ANALYTICAL CHEMISTRY
182
Effect of pH. The reaction of the medium has a great influence on the precipitation of the mercuriamidines. At p H 4 precipitation is partial and a t p H 3.5 it may not take place a t all. When the pH is 7 or above, mercuric oxide starts to separate and the results are irregularly increased. Precipitation takes place quantitatively when the pH is kept between 5.2 and 7.0. It is most accurate a t pH 6.2. The precision and accuracy of the method are shown in Table 11. Properties of the Mercuriamidines. The mercuriamidines are white, bulky, very gelatinous, and insoluble in water and in alcohol. They are dissolved by dilute acids. Dried over sulfuric acid and under reduced pressure (8 to 10 mm.) they occur as white fine amorphous powders. The precipitates dried over sulfuric acid under reduced prcssure,and analyzed for their mercury, nitrogen, and acetate content gave results that agree n-ith the following suggested general structure : CHsCOOHg--S N-HgOOCCHs
H2N CONCLUSION
These two methods or a t least one of them can be used for the
determination of the purity and of the molecular weights of the amidines. ACKNOWLEDGMENT
The author wishes t o thank Raymond Delaby, general secretary of the International Union of Chemistry and profeesor of Pharmacie Chimique in the Faculty of Pharmacy, Paris, under whose general direction this work was conducted. The author wishes to acknowledge also Robert Damiens, Soci6t6 des Usines Chimiques RhGne-Poulenc, and Rudolph J. Pauly for providing samples an? for their many helpful suggestions. LITERATURE CITED
(1) Bougault, J., J . pltarm. chim., 16,33 (1917). (2) Bougault, J., and Robin, P., Compt. rend., 17.1, 38 (1920) (3) Ibid.. 172. 452 11921). (4) Delaby, Raymond, Biologie mbdicale, 35, 3 (194G). (5) Drake, N., et aL., J . Am. Chem. Soc., 70, 168 (1948). (6) Kolthoff, I., “Volumetric Analysis,” Vol. 11, pp. 99, 205, S e w York, John Wiley & Sons, 1929. (7) Robin, P., Ann. chim.,16, 122 (1920). (8) Robin, P., Compt. rend., 173, 1085 (1921). (9)Ibid., 177, 1304 (1923). RECEIVED October 25, 1950. Abstracted from a thesis submitted by Fouad H. Stephan to the Faculty of Pharmacy, Vniversity of Paris, in partial fulfillment of the requirements for the doctorate degree of the Univeraity of paris.
Application of Ion Exchange to Determination of Boron J. ROBERT MARTIN’ AND JOHN R. HAYES The Pennsylvania State College, State College, Pa.
Conventional methods for the determination of boron require a rather lengthy separation from interfering elements. In an effort to improve this, an attempt was made to substitute ion exchange for the usual distillation. The method developed employs a cation exchanger to separate boron from large amounts of cation-forming elements, The measurement of the boron content, accomplished
C
OLORIMETRIC, gravimetric, and volumetric methods of measuring boron concentration have been propored, but their application to many materials received for analysis requires a difficult and frequently lengthy separation of the boron from interfering elements. The most general method of separation involves distillation of boric acid as its methyl ester ( 2 ) . When this separation is followed by a titrimetric determination, a blank of 0.3 mg. or less of boron is introduced, which is due in part to the extraction of boron from glassware and in part to a titration error. Incomplete volatilization is obtained from large amounts of aluminum salts and a double distillation is recommended in the presence of both iron and silicon. The time required for this procedure is a further disadvantage. Separation of boron from some elements which form weak bases has been attempted by precipitation of the hydrous oxides using various alkali or alkaline earth hydroxides, carbonates, or oxides. I n all cases boronis adsorbed by the hydrous oxide precipitate and a reprecipitation is necessary except for very dilute solutions. Schafer and Sieverts (6) precipitated zinc, nickel, iron, and aluminum with 8-quinolinol without adsorption of boron. Electrolysis of boron solutions using a mercury cath1
Present address, E. I. d a Pont de Semonrs B Co., Wilmington. Del.
by a modification of the familiar titrimetric procedure in the presence of invert sugar, permits the presence of moderate amounts of such anions as silicate, arsenate, molybdate, and phosphate. The method gives accuracy equal to or better than conventional procedures in the analysis of a variety of types of samples with boron contents ranging from 0.03 to 20.0%. An appreciable saving in time is possible.
