Determination of Relative Color Density of Liquids - Analytical

Louis Lykken and John Rae Jr. Anal. Chem. , 1949, 21 (7), pp 787–793. DOI: 10.1021/ac60031a008. Publication Date: July 1949. ACS Legacy Archive. Not...
0 downloads 0 Views 929KB Size
V O L U M E 21, NO. 7, J U L Y 1 9 4 9

787

211 Slaaj, A. R., “Poisons, Their Detection and Estiinatiori,” in N. H. Furman, “ S c o t t ’ s Standard Methods of Chemical ,&)

Analysis,” 5th ed., New York, D. Van Nostrand Co., 1943. McNally, W. D., “Toxicology.” p. 3, Chicago, Industrial Medicine, 1937.

23) I b i d . , p. 21. :24) Merck’s Index, 5th ed., p. 601, Itahway, N. J., 1940. ‘25) Sollmann, T. H., “Manual of Pharmacology,” 4th ed., Philadelphia, W. B. Saunders Co.. 1932.

(26) Tanner, E‘. \Y.,“Food-Borne Infections arid Intoxications.” Champaign, Ill., Twin City Printing Co., 1933. (27) Underhill, F. P., “Toxicology or Effects of Poisons,” 2nd ed.. Philadelphia, P. Blakiston’s Son & Co., 1936. (28) U.S.Pharmacopoeia, XIII, p. 20, 1947. (29) I b i d . , p. 871. (30) Witthaus, R. A., “Manual of Toxicology,” New York, William Wood & Co., 1911. RBCEIVED August 19, 1948.

Determination of Relative Color Density of liquids A Rapid Photoelectric Method LOUIS LYKICEN AND JOHN RAE, J R . ~ Shell Derelopment Company, Emt.ryt.ille, Calif. 4 rapid method for the reproducible photoelectric determination of the relative color density of liquid products involves measurement of the optical density of the test material relative to that of an easil? prepared liquid color standard selected to approximate the test material in hue and color density. The method is particularly applicable to the determination of colored substances present in the w-idr variety of liquid products exhibiting spectral characteristics like those of normal lubricating oils and other “naturally colored” liquids having smooth.

Y

A S indication of the purity or degree of refinement of niaii) products, color is often the only criterion that is readill available to and appreciated or understood by the consunirr The measurement of color is a very useful research tool, particv larly for following the course of chemical reactions in purification, sging, and oxidation studies. Moreover, with modern improved ,)hotoelectric colorimetric equipment, color considerations ai e wcoming increasingly important and useful to the research chrm1st. For these reasons, methods for measuring color are of wide.pread interest to the chemical industries. The several methods available for the complete description of d o r (3) require the use of unnecessarily expensive equipment 01 rivolve somewhat complex visual measurements. Moreover -he results obtained by these comprehensive methods are unnecriwarily complicated for many purposes. I n many instances, B neasure of the total transmittance of a given test material in the visible region of the spectrum is sufficient for the color descriptioii required and, indeed, is often more useful than spectrophotometric analysis in the study of certain types of chemical reaction. wch as oxidation. In the American petroleum industry, the most widely used ‘total transmittance” or color density method is the National Petroleum Association (NP.4) or A.S.T.M. union color number method ( 1 ) for measuring the color density of lubricating oili This method depends upon visually matching the color density oi .he light transmitted by the test material with one of a series of d o r e d glasses. However useful this method has been in thr mst, it is limited by poor reproducibility because different obmvers seldom strike the same compromise among the threr -tttributes of color sensation-hue, saturation, and lightness ( 3 ) . Furthermore, the glass color standards are difficult to duplicate 3pectrally from set to set and they provide a discontinuous scale ahich can be evaluated only a t fixed intervals; color values falling qetween these points can only be roughly estimated. Other b

Present address, Shell Oil Company, Inc., Houston, Tex.

bpectrophotonirt ric absorption curvea with maximum absorption in the near ultraviolet. The results obtained are easily visualized in terms of real psychophj sical significance and readily correlated with such phenomena as oxygen degradation of lubricating oils, gasolines, motor fuel blends, solvents, polymers, and other materials. This method is especially useful as a specification method for measuring color densit) which is often taken as a criterion of purity. or :is a control method for following manufactnrinp or refining processes.



visual iiiethods such as the Saybolt ( 2 ) and Hazen (7) methods for measuring color density are subject to errors similar to those mentioned in connection with the National Petroleum Association method. Among the first to suggest overcoming these difficulties t)y the use of a photoelectric colorimeter were Story and Kalichevsky (9‘1 who, in 1933, described a photoelectric method for measuring the color of liquid petroleum products. However, the photoelrrtri~ method was slow to gain acceptance by industry. Perhaps the chief contributing factor was the lack of precise photoelectrir volorimeters which have come into common use only during the past decade and even today are not being used t,o best advantage in many laboratories. I n 1942, Diller, DeGray, Rilson, and co-workers (4, 5 ) demibed a photoelectric method based on absolute transmittance and using a tungsten filament. light source, a broad band filter, and a barrier-layer type of photocell to measure the amount of the light transmitted through the test material contained in a specified round cell. In this method, no provision was made for inevitable changes in instrument response resulting from variations in spectral characteristics of the light source, in photocell response, and in spectral characteristics of the light filter employed. For these and less obvious reasons, the method has not been generally accepted. A recent contribution to photoelectrir methods for color density evaluation has been made by Osborn and Kenyon ( 8 ) . I n their thorough treatment of the subjert,, these authors have rendered an excellent discussion of the theoretical aspects and have described a rather elaborate method for setting up color standards for instrument calibration, preparatory to measuring the color density of a given line of spectrally similar products. Another interesting contribution to the field of photoelectric color measurement has been made by Whyte ( I O ) , who ckvised several color grading systems for tallows and greases. Any specification method should give the same results regardless of the peculiarities of the particular apparatus or instrument

ANALYTICAL CