840
ANALYTICAL CHEMISTRY
(expressed in units) on the donor side or the increase in quantity on the recipient side, and d is the mean concentration difference (expressed in units per milliliter) during the diffusion period. The value of d is obtained by plotting the observed concentrations on the donor and recipient sides on graph paper in relation to time. Smooth curves are then drawn to connect the values on each side, and the mean concentration difference is calculated from the area limited by the curves. Correction for Tissue Respiration. I n diffusion studies on living membranes some oxygen will be used by the respiration of the tissue. If the experiment is carried out by withdrawal of samples in immediate succession from the two compartments, no correction for the tissue respiration is necessary if the volumes of fluid on the two sides are the same, the supposition being made that an equal amount of oxygen is consumed on both sides of the membrane. I n experiments in which a time interval elapses between the withdrawal of samples from the two compartments, the value of p in Formula 2 must be corrected for the amount of oxygen consumed by the tissue. If it is aPsumed that the oxygen used for the tissue respiration on the average has diffused through half of the membrane, the correction to be applied is equal to one half of the difference between the decrease in quantity of oxygen on the donor side and the increase in oxygen on the recipient side. If a solution aerated with pure carbon dioxide is used in the donor compartment of the apparatus, the respiratory carbon dioxide can be disregarded.
DISCUSSION
The main advantages of the present procedure as compared with previous methods are the following: It permits determination of the diffusion coefficients of several solutes in the same experiment; it provides for the withdrawal of samples for analysis a t any time during an experiment,, thus making it possible to carry out successive diffusion periods within a single experiment; it permits an accounting for the diffusing quantities of a compound by comparison of the amounts disappearing on one side of the membrane with those appearing on the other side; and it involves the use of easily movable syringe plungers, thus ensuring against differences in total pressure on the two sides of the membrane. LITERATURE CITED
(1) Hill, A. V., Proc. Roy. Soc. London, 104B,39 (1928-29). (2) Kirk, J. E., and Hansen, P. F., J . Bid. Chena., 199, 675 (1952). (3) Krogh, A,, J . Physiol., 52, 391 (1918-19). (4) Pletscher, h.,Staub, H., Huneinger, W., and Hess, W., H e h . Physiol. et Pharmacal. Acta, 8 , 306 (1950). (5) Wright, C. I., J. Gen. Physiol., 17, 657 (1933-34). RECEIVED for review September 7, 1954. Accepted December 13, 1954, Investigation supported in part b y a research grant (PHS-891) froin the National Heart Institute of the National Institutes of Health, Public Health Service. A demonstration of the technique was given a t the 43rd meeting of the American Society of Biological Chemists, April 14 t o 18, 1952 S e w York. N. Y.
Effect of Particle Size on the Characteristics of Silicic Acid Chromatographic Adsorbent EARL W. MALMBERG McPherson Chemical Laboratory, The O h i o State University, Columbus, O h i o
The effect of particle size on the efficacy of silicic acid in affording chromatographic separations has been studied with the purpose of obtaining the best adsorbent from commercially available materials. Procedures of sieving and ball-milling have been used to provide samples whose effectiveness was tested by their ability to separate the 2,4-dinitrophenylhydrazonesof acetaldehyde and formaldehyde and the dinitrophenylosazones of glyoxal and methyl glyoxal. A procedure for obtaining a satisfactory adsorbent from a commercially available silicic acid is described.
T
HE influence of physical characteristics on the behavior of
chromatographic adsorbents is geiierally ignored until difficulties arise which necessitate bringing certain variables under control. T h e influence of particle size on the chromatographic behavior of silicic acid is t h e subject of t h e present investigation. For some years commercially available silicic acid has been used as a chromatographic adsorbent and has given uniform characteristics except for certain variations in adsorptive strength which are to be expected. Recently t h e samples of silicic acid from t h e usual source have been completely different in particle size and apparently in their adsorption characteristics in general. Studies of particle size distribution, grinding, and chromatographic separations were made on various samples of silicic acid. A procedure for preparation of a very satisfactory adsorbent from a commercially available Mallinckrodt silicic acid is described as devised from these investigations.
