the trimethylphenylethylamine complex) the general order of elution of various guests although seeming to conform to CTeffect (p-xylene is eluted earlier than p-dichlorobenzene) can not be taken as an unambiguous proof of this effect, because here vapor pressure of the guest and electronic interaction operate in the same direction. Based on these arguments and on the additional data of Table I our previous interpretation of the elution order of various molecules from the Ni(1phenylethylamine)4(NCS)2column must be modified ( I ) . In contrast, gas chromatography can be a very diagnostic tool for shape-selective stationary phases like liquid crystals (19), tri-o-thymotide (20), desoxycholic acid (21), and M(4MePy)4(NCS)2as shown in the present work.
ACKNOWLEDGMENT
Authors are grateful to Pierre De Radzitzky for his kind help. Thanks are due to A. N. Basu for his encouragement throughout this work and to A. Lahiri for permission to publish this paper.
RECEIVED for review May 6, 1969. Accepted August 7, 1969. (19) M. J. S. Dewar and J. P. Schroeder, J . Amer. Chem. SOC.,86, 5235 (1964). (20) A. 0.S. Maczek and C. S.G. Phillips,~"GasChromatography," 1960, R. P. W. Scott, Ed., Butterworths, London, 1961, p 284. (21) A. 0. S. Maczek and C . S. G. Phillips, .I. Chromatogr., 29, 7 (1967).
eparation of Aluminum from ther Elements by Chromatography in Oxalic-Hydrochloric Acid Mixtures and Its Application to Silicate Analysis F. W. E. Strelow, C. J. Liebenberg and F. von S. Toerien National Chemical Research Laboratory, Pretoria, South Africa
ONLYA FEW ion exchange procedures for the separation of aluminium from the silicate forming elements iron, titanium, zirconium, magnesium, calcium, manganese, sodium, and potassium have been described in the literature. Yosimura and Vaki ( I , 2) used ammonium acetate for the cation exchange separation of aluminium from calcium and magnesium, but hydrolysis of aluminium caused difficulties. In the method of Oki et a f . (3) the quantity of aluminium and other elements is limited because of the relatively small separation factor cy;: ,E 2 in 0.8M HC1. Maines (4) has adsorbed Fe(III), titanium and aluminium selectively on an anion exchange resin from sulfosalicylate solutions. However, aluminium is not very strongly adsorbed under the experimental conditions and large columns have to be used to avoid losses, Furthermore, the destruction of sulfosalicylate is rather tedious. Oxalic acid should have promise as an eluting reagent for the separation of the above elements because it forms considerably more stable complexes with tri- and quadrivalent elements than with divalent ones, and because it is a moderately strong acid. Its eluting properties therefore can be modified simply by selecting and mixing the right concentrations of oxalic and a strong mineral acid. It also is relatively easily destroyed. The cation exchange separation of aluminium from gallium reported by Tsintsevich et al. ( 5 ) apparently is the only procedure described for aluminium in oxalic acid, besides the investigation of the separation of aluminium from oxalate by Djurfelt et al. (6). (1) J. Yosimura and H. Vaki, Japan Analyst, 6, 362 (1957). (2) J. Yosimura and H. Vaki, 2. Anal. Chem., 161, 393 (1958). (3) Y. Oki, S. Oki, and S . Hidekata, Bull. Chem. Soc. Japan, 35, 273 (1962). (4) A. D. Maines, Anal. Chim. Acta, 32, 211 (1965). ( 5 ) E. P. Tsintsevich, I. P. Alimarin, and L. F. Marchenkova, Ve'estnik Moskoti. Uniu., Ser. Mat., Mekh., Astron., Fiz. Khim., 13, 221 (1958), Chem. Abstr., 53, 10898 (1959). (6) R. Djurfelt, J. Hansen, and 0. Samuelson, Stiensk Kem. Tidskr., 59, 14 (1947). 2058
e
A systematic investigation of the anion exchange distribution coefficients of elements in oxalic-hydrochloric acid mixtures showed the expected differences in the adsorption of divalent and higher valent elements. Furthermore, it was shown that aluminium is less strongly adsorbed at higher hydrochloric acid concentrations than Fe(III), Ti(IV), and other multivalent elements. A method for the quantitative separation of aluminium from Fe(III), Ti(IV), Zr(IV), Mo(VI), V(V), Ca, Mg, Mn(II), K, Na, and other elements was developed which has definite advantages over other methods. The separation has been combined with a complexometric procedure which uses excess DCyTA, back-titration with standard zinc solution, and xylenol orange as indicator at pH 5.5 for determination (7, 8). The method was applied to synthetic mixtures of elements and standard silicates and was found to give accurate and precise results. EXPERIMENTAL Reagnets and Apparatus. Analytical grade reagents were used throughout, excepting DCyTA (1,Zdiaminocyclohexane-tetraacetic acid). DCyTA was obtained from E. Merck A.G., Darmstadt, Germany, and standardized against 99.99% pure aluminum foil Diu zinc sulfate. Borosilicate-glass tubes of 2.0-2.5 i.d., fitted with B19 groundglass joints at the top and fused-in No. 2 porosity sinterplates and taps at the bottom were used as columns. AGl-X8 anion exchange resin and AG50-X8 cation exchange resin (200-400 mesh) were supplied by Bio-Rad Laboratories of Richmond, Calif. Bromic acid was prepared by passing an aqueous solution of KBr03 through a cation exchange column. The main eluting agents were: Solution 1: 0.05M oxalic acid-0.10M HCl-0.92 HzOZ. Solution 2 : 0.05M oxalic acid-0.50M HC1-0.02x HzOz. Solution 3 : 0.25M oxalic acid-0.10M HC1.