ode provides an excellent Yepwation but does not remove alunlinum, beryllium, titanium, zirconium, vanadium, arsenic, or phosphorus (8). Other less attractive methods, including an ether extraction, have been proposed. Boric acid can be separated from interfering cations without many of the limitations and disadvantages of the above methods by employing the ion exchange technique for the removal of interfering ions. The proposed method involves passage of the neutral or slightly acid solution through a strong acid type of cation exchanger. After carbon dioxide has been removed from the effluent by boiling, the pH is adjusted to 6.90, invert sugar is added, and the boron-containing complex acid is titrated back to pH 6.90 with standard alkali. REAGElVTS
Standard boric acid solutions were prepared by dissolving weighed amounts of anhydrous boric oxide in water to give concentrations of 0.2 and 1.0 nig. of boron per milliliter, respectively. The boric oxide was prepared by fusion of reagent . . -~ grade boric acid in a platinum dish. Carbonate-free 0.1 N and 0.02 .V sodium hydroxide solutions were prepared from saturated sodium hydroxide solution ( 3 ) and were standardized by potentiometric titration of aliquots of the standard boric acid solutions using the procedure described for the
V O L U M E 24, NO. 1, J A N U A R Y 1 9 5 2 titration of samples. An unstandardized 2 .I' solution was also prepared for use in preliminary neutralization. Invert sugar solution was prepared by the method of Mylius ( 5 ) . Commercial sucrose (1 kg.) was dissolved in 650 ml. of prevlously boiled hot water and 8 ml. of 1 N hydrochloric acid were added. The solution was heated at 80" to 90"C. for 2 hours and then adjusted to pH 6.90 (measured electrometricallv using a pH meter) with alkali just before use. The folloming reagent grade salts mere used to provide ions for testing the efficacy of the ion exchange separations: AICI,.GH,O, BeS04.H20; MgO; ZnO; CdSOd.4H?O; Co(SOs)?.BH,O; CuSO4.5H20, HgSO,; NiS04.6H20;Th(SOa)a.4H,0;SnCI4.5H20;Ticla; UO2S04.3H,O; Zr(SO&.4H20; K2Cr207; KXInOd; NalVOa; CrCI3.6H2O; ~fnci~.4H20. Fea(S04)S.xH20and Sa20.zSi02; were used after analyses to determine the 11'1 n and silicon contents, respectively. APPARATUS
Both Dowex 50 and Amberlite 111-100 A.G. were found to provide a satisfactory Reparation of horic acid from interfering cations. Because Amberlite IR-100 vas available in the analytical grade, it, was used in all cases except when high capacity was required-viz., for samples weighing more than 1 gram. The resin columns were of the Jones reductor type, consisting of a tube terminated at the top with a reservoir of 75-ml. capacity and at the bottom with a stopcock and outlet, tube. Two sizes were employed: 1.9 cm. (inside diameter) X 30 cm., 80-cc. capacity, and 2.6 cm. (inside diameter) X 33 cm., 175-cc. capacity. The resin columns were supported by 4-cm. plugs of tightly compressed glass TWOI. I n order to effect economy of time and reagents it was found advantageous to replace the used resin with previously regenerated resin, so that a new sample could be analyzed immediat,ely. The used resin as allowed to accumulate and was then transferred to a large column and regenerated with 10% hydrochloric acid until the effluent gave no test for the ions that had been adsortled. After thorough washing the resin was stored for furt8heruse. A Beckman .\Tonel H 2 line-operated pH meter mas used in conjunct,ion wit,h a glass electrode and saturated calomel cell for the potentiometric titrations and for pH adjustment. The titration \vas carried out in a carbon dioxide-free atmosphere by using a Berzelius beaker (spoutless) as the titration vessel, and providing if with a stopper bored to receive the glass and calomel electrodes, the motor-driven stirrer, and an inlet tube for an air stre:tm T h e air s h a m had been passed through successive bot,tles of concicntrated sulfuric acid, soda lime (two bottles), and water. PROCEDURE