GENERAL EXPERIMENTAL PROCEDURES
Throughout this work two rather sensitive chromatographic problems, the separation of the 2,4-dinitrophenylhydrazones of formaldehyde and acetaldehyde and of the 2,+dinitrophenylosazones of glyoxal and methylglyosal, were used as criteria of the chromatographic power of a n adsorbent. These tests are referred to below simply as the formaldehyde-acetaldehyde and glyoxal-methylglyoxal separations, respectively. T h e procedure (1) for the formaldehyde-acetaldehyde separation requires a prewash of Vlj0ml. ( 2 ) of 10% of acetone in ligroin (Skellysolve B), 0.5 mg. each of hydrazones in 5 ml. of 1 : 4 chloroform-ligroin (on a column of 14-mm. inside diameter), and development with 10% of ether in ligroin. Some of the best adsorbents did not, require the prewash, but even with the best the separations were improved. The osazones, 0.1 mg. of each in 1 to 14 nitrobenzene-benzene, were placed on a column wet with benzene and developed viith a solution of 5% of ether in 1 to 1 benzene-ligroin. This system has a complicating feature: Some adsorbent,s n 3 l not afford a satisfactory separation under any conditions, while other samples ail1 give a good separation a t slow flow rates but not a t fast. From the profiles of the zones, this rat,e effect seems to be a result of a slow rate of desorption for some of the adsorption sites. The ball-milling of the adsorbent was done in a Paul 0. Abbe mill of 6-gallon size, 55 r.p.m. The mill xvas charged wit8h 0.5 kg. of silicic acid and 1.5 kg. of 1-inch porcelain balls. Combinations of larger amounts of material and balls were found to be ineffective. I n all cases the adsorbent was heated 4 hours a t 200" C. before testing. Each column was packed with the tube in position on a suction flask, the adsorbent being poured in with a stopcock on a safety bott.le open to the atmosphere; after initial set'tling was completed, the stopcock was closed, and as the vacuum was established, the column further contracted. T o prevent surface spreading, the upper surface of the adsorbent must be pressed
V O L ’ U M E 27, NO. 5, M A Y 1 9 5 5
b
841
hyde-2,4-dinitrophenylhydrazonefrom one of t h e more satisfactory samples gave only 82% compared with 98 to 100% which is obtained from good adsorbents, further investigation was made on silicic acid from other sources.
20
E/ 5 W‘
SAMPLES OF SILICIC ACID ESPECIALLY PREPARED
Investigations which were made of t h e ordinary Mallinckrodt reagent grade silicic acid showed t h a t t h e original material was not suitable for chromatographic purposes, and with ball-milling, etc., t h e properties could be improved somewhat but a satisfactory formaldehyde-acetaldehyde separation could not be achieved. A special, finely divided silicic acid from t h e Mallinckrodt Chemical Co. gave very slow rates of filtration even when mixed 1 t o 1 with Celite 535 but gave good separations in both the formaldehyde-acetaldehyde and glyoxalmethylglyoxal tests. Nothing further is known about t h e history of this sample.
70
00
l&
0
z 90 0 I
3
2
6
3
9
1
4
5 A 5 8
2
0
0
TIME OF GRINDING, H R S .
Figure 1.
Zone positions with variations in particle size of adsorbent
z
5
Test of Mallinckrodt Ramsay and Patterson Jilicic acid with separation of dinitrophenylhydrazones of formaldehyde and acetaldehyde. Original material and samples ball-milled a n d mixed with Celite 1 to 4. Silicic acid mixed with 0.5 Dart of Celite 535 5A t o 5 8 . Silicic acid as received; n d Celite 1 t o 5.4. Identical procedure of development 5B. Additional development t o improve comparison ~~
Table I.
Particle Size Distribution of Old Silicic Acid and New Coarse iidsorbent (Percentage particle size classified with U. S. standard sieves) Mesh Size
+80
- 80, + 150
Sample I“
Sample I16
2 5
52 12
-150, +200 -200, +300 - 300
12 8 73 1B 12 8 a Sample I old silicic acid, satisfactory rhromatographic properties. Sample f I , new coarse ‘eilicic acid: unsatisfactory chroinatographic properties.
