harrison, Pharm. J., 88, 12 (1911). I. M. Kolthoff, J . Chem. SOC.,110, II, 490 (1916). K. Burger and L. Ladanyi, Acta Pharm. Hung., 30,80 (1960). A. Schwicker, 2.Anal. Chem., 110, 161 (1937). (7) D. N. Bernhart, ANAL.CHEM., 26, 1798 (1954). (8) G. G. Guilbault and W. H. McCurdy, Anal. Chem. Acta, 24, 214 (1961). (9) G. Gutzeit, U.S.Patent 2,697,651(1954). (10) M. N. Sastri and Ch. Kalida, Rec. Trav. Chim., 75, 1122 (1956). (11) T. D. Verbitskaya and N. K. Romanova, Zuvodsk. Lab., 26, 818 (1960). (1) (2) (3) (4) (5) (6)
1014
ANALYTICAL CHEMISTRY
methods are based on the amount of hydrogen formed during reduction in a plating bath (I2), or simply on the rate at which a certain amount of metal is deposited from a plating bath containing a metal salt and the reducing agent (13). The methods of determination either take considerable time, require expensive equipment, or lack precision. Little is known about the determination of dimethylamine borane and other boranes. With slight modifications, this new method is also suitable for determining the hypophosphite concentration in similar solutions, although certain reducing agents, in particular phosphite ions, interfere with the determination. The method is fast, simple, and requires little equipment. For absolute determinations, standardization with a DMAB or hypophosphite solution of known concentration is required. The relative standard deviation in a series of 12 DMAB determinations is 0.75 %. EXPERIMENTAL
The solution to be tested is titrated in a vessel containing a known amount of palladium salt solution, a palladium electrode, and a reference electrode such as a saturated calomel (12) S. Greenfield and R. M. Cooper, Talunra, 9,483 (1962). (13) A. A. D'yakow and R. G. Rosenblum, Zacodsk. Lab., 30 (lo), 1216 (1964).
n
> E
-I
.01
4
c
z Q
3.78 ml DMAB-sol (ENDPOINT OF TITRATION)
30 0 .. 200 -.
Figure 2. Concentration of PdClz solution
Figure 1. Titration curve electrode. The electrodw are connected to a high-resistance millivoltmeter (pH meter). The DMAB reduces the palladium salt to palladium metal:
+ 3 Pd+' + 3 Hz0
+ + RzHN + 6H+
3Pd0 H3B03 +
cat
. BH3 + 3 HzCI + 3Hz t
+ H3B03 + RzHN
Hypophosphite solutions can be determined similarly, with the following exceptions: The sample to be titrated should contain approximately 3 grams/liter of hypophosphite ions or approximately 5 grams/liter sodium of hypophosphite. The palladium solution contains no ferrous chloride and is heated to 80-90" C before the titration. The precision of the DMAB determination as described above with ferrous chloride in the palladium chloride solution corresponds to a standard deviation u = 0.03 ml of 0.75x. The standard deviation was u = 0.155 ml or 3.9% without the ferrous chloride. The precision of the hypophosphite determination without ferrous chloride was CT = 0.05 ml or 1.0%. The addition of ferrous chloride lowered the precision in the hypophosphite determination. LIMITATIONS OF THE METHOD
(1)
Surplus DMAB frees hydrogen on catalytic surfaces such as the palladium electrode: RaHN
.03
(%PdC12 1
DIMAB-sol (ml)
RzHNeBH,
.o 2
(2)
The potential at the palladium electrode changes abruptly as hydrogen is formed, thus yielding a sharp end point for the titration. The precision of the DMAB determination can be improved if ferrous ions-e.g., 1 gram/liter FeClz 4H20are added to the palladium solution. In hypophosphite determinations, the palladium solution must be heated to 8&90" C, because the reduction of palladium ions by hypophosphite is relatively slow at room temperature. Ferrous ions do not seem to improve the hypophosphite determination. METHOD
T o obtain a palladium-ferrous solution, dissolve 1 gram of purified palladium chloride (PdClz), 16 grams of ferrous chloride (FeCh . 4 HzO), and 5 ml of concentrated HCl in 5 liters of water. Place 250 ml of this solution into a 400-ml beaker containing a palladium electrode (palladium sheet, 0.25 X 3 inches) and a calomel reference electrode. Connect the electrodes to a pH meter, and observe the meter reading. If the meter has a zero adjustment, set the meter to a reading of 0 mV. While stirring vigorously, slowly (at a rate of approximately 1 ml/minute:i titrate the solution to be tested (which should contain 2.fi grams/liter DMAB or less) from a microburet into the beaker. Note the increase in potential on the pH meter. An increase of approximately 350 mV should be taken as the end point of the titration. A typical titration curve is shown in Figure 1 . If the palladium electrode is to be used again immediately, heat the palladium electrode to 150-200" C in air for a short time, in order to obtain a steady potential reading before the titration. This treatment is not necessary if the palladium electrode has been left at room temperature for more than 1 hour.
