lntercomparison of Beckman Spectrophotometers GALEN W. EWING’ Sterling- Winthrop Research Institute, Rensselaer, N, Y . AND
THEOPHILUS PARSONS, JR. E. 1. d u Pont de Nemours and Co., New Brunswick, N . J . A comparative study has beeu made of ten Beckman quartz spectrophotometers using potassium acid phthalate as a standard reference material. The samples have been weighed and dissolved by an identical procedure in the several laboratories. It is found that the precision to be expected of any one
instrument is considerably better than the agreement with other similar instruments. A n assay to be valid should be compared with a standard determined on the same spectrophotometer, or corrected by a factor determined in advance for that instrument.
I
S THE course of routine analyses on chemical intermediates, it was found that precise checks could not be obtained between determinations in two laboratories using standard solutions and presumably identical Beckman spectrophotometers ( 1 ) . To track down the source of the discrepancies, a series of careful measurements was made on one sample of standard material on both instruments, and this work was later extended to other laboratories t o cover a total of ten instruments. Potassium acid phthalate, which was chosen as the reference material, shows the following advantages: It is easily purified and is stable enough to retain its purity for an indefinite period under usual laboratory conditions. It has a single, rather broad absorption maximum a t approximately 281 mp wave length, and H well defined, broad minimum a t 264 mp, both in the range of convenient operation of the spectrophotometer, and both sufficiently broad that small differences in slit widths and small inaccuracies in the wave-length scale will not have an appreciable effect on the result. The extinctions of the maximum and minimum are close enough together, so that both points can be measured on the same solution, with optical densities all within the range 0.4 to 0.7 ( 2 ) . The sample studied was Mallinckrodt’s primary standard analytical reagent. It was checked spectro1
photometrically against Bureau of Standards material, from which it showed no appreciable difference. Portions of this material in the solid state were sent t o each participating laboratory where they were weighed out and dissolved by an identical procedure Weigh out accurately, directly from the bottle without drying, a sample of approximately 0.4000 * 0.0015 gram. Dissolve in 100-mi. of distilled water in a 100-ml. volumetric flask. After thorough shaking withdraw by pipet a 10-ml. sample and place in another 100-ml. volumetric flask. Make up t o volume with distilled water and shake well. From the second volumetric =flask withdraw by pipet a 25-m1. sample and dilute by a similar procedure t o 100 ml. Thus the final concentration is about 0.1000 gram of potassium acid phthalate per liter. Run this last solution in the Beckman spectrophotometer against a water blank, using 1-em. silica cells. Use wave lengths of 264, 265, 266, 280, 281, and 282 mp. Adjust the slit t o give a band width of 1 mp, or as near this value as possible. Calculate a constant, K , for each wave length by the formula
K = D/c where D is the optical density and cis the concentration in grams Der liter.
In each case corrections for the particular absorption cells used were obtained by filling both reference and test cells with distilled water and determining - the oDtical density of the test with reqpcct to the reference cell at both 264 and 281 mp wave lengths. The figure so obtained was subtracted from the observed densities of the phthalate solutions. KOattempt way made t o control the temperature of the appaIatus and solutions. A rough calculation shows that the effect of normal variations in room temperature with 1037 respect to the expansion coefficients of glassware, etc., will be negligible. To test the change in the absorption uf phthalate with small temperature changes, an experiment was performed with a solution of potassium acid phthalate in special absorption cells provided with jackets for the circulation of water of controlled temperature (Table I). These cells were 100 mm. in length; hence the solution used was one tenth the concentration specifitsd for the comparative experiments. S n y differences due t o temperature changes are entirely within the limits of experimental error. The actual values of K,,,. in thm experiment are considerably lower than in the comparative experiments described. This is due to failure of Beer’s lam over the tenfold concentration change in-
Present address, Department of Chemistry, Union College, Soheneotsdy.