--___ down firmly with a dowel. All work was done with a so-called No. 1.5 column, of 14-mm. inside diameter. A flow rate of 3 minutes for VlSoml. of ligroin for a 150-mm. column is regarded as fast, B to 8 minutes is satisfactory, and 10 to 12 minutes is slow. These rates are with the column completely wet. Positions of zones were measured on extruded columns, so that the results are completely free of any ambiguity which could result from distortions. INVESTIGATIOK OF COARSE SILICIC ACID
A comparison of the particle size distribution of the older satisfactory adsorbent and the newer coarse material is shown in Table I. Chromatographic characteristics of individual paiticle size fractions of sample I1 were studied with the following results. The -200, $300 fraction mixed with 0.5 part of Celite gave a satisfactory f l o ~rate but no separation of the glyoxal-methylglyosal system. K h e n an amount of the -300 fraction equal t o the amount of -200, +300 fraction was mixed in, the filtrat,ion was very slox but a satisfactory glyoxal-methylglyoxal separation was obtained. K h e n t h e +80 fraction was ball-milled and mixed with Celite to give a satisfactory flow rate, a satisfactory formaldehyde-acetaldehyde separation was obtained but t h e glyoxalmethylglyosal system was spread very badly. When the original material of sample I1 was subject,ed to various lengths of time of ball-milling and the products were mised with Celite, usu:tlly one of the two test separations could he obtained satisfactorily, but results were generally unsatisfactory. TT’hen a test of quantitative recovery of n-butyralde-
k
I
2
3
4
5 A 5 B
P 3
6
9
1
2
0
0
TIME OF GRINDING, HRS.
Figure 2.
Zone positions with variations in particle size of adsorbent
Test of Mallinckrodt Ramsay and Patterson silicic acid with separation of dinitrophenylosazones of glyoxal and niethylglyoxal. Original material and samples ballmilled and mixed witli Celite 1 t o 4. Silicic acid mixed with 0.5 part of Celite 535 5 A t o 5B. Silicic acid as received: no Celite 1 t o 5A. Identical procedure of development 5R. .Idditional development t o improve comparison
When the Mallinckrodt reagent grade silicic acid which is specified as “Prepared for chromatographic purposes according to t h e method of Ramsay and Patterson” was studied, very promising results were obtained. T h e “chromatographic purposes” referred to are for the partition methods developed by Ranisay and Patterson (3),and t h e particle size of the material ab sold is convenient for that purpose. T h e material referred to hereafter as “chromatographic silicic acid,” when used as received for adsorption chromatography gives a satisfactory flow rate, but the extrusion characteristics and the occurrence of a certain amount of distortion leave room for improvement. I n addition, when the material can be mixed Kith an inexpensive bulky material such as Celite 535, t h e cost is considerably reduced. When some experiments on ball-milling showed that considerable improvement in separations and in these characteristics would result, an investigation was made to determine the optimum conditions. T h e variations in particle size in different lots of silicic acid are shown in Table 11; the effect of t h e grinding procedure is also shown. The chromatographic properties of this adsorbent as received and after certain procedures of grinding were tested with the formaldehyde-acetaldehyde and g13 oxal-methylglyoxal pro-
ANALYTICAL CHEMISTRY
842
Table 11. Particle Size Characteristics of Chromatographic Silicic Acid as Received and after Grinding (Percentage particle size classified with U.S. standard sieves) Lot 1
Mesh Size
Lot 2
~
Lot 2 After Grinding, Hours
4
8
cedures; the results are presented in Figures 1 and 2. The silicic acid as received, without grinding or mixing with Celite, can be used. The very much sharper zones in the samples ground and mixed with Celite, the easier extrusion, lower cost, and other factors all indicate that the work of grinding the adsorbent is justifiable where chromatographic work of any difficulty is encountered. The amount of grinding must be selected by balancing the slight improvement in separations between for ex-
ample the 9-hour and 12-hour samples against the very much slower rate of flow with the finer grinding. A procedure for grinding chromatographic silicic acid ( 3 ) for 8 to 10 hours, mixing with 0.5 part of Celite 535, and heating for 4 hours a t 200' C. has been adopted for the author's chromatographic work. With this adsorbent, vcry good separations have been attained in work which has included a very wide variety of functional groups and organic molecules. The unsatisfactory results with the e a lier samples of ordinary silicic acid show that particle size of an adsorbent is not the sole critical factor. However, the performance of a material which has suitable adsorption properties can be improved by decreasing the particle size.
