vided better separations among mono-, di-, and tri-substituted amines. Copper fixed the lower members of the mono-substituted amines, while silver did not, in agreement with the work of Bjerrum (1, f?),who showed that the first stability constant for copper(I1) was larger than that for silver. Bjerrum also showed that the constants for these two elements are generally larger than that for cobalt(I1). In Table I11 the RL values indicate that cobalt forms more stable complexes than silver. This anomaly may be due in part to the fact that no determination was made of the upper limits for the amounts of sorbate that could be used without saturating the sorbent on the first part of the column, or the presence of different amounts of chemically reactive substances on the impregnated sorbents. It should be possible to correlate values of RL with stability constants either for the first coordination step or the over-all complex involving each amine in a series in combination with a single metal cation. The constants for silver-amine complexes in benzene differ not only in magnitude, but possibly also in order from those in water. Thus the stability values for the mono-alkyl amines do not differ greatly in magnitude in water ( I ) , whereas the RL values in benzene for m-propyland n-butylamines on silver nitrate were different. More complete studies of such systems as those in Tables VI1 and VI11 would be of both theoretical and practical interest. Evaluation of the following equilibria is required: sorption of the salt used for impregnating the sorbent (or,
in the case of a pure salt, its solubilit.y), stability of the chemical entity formed by the salt and the sorbate, competitive equilibria involving the solvent, and rates of the reactions, particularly those involving reactions between salt and sorbate. LITERATURE CITED
(1) . . Bjerrum,. J.,, Chem.
Revs. 46. 381
(1950). (2) Bjerrum, J., “Metal Amine Formation in A ueous Solution,” P. Haase and Son, &openhagen, 1941. (3) Brunauer, S., Emmett, P. H., J. Am. Chem. SOC.59, 2682 (1937). (4) Cassidy, H. G., “Adsorption and Chromatography,” pp. 178-80, Interscience, Xew York, 1951. (5) Deitz, V. R., Ann. N . Y . Acad. Sci. 49. 315 11948). - ~ (6, DeSesa, hl. A., Rogers, L. B., Anal. Chim. Acta 6,534 (1952). (7) . . DeVault, D., J . Am. Chem. SOC. 65, 532 (1943). ‘ (8) Dickey, F. H., J . Phys. Chem. 59, 695 il9.55). (9) Erlenmeyer, H., Dahn, H., Helv. Chim. Acta 22, 1369 (1939). (10) Feigl, F., “Quantitative Analysis by Spot Tests,” p. 307, Nordemann, New York, 1939. (11) Ibdd., p. 317. (12) Forstner, J. L., Ph.D. thesis, Massachusetts Institute of Technology, 1952. (13) Gapon, E. K.,Belen’kaya, I. M., Kolloid. Zhur. 14, 323 (1952). (14) Gliickauf, E., Nature 156, 748 (1945). (15) Ibid., 160, 301 (1947). 116) Hildebrand. J. H.. Rotariu.‘ G. J.. ANAL.CHEW24, 770 (1952). (17) James, T. H., Vanselow, W., J . Am. Chem. SOC.73, 5617 (1951); 74, 2374 (1952). (18) Jung, W.,Cardini, C. E., Fuksman, hl., Anales asoc. qufm. arg. 31, 122 (1943). (19) Jura, J., Grota, L. Hildebrand, J. H., 118th Meeting, AkS, Chicago, Ill., September 1950. I
\
\--
- I
- - - I
~
(20) Khym, J. X., Zill, L. P., J . Ani. Chem. SOC.74, 2090 (1952). (21) Langmuir, I., Phys. Rev. 6 , i 9 (1915). (22) LeRosen, A. L., Vonaghan, P. H., Rivet, C. A., Smith, E. D., ANAL. CHEM.22,809 (1950). (23) hlartell, A. E., Calvin, &I.,“Chemistry of Metal Chelate Compounds,” Appendix I, Prentice-Hall, Xew York, 1952. (24) Mayr, C., Prodinger, W.,Z. anal. Chem. 117, 334 (1939). (25) O’Connor, D. J., Bryant, F., Nature 170,84 (1952). (26) Pauling, L., “Sature of the Chemical Bond,” Cornel1 University Press, Ithaca, N. Y., 1940. (27) Reichl, E. R., Loffler, J. E., Mikrochim. Acta 1954, 226. (28) Robinson, G., Discussions Faraday SOC.7. 195 11949). (29) Russell, A. ’S.,Cochran, C. K., Ind. Ens. Chem. 42, 1336 (1950). (30) Sacconi, L., Dzscussions Faraday SOC.7, 173 (1949). 1311 Samuelson. 0.. Z. Elektrochem. 57. 207 11953). (32) Smith,’ E. D., LeRosen, A. L., ANAL. CHEM. 23, 732 (1951). (33) Strain, H. H., “Chromatographic Adsorption Analysis,” p. 93, Interscience, New York, 19+2. (34) Rielund, T., Fischer, E., Naturwissenscha ten 35, 29 (1948). 1351 Willardf H. H.. Merritt. L. L.. Jr.. ’ Dean, J. A,,“Ins&umental’Metho’ds of Analysis,” p. 140, Van Kostrand, New York, 1948. I