N. Y. tj.45
6.40
Lo
6.35
355
. MEAN
2
B
q 6.30 F
6.25
+a4
:-*
--- --------___-__---__--_-_ - ....................
to-
-
;
_________-__________-----_____-__-____----------
=-
6.20
949
I
4
-
*77
Figure 1.
7+
ii.8
Extinction Coefficients of Potassium Acid Phthalate at Wave Length of Maximum Absorption
Serial number of the spectrophotometer is indicated adjacent to each group of determinations
423
tion. The maximum error which might be introduced into the measurements through cumulative errors in glassware and
424
A N A L Y T I C A L CHEMISTRY
4.25
4.20
d
’B
4.15
4.10
4.05 i 0
5
I
10
15
20
25
30
35
Reading Kuinber
Figure 2.
Extinction Coefficients of potassium Acid phthalate at Wave Length of Minimum Absorption
Serial n u m b e r of spectrophotometer is i n d i c a t e d adjacent to each g r o u p of
ferences of manufacture inevitable in a sensitive instrument of this type. Error from this source may be avoided by comparing each instrument with the statistical mean of a large number of instruments, to determine a correction factor. In the absence of any such general statistical mean, two laboratories may eliminate these instrumental errors by using an appropriate secondary standard, such as potassium acid phthalate, as a basis for comparison. A continuing comparison would also furnish evidence of procedural or instrumental changes taking place in either of the laboratories. The present study is in general agreement with the findings of the Vitamin Oil Producers Institute (S), which has reported data taken on nineteen Beckman sDectrophotometers as well as other types of apparatus. Of these nineteen, six fall consistently within k1.570 of the mean, six betX-een and *3.O%9 and the remaining seven are erratic in that some of the five samples studied showed good agreement and some poor.
determinations
ACKNOWLEDGMENTS
analytical balance is considerably less than 0.5%, based on the following specifications: 10-ml. pipet, tolerance 25-ml. pipet, tolerance 100-ml. flask, tolerance Balance, precision
1 0 02 ml.
I t is a pleasure to acknowledge the cooperation of the following individuals and of their organizations in making determinations for this study; in many cases they have made valuable sugges-
0 03 ml. 0.08 ml. 0 05 mg.
The data obtained on the ten instruments are given in ~ ~ 11, arranged according t o serial numbers. The original data are
bTable l I.~ Effectof Temperature on Extinction Coefficients of Potassium Acid Phthalate Wave Length, mp
K200
K”=
given for each of the several solutions made up from the crystal264, min 4 24 4 27 line material in each laboratory, averaged in each case in the 5.72 281, max. 5 70 columns headed “Mean.” The columns headed “Dev ” show the deviation of the mean of the readings for each instrument from the mean of the individual means for all the instruments. Table 11. Extinction Coefficients of Potassium Acid Phthalate Determined The average deviations of the individual on Various Beckman Spectrophotometers means from the mean of the means are, Ratio Kmsx./Km Dev. in. Rmsx. KIlllll. Serial Sample for the minimum, 0.77%, for the Detna. Mean Dev. Detns. Mean Dev. so. No. maximum, 0.95%, and for the ratio, 0.071 1.517 0.002 6.45 6.365 4.27 4.210 0.035 305 1 6.41 0.63%. 4.21 2 6 37 4.19 3 These thirty-six individual readings 6.31 4.17 4 have been plotted in Figures 1 and 2 for 0,006 6.24 6.240 0.074 1.506 0.031 4.14 4,144 316 1 6.23 4.15 2 the maximum and minimum, respec6 . 2 4 4.13 3 tively. The values of the arithmetic 6.23 4.15 4 6.24 4.15 5 means and of the standard deviations 0.063 1.518 0,008 6.377 4.200 0.025 6.36 4.21 3.55 1 -Le., =tu-are indicated by horizontal 6.37 4.19 2 6.38 4.19 3 lines. The standard deviations were 6.39 4.21 4 calculated according to the method of 0.111 1.520 0.008 6.21 6.203 0.095 4.11 4.080 377 1 Rorthing and Geffner (4). 6.17 4.07 2