x
(7) R. Pfibil and V. VeselY, Talanfa, 10, 1287 (1963). (8) K. E. Burke and C. M. Davies, ANAL.CHEM., 36,172 (1964).
ANALYTICAL CHEMISTRY, VOL. 41, NO. 14, DECEMBER 1969
0 0 5 M oxalic t 0 50M H c l IO
-+-;; j Hcl
f58
!
TI(E!)
!
ml eluate
Table I. Results of Separations of Binary Mixtures. 54.12mg of A1 mg Taken mg Found Other element A1 Other element 54.10f 0.11 41.18 i 0.04 41.19 Ca 54.13 f 0.09 24.50 i 0.03 24.51 Mg 54.14 i 0.08 54.89 =k 0.06 54.88 Mn(I1) 54.09 f 0.13 39.65 f 0.28 K 39.55 54.08 f 0.10 23.17 f 0.19 Na 23.14 54.10 f 0.09 48.49 f 0.09 48.45 Ti(1V) 54.15 =k 0.11 56.05 f 0.06 56.08 Fe(III> 54.12 f 0.10 92.59 f 0.12 Zr 92.56 54.15 f 0.13 96.93 f 0.22 96.85 Mo(V1) These results are means of triplicate determination. Q
Figure 1. Elution curve for Mn(II)-Al-Ti(IV) mixture 46 ml of AGI-XS, 200-400 mesh oxalate-chloride form. 2.0 X 14 cm column. Flow rate 3.0 f 0.3 ml/min
Elution Curve Mn(I1)-ATI-Ti(1V). Distribution coefficients indicated that Mn(II), Mg, Ca, K, and Na are not absorbed by AGl-X8 resin from solutions containing between 0.104.00M HC1 and 0.05-0.50M oxalic acid. Aluminum is absorbed from 0.10M-0.30M HC1 at oxalic acid concentrations of 0.05-0.50M, but can be eluted with 0.50M HCl containing 0.05M oxalic acid, while Ti(IV), Fe(III), Zr, and Mo(V1) are still fairly strongly adsorbed. The elution curve in Figure 1 was prepared to observe the elution behavior of these elements. Manganese(I1) was chosen because it forms the most stable oxalate complex of the nonabsorbed group, and Ti(IV), because its distribution coefficients are lower thzn those of Fe(II1) and Zr under the experimental conditions chosen. A 46-ml AGl-X8 resin column (2.0 X 14 cm) was equilibrated with 100 ml of solution 3, and a solution containing 1 mmole each of Mn(II), Al, and Ti(1V) in about 200 ml of 0.20M HC1-0.25M oxalic acid-O.02Z H 2 0 2 was passed through it. Manganese(I1) was eluted with 300 ml of solution 1, aluminum with 500 ml of solution 2, and finally Ti(1V) with 250 ml of 6.OM HC1. A flow rate of 3.0 0.3 ml per minute was maintained throughout. Fractions of 25 ml were taken from the beginning of the absorption step and analyzed. Quantitative Separation of Binary Mixtures. A 2.0- X 14-cm column containing 46 ml of AGl-X8 resin was prepared and equilibrated with 100 ml of solution 3. One millimole of aluminum and of one other element were passed through the column in 100 ml of a solution of 0.10M HCI0.10M oxalic acid-0.02z HzOz. Mn(PI), Mg, Ca, K, or Na were eluted with 150 ml of solution 1, taking the eluate from the beginning of the adsorption step. Aluminum was then eluted with 50 ml of 0.50M HC1 followed by 400 ml of 6.OM HCl. From binary mixtures with the oxalate complex forming elements, aluminum was eluted first with 500 ml of solution 2. Titanium and zirconium were then eluted with 400 ml of 6.0M HCl, while in the case of Fe(II1) and Mo(VI), the oxalate was eluted first with 200 ml of 6.OM HCl, before Fe(II1) was eluted with 250 ml of 0.50M HC1 and Mo(1V) with 250 ml of 1.OM ammonium nitrate containing 0.50M ammonia solution. A flow rate of 3.0 ==I 0.3 ml per minute was maintained throughout. After the oxalic acid had been removed by a cation exchange step or destroyed by nitric acid-bromic acid oxidation, the amounts of the elements were determined. The results are presented in Table I. Determination of Aluminum in Silicate Rocks. About 1 gram of silicate (0.5 gram when A1203was more than 20 per cent) was dissolved and silica volatilized by heating in a platinum dish or beaker made of Teflon (Du Pont) with a mixture of HF, HC1, and H2S04. The sample was taken to fumes of HzS04 and care was taken to remove fluoride as completely as possible. Most of the HISO( was expelled.
