1784
ANALYTICAL CHEMISTRY
test'cd iii several sugar laboratorirs to determine reproducibility of readings between instrumentp. -1s an example of the precision nliich might be expect,ed from t.liis illstrumnit, the rcaults of preliminary test.s on one of t,he interim models of the cdorimeter are discussed. In these tests, 28 sets of white sugar samples were run o i t the colorimet.er, using the color method outlined. The -log t d u e s for earh wave length, together with calculated color index values for these samples, are summarized in Table V. The standard deviation of color index for these samples is +0.6 unit, which appears t.o be accurlito enough for routine color det,erminations. The color indes values from readings on the Beckman colorinieter are coinpared in Figurc. 2 with those measured on the same samples in the Beckman D L spect,rophotoineter. The colorimeter readings gerierally are in . good agreement with the color indes values obtained on the spectrophotometer, averaging only about 0.4 unit higher. Some sinal1 variations can lie espected hecause of differences in phototuhe sensitivity, in spectral band widths, etc. These preliminary tests indicate that t,his instrumeiit, niay hal-r the accuracy required for routine white sugar color detiwninations. Further testing on i,eproducibility of readiiigi: between final product,ion niodols is required, hovxvrr, before a f i l i a l decision can be made. SU3IMARY AND CONCLUSIONS
Considerable progress has been made in t,he standardization of inethode and instrunient,s for determiriat,ion of color of white sugar solutions. ; i survey of methods s h o w d the need for uniformity. .li st,andardized method has been accept,ed as a n interim procedure k)). the U.S. Sational Committee of the Intcrnntional Coniniiasion for Uniform Methods of Sugar Analysis. .'In iniprovd rvhite sugar colorimeter has also heen developrd xiid i- iion- !wiiig tested.
ACKNOB'LEDGMEST
The authors gratefully wish to acknowledge the assistance of the associate referees of the Subcommittee on Subject 13, "Color and Turbidity in Sugar Products," who have part,icipated in most of the work reported in this paper. They also wish to expreeappreciation to R. P. Connor of the Beckman Instrument Co., F. K . R a d i n g s and J. R. Johnson of the Amalgamated Sugar Co., and J. E. Doyle of the California and Hawaiian Sugar Refining Corp., for the assistance given in this investigative project. LITERPlTURE CITED
(1)
Broeg, C. B., and Walton, C. F., Jr..
;~s.ionnor the nbutyl alcohol should be mised with the sample until color formation is complete in the aqueous layer.
1
I
I
I
I
I
1
I
-25
PYRIDINE-PYRAZCLCNE In-BUTYL
;r
I l O m l n-BUTYL
450
475
O5 nn ""
t
I
I
I
I
I
2
3
4
5
6
7
525
550
575
BENZIDINE-PYRIDINE In-BUTYL
METHOD
1
I
\
bLCOHOL EXTRACT1
I
I
I
I
8 9 IO A D O I T I O N OF R E A G E N T S
SAMPLE pY BEFORE
11
Effect of pII on Color Formation
Figure 2.
pH adjusted with acetic acid or Sodium hydroxide
The data of Figure 3 show that in the presence of the buffers used, both reactions are not reliable. However, the presence of sodium acetate, as a result of neutralization of the caustic solution from the distillation step with acetic acid, does not have a detectable effect upon color production within the concentrations normally encountered. There appear to be other variahles affecting color production such as the pH of the solution during halogenation, but these variables have not been investigated.
ALCOHOLI
500
METHOD EXTRACT)
L
1
DYRIDINE-PYRAZOLONE, EXTRACTED COLOR
ALCOHOL
'C
1
-I
I
I\+ --/;
I
I
425
tained 60Cj, by volume of the Clark and Lubs buffer of cmrcsponding pH. Figure 2 shows that in unbuffered solutions, the pyridine-pyi azolone procedure is satisfactory for solutions of pH varying from 3 to 8 ; the benxidine-pyridine method shows results that ale slightly inferior for the same rang?. With the pyridinepyrazolone procedure the p H of the final reaction mixture was close to i on those samples within the region of maximum absorbance.
