cobalt-indicator complex is so strong that it can cause errors. Apparently, as the end point of the back-titration is neared and the concentration of the excess EDTA becomes very small compared to that of the indicator, the cobalt tends t o leave the EDTA complex and to form a complex with the indicator. This may lead to a premature end point and high results. By back-titrating rapidly, however, satisfactory results can be obtained (3, I S ) . To determine the effect of the total time of the backtitration on the accuracy of the method, nine samples of a cobalt drier were titrated for varying lengths of time and a t somewhat different rates. Results are given in Table 11. Samples 1, 2, 3, 4, 5, 6, and 8 were back-titrated a t a fairly constant rate from start to end point. Sample 7 was back-titrated by first adding 80% of the titrant as quickly as it would leave the buret, and then by adding the remainder dropwise until the end point was reached. Sample 9 was titrated by adding all of the bark-titrant dropwise. Data in Table I1 show that the average of the results of the first seven determinations is 6.07%. The average of earlier results which were obtained before the time dependency was realized is 6.09% (Table I). Apparently, as long as the back-titration is completed within about 21/* minutes, the results agree well with those obtained by other methods. Solvent. The work of Gerhardt and Hartmann (4) illustrated that EDTA titrations can be carried out in a solvent which is largely nonaqueous. Acetone, the solvent used by Gerhardt and Hartmann for lubricating oils, was not satisfactory for driers because most driers are insoluble in acetone. Ethyl alcohol was tried and found to be satisfactory for all but the lead driers. Occasionally, a manganese drier would not dissolve in the alcohol and dioxane was used for lead and manganese driers. The mixtures had t o be heated to effect
solution in alcohol or dioxane. The best combination of solvents for all the driers proved to be benzene and alcohol. By adding 2 ml. of benzene and then 50 ml. of alcohol, the heating step could be eliminated. The benzene apparently holds the drier in a true solution which can then be diluted with alcohol. The disadvantage of this combination is that the benzene is responsible for the cloudiness which develops after the aqueous titrant is added to the sample solution. The cloudiness apparently makes the end point a little more difficult to detect. However, the results obtained in the alcohol-benzene solution agree well with those obtained in alcohol alone or in dioxane. Precision and Accuracy. On the basis of the cobalt and lead results, the coefficierft of variation of the method is estimated to be 0.8%. This figure is comparable with that obtained by Gerhardt and Hartmann (4) for the titration of zinc additives in lubricating oils. The precision of the EDTA titration in alcohol-benzene, as measured by the standard deviation, is significantly poorer than the same titration in aqueous media. The poorer precision is attributed to the fact that the end point in the alcohol-benzene titration must be detected in the presence of a turbidity. It is difficult to determine the accuracy of the method because of the lack of standards. The results obtained by various methods are shown in Table I. If it is presumed that the gravimetric method gives the true metal content of the driers, then a comparison of the averages of the EDTA alcoholbenzene method Kith those of the gravimetric method leads to the relative accuracies: calcium, +0.73%; cobalt, -0.49%; lead, -0.24%; manganese, -2.4%; and zinc, +1.4%. ACKNOWLEDGMENT
The authors wish t o acknowledge
Table II. Effect of Total Time of BackTitration on Cobalt Results
Sample KO.
Time, See. 30
1 2
Cobalt,
%
6.07 5.96 6.04 6.09
40 60 80
3 4
the assistance of N. K. Kaprielyan who obtained most of the flame photometric data. The helpful suggestions of J. C. Weaver and the encouragement of R. F. Schneider, K. R. Brown, and Richard Tobin are also gratefully acknowledged. LITERATURE CITED
(1) Am. SOC.Testing Materials, Designa-
tion D 56447, “Standard Methods of Testing Liquid Driers,’’ 1947. (2) Cox, D. S., Paint and V a r n i s h Production 45 (1955). ( 3 ) Flaschka, A., Barnard, A. J., Jr., Broad, W. C., Chemist Analyst 47, 25 (1958). (4)Gerhardt, P. B., Hartmann, E. R., ANAL.CHEW29, 1223 (1957). (5) Harris, W. F., Sweet, T. R., Ibid., 26, 1649 (1954). (6) Leggieri, G., Chimica ( M i l a n ) 10, 287-8 (1955). (7) Lucchesi, C. A , , Ofic. Dig.Federation
P a i n t & Varnish Production Clubs 30,
212-30 (1958). (8) Mattiello, J. J., “Protective and Decorative Coatings,” Vol. I, p. 499, Wiley, Kern York, 1942. (9) Pap:: H. F., “Organic Coating Technology, Vol. I, p. 227, Kiley, New York, 1954. (10) Ibid., p. 240. (11) Pokorny, J., Pribyl, J., Chem. zvesti. 9, 20-6 (1955). (12) Welcher, F,. J., “Analytical Uses of Ethylenediamlne Tetraacetic hcid,” Van Sostrand, Ken- York, 1958. (13) Ibid., p. 231. RECEIVEDfor review May 8, 1958. Accepted July 29, 1958.
Simple Microspectrophotometer DONALD
F. tl.
WALLACH and DOUGLAS
M. SURGENOR
Department of Biological Chemistry, Harvard Medical ‘School, and Protein Foundation, Boston, Mass.
b A spectrophotometer is described which permits rapid and accurate absorption measurements on 10 PI. of solution. A light beam, 0.4 mm. in diameter, is obtained from a glass prism monochromator. The cuvette is fabricated from capillary tubing, 1 mm. in inside diameter. It is positioned with the aid of a microscope, after which the transmitted light is
deflected to a photomultiplier. The ratio l o / / is measured with a balanced photomultiplier bridge circuit.
an investigation into the chemistry of human blood platelets and other tissue cells, the need arose for rapid and simple spectrophotometric measurement of several constituents, particularly nitrogen, calcium, and URIXG
phosphorus, in very limited amounts of tissue. Although there is considerable literature on the spectrophotometry of cells and tissues ( 5 ) , techniques for simple microspectrophotometry are relatively few ( I ) , and, when applied to truly small volumes, difficult and costly. A simple microspectrophotometer on the general principle of the microcolorirneter of Holter and Lgvtrup VOL. 30, NO. 1 1 , NOVEMBER 1958
0
1879
( 2 ) and Krueglis ( 3 ) has been constructed, permitting rapid and precise absorption measurements in the visible range of the spectrum, utilizing volumes of about 10 d. APPARATUS
d schematic diagram of the apparatus is shown in Figure 1. I n essence, a very small beam of monochromatic light from a monochromator is directed through a capillary cuvette. Absorption of light by material in the cuvette is measured by a balanced photomultiplier bridge circuit. The light source is a Sylvania C-10, lO-rratt zirconium arc lamp with a brilliance of 18 cp. per sq. mm. and a diameter of 0.53 nim. The collective lens, L1,is a 60-mm. lens which forms an image of the arc on the entrance slit, SI. A modified Hilger glass spectroscope is used as the monochrometer. Light from SIis collimated by lens Lz (fll0; j = 250 mm.) and dispersed by a constant-deviation glass prism. The dispersed light is focused by telescope lens Lp(f/lO; f = 250 mm.) via a rightangle prism upward onto exit slit Sz. The entrance and exit slits are mounted in brass holders which permit precise positioning of the slit and interchangeability of slits-e.g., of different widths. Each slit consists of a clear line 1 mm. long and of the desired width, ruled centrally on the blackened surface of a 19 x 0.5 mni. optical flat. Slits varying in width from 25 to 250 microns are used (prepared by D. W. l l a n n , Lincoln, Mass.). The beam from the exit slit is split by a 0.5-mm. quartz plate a t 45" to the optical axis, deviating about 5% of the light onto a reference photomultiplier, Pill, (Figure 2). The image of the exit slit is focused a t a point halfway through the bore of the cuvette by a lens, L4 (f = 20 mni.). At this point the beam is about 0.2 mm. in diameter and, when the cuvette is properly
aligned, reflections from its wall are eliminated. The maximum diameter of the beam within the cuvette is 0.4 mm. The beam emerging from the cuvette is picked up by a 5X, coated microscope objective and thence projected onto a measuring photomultiplier, P X 2 . A movable mirror, M , allon-s deflection of the light beam into a 1OX ocular for alignment of cuvettes. -4 movable cover over the ocular prevents extraneous light from reaching the photomultiplier. Cuvettes and Cuvette Carrier (Figure 3). Cuvettes are made from borosilicate glass tubing of 1-mm. bore and about 6-mm. exteinal diameter. They are precisely 10 mm. long and have ends which are parallel, perpendicular to the bore, and optically polished. Carefully polished quartz disks 6 mm. in diameter and 0.5 mni. thick served as end viindoiw. Cuvettes and AT indows n ere made by the A. D. Jones Optical Co., Cambridge, Mass. The cuvettes must be scrupulously clean and free from marks and scratches. The cuvette carrier is a piece of Bakelite, 50 X 25 X 6 mm., containing two rows of holes 6 mm. in
V
I
diameter and 5 mni. deep. A centered 2-mm. hole passes t,hrough the bottom of each hole. To load cuvettes, bobtom windows are placed in each hole! so that they rest flat on the bottom; the cuvettes are then placed gently on top. They are filled by placing a 10-pl. drop on the top of the bore. This flons down into t'he cuvette, displacing the air through the bottom, and forms a capillary seal between the window and the cuvette. Another end window is pressed down on top, creating a second capillary seal. This procedure takes only a few seconds; once formed, the seals hold for over 30 minutes. For certain organic solvents, it is necessary to seal top and bottom lightly with silicone grease and to use a fine polyethylene catheter for filling. After filling, the curette carrier is placed inside a light tight box with removable lid on t'op of the microscope platform. A mechanical stage operated by controls brought to the outside of the box permits precise alignment of cuvettes under visual control, when the observation mirror is positioned to reflect the light beam into the ocular. Electronics (Figure 1). -4balanced
"
LIGHT S O U R C E
i
A - TELESCOPE LENS (131 0 90'PRISM C . E X l T SLIT 1521 D - QUARTZ PLATE E . REFERENCE PHOTOTUBE HOUSING F . BAFFLES
0 - LENS I L q l H ' CUVETTE
I
I - S X OBJECTIVE J. MOVABLE MIRROR
'
Figure 3. and carrier
K - 10 X OCULAR L - PHOTOTUBE HOUSIW
Cuvette
M. WAVE LENGTH KNOB
Left. Cuvette and end
%&LE
vindows Assembled cuvette in carriers. Lower. C u v e t t e i n Bakelite carrier ready for filling
u 0 2 4 6 C M
Upper.
..._....._.._.. 7 I
Figure 1. apparatus
1880
Schematic
diagram
ANALYTICAL CHEMISTRY
of
BAL A N C E PHOTO MULTlP LIE R ANODE
CATHODE
AYRTON SHUNT
0 0
I3v
DETECTOR TUBE
Figure 4.
DARK CURRENT CONTROL
Balanced bridge circuit for measuring photomultiplier output
90
Figure 5 . Absorption spectrum of Corning No. 5120 didymium glass (5mm.)
0.7 80
70
0.6
60
3
0.5
I
-
50
,
0.4
z a 0
40
m
9
30
m
0.3
q
IO *O 0
I 400
0 2
0.I
500 WAVE
600 LENGTH , M I L L I M I C R O N S
700
0 1.5 3.0 4.5 OfBROMOTHYMOL BLUE
in
BARBITAL B U F F E R pH E 6 , r/2 0 I
photomultiplier bridge circuit similar to that of Oldenberg and Broida ( 4 ) is employed. TKO1P21 photomultiplier tubes are used as detectors. Power to the photomultipliers is provided by dry cells; multiple taps provide a coarse sensitivity control. Fine control is obtained by variable resistances r-11 and r-12. By means of suitable substituting switches either tube may be read by itself. Although both photomultipliers have appreciable and different dark currents a t high supply voltages, these are steady to within less than 1%. Therefore the net dark current may be bucked out by a small external voltage. This adjustment is made (Rl’,Rz’, R3’) prior to the other adjustments and requires only occasional checking. The bridge measures the difference between the light intensities on the photomultiplier tubes. Measurements are independent of source
fluctuations in the light source, as these affect both detectors equally. With a blank cuvette in position, R,, which is a 0.1% precision, calibrated, 10-turn, direct-reading potentiometer, is set to 100 and the galvanometer is adjusted to 0 by R. When an absorbing material is placed in the light path, the resulting galvanometer deflection may be read directly. For more precise measurement the galvanometer may be brought back to 0 by adjusting R,. The change of reading on R, is a measure of the transmittance of the sample. Although well compensated for other extraneous variations, the system shows small deviations from 0, which are ascribable to differences in the rate of fatigue and recovery between the two photomultipliers. These variations are a function of the anode current and amount to less than 1% in the present apparatus.