Saniples of suitable size are dissolved in water or in a minimum amount of hydrochloric acid. When the application of heat is necessary to effect solution, the flask is connected to a reflux condenser to prevent loss of boron by volatilization. The solution is diluted to 50 ml. and is then passed through the resin bed. The resin is washed with 200 ml. of wat'er in small portions. The rate of flow is adjusted so that. the time for passage of the sample and washings is approximately 15 minutes. The effluent is n p r l y neutralized t'o methyl orange with 2 S sodium hydroxide and is boiled under a reflux condenser for 5 minutes. S o volatilization of boron was apparent when the solutions were boiled gently in covered beakers, but a reflux condenser is a necessary precaution in case the time o r rate of boiling is not carefully controlled. After c oling to room temperature in a water bat,h, the solution is adjusted t'o pH 6.90 with 0.02 'V alkdi. Sufficient invert sugar is then added to make the resulting ncentration 0.6 &I and the sample is tit'rated immediately to pII 6.90 with standard sodium hydroxide solution. The volume of standard alkali required to restoie the solution t,o a pH of 6.90 after the addition of invert sugar is multiplied by the boron titer of the alkali to calculate the amount of boron present. is(
DISCU SSIOS
Titrimetric Determination. After tioron has been isolated free of interfering substances by an ion exchange separation, the titrimetric and colorimetric procedures appear most attractive for measuring boron concentration. In general, colorimetric methods are more sensitive and selective and could undoubtedly be applied advantageously in many instances, especially when extremely small amounts of boron are to be measured. The titrimetric procedure was chosen for this work, however, berause it is accurat.e and relatively rapid, does not require a preliminary
183
evaporation of the effluate, and does not involve handling of hazardous concentrated sulfuric acid solutions. Boric acid is very weakly ionized, but the addition of certain polyhydric alcohols or large concentrations of calcium salts promotes the formation of stronger complex acids which can be titrated. As the pH of a 0.1 N boric acid solution is approximately 5, the interference of strong acids can be eliminated by preliminary neutralization to the methyl orange end point. This classical titration procedure has been modified by Foote ( 1 ) and Schafer and Sieverts (7), who titrated the boric acid solution to the same pH before and after the addition of a polyalcohol, thus eliminating the interference of weak acids and bases which were present in such small amounts that their buffering action was negligible. The principle involved is that with no polyalcohol present boric acid is so weakly ionized that it is only slightly neutralized a t p H values up to 7.6, but in the presence of sufficient polyalcohol it is completely neutralized at or below this pH. This modified type of titration was chosen for the determination of the boron concentration following an ion exchange separation because it did not necessitate the removal of small quantities of weak acids which are formed by elements such as sulfur, phosphorus, silicon, arsenic, and molybdenum. Small amounts of these anion-forming elements are present in many boron-containing metallurgical materials.and are not separated from boron by means of a cation exchanger. The proposed titration procedure also eliminated inaccuracies due to variation in the small resin blank-i.e., a slight, amount of residual acid washed out of the ion exchange column after the resin had been thoroughly rinsed. The adopted procedure represents a compromise between the modifications that had been proposed. The apparent ionization constant of the complex boric acid increases with polyalcohol concentration and the pH a t the titration equivalence point is thus dependent on this factor. I t is desirable t o have the polyalcohol concentration as high as possible, so that, the equivalence point will come a t a relatively low pH and the volume of standard alkali required for neutralization will then more nearly approach thc theoretical value, as less boric arid is neutralized before addit,ion of the polyalcohol. The polyalcohol concentration is effectively limited, however, by the volume of solution which accumulates during washing of the ion exchange column unless a time-consuming evaporation is included. Investigation of the potentiometric titration curves for boric acid in the presence of varying conceritrat,ions of invert sugar showed that 0.6 31 \vas a suitable concentration for invert sugar and in this medium the neutralizat,ion of boric acid was complete a t a pH of 6.90. The slope of the curve a t the equivalence point was 1.4 pH units per ml. and the volume of alkali required deviated from the theoretical value by less than 3%. This deviation \vas taken into account by standardizing the sodium hydroxide solutions against pure boric acid, using the procedure