I
1 1 1
I \
600
625
650
675
WAVE LENGTH, mp
Figure 1. Ahsorption Curl es for Colorimetric >lethods 2 y of c,anidc in 25 ml. of water, final volume
BENZIDINE-PYRIWNE E X T R 4 C T E D n- B U T Y L A L C O H O L I
5lI l O m l
In using Epstein's procedure, it was found that the volume of mixed reagent could be varied within wide limits without affecting color development. I t is only necessary to assure that sufficient. reagent is present as a safety factor in case an excess of chloramine T is added. I n this study 5 ml. was found to be adequate. Results. Figure 1 presents the absorption curves produced by application of the two colorimetric methods to samples containing 2 y of cyanide as CX- in a final volume of 25 ml. The ileak absorbance for the benzidine-pyridine method a t 480 mfi is identical to that indicated by Susbaum and Skupeko (6). The absorption maxima for the pyridine-pyrazolone color shifted toward a longer wave length in changing from aqueous to eytracted absorbances. Extraction of the color resulted in an increase in sensitivity as xell as the avoidance of turbidity effeck At the same ratio of sample to solvent, the absorbance obtained in extracting the pyridine-pyrazolone color is almost double that of the benzidine pyridine. Since the peaks are broad, a small error in wave length setting or the use of filters will make little difference in the results obtained. Effect of pH on Color Production. The two colorimetric procedures were applied to solutions of variable p H and constant cyanide content. I n one series (Figure 2), initial pH values of the samples were obtained by addition of sodium hydroxide or acetic acid. In the second series (Figure 3), the solution con-
e
I I
-
'
2
PYRIDINE-PYRAZOLONE lAaUEOUS 2 5 m l VOLUME1
155):pH
Figure 3.
I
-
b F T E R b D D l T l O N OF R E b G E N T S
OF b Q U E o U S M I X T U R E
I
~' \
I
I
I
I
I
Effect of pH on Color Development
15 ml. of Clark and Lubs buffer used
If visual color comparisons were made on the pyridine-pyrazolone method, much closer control of the pH would be necessary An apparent peak in color due to a change in hue occurs a t p H values b e h e e n 4 and 5 This effect is not evident in the absorbance data. From this experiment, it was concluded that direct color application on a highly buffered sample ha3 little significance unlev both samples and standard? have comparable buffer char-
1786
ANALYTICAL CHEMISTRY
acteristics. Distillation appears to be the easiest means to equalize this effect. Effect of Time. The effect of time upon color development is shown in Figure 4. All readings are based on 25-ml. samples containing 2 y of cyanide. The aqueous pyridine-pyrazolone color required 15 minutes to reach a maximum. Since more dilute solutions generally required slightly more time, 20 minutes was accepted as the minimum time for color development. The benzidine-pyridine color development was a t or near the maximum in 5 minutes. For practical purposes any time after 10 minutes could be used for reading the absorbance. Both colors in aqueous media faded a t a steady rate after maximum absorbance was reached. However, it would be possible to read after longer time intervals than those suggested provided a blank and standard were included with each series. The extracted color for pyridinepyrazolone showed no fading in over 3 hours.
PYRIDINE -PYRAZOLGNE
10 EO 30 40
60
80
I 100
TIME I Y M N V E S AFTER
Figure 4.
1 -
EXTRACTED
I
I
120
140
AQUEOUS I 160
I
I
I80
ZOO
REAGENT ADDITION
Effect of Time on Color Development
Reagent Stability. A series of calibration curves for the benzidine-pyridine method is presented in Figure 5 . Good linearity was evident with each reagent, but substantial difference in slope was apparent. The advantage of fresh reagent can be seen from the difference in slope of the two curves, 2 and 2A. The first was based on fresh reagent, the second qn the same reagent 1 hour later. As the reaction shows such variable color production quantities for a given increment of cyanide, a blank and standards must be set up with each series of samples. No effect on absorbance was observed when the color was produced in the dark or in a sunlit portion of the laboratory; therefore, color development was allowed under average lighting conditions for all samples. Reagent blanks generally did not increase in color within the normal time interval, and were fairly uniform from one reagent to another, although a small amount of color was produced in each. Figure 6 includes essentially the same type of data for the pyridine-pyrazolone method. One of the reagents was prepared from recrystallized pyrazolone and the others were prepared with stock Eastman 3-methyl-1-phenyl-5-pyrazolone.Within experimental error these curves are equivalent. Sensitivity of the aqueous color was approximately 0.5 y as cyanide. When 25-ml. samples are used, quantitative results are possible in the presence of 1 to 5 y of cyanide. The mixed reagent was prepared daily because one-day-old reagent showed a higher blank absorbance, while the color produced for a given increment of cyanide was less. The extraction of the pyridine-pyrazolone color with the solvent sample ratio used (25 ml. of sample to 10 ml. of solvent) Increased the sensitivity to 0.1 y of cyanide. The effective range for quantitative purposes included from 0.2 to 2.0 y of cyanide ion for the volume indicated.