6.20 4.08 6.23 4.06 0.056 1.520 0,008 6.370 0.015 6.33 4.190 576 1 4.19 6.41 4.21 2 6.41 4.22 3 0 6.33 4.14 4b 1.488 0.024 6,250 0.064 6.22 4,200 0.025 4.20 598 1 6.28 4.20 2 1.517 0.003 0.018 6,296 6.28 4.160 0,025 4.16 761C 1 6.30 4.16 2 6.31 4.13 3 1.524 0.012 0.034 6.34 6.346 4.165 0.010 4.11 949d 1 6.36 4.17 2 6.38 4.21 3 6.31 4.17 4 0.018 1,494 0,032 6.26 6.262 4,205 0,030 4.22 1 988 e 6.31 4.19 2 0.007 1.619 0.076 6.390 0.032 6 39 4.207 4.22 1037f 1 6.38 4.21 2 6.39 4.19 6.40 4.21 Same weighed sample used for determination of d a t a for Table I, appropriately diluted for 10-mm. cells. b Average of four determinations made 7 months later than others reported for this instrument. Same d Same solutions as on No. 305. C Same solutions as on No. 377, same cells, same operator. solutions as on No. 598, different operators. I Same solutions as on No. 355. 3 45
CONCLUSIONS
Both table and graphs indicate that the agreement among the several readings taken on any one instrument is usually much better than the agreement among the various instruments. The constancy of the ratio of maximum to minimum extinction is only slightly better than the absolute values themselves. It is evident from these results that any absolute assay which is undertaken using the Beckman spectrophotometer will be subject to an uncertainty greater than that indicated by the limits of precision of each individual instrument. This is presumably due to slight dif-
f
V O L U M E 20, NO. 5, M A Y 1 9 4 8
425 LITERATURE CITED
tions as well. The serial numbers of the spectrophotometers are included in parentheses.
(1) Cary, H. H., and Beckman, A. O., J . Optical SOC.Am., 31, 682
(1941).
A. Black, E. R. Squibb and Sons, New Brunswick, N. J. (59% 988) Irving iM.Klotz, Northwestern University, Evanston, 111. (318) Kenneth Morgareidge, National Oil Products CO., Harrison, N. J. (305,949) J. M. Vandenbelt, Parke, Davis and CO., Detroit, Mi&. (355, 1037) Instruments 377 and 761 are in the laboratory of the SterlingWinthrop Research Institute, and No. 576 is in that of E. I. du Pont de Nemours and Co.
(2) Vandenbelt, J. M., Forsyth, Jean, and Garrett, iinn, IND. ENQ,
CHEM.,ANAL.ED., 17, 235 (1945). (3) vitamin oil Producers Institute, 149 California St., Sen Francisco, Calif., Collaborative Assay KO.1 (1943). (4) Worthing, A. G., and Geffner, J., “Treatment of Experimental Data,” pp. 168-70, S e w York. John Wiler & Sons, 1943. RECEIVED J u l y 18, 1946.
Preparation of Standard Chromous Sulfate or Chromous Chloride Solutions of Determinate Concentration JAMES J. LINGANE AND ROBERT L. PECSOK Department of Chemistry, Harvard Unicersity, Cambridge 38, Mass. from pure potassium dichromate. The potentiometric titration of +2 copper in 4 to 6 N hydrochloric acid with 0.1 N chromous ion is discussed, and data show that i t is accurate to *O.lqo or better. The direct potentiometric titration of dichromate ion with chromous ion in dilute sulfuric acid is not a satisfactory standardization method, but excellent results are obtained by adding excess ferrous ion and titrating the resulting ferric ion.