The pH of the palladium chloride solution is approximately 2.8 in the above procedure. Increasing the pH above 3-3.5
by adding sodium hydroxide leads to erratic results. Lowering the pH to pH 1 by adding more hydrochloric acid leads to slightly higher values in the titration. A pH of approximately 2.8 is preferred. Changing the palladium content of the palladium solution affects the results as shown in Figure 2. A palladium solution of constant concentration should be used for consistent results. The addition rate of DMAB or hypophosphite solutions in the titrations can be varied from as low as 0.2 ml/minute up to 2 ml/minute without noticeably affecting the results. An addition rate of approximately 1 ml/minute or 20 drops/ minute (approximately 2.5 mg of DMAB or 5 mg of sodium hypophosphite per minute) is preferred. Phosphite ions and any other ions oxidized by palladium ions (such as sulfite, hydrazine, etc.) interfere with the determination. Solutions to be tested should be brought to a pH of approximately 8 or less before the analysis. A pH above 8 would increase the pH of the palladium solution to above 3.5 and lead to erratic results. This is particularly important if the solution to be tested is ammoniacal; palladium ions form a diammine palladium precipitate at pH above approximately 4. Very large amounts of nickel or cobalt salts (1 mole/liter or more) in a sample containing 2.5 grams/liter of DMAB or 5 grams/liter of NaHZPOz.H 2 0(approximately 50 mmoles/liter) seem to lower the precision of the method. DISCUSSION
Palladium ions were chosen as the reductant because palladium is positioned high enough in the electromotive force series of elements (normal potential of the reaction Pd+2 e + PdO is $0.820 V) to permit reduction by DMAB and
+
VOL. 39,
NO. 8,
JULY 1967
1015
hypophosphite without the simultaneous formation of hydrogen. In addition, palladium seems to react faster than other noble metals (such as gold or platinum) and in solution is more stable than silver. The price of a palladium solution used for one determination is approximately 9 cents. During the titration, two stable potential levels are observed at the palladium electrode. One level is observed as long as palladium ions are in the solution and no hydrogen gas is formed; the other level occurs when all palladium ions are reduced and hydrogen gas forms. The levels are approximately 750 mV apart. Because, during the titration, formation of hydrogen before the end point means a loss of reducing agent, a temporary surplus of reducing agent must be avoided. Adding the reducing agent at a correspondingly slow rate and stirring vigorously during titration help to avoid a temporary surplus of reducing agent. The precision of the DMAB analysis is improved if ferrous ions are present in the solution. When there is a local temporary surplus of reducing agent during the titration, it appears possible that first all locally available palladium ions are reduced, then both iron metal and hydrogen gas are formed simultaneously. The iron metal is deposited on top of the palladium metal. Iron is a much poorer catalyst for the reduction than palladium. Therefore, both further iron deposition and hydrogen formation are slowed down. This would decrease any loss of reducing agent during the titration. On the other hand, hypophosphite reacts more slowly than DMAB. No iron is needed in this case; iron may even delay the reduction too much, leading to a less well defined end point of the titration.
4.5 r
-
7
4.0
m
s
n
3.5 1
-
I
150
I
I
250
350
(ml)
Figure 3. Volume of PdC12 solution containing 50 mg
of PdClz
The concentration of the palladium chloride solution affects the determination as shown in Figure 2. A possible explanation is that a certain surplus of reducing agent must be added before hydrogen forms-for instance, oxygen dissolved in the palladium solution may have to be reduced first. If this is so, the surplus of reducing agent should be proportional to the volume in which a given amount of palladium chloride is contained. Plotting the volume containing 50 mg of palladium chloride against the amount of reducing agent used (see Figure 3) shows that the amount of reducing agent, indeed, is almost a linear function of the volume of the palladium solution. RECEIVED for review January 26, 1967. Accepted April 21, 1967.
Quantitative Determination of Benzenehexachlorides by Gas Chromatography Abram Davis and Helmut M. Joseph
Hooker Chemical Corp., Niagara Falls, N . Y. 14302 NUMEROUS GAS CHROMATOGRAPHIC methods for the determination of y-benzenehexachloride have been described in the literature but very few concern the determination of the other benzenehexachloride (BHC) isomers. Previously reported methods (1-3) use thermal conductivity detectors and cannot be used for trace analysis as such. Most benzenehexachlorides and heptachlorocyclohexanes have unique absorption bands in the infrared region. An infrared analysis procedure ( 4 ) has been devised and used routinely in our laboratory for many years. However, when all isomers have to be determined, infrared analysis becomes time-consuming . A gas chromatographic analysis using an electron capture detector appears very desirable from the point of view of sensitivity. However, its response is not linear over a wide
(1) W. Esselborn and K. G. Krebs, Pharm. Zrg., Ver. ApothekerZtg., 107,464 (1962); C.A., 10,2059 (1963). (2) H . Furst, H. Kohler, and J. Lauckner, Chem. Techrz., 16, 105 (1964). ( 3 ) C. 1. Guillemin, Aiial. Chim. Acta, 27, 213 (1962). (4) L. E. Tufts and R. H. Kimball, Hooker Chemical Corp., Niagara Falls, N. Y . , private communication, May 15, 1949.
1016
ANALYTICAL CHEMISTRY
range of concentrations and, therefore, it is not suitable for the determination of all the components of a typical BHC sample during one run. This does not detract from an electron capture detector used in residue analysis of y B H C (lindane) or any other BHC isomer. The method presented combines sensitivity, speed, and applicability to widely varying mixtures of a-,p-, y-, 6-, E-, and Q-BHC and y- and eheptachlorocyclohexane; traces of chlorobenzene impurities can also be determined. Adherence to experimental conditions reported below is important to achieve reproducible and complete elution of the above components. We observed sample decomposition when operating at higher column temperatures or with different column support types. The 6-heptachlorocyclohexane cannot be determined in the described gas chromatographic system because of its extreme thermal instability. The V-BHC and the a-hepta yield only small elution areas compared to the amounts of injected sample. The latter component is only incompletely resolved from p- and E-BHCand will not be observed in small concentrations. These shortcomings do not impair the determination method as they concern only very minor components in benzenehexachloride samples,