'
1
hl
-
* # 2
02
I 75
I 50
25
I IC0
Figure 5. Calibration Curves for Benzidine-Pyridine Method 2A.
BENZIDINE -PYRIDINE
I
/o
Color developed on curve 2 reagent 1 hour after mixing. All others were fresh preparations
... . . . . PYRIDINE - PYRAZOLONE EXTRACTED
I
I
[Bispyrazolone, heretofore prepared in the individual laboratory according to Epstein's inst,ructions, can now be obtained from Distillation Products Industries, Rochester 3, X, Y., listed as chemical S o . 6969, 3,3'-dimethyl-l,1 '-diphmgl-(l,i'-bi-2-pyrazolene)-5,5'-dione. ] Precision. Statistical data for the three procedures are given in Table I . The aqueous pyridine-pgrazolone series indicated a high degree of precision. Blthough precision of the extracted pyridine-pyrazolone procedure was inferior, this method has a definite advantage in addition to improved sensitivity bemuse on turbid samples of river u.at,er the estracted series showed no loss of precision, while the aqueous series had appreciably greater variation. The benzidine-pyridine test has relatively poor reproducibility, and in most cases in it series of samples at least one exhibited abnormal behavior. This is reflected in a standard deviation of more than three times that obtained in the estracted pyridine-pyrazolone method. In both colorimetric procedures for this series, the reagent was mised during the halogenation period of the samples. An adequate quantity was prepared for the 10 replicat,es and was mised with the sample within 5 minutes. 12,
I
I
I
I
Y
I
I
-
-
IO
I
-
084OJEOUS
W U
COLOR
REID
z
. l d
10 '
m
-
n n " V
0
I 1
Figure 6.
I 2
I 3
I 4
I 5
I
' I 6
7
8
Calibration Curve for PyridinePyrazolone Method
Range of absorbanoes obtained with four different pyridinepyrazolone reagents, one with recrystallized pyrazolone
A4ttemptsto locate the source of the erratic values obtained in the benzidine-pyridine method were not very successful. On industrial wastes containing a high concentration of organic matter and showing a large and variable bromine demand, it was difficult to brominate the sample and still more difficult to
1787
V O L U M E 2 6 , N O . 11, N O V E M B E R 1 9 5 4 Table I.
Statistical Data on Cyanide Color Testsa Ahnorhanre . - ~~ ~_ . ~ __ ~~
~
Aqueous Pyridinepyridinepyrazolone pyrazolone extracted 0.340 0.920 0.348 0,860 0,880 0 360 0.926 0.346 0.935 0.348 0.938 0.338 0.958 0.350 0.880 0.352 0.960 0.348 0.950 0.348 hfean ( 2 ) C N - . y 2 000 2 000 Standard deviation (s) CS-.y 0.035 0.078 Coefficient of variation ( C V ) . '7 1 7 3.9 [loo ( S / f ) = C V ] 2 y uf cyanide in a final aqueous volume of 25 rnl
Benzidinepyridine extracted 0.290 o 386 0 480 0 450 0 434 0 416 0 452 0 412 0 442 0 456 2,000 0 2.32 12 6
aeration schedule for the C and D seriea were as nearly identical as possible, yet the cyanide removal varied within wide limits. The more complete removal of cyanide in those samples for the pyridine-pyrazolone color method was strictly fortuitous as other samples of sewage, distilled water, and wastes indicated no pattern other than incomplete and variable removal of cyanide. Buffering and complexing action in the sample have a significant effect upon removalof cyanide by aeration, therefore the procedure cannot be recommended except where rough checks will suffice. Cyanate interference is relatively slight. As indicated in Table 111, 10 mg. of potassium cyanate produced a color equivalent to 1 y or less of cyanide with either of the colorimetric methods. I n coke waste cyanide determination, cyanide recovery shon-ed no measurable difference, with or without 500 p.p.ni. of added cyanate or thiocyanate (total thiocyanate, 725 p.p.m.). Within the range indicated, distillation effectively separated both of these ions from cyanide. Interference from Amino Acids. There has been some indication that these two methods were subject to interference by various organic nitrogen compounds. Glycine, valine, alanine, cystine, and urea were used with each method in amounts varying from a trace to 10 mg. in the reaction tube. Both methods were comparable in the degree of interference encountered. Glycine was the only amino acid tried which produced an effect greater than the experimental error of the test. The interference due to 1 mg. of glycine resulted in a color equivalent of 0.5 y of cyanide as CS-. Thr presence or absence of added cyanide made no detectable difference in the degree of interference encountered. Urea produced a small positive interference that was insufficient to estimate quantitatively rn ith as much as 10 nig. per reaction tube A4sindicated in Table 111, distillation eliminated the positive interference due to glycine or urea. Other organic compounds which contain or produce the cyanide ion under the analytical conditions employed may result in a positive interference. There is no known general test which will show the magnitude of this effect, hence the presence of organic nitrogen introduces an uncertainty into the analysis for cyanide.