described for titration of the samples. The titration curve for boric acid without polyalcohol has a greater slope but is sufficiently parallel to the curve for boric acid in the presence of invert sugar to minimize errors due to possible pH meter drift. Effect of Electrolytes. The apparent ionization constant of boric acid in aqueous solutions increases with boric acid concentration and with the concentration of certain ions such as lithium, barium, and calcium ( 4 ) . The effect of electrolyte concentration variation on the accuracy of the titrimetric determination was therefore studied by titrating boric acid solutions containing known amounts of sodium chloride ranging up to a concentration of 0.25 LV. The amounts of 0.02 S Podium hydroxide required for neutralization to a pH of 6.90 varied by 0.12 ml., thus indicating that the effect of electrolyte concentration variations is rirgligible under ordinary conditions. Blank. Inherent in the proposed method was the possibility of a blank arising from residual acid in the resin and in the polyalcohol reagent. The modified titration procedure eliminated
ANALYTICAL CHEMISTRY
181 the blank arising from the first source, and any blank arising from the polyalcohol reagent was removed by preliminary neutralization of the invert sugar solution to the equivalence point pH. Thus the blank was completely eliminated except for any boron which could conceivably be introduced from glassware or as a contaminant in reagents. The blank was experimentally shown to be nil. Resin Rinse Requirements. The amount of water required to remove all boric acid from the resin bed was ascertained by passing boric acid solutions through the column, washing with varying amounts of water, and titrating the total effluent. Quantitative recovery of 3.08 nig. of boron in 50 ml. of solution was effected by washing the resin with 150 ml. of water in small portions (ca. 25 ml. each). I n order to provide a safety factor, 200 ml. were chosen as the volume of rinse solution. Polyalcohol. From the many polyalcohols which have been suggested for increasing the apparent ionization of boric acid, invert sugar solution was chosen because of its relative effectiveness, availability, low cost, and freedom from a reagent blank correction.
T a b l e I.
Effect of Various Ions on D e t e r m i n a t i o n of Boron
Sample Compositiona Fe Al, Be, hrg, Zn Cd, Co, Cu, Hg. Xi T h Sn T I U Zr c&-'-, h i n o r - , V O ~ -
__
P
As
Mo P AS
Mo
Si Si
Mg. 205 3leach 3leach 3leach 62each 0.9 0.3 0.3 1.7 0.5 0.5
25 50
200
Si a
KO.of Drtns. 4
5 6 G
Ar. Error, ,1Ig
0.00 0.00
-0.02 0.00
muin
Error, Alg.
+0.03 10.01 -0.05 +0.03 -0.02
Relative Standard Deviation, 70 0.6 0.3 0 8
+0.04
0.7 0.3 0.4
+0.07
+0.09
0.4
+0.03
+0.06 +0.04 +0.04
1.5 1.3 1.3
5
-0.01 f0.03
5
6 2 2
4
Rlaxi-
+0.03
+0.03
Each sample contained 3.08 mg. of boron.
Separation of Boric Acid from Cations. The efficacy of the ion exchange method for isolating boron from various interfering cations was tested by analyzing solutions of known composition. Considering the studies that had already been made of exchange of inorganic cations, it was considered sufficient to pick representative groups of ions for investigation without attempting to investigate all possibilities. The separation of boron from various ions which are known to interfere in the titrimetric determination and those which are most likely to be found in metallurgical materials was accomplished as indicated by the data of Table I. In general, the solutions were prepared by dissolving boric acid and the designated salts in 50 nil. of water containing the minimum amount of hydrochloric acid necessary to prevent hydrolysis. The described procedure proved capable of separating boric acid from a t least 60 times its weight of iron and 10 times its weight of aluminum, beryllium, magnesium, zinc, cadmium, cobalt, copper(II), mercury(II), nickel, thorium, tin( IV), titanium, uranium, and zirconium. These values were experimentally verified, but do not necessarily represent the upper limit of concentration that may be tolerated. In fact, it is probable that considerably higher amounts of these cations can be quantitatively separated from boron if sufficient resin is used. Interference of Anions. XIany boron-containing materials consist largely of elements which form cations when the samples are dissolved, yet small amounts of anion-forming elements are often present. Anions containing elements that can be reduced to a valence state in which they exist as cations may then be removed by the cation exchanger. Boric acid was separated from chromate, vanadate, and permanganate by adding 3% hydrogen peroxide to the hot 0.7 N