A simple technique for preparing standard chromous sulfate solutions by complete reduction of chromic ion in very dilute sulfuric acid with zinc is described. The solution is stored under hydrogen in contact with amalgamated zinc in the same vessel in which it is reduced and is stable for several weeks. Chromous sulfate (or chromous chloride) solutions of an exactly specified concentration, which do not require standardization, may be prepared determinately
+
T
standard solutions. Buehrer and Schupp ( 5 ) started with potassium dichromate which they first reduced to the chromic state wit,h hot, concent,rated hydrochloric acid; the diluted chromic solution was finally reduced to the chromous state with pure zinc and then transferred t o the storage bottle under a layer of kerosene. Buehrer and Schupp reported that their solution decreased in titer at the rate of about, 1we per week. Rienacker ( I d , 1 7 ) first prepared pure chromous acet’ate by reducing chromic solutions with zinc and then precipitating the resulting chromous solution x i t h sodium acetate. The chromous acetate was then x-ashed t,horoughly Kith water, finally dissolved in only a very slight excess of hydrochloric acid, and transferred to the storage bottle under hydrogen. Rienacker found that a solution thus prepared decreased in titer only about O.1ye per week. The Rienacker procedure appears to be the best of those recommended up to 1932, but the fact that all the manipulations involved must be carried out under hydrogen renders it inconvenient. Thornton and Sadusk ( 1 4 ) prepared chromous sulfate solutions in dilute sulfuric acid (approximately 0.18 S)by reducing solutions of potassium dichromate n-ith amalgamated zinc in a Jones-type reduct,or; t,hey obtained only 67y0 reduction to the chromous state, but reported that the solution “did not undergo an appreciable change of titer during a period of 2 months” when stored under carbon dioxide in the storage apparatus of Thornton and Wood (15). The reduction of chromic solutions by flowing them under carbon dioxide through a Jones-type reductor containing amalgamated zinc m-as later investigated by Stone and Beeson (13),who found that violet chromic solutions in dilute sulfuric acid were reduced more rapidly by zinc than solutions containing the green modification, and that 90 t o 100% yields of chromous chromium could be obtained lvhen violet chrome alum solut,ions\-,we used instead of dichromate solution.
H E standard potential of the couple Cr+++ e = Cr++is -0.40 volt us. the standard hydrogen electrode (6, 6), and chromous ion is the most powerful reductant used in the form of a standard solution in volumetric analysis. Because of this fact the preparation and storage of chromous solutions necessitate special care, and the preparation of a standard solution of an exactly determinate strength has not heretofore been described. Chromous ion is so very easily air-oxidized 4Cri-t
+ 02 + 4 H + = 4Cr++f + 2H20
(1)
that the solution must be stored and delivered from the measuring buret under an inert gas (usually hydrogen), and the solution being titrated must also be scrupulously freed from dissolved air. Furthermore, chromous ion is a powerful enough reductant to reduce hydrogen ion 2Cr
t+
+ 2H+ = 2 C r + f + + H2; R
= lO+l3
(2 1
and acid chromous solutions arc metastable. The oxidation of chromous ion by hydrogen ion is very slo~vwhen the solution in dilute sulfuric or hydrochloric acid is composited from very pure materials, but Reaction 2 is strongly catalyzed by various substances, particularly platinum and other finelv divided metals (1, 9). Hence chromous solutions must be prepared from very pure materials that are free of foreign heavy metals. Forbes and Richter (j), m-ho made the first reliable measuiement of the chromic-chromous potential, prepared chromous chloride by reducing resublimed chromic chloride with hydrogen a t 400” C. and dissolving the solid salt in appropriate acid solutions under carbon dioxide. Dimroth and Frister (4),who introduced standard chromous solutions in volumetric analysis, also employed chromous chloride. Traube, Burmeister, and Stahn (16), Grube and Schlecht ( 7 ) , and Asmanow (1) prepared chromous salts by electrolytic reduction of chromic solutions; Asmanow precipitated chromous sulfate pentahydrate from the reduced solution with alcohol, and used this salt to prepare
.
The method described herein employs reduction of chromic solutions by amalgamated zinc, but it possesses the following advantages over previous procedures: The solution is reduced and stored in the same vessel, and the apparatus is much simpler than any previously described; a solution containing the chromium entirely in the +2 state is obtained; and by starting with pure potassium dichromate (or any chromic salt of known purity) L