Figure 6. Concentration-absorption curve for bromothymol blue
PERFORMANCE
The performance of the instrument as a spectrophotometer is illustrated in Figure 5 . The curve represents the absorption spectrum of a piece of standard Corning KO. 5120 didymium glass filter 5 nim. thick, masked to gi\-e an optical path 0.5 mm. in diameter. Slit width used throughout was 50 microns, equivalent to a nominal band width of 20 A. a t 4000 A. and 70 A a t 7000 A . The instrument has proved particularly useful as a colorimeter for the quantitative analysis of colored materials in millimicrogram amounts. VOL. 30, NO. 1 1 , NOVEMBER 1958
1881
In Figure 6 the transmittance of bromothymol blue a t 6150 A. in pH 8.6 veronal buffer is plotted as a function of concentration. ACKNOWLEDGMENT
The authors are most grateful t o J. L. Oncley for his advice and encouragement, and to Frederick Gilchrist and Ross Schubarth for assistance in
fabrication of parts of the apparatus and electronic circuits. LITERATURE CITED
(1) Craig, R., Bartel, A,, Kirk, P. L., Rev. Sci. Instr. 24, 49-52 (1953). (2) Holter, H., L@vtrup,S., Compt. rend, trav. lab. Carlsberg, SBr. chim. 27,
27 (1949). (3) Krueglis, E. J., Ibid., 27, 203 (1950). (4) Oldenberg, O., Broida, H. P., J. Opt. SOC.Am. 40, 381 (1950).
(5) SR-ift, H., Rasch, E., in “Physical Techniques in Biological Research,” Oster, G., Pollister, A. W., eds., Vol. 111, pp. 353-400, Academic Press, New York, 1956. RECEIVED for review August 12, 1957. Accepted October 11, 1957. WORKsupported by grants from the Milton Fund of Harvard Universityand the U. S. Public Health Service (H-2440); and b a fellowship of the American Cancer Sbciety to D. F. H. W.
Determination of Trace Amounts of Total Nitrogen in Petroleu m Distillates by Adsorption Improved Modification SIR: Since publication of the original paper on “Determination of Trace Amounts of Total Kitrogen in Petroleum Distillates” [ANAL. CHEM. 29, 177 (1957)l by Bond and Harriz, several analysts have reported difficulty in controlling violent bumping during the Kjeldahl digestion, with occasional breakage of the flask. This difficulty has been investigated further and traced largely to the fact that some operators fail to cut the silica gel column into lengths of less than 1 inch (preferably 0.75 inch) as specified in the method. Longer sections tend to bridge across
the bottom curvature of the flask and bumping then occurs, whereas with the shorter lengths digestion proceeds smoothly. These observations have also led to a simplification of the procedure for blank determinations. Early attempts to perform a blank determination on the loose gel were unsuccessful because of bumping, necessitating the use of a magnetic stirrer. The gel for the blank is now placed in a regular adsorption column and then wetted with about 5 ml. of nitrogen-free iso-octane or other suitable hydrocarbon previously percolated
through silica gel. This technique completely eliminates the need for magnetic stirring during digestion. The usual blanks of 0.3 to 0.6 ml. of 0.01N acid are equivalent t o 0.1 to 0.3 p.p.m. of nitrogen for 500 ml. of the typical naphthas tested. The authors wish to thank John C. Tomlinson, of these laboratories, for his suggestions for improvements to the method. GEORGE R. BOND,JR. CLIFFORD G. HARRIZ Houdry Process Corp. Marcus Hook, Pa.
.
c
..I*
Volumetric Determination of Potassium in t-ertilizers Modifications and Comments SIR: Schall (2) recently described a volumetric method for determining potassium in fertilizers. The method consists of precipitation of potassium with sodium tetraphenylborate, removal of the insoluble salt, and titration of excess tetraphenylborate with standard quaternary ammonium salt solution, using bromophenol blue as indicator. The method, because of its speed and relative ease as compared with other potassium methods ( I ) , has stimulated considerable interest among fertilizer chemists; however, as with any new method, minor modifications such as indicator and strength of solutions warrant investigation. In our laboratory, 1882
ANALYTICAL CHEMISTRY
the following refinements and modifications have been made: Dilute tetraphenylborate solution to one half recommended strength. No dilution of quaternary ammonium salt is needed. Add 7-ml. excess of tetraphenylborate solution instead of 2 ml. Use a 5-ml. microburet instead of the recommended 10-ml. semimicroburet. Carry out precipitation in 100-ml. rather than 50-ml. volumetric flasks, with no increase in complexing reagents. Take a 50-ml. aliquot portion for titration. Where sample contains more than 50% KzO, use one-half sample weight (1.25 grams). This is in lieu of doubling
the flask size, which would be from 100 to 200 ml. for us, or from 50 to 100 ml. in the original procedure. Bromophenol blue, although an acceptable indicator, does not give an easily detectable end point. Fertilizer mixtures containing large amounts of natural organic materials and so-called minor or secondary elements often give dark-colored solutions in which the end point is partially obscured. An investigation of indicators has shown that a 0.04’% solution of Clayton Yellow (C.I. 813), also called Titan or Mimosa yellow, in distilled water gives a sharp color change from pale yellow to intense pink. In the presence of