remove t,he excess bromine in a consistent nianner. Color or turbidity in the sample int,ensified this difficulty. The analyst was never sure that the sample was a d e q ~ a t ~ e lbyr o n h a t e d or that the excess had been neutralized. Interference from Cyanate and Thiocyanate. Colorimetric methods are subject to interference by both cyanate and thiocyanate. Thiocyanate interference is great enough to preclude the direct determination of cyanide by both methods on samples such as coke plant wastes, where the concentration of thiocyanate may be more than 100 times that of cyanide. The effect of thiocyanate on t.he pyridine-pyrazolone met.hod did not present a consistent pattern. In Table I1 thiocyanate color production was fairly consistent, but a t times thiocyanate produced very little color. The time of reaction in the halogenation st.age is not the entire answer nor is the pH of the react'ion media, although a high pH tends toivard the formation of less color due to thiocyanate. Unfortunately, the pH range where thiocyanate color decreases is nearly the same as that where cyanide color diminishes, hence this means of control would be too crit,ical for routine use. Epstein ( 3 )suggested the use of frrric iron to catalyzc the reaction of thiocyanate and the pyridine-pyrazolone reagent. Experimental attempts to use ferric iron as a catalyst Table 11. Results with Nusbaum and Skupeko's Acidification and Aeration failed to produce an increase in differfor Separation of Cyanide from Thiocyanaten ential color between the blank and a Benzidine-Pyridine Pyridine-Pyrazolone standard containing thiocyanate when z of z of Absorbseparate AbaorbPeparate both contained ferric iron. ance a t combance a t comPainyle 480 nip Av ponents Table I1 indicates the performance of Composition 630 mir Av. ponents 1 the colorimetric methods when applied 0 247 0.240 Cyanide, 1 y 0.355 0.357 0 233 0.359 to Susbaum and Skupeko's (6) proce2 0 268 0 257 Thiocyanate, 2 y 0.202 0.210 0 246 0.218 dure for the determination of combined 3 Cyanide, 1 y + Thiocyanide and thiocyanate This proce0 450 0.465 0.497 cyanate, 2 y 0.603 0.598 0.567 0.480 0.592 dure measures both constituents; hence it 0.240 0.238 .4i S o . 1 + boric acid 0.219 0.235 was proposed to make one determination 0.237 0,250 .4 a 0.260 0.262 S o . 2 + boric acid 0.134 0.140 for the total quantity of the two compo0.263 0.158 0.488 0.483 0.500 S O3 + boric acid ncnts and a seconddeterminationafter re.4 3 0.419 0 401 0.381 0.480 0.382 moval of cyanide by acidification and S o . 1 + phosphoric acid 0.411 0.398 0.210 0.220 BI aeration of the sample which then should 0.230 0.386 0.226 0.233 S o . 2 + phosphoric acid 0.391 0.400 B? in,.lude only the thiocyanate. Both boric 0.240 0,410 0.450 and phosphoric acids were recommended 1.0,3 + phosphoric acid 0.796 0.782 0.798 0.450 0.453 B3 0.450 0.769 as acidifying agents depending upon 0.125 0 126 c i Ai aerationh 0.015 0.023 sample type and purpose of the analysis. 0.126 0.031 C'? 0.255 0.263 A? T aerationb 0.239 0.267 Table I1 includes the results of the two 0.270 0.295 0.233 0,233 0.389 color methods on cyanide and thioC3 .id + aerationb 0.341 0.310 0.290 Off cyanate in separate and combined form, 0.278 color with and without acids and aeration In0 046 0 041 D1 BI aerations 0 0 0.037 0 spection of the data does not indicate 0 209 0.214 D2 B2 t aerationb 0.364 0.343 Q 218 0.321 that this technique viill give dependable 0.380 0.315 0 255 B3 I aerationb 0.319 0.339 0.343 D3 rewlts using either color procedurr. The 0.250 0.358 " All samples adjusted t o 26-ml. aqueous volume then extracted with 10 ml. of n-butyl alcohol betwo components in separate form did not fore reading All samples taken from three stock s h u t i o n s of Ohio River water plus reagent. check with the combined samples in sevb 10 minutes of aeration a t 90' C. eral instances. The sample tubes and