Table 11. Results for Determination of A1 in Standard Silicates Number of deterZ AhOa minaSample std value AlzOa found tions 15.00 15.01 f 0.03 4 Diabase W-1 AGV-1 17.01 17.07 f 0.02 4 0.65-0.85 0.711 2~ 0.003 4 PeriodotitePCC-1 Dunite DTS-1 0.25-0.55 0.211 f 0.003 4 Synthetic silicate 1 9,804 9.811 f 0.022 1.7 Synthetic silicate 2a 0.490 0.489 f 0.003 6 98.04 mg and 4.90 mg of AlzOa present, a 1-gram sample weight was assumed. Q
About 30 ml of 1M HCl were then added, and the solution was warmed until all soluble salts had dissolved. Any remaining insoluble material was separated and dissolved by heating in a platinum crucible with a mixture of H s P O ~ , HC104, and H F in a ratio of 1:3 :1. HClO4 and H F were expelled by heating and the residual viscous mass was dissolved in 5 ml of 0.5M HC1. This solution was added to the first filtrate and the combined solution was diluted to about 250 ml, adding 80 ml of 5 % oxalic acid, 10 to 12 grams of boric acid, and 10 ml of 0.5% € 3 2 0 2 . The solution immediately was passed through a 2.0- X 14-cm column containing 46 ml of AG1-X8 resin which had been equilibrated with 100 ml of solution 1. The sample was washed onto the resin with 50 ml of solution 1 and Mn(II), Mg, Ca, K, Na were then eluted with 150 ml of the same solution. A 2.5- X 17-cm column containing 75 ml of AG50-X8 cation exchange resin was connected to the anion exchange column, and aluminum was eluted with 550 ml of solution 2 and adsorbed on the cation exchange column. The columns were disconnected and oxalic acid and V(V) were washed from the cation exchange column with 150 ml of 0.50MHCl followed by 100 ml of 0.01M HNO, containing 0.02% Hz02. Subsequently aluminum was eluted with 3.OM HC1. The excess HCl from the whole eluate or a suitable fraction was removed by evaporation, an excess of 0.05M or 0.005M DCyTA was added, and after pH adjustment, the excess DCyTA was back-titrated with 0.01 or 0.005M zinc sulfate solution at pH 5.5 using xylenol orange (solid 1% mixture with KNO,) as indicator. The method was applied to two synthetic standards which contained 26.00 mg of Na, 48.40 mg of K, 39.42 mg of Ca, 27.61 mg of Mg, 13.47 mg of Mn, 64.43 mg of Fe, 12.63 mg of V, 12.20 mg of Ti, 29.68 mg of Zr, and 51.89 and 2.595 mg of Al, respectively, plus 5 ml of 5 M HzS04, 20 ml of 2 M HC1, 10 ml of 0.5% H~OZ, and 4 grams of oxalic acid in a volume of 200 ml, and to 4 standard silicate rocks of the U. S. Geological Survey. The results are presented in Table 11.