-
LITERATURE CITED
lialmberg, E. W., J . Am. Chem. SOC.,76, 980 (1954). (2) llalmberg, E. W., Trueblood, K. i Y , and Waugh, T. D., A x . ~ L . CHEX.,25, 901 (1953). (3) Ramsay, L. L., and Patterson, W. I., J . Assoc. Ofic Agr. Chemists, 28, 644 (1945). (1)
RECEIVED for revie- June 29,
-4ccepted November 10, 1934.
1954
An Improved Acidimetric Determination of Fluoride J. M. CHILTON
and
A. D. HORTON
A n a l y t i c a l Chemistry Division, O a k Ridge N a t i o n a l Laboratory, O a k Ridge,
In an unbuffered solution, the titration of a neutral solution of fluoride with aluminum ions results in an abrupt change of pH at the stoichiometric end point. Details of the separation of fluoride from interfering substances are presented; only phosphate and highly associated fluoride complexes require distillation. With the outlined procedure, in which a recording pH meter was used, fluoride was determined in a sodium fluoride solution with a standard deviation of less than 0.270, and in a fluosilicate solution with a standard deviation of 0.6%. -4nalysis of four samples of NBS phosphate rock No. 120 yielded an average recovery of 99.1%. The method is applicable in the range of 0.1 to 3.5 mg. of fluoride per ml. of water.
T
HE determination of fluoride by distillation from a perchloric
acid solution, followed by the titration of the fluosilicic acid n i t h thorium nitrate in a buffered solution, was introduced by Willard and Winter (9). However, when performed by an inexperienced analyst, this procedure lacks the precision which a recording instrument will provide. All proposed methods for the determination of fluoride, both volumetric and colorimetric, depend on the reaction of the fluoride 1%-itha metal ion, slich as thorium, zirconium, iron, or aluminum, to form a highly associated complex. The colorimetric procedures depend on the determination of the amount of the metal in excess of that complexed by the fluoride. I n the Willard-\Tinter titration, the appearance of an excess of thorium is indicated by the formation of a lake. A procedure by Willard and Horton (8) modifies this to determine the appearance of an excess of thorium bv a fluorescent indicator. Greeff, in 1913, proposed the titration of fluoride with ferric iron, using thiocyanate as an indicator (3).
F + ++
+ 6F-
+
FeFc---
(1)
He also proposed the titration of fluosilicic acid with standard hydroxide
Term.
HeSiFs
+ 2KOH
+
K2SiFe
+ 2H20
(2)
I n 1926, Treadwell and Kohl titrated fluoride with both aluminum and iron, using a potentiometric indication of the ferrous-ferric couple as an end-point indicator ( 7 ) . FeFG---
+ Al-++
-*
.41F,---
+ Fe+++
(3)
In 1930, Kurtenacker and Jurenka investigated several reagenta for the titration of the fluoride ion, using p H indicators to detect the end point (8). Among the titrants used were boric acid, cerous nitrate, and aluminum chloride. Aluminum ions, added to a neutral solution of fluoride ion, react with the fluoride to form a complex anion.
AI++++ 6F-
AlF,---
(4)
Any excess aluminum which is added to the solution R i l l hydrolyze, causing an increase in hydrogen ion concentration.
Al+++
+ n(OH)-
-P
Al(OH),3-
(5)
Batchelder and Meloche made further investigations n-ith the cerous titration, using methyl red as an indicator, and concluded that the color change a t the end point was an adsorption phenomenon ( 1 ) . All of these methods JTere checked to determine which was best adapted to use with a recording, automatic titrator-e.g., now-Precision recording titrator. The titration of fluosilicic acid with standard base 17 as found to be unsatisfactory because the reaction is not instantaneous and does not give a sharp break unless the titration is performed verv slomly, and the presence of other strong acids in the sample interferes. Another hydroxide reaction. given by Treadwell ( 6 ) )is also very slow, and seems to go to completion only in the presence of an excess of hydroxide. SiF,--
+ 6OH-
+
6F-
+ Si(OH), + 2H20
(6)
Solutions of reagent grade sodium fluoride were titrated with a solution of reagent grade potassium alum. Both the procedure