,
RECEIVED for reviea December 19, 1957. Accepted October 13, 1958. Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Pittsburgh, Pa., March 1953. Taken from a thesis submitted by James L. Forstner in partial fulfillment of the requirements for the degree of doctor of philosophy, Massachusetts Institute of Technology, Cambridge, Mass., December 1952. Work supported in part by the Atomic Energy Commission under contract .;1T(30-1)-905
Improved Spent Sulfite Liquor Determination by the Nitrosolignin Method OTTO GOLDSCHMID and L. FRANK MARANVILLE Olympic Research Division, Rayonier lnc., Shelton, Wash.
b The Pearl and Benson determination of spent sulfite liquor with a photoelectric colorimeter has been improved by a filter combination which gives more nearly monochromatic light of about 430 mp, This eliminates the undesirable light absorption due to the nitrite reagent and improves both sensitivity and reproducibility. Interference of certain materials such as water-soluble extractives from bark and wood can b e reduced but not eliminated by the use of an alkaline blank solution of the water sample of 370
ANALYTICAL CHEMISTRY
the same pH as the reagent-treated sample. Agreement or disagreement between the results obtained using both a neutral and the alkaline blank indicates the probable absence or presence of substances other than spent sulfite liquor.
T
HE determination of spent sulfite liquor (SSL) in sea water by the nitrosolignin color reaction was originally developed by Pearl and Benson (5) as a visual colorimetric method. Nitrite and acetic acid were added to
the water sample in a Sessler tube; after addition of ammonia to enhance the color by conversion of the nitrosolignin formed into the quinoidal salt form, the color was compared visually with that of similarly treated known spent liquor dilutions or platinumcobalt standard solutions. ilnderson ( I ) , who investigated the effects of reagent concentration, pH, and temperature on the rate of nitrosolignin color formation in detail and established optimum conditions for the formation of a stable, reproducible
color, determined color by a photoelectric colorimeter. He eliminated the interference due to the natural color of the water by subtracting a blank, determined on the untreated water sample, in addition to the reagent blank, from the measured nitroso color. Honever, because the nitroso color reaction is not specific for lignin but is given by phenolic materials in general, the method must be calibrated with spent liquor dilutions prepared with water containing the color-forming material typical of the area. IllcKernan, Tartar, and Tollefson (4) adapted the method, essentially as modified by Anderson, for use iyith the Lumetron photoelectric colorimeter (Model 402-E, Photovolt Corp.) equipped nith Lumetron filters 11420 and B420. To eliminate the natural water color, they measured the light transmittance of the reagenttreated water sample in per cent of that of an untreated portion of the same sample, the colorimeter being set a t 100% transmittance with the untreated sample. However, the reagent blank (mainly due to nitrite) reduced the light transmitted by a sample not containing any spent liquor to about 80%. This reagent blank \\as considered constant and was included in the calibration curve. The Lumetron colorimeter is eminently suitable for routine determinations of extremely ion- spent sulfite liquor concentrations by this method because its construction andoperation are simple, and because the 15-cm. path length of the sample cells provides an increase in absorbance readings of 50Yc over instruments with 10-cm. cells. After a spectrophotometric study of the nitroso color of lignin and of certain phenolic materials that mag be present in water samples and may interfere with the determination, an improved combination of light filters for the Lumetron was selected. The filter combination eliminates the reagent blank almost completely, and improves both sensitivity and reproducibility. A modification of the method indicates clearly whether or not certain interfering substances are present in the water sample. METHOD