hydrochloric acid solution of the sample until the samples became yellow. After the samples had boiled for 5 minutes in a flask fitted with a reflux condenser, stannous chloride was added until PO further color change took place and then 5 drops were added in excess. The cooled solutions were analyzed by the usual procedure, with the results shown in Table I. Other anions such as arsenate, phosphate, molybdate, and silicate cannot be separated from boron in the foregoing manner. Small amounts of these substances can be tolerated when the described modified titration procedure is used. The effect of these anions was to reduce the slope of the titration curves because of their buffering action and thus to decrease the accuracy of the determination when appreciable amounts were present. The data of Table I indicate that boron may be determined in solutions containing less than 0.9 mg. of phosphorus, 0.3 mg. of arsenic, and 0.3 mg. of molybdenum with a relative error of approximately 1%. As arsenic and phosphorus in steel seldom exceed 0.03 and O.l%, respectively, a 1-gram sample could be used for the determination of boron without requiring the removal of these elements. The effect of 200 mg. of silicon was approximately in the same order of magnitude. DETERMINATION IN COMMERCIAL MATERIALS
As no satisfactory standard boron-containing materials were immediately available, synthetic samples were made to simulate typical sample compositions. The synthetic steel samples were prepared by dissolving appropriate sized samples of National Bureau of Standards Bessemer steel 8e in 15 to 20 ml. of 1 to 1 hydrochloric acid after adding measured volumes of standard boric acid solution. A 1.5-gram sample was used for mixtures containing less than 0.25% boron and a 1.0-gram sample was used when the boron content exceeded that amount. The sample was dissolved by heating in a flask that had been fitted with a reflux condenser. After the sample had dissolved, 1 ml. of 30% hydrogen peroxide was added to oxidize carbides and the solution was boiled 15 minutes. The samples were cooled, diluted to 75 ml., and analyzed uEing the described procedure. The other samples were prepared by dissolving suitable salts in water containing the minimum amount of hydrochloric acid to prevent hydrolysis, when necessary. The procedure was applied to the analysis of synthetic standards simulating steel, ferroboron, a casting alloy, and nickel plating solutions with the boron content ranging from 0.025 to 20% (Table 11). The average relative error for determination of amounts of boron exceeding 2 mg. was -0.03% and the maximum
Table 11. Analysis of Commercial Materials Sv. Sample Composition
No. of
Absolute Error,
Maximum Relatire Absolute Standard Error, Devia70 tion, %
Detns.
%
Steel N.B.S. steel 8e + 0 . 0 2 5 7 B N.B.S. steel 8e +0.4lOd B
4 4
+0.001
Ferroboron Fe88%,.1147,,Si2.5%,B4.93% Fe 73 6%. A1 3.170, SI 3.1%, B 20.26c7,
4 4
4
-0.01
-0.01
0.G
4
+0.01
+0.04
0.3
Casting alloy A. Cr 20%, Co 40%, Ni 39%, B 1.03y0 B.
Cr 2076, Co GO'%, i i i 9.87,, hIn1.0%, Sil%,B7.937,
+0.003 +0.003
8.3
-0.01
-0.02
0.2
+0.06
+O
0.2
+0.002
11
0.5
1as
V O L U M E 2 4 , NO. 1, J A N U A R Y 1 9 5 2 relative error was 1.0%. The standard deviation of a single value was 3 parts per thousand. Advantages. The proposed method of analysis possesses the follou ing advantages over currently available procedures applicable to the same concentration range: freedom from coprecipitation or occlusion of boric acid; freedom from a blank correction; no requirement of boron-free glassware, because only acid solutions are boiled and then only for a fel\, minutes; minimuin opportunity for loss of boron by volatilization because lengthy evaporation is not necessary; simple equipment; and increased rapiditv and good accuracy.
LITERATURE CITED (1) Foote, F. J., IND.ESG. CHEM.,* 4 ~ 4 ED., ~ .4 , 3 9 (1932). (2) Furman, N. H., “Scott’s Standard Methods of Chemical iln:ilYsis,” p. 179, New York, D. Van Yostrand Co., 1939. (3) Kolthoff, I. M., and Stenger, V. A , , “Volumetric Analysis.’ p. 70, New York, Interscience Publishers, 1947. (4) I b i d . , p,114. (5) Mylius, W., Chenl.-Ztg , 57, 195 (19333. (6) Schafer, H., and Sleverts, A.9 2’. anal. Chenl., 121,161 (1941). (7) Ibid., 121, 170 (1941). 18, 607 (1926). (8) TschischeTVskl, s , ,~ ~ d cjLern,, .
RECEIVED April 21, 1951.
Automatically Increasing Solvent Polarity in Chromatography KENNETH 0. DONALDSON, VICTOR J. TULANE, AND LIWRENCE 11. 3IARSfIALL Howard L’niaersity, Washington, D. C . The determination of organic acids separated on silica gel columns is limited in sensitivity by the reduced resolution experienced for the more watersoluble acids. Although the literature contains procedures for improving resolution by manipulation of the solvent, the changes in polarity which accomplish this end arc abrupt and are obtained by frequent manual operations. The widespread use of
I
X SEPARATIOSS employing partition chromatography for organic acids, the resolving power of silica gel systems decrease8 with increasing water solubility of the solutes ( 2 , 4, 5 ) . Isherwood ( 2 ) proposed the use of several solvents, but was aware that no one, or even two, ITould suitably resolve an organic acid mixture of the complexity used in his study. I n a procedure for the determination of furmaric acid ( 4 ) , this was again recognized
5 4.
1
acetic + fumaric
P 7)
L x
E32
-0
si
620
*
9
oconitic
oxalic
10 20 30 4050 60 70 80 9 0 fraction rJrber
f r a c t i o n nJrnber
Figure 1. Chromatograms Typical curve (left) resulting when chromatographic solvent progressively increases in n-butyl alcohol content. Typical results ( r i g h t ) , when a eolvent of fixed ooinposition (10% n-butyl alcohol in chloroform, v / v ) is employed. T h e difference between t h e character of t h e curves as well BS t h e resolving power of the two systems with respect t o fumaric a n d acetic acids contrasts here t h e two types of chromatographic influent. Fraction number X 2 equals ml. of effluent.
apd effluent fractions on the chromatogram were collected by geometrically increasing volumes. I n effect, such a procedure gradually increased the volume of polar mobile phase in each effluent fraction. Marvel and Rands ( 5 ) successfully employed a series
mechanical apparatus for the collection of chromatographic fractions necessitates a procedure that requires no attention once the separation has begun. A simple device permits the delivery of a solvent automatically but gradually increasing in polarity. Recovery studies indicate that, with the exception of oxalacetic acid the method is suitable for the quantitative determination of organic acids.
of solvents progressively increasing in composition with respect t o n-butyl alcohol. Their solvents were added manually to the column in order of increasing n-butyl alcohol concentration. This procedure, like the first, required constant attention and mas, therefore, more noticeably disadvantageous when mechanical apparatus was employed for the collection of effluent fractions. The above procedures reduced the number of effluent fractions required to collect a given acid by manually but not gradually increasing the polarity of the mobile phase. When the concentration of each fraction n as plotted against its number, the curves on the chromatogram ivere sharper than those obtained Kith a solvent of fixed polarity. The present report, combining the principles underlying the determination of fumaric acid and those of Marvel and Rands, describes a procedure for gradually and automatically increasing the alcohol concentration and thereby the polarity of the solvent entering the column. With this method no false chromatographic peaks or shoulders on the peaks occur, as they sometimes do when the solvent polarity is increased in discrete steps. The applicability of this approach for the separation of certain acids of physiological interest is examined. A typical chromatogram employing the new procedure appears in Figure 1 (left). APPARATUS
A device was arranged so that the stem of the solvent reservoir ( A , Figure 2 ) extended to xithin 3 em. of the bottom of the mixing vessel, B . When 4 was filled Kith 50% (v./v.)n-butyl alcohol-chloroform and B with 175 ml of pure chloroform, the solvent flouing from column C created a fall in pressure within the system which was relieved by solvent delivered by the side arm. The 50y0 mixture from A flowed into B and because of the difference between the density of n-butyl alcohol (0.804 gram per ml.) and chloroform (1.497 gram per ml.) mixing was obtained, as demonstrated below, by the time the ~olvententered C. As a result, the solution entering the column gradual11 increased with respect to alcohol concentration. n-Amyl alcohol (density, 0.810 gram per ml.), where indicated be!ow, was substituted for n-butyl alcohol.