-
-
ANALYTICAL CHEMISTRY
1788
Results. Direct titration of river water and other relatively clear samples was effective, proDirert Direct Titration Titration after vided sulfides and heavy metals Pyridine BenzidineDirect after Serfass Tartaric Arid hIaxiinum Pyrazolone, Pyridine, Titration, Reflux, Distillation, were absent. C o m p l e x i n g Theoretical Conipound hIg Jlg. hlg. JIg 1 Rlg. CN -, 31g.h materials generally caused lon. Potassium zinc cyaresults (Table 111). nide 23.0C 31 16.3 16 70 10.69 10 8 Potassium cadmium Table I V includes the statis14.8c cyanide 16 4 c 12.3 12.45 12.35 13 Fc Potassium silver cvatical parameters of combined 11.7C 18 4 c nide 0.02 10.30 10 20 10 5 d i s t i l l a t i o n and titration of Potassium nickel cya2.55 2 43 nide 11 77 0.10 11 57 12 2 Ohio River water with added Potassium ierro0,003 cyanide 0 003 2.01 0.001 p o t a s s i u m cyanide. While 1 96 2 0 Potassium ferrierrors in the final result include 0,002 0.002 cyanide 1.11 0.001 1 12 1 23 1.57 2.68 Copper(I)-copper(II) 0.82 6.XO t i . 02 10 8 those due to tartaric acid distilcyanide lat’ion as well as to titvation, a Copper!I)-copper(II), cyanide, ammonlcoefficient of variation of 2.0% 0.08 n . 15 ated color 8.40 3.71 11.2 masked was obtained. I n an earlier end point study (6) the same standard Potassium cobalticyanide 0.15 0 li 0 08 2.02 0.68 10.8 d e v i a t i o n w a s obtained on .. .. 7.20d ... 10.8 ... 9.718 16.8 .., direct titration of Great Miami 0 . .50 nil Potassium thiocyanate variable nili nil/ 1 nu River water. The titration of 0.001 Potassium cyanate 0.001 nil { nil/ nil 10.00 Glycine nili nil 0.0005 0.0005 nil/ 1.09 cyanide with silver nitrate in nil nil/ Urea trace trace 1o.ov nil/ this manner results in excellent a 1-hour reflux except where indicated. b Calculated from the assumed chemical formula. precision as long as the cyanide C Quite variable. Value given is that calculated from tlir lowest dilution on which the absorbance could he concentration is above the recdeterniined. d 7-hour re0ux ommended 1 p.p.m.-for ex* 28-hour reflux. f 500 p , p , m , added t o the distillation flask. ample, it was the authors’ exa LIilligrams of interfering substance per color reaction tube perience that a t 0.4 p.p.m. cyanide, the standard deviation was approsiniately four times DETERMINATION OF CYANIDE BY SILVER NITRATE that a t x conrentration of cyanide above 1 p.p,m. TITRATION Khen the sample contains free fatty acids the end Doint will The Liebig titration as modified by Ruchhoft et al. ( 7 ) and by be masked by the high concentration of soap formed in the misSei,fass et a2. (9) continues to be one of the basic procedures for ture. If the soap concentration is moderate, t,he end point can estimating cyanide. Several methods are in use for determining be detected by allowing the froth t,o rise to the top after addition the end point, including the turbidit,y of the silver cyanide comof reagents and mixing. F a t t y acids are not separated from cyples, turbidity and color of the complex in t,he presence of iodide anide by dist.illation, hence extraction mag be necessary to avoid ion, and the rhodanine indicator of Ryan and Culshaw ( 8 ) . difficulties due to a high concentration of such acids in the sample. Table 111. Results of Interference and Complexed Cyanides with Different Analytical Procedures
? C
SELECTION O F MEASUREMENT TECHNIQUE
Table IV. Statistical Parameters of Combined Distillation and Titration of Ohio River Water with .4dded Cyanide Direct Titration, M g C S Distilled, Mg. (2s2.08
2.05
2.03 2.03 2. = 2.05
i
a
cv
Procedure.
APPARATUS.
1.99 1.96 1.98 1.99 2.00 1.98 2.05 2.05 2.09 1.98 2.01 = 2.01 = 0.04 = 2.0%
I