This method differs from that of McKernan et al. (4)in the type of light filters used with the Lumetron colorimeter, and in the fact that in Method B a portion of the sample made alkaline by the addition of acetic acid and ammonia is used as a blank. Apparatus. Lumetron photoelectric colorimeter (Model 402-E, Photovolt Corp.) equipped n i t h a n absorbance dial; two 15-cm. precision-type cylindrical absorption cells marked 1 and 2; adapter for cylindrical cells; light filters Nos. 5113 and 3389 (Corning Glass Works), color speci-
\
\ \
h
i 4
n 0 C 0 C
0
3 Figure 1 . Absorbance spectrum of nitrosolignin Spent liquor solids concentration 0.256 gram per liter, 1 -em. cells 1. Reagent-treated sample 2. Reagent blank 3. Alkaline blank A. Nitroso color 5. Neutral blank
fication S o s . 5-55 and 3-73, respectively, polished surfaces, 2 inches square. Reagents. 10% sodium nitrite C.P. solution, made up fresh before each series of tests; l o % acetic acid; 2AT ammonium hydroxide. Procedure. Collect a 500- t o GOOml. sample of water and unless perfectly clear, let stand overnight to allow all suspended material to settle. However, the test should be completed within 24 hours after sampling. Decant 450 ml. of the sample into a GOO-ml. beaker without disturbing any settled material. If decanted sample is not perfectly clear, stir with a glass rod to ensure a uniform sample. Into each of three 250-ml. beakers measure a 150-ml. portion of the sample. By graduated pipets, add to the first beaker 3 nil. of 10% sodium nitrite solution and 3 ml. of 10% acetic acid, and stir thoroughly; let stand 15 minutes, add 6 ml. of 2N ammoniuni hydroxide, and stir again. To the second beaker add 3 ml. of 10% acetic acid and 6 ml. of 2-47 ammonium hydroxide, and stir (alkaline blank), The third beaker contains the neutral blank. METHOD A. Rinse cell 1 with a small portion of the neutral blank. then fill with this solution. Similarly, rinse and fill cell 2 with the reagent-treated sample. Place cell 1 in Lumetron cell compartment and zero the instrument. Replace cell 1 nith cell 2, rebalance the instrument, and read absorbance. METHODB. Emptv cell 1, rinse Lvith small portion of alkaline blank, and fill cell ith this solution. Zero the
lo. Figure 2. Light transmittance of Lumetron filters
+ B420 + Wratten 2A Corning 3 3 8 9 + 51 1 3
1.
M420
2. 3.
MA20
A. 5.
Nitrosolignin color Nitrite reagent in 15-cm. cell
instrument with the alkaline blank solution in cell compartment, replace cell 1 with cell 2, rebalance the instrument, and read absorbance. Calibration. Adjust a representative sample of spent sulfite liquor to exactly l o yo total solids content. Prepare a series of dilutions containing 3, 5, 10, 20, etc., p.p.m. (by volume) of the adjusted sample in spent liquor-free, filtered sea water. For each dilution determine absorbance by both Methods A and B as outlined under Procedure. For each method, plot absorbance us. concentration of the dilutions in parts per million of spent sulfite liquor of 10% total solids content. From the best straight line or by the least squares method, determine the slope (absorbancejparts per million) for Methods =1 and B, respectively. The factors for converting absorbance values to spent liquor concentrations are the reciprocals of the slopes. Reagent Blank. Water from different sources, and even distilled water, gives the nitroso color reaction to a different, and often appreciable, degree. The reagent blank must be determined so that the absorbance of the nitroso color of the particular 17-ater used is cancelled out. If the absorbance of a solution containing a n excess of all the reagents is determined against another solution containing the normal amount of the reagents, the nitroso color of the water forms in both solutions and only the absorbance of the excess of the VOL. 31, NO. 3, MARCH 1959
371
Table I.
Solution
SampleQ, MI.
A
50
B
C
D
Composition of Solutions
HzO, MI.