ANALYTICAL CHEMISTRY, VOL. 41, NO. 14, DECEMBER 1969
0
2059
DISCUSSION
The described method provides an excellent means for the selective separation of aluminium from many other elements. As much as 100 mg of aluminium (200 mg of AlzOa)can be separated quantitatively from even larger amounts of other elements, provided the quantities of Fe(III), Ti(IV), Zr, V(V), and Mo(V1) are not excessive ( < l o mmoles together with Al). Only up to about 50 mg of calcium may be present in the absence of boric acid, but in the presence of boric acid, this is increased to about 150 mg. Slow precipitation of calcium oxalate occurs with larger quantities of calcium which leads to mechanical retainment on the column. With samples high in calcium, separation should be started immediately after addition of oxalic acid. Large quantities of V(V) and Mo(V1) are also precipitated when present together with large quantities of calcium, but these elements are only minor components in silicate analysis. The elution curve of aluminium shows a double peak (Figure 1) which is probably due to the presence of two different aluminium oxalate complexes with fairly slow conversion reaction rates. Ga, In, Zn, Cd, and Sn(1V) are also retained by the resin quantitatively when elution with solution 2 is carried out and can be separated from aluminium by this method. Cu(II), Ni(II), and Co(I1) are eluted with solution 1 together with
Mn(I1). However, Cu(I1) especially tends to form an insoluble Cu oxalate ppt, and only a few milligrams can be separated satisfactorily. Vanadium is partially reduced by oxalic acid and accompanies aluminium partially, but can easily be eluted from the cation exchange column with 0.01M HNCPa-0.02 HzOz, while aluminium is retained. The method is well suited for the accurate determination of aluminium in silicate rocks and should be useful for reference analysis. Results for large quantities compare favourably with the classical gravimetric method, while results for small quantities (a few milligrams) are considerably more accurate (Table 11). The results obtained for the synthetic mixtures show that the method is also more accurate than spectrophotometry, atomic absorption spectrometry, and X-ray Auorescence spectrometry. About 30 separations and determinations can be carried out by one analyst in 3 days provided sample dissolution is routine. ACKNOWLEDGMENT
This work is part of a D.Sc.-thesis by the third author at the Department of Inorganic and Analytical Chemistry of the University of Pretoria.
RECEIVED for review April 29, 1969. Accepted July 14, 1969.
Kinetic Ana lysis of Phermogravimetric Data J. M. Sharp and Sally A. Wentworthl Department of Ceramics with Refractories Technology, The University, Shefield, England
DURING THE PAST ten years, several methods have been developed to allow kinetic analysis of thermogravimetric data, but few attempts have been made to compare them critically. Both experimental and analytical errors of kinetic data have been investigated by Sestak (1, 2) who has given major attention to the mathematical description of individual methods. Inherent errors in several methods of mathematical analysis were evaluated by comparing the values of the kinetic parameters obtained from each method with the values of those for a theoretical TG curve computed from chosen kinetic data (1). The so-calculated values of E did not differ by more than 10% from the originally chosen real values allowing the conclusion that the mathematical accuracy of all methods studied is satisfactory for the computation of kinetic data. Flynn and Wall (3) also have reviewed methods of kinetic analysis placing emphasis on the thermogravimetry of polymers. They discuss five categories of analytical methods: “(a) ‘Integral’ methods utilizing weight loss us. temperature data directly, (b) ‘Differential’ methods utilizing the rate of weight loss, (c) ‘Difference-Differential’ methods involving differences in rate, (d) Methods specially applicable to initial rates, and (e) Nonlinear or cyclic heating rate methods.” 1 Present address, Department of Soil Science, Ontario Agricultural College, University of Guelph, Ontario, Canada
(1) J. Sestak, Talanfa,13, 567 (1966). (2) J. Sestak, Silikaty, 11, 153 (1967). (3) G. H. Flynn and L. A. Wall, J. Res. Nut. Bur. Stand., A , 70, 487 (1960). 2060
In our present work, a critical comparison is made of three methods which have been proposed for analyzing TG curves. The methods are representative of categories (a), (b) and (c) of the classification by Flynn and Wall (3). The most frequently used method of obtaining kinetic parameters from T G curves is that developed by Freeman and Carroll ( 4 ) (henceforth referred to as method I). It is a Difference-Differential method and attempts to determine both the order of reaction and the activation energy. This method has several disadvantages (5) and often leads to uncertain or meaningless values for the order of reaction. Since in solid state reactions there is theoretical justification only for orders of reaction of 0,1/2,2/3, and 1, Coats and Redfern (6)(method 11) developed an Integral method which is applied to T G data assuming in turn each of these orders of reaction. The correct order is assumed to lead to the best linear plot from which the activation energy is also determined. Many solid state reactions cannot be classified in terms of an order of reaction, among which are those following diffusion equations or giving sigmoidal curves under isothermal conditions. To overcome this difficulty, Achar et al. (7) developed a Differential method of analyzing a T G curve (method 111) which applies generally to all reaction mechanisms, provided that the correct mechanism is already known. Although this may seem to be a serious limitation of the (4) E. S. Freeman and B. Carroll, J. Phys. Chem., 62, 394 (1958). (5) R. L. Bohon, Proc. First Toronto Symp. Thermal Anal., H. G. McAdie, Ed., pp 72-73 (1965). (6) A. W. Coats and J. P. Redfern, Nature, 201, 68 (1964). (7) B. N. N. Achar, G. W. Brindley, and J. H. Sharp, Proc. Znt. Clay Con$, Jerusalem, 1, 67 (1966).
ANALYTICAL CHEMISTRY, VOL. 41, NO. 14, DECEMBER 1969