..
10% CHICOOH, MI.
NaNOz, M1.
1
1 1
..
50
..
..
1 1
50
1
50
10%
E 50 1 a Spent sulfite liquor, 0.256 gram per liter total solids. Added 15 minutes after nitrite addition, c Figure 1.
reagents is determined. The reagent blank is so small that a fourfold excess of the reagents is advantageous. To 150 ml. of the water used for making up the reagents, add 3 ml. of 10% sodium nitrite solution, 3 ml. of 10% acetic acid, and after 15 minutes, 6 ml. of 2N ammonium hydroxide, To a 114-ml. port:on of the same water, add 12 ml. of 1Oy0sodium nitrite solution, 12 ml. of 10% acetic acid, and after 15 minutes, 24 ml. of 2N ammonium hydroxide. Rinse and fill cells 1 and 2, respectively, with the two solutions, and determine the absorbance of the second solution us. that of the first solution as described under Procedure, Divide the absorbance by three to obtain the reagent blank for Method A. Obtain reagent blank for Method B in the same manner, but make up the second solution as follows: To a 141-ml. portion of the same mater, add 12 ml. of 10% sodium nitrite solution, 3 ml. of 10% acetic acid, and after 15 minutes, 6 ml. of 2N ammonium hydroxide. Calculation. The apparent spent sulfite liquor concentration in parts per million of liquor of 10% total solids content is calculated as follows: Parts per million = ( A
- B) X F
where sl is the absorbance, B the reagent blank reading, and F the calibration factor. The calculation is carried out with the appropriate values from both Methods A and B. In the Pacific Xorthrvest results of this determination are customarily expressed in parts per million of a hypothetical digester strength spent liquor of 10% total solids (6). It is hoped that a more rational basis will be adopted in the future, because both the solids content of the liquor and the composition of the solids vary from mill t o mill and depend on the grade of pulp produced. Some of the experiments in the following section were carried out with actual digester strength liquors of different total solids contents. EXPERIMENTAL
Spectral Characteristics of Nitrosolignin Color. The absorbance spectrum of the nitrosolignin color (Figure 1, curve 4) was obtained with a Cary 372
ANALYTICAL CHEMISTRY
..
2N
",OH, MI. 2b 2b 2 2
..
Curves No. 1 2 3
,. 5
3Iodel 11 recording spectrophotometer. A sample of spent sulfite liquor previously adjusted to lOyo total solids content was diluted 1 to 400 with distilled water. Five solutions, A to E, mere prepared (Table I). The nitrosolignin spectrum (curve 4) was obtained by placing solutions A and D,in separate 1-cm. cells, in the sample beam of the spectrophotometer, and solutions B and C, in separate l-em. cells, in the reference beam. This procedure correctly eliminates all but the absorbance due t o the nitroso color (3). If solution E (neutral blank) had been used in place of solution C, which has a pH of about 9.0, a somewhat higher absorbance would have resulted for curve 4, because the difference between the alkaline and the neutral blank solutions would have been added to the nitrosolignin spectrum, Figure 1 shows that the maximum of the nitrosolignin spectrum occurs at 430 mp; the reagent blank absorbs strongly a t wave lengths below 400 mp and above 420 mp its absorbance is very low; spent sulfite liquor solutions show an indicator effect, their absorbance being somewhat higher in alkaline than in neutral solution. Spectral Characteristics of Lumetron Filters. Transmittance curves for various light-filter combinations for the Lumetron colorimeter determined with the Cary spectrophotometer are shown in Figure 2, together with the nitrosolignin spectrum (Figure 1, curve 4), and the spectrum of the nitrite reagent blank in a 15em. cell (curve 5 ) . The latter curve was determined in a 10-em. cell by increasing the concentration of the reagents by a factor of 1.5. Maximum light transmittance of the Lumetron ?11420-B420 filter combination (curve 1) occurs a t 407 mp, a considerably Ion-er wave length than the maximum of the nitrosolignin spectrum a t 430 nip. hIoreover, this filter combination transmits appreciably at wave lengths below 400 mp where nitrite absorbs strongly. This causes the high reagent blank reading in the Lumetron method of McKernan et al. (4) and was corrected by replacing the Lumetron B420 filter with a Wratten 2A filter which has a sharp cutoff a t
about 410 mp. The filter combination Lumetron M42O-Wratten 2A (curve 2) has a transmittance maximum a t 423 mp and a much narrower band width than curve 1. Because the Wratten 2 h filter is no longer available, an equivalent or better filter resulted from the combination of Corning glass filters Nos. 3389 and 5113 (curve 3), which has a transmittance maximum a t 430 mp, a short wave length cutoff (0.1% transmittance) a t 414 mp, and which transmits almost exactly twice as much light as the Lumetron M420-Wratten 2 4 combination. Spectral characteristics for the three filter combinations are summarized in Table 11. Lumetron Performance with Different Filter Combinations. T o establish the effect of the different filter combinations on the spent liquor determination with the Lumetion colorimeter, dilutions containing 5 , 10, 20, and 40 p.p.m. of a sample of spent sulfite liquor (13.92% total solids) were prepared in sea water (Hood Canal) free of spent liquor. Solutions for the nitroso color determination were prepared as described under Procedure. Absorbance m-as measured by both Method A (neutial blank) and Method B (alkaline blank', using the h1420-B420 and the M420Wratten 2A filter combinations. Results of these determinations (Figure 3) show that by limiting t h e short wave length transmittance, the modified filter combination (M420-2.4) decreases the intercept of the absorbance-concentration plot by a factor of
IO 20 30 40 SSL CONCENTRATION, ppm.
Figure 3. Effect Lumetron readings
of light filters on
P.p.m. of spent sulfite liquor of 13.92% total
solids Method A. Method 8.
Neutral blank Alkaline blank
about 8, from 0.125 to 0.016. This intercept represents the absorbance of the reagents plus any nitroso color of the water used in making the dilutions. With the modified filter combination, the absorbance of the reagents in 15 em. cells is less than 0.005. Thus, part of the intercept obtained with the modified filters is probably due to the nitroso color of the dilution water. The modifled filter combination improves the sensitivity of the determination by increasing the slope of the absorbance-concentration lines. Table I11 summarizes slopes (s) and intercepts (i) obtained by the method of least squares for the equation of the straight lines of Figure 3: Absorbance
Filters
Filter M420 h1420 3389 Combination B420 2A 5113 Transmittance maximum Wave length, mp 407 423 430 Transmittance,
7%
Band width a t T,,,, rnw Cutoff, 0.1% T Low wave length, mp High wave length, mp
= s X (parts per million spent sulfite liquor) i
+
The table also contains similar data comparing the filter combinations -344202A and Corning 3389-5113 obtained with dilutions of a different sample of spent liquor. Precision. T o determine the precision of the Lumetron method in the range of low spent liquor concentrations, five replicate determinations were carried out on each of three dilutions (1, 5, and 10 p.p.m.) of a sample of spent liquor in distilled water. The determinations included the whole procedure starting with the reagent addition. Individual absorbance values were converted into parts per million using a n equation obtained by the method of least squares
II. Spectral Data for Light
Table
Table 111.
Filter Combination >1420-B420' L1420-2Aa hZi20-2Ab 3389-51 13* a
10 3
6.0
23.8
42
28
28
363
406
414
461
466
480
Interference. Pearl and Benson (5) obtained a slight coloration from extracted tannins or phenolic compounds when they tested an infusion of sawdust and bark. Such compounds may be present in water samples either from natural sources or as the result of log storage in sea water or other industrial operations with wood. Spectral characteristics of the nitroso colors of Western hemlock bark extract and a Douglas fir wood hydrolysis liquor were investigated. Western hemlock bark was hammermilled, steeped overnight in distilled water at room temperature, and the filtered extract (0,46y0 total solids)
Slope (s) and lntsrcept ( i ) with Different Filter Combinations
Method I3
Method A s X lo3
2
5 85 6 13 4.83 5.06
0.122 0.016 0.012 0.014
x
s
5 5 4 4
103 31 84 52 80
i 0 125 0 017 0.012 0 013
Kith spent sulfite liquor of 13.92% total solids. 10.Oyototal solids.
* With diflerent spent sulfite liquor of
from the average absorbance values at the three nominal dilutions (Table IV). Three-sigma confidence limits for a single determination, calculated from the standard deviation of about 5% in the lorn-est concentration range, are kO.15 and +0.75 p.p.m., a t 1 and 5 p.p.m., respectively. Thus, the Lumetron method is capable of a fairly high precision, even at extremely low spent liquor concentrations. The same degree of precision is not always obtainable with sea xater samples, probably because of turbidity effects. These results give no indication of the accuracy of the test which depends entirely on the absence of interfering substances.
diluted 1 to 100 with distilled m-ater. The absorbance spectrum of the nitroso color of the diluted solution was determined with the Cary spectrophotometer in 1-cm. cells in the same way as the nitrosolignin color. Figure 4 shows the spectra of the untreated diluted solution
I.
W 0
z a m
[LI! 0 m m
a
0.
3 Figure 4. Absorbance spectrum of bark extract Extracted solids concentration 0.46 gram per liter, 1 -cm. cells 1. Neutral blank 2. Alkaline blank, after mixing 3. Same, after 1 0 0 minutes 4. Same, after 24 hours 5. Reagent-treated sample
Figure 5. Absorbance spectrum of bark extract nitroso color Extracted solids concentration 0.46 gram per liter, 1 -cm. cells 1. After mixing 2. After 100 minutes 3. After 2 4 hours
Figure 6. Absorbance spectrum of wood hydrolysis liquor nitroso color Extracted solids concentration 0.92 gram per liter, 1 -cm. cells 1 . Reagent-treated sample 2. Alkaline blank 3. Nitroso color (alkaline blank) 4. Neutral blank 5. Nitrosa color (neutral blank)
VOL. 31, NO. 3, MARCH 1959
a
373
Table IV.
Precision of Lumetron Method
in a n autoclave for 1 hour with water at 175' C. The cloudy solution drained from the chips was cleared in a high speed centrifuge. The clear solution (4.51y0 total solids) was diluted 1 to 50 with distilled water. Figure 6 shows that the nitroso color absorbance a t 430 nip obtained with the alkaline blank (curve 3) is much lower than that obtained with the neutral blank (curve 5).
Nominal Dilution, p.p.M.5
Av. of 5
Table V.
Apparent Spent Sulfite Liquor Concentration in Natural Water Samples
Tests, Std. Dev. '.Pm. '.Pm. Av. % 1 0 94 0 044 4.7 5 5 14 0.263 5 1 10 9 94 0,171 1.7 a Spent sulfite liquor of 15.3% total solids.
Spparent Spent Sulfite Absorbance in 15-Cm. Cells Liquor Concentration, Nitroso Color P.P.M. Untreated Method Method Ba &lethod Method Ba Source of Sample sample A 1 hr. 24 hr. A 1 hr. 24 hr. Forest Pool 0 235 0 073 0 037 0 015 14 6 1 Kennedy Creek 0 194 0 050 0 026 0 00'7 9 4 0 Skookum Creek 0 090 0.014 0 007 0 009 2 0 0 Coffee Creek 0 291 0 113 0 052 0 039 22 9 6 __ Johns Creek 0 103 0 031 0 027 0 019 5 5 2 Cranberry Creek 0 124 0 039 0 024 0 017 7 3 2 Deer Creek 0 171 0 042 0 028 0 020 7 4 2 a Time elapsed betn-een reagent addition and Lumetron reading. (curve 1, neutral blank), of the same solution containing acetic acid and ammonia only (alkaline blank) immediately after mixing with these reagents (curve 2 ) , after 100 minutes (curve 3), and after 24 hours (curve 4), and the spectrum of the reagent-treated solution (curve 5 ) which was unchanged after 24 hours. The spectrum of the nitroso color obtained with the alkaline blank and the other necessary blank solutions in separate cells, is shown in Figure 5 , immediately after mixing with the reagents (curve I), after 100 minutes (curve 2 ) , and after 24 hours (curve 3). Maximum absorbance of the bark extract nitroso color occurs at about 330 mp, but absorbance a t 430 mp is considerable. From the absorbance values a t 430 nip of Figure 1, curve 4, and Figure 5 , curve 1, it can be calculated that the absorptivity (specific absorbance per gram of solids), a t 430 mp of the nitroso color of bark extract is almost four times higher than that of spent liquor solids: AbsorDAbsorb- Concen- tiviti, ance, tration, L./G. 430 hfp C./L.-l Cm.-' Bark extract 0.50 0.046 10,9 Spent liquor 0.75 0.256 2.9 The apparent decrease with time of the bark extract nitroso color (Figure 5 ) is caused by the progressive increase in absorbance of the alkaline blank solution (Figure 4, curves 2 to 4). This behavior in alkaline solution, typical of polyphenolic materials with catechol groupings, is probably due to o-quinone formation. Chips of Douglas fir wood were heated 374
ANALYTICAL CHEMISTRY
DISCUSSION
Results obtained by this method represent apparent spent sulfite liquor concentrations because they include the natural nitroso color of the water. Samples of sea water from locations free of spent sulfite liquor give nitroso color readings corresponding to apparent sulfite spent liquor concentrations of about 3 p.p.m. With samples that contain spent sulfite liquor, absorbance values obtained by Methods A and B differ slightly as a result of the indicator effect shown by spent sulfite liquor in neutral and alkaline solution (Figure 3). Conversion of absorbance values obtained by the two methods into spent sulfite liquor concentrations by the appropriate calibration factor eliminates the difference, and the two methods give identical results. If the sample contains interfering substances that exhibit larger indicator effects than spent sulfite liquor, Method B gives lower results than A upon conversion of absorbance into concentration. Moreover, when polyphenolic materials similar to hemlock bark extract are present, results by Method B are not only lower than those of A, but decrease with time because the absorbance of the alkaline blank solution increases. Interfering substances of this type may be present in natural waters. Results obtained by Methods A and B with samples of water from a forest pool, highly colored by decaying vegetable matter, and several creeks are shonn in Table V. Although none of these samples contained any spent sulfite liquor, Method A gave positive
readings with all of them. The two samples with the highest absorbance before reagent addition gave especially high readings. Khile the results by Method B were substantially lower, all except one sample still gave a positive reading. However, the decrease after 24 hours indicates that the nitroso color was probably caused by polyphenolic materials similar to those in bark extract. Although interference is greatly reduced by Method B, neither method bjitself gives correct results in the presence of interfering substances. Honever, if the determination is carried out by both Methods A and B, it is clearly indicated that interfering substances are present whenever the results of Method B are appreciably lower than those of A. Aside from interference from substances other than spent sulfite liquor, the most serious source of error is probably turbidity of the sample. Experiments with solutions of known spent liquor concentration have shown that filtration lowers the concentration, presumably by adsorption of lignin sulfonates on the filter medium. Filtration of the sample has, therefore, not been included in the method. Experience has shown that in routine determinations, the performance of the photoelectric colorimeter with the improved filter combination compares favorably with that of more elaborate spectrophotometers in reproducibility and sensitivity. Simplicity and inherent stability of the photoelectriccell compensating circuit (8) improve the reproducibility, and the long light path iniproves the sensitivity a t very low spent liquor concentrations. ACKNOWLEDGMENT
The authors thank D. W. Balkema, R. E. Boyd, T. C. Townsend, and Elisabeth V. Coffman for assistance with the experiments, and J. E. Jefferj for valuable suggestions. LITERATURE CITED
(l),iinderson, A. A., B.S. thesis, University of Washington, Seattle, Wash.. 1942. (2) Brice, B. A,, Rev. Sci. Instr. 8, 279 (1937). (3) Lvkken. Louis. Treseder. R. S , Z a h , Victor, IxD. ENG.CHERI.,A s . 4 ~ . ED. 18, :103 (1946). 'nrln. _ r). I,.. Tartar. V.. Tollef(4) McKer.-..--, _ ~. son, R., State of '\\Tishiniton, Dept. of Fisheries, Biological Rept. 49A, llL5 (April 1F14F1) \ - 2
T r a d e J . 111,235 (1Y4U). (6) Williams, R. W., Eldridge, TT. E.,
Mains, E. M., Lasater, J. E., ed. by Holland, G. A., State of Washington, DeDt. of Fisheries. Research Bull. 1, 19 (December 1953). RECEIVEDfor review August 12, 1958. Accepted November 12, 1958. Contribution No. 43 from the Olympic Research Division, Rayonier Inc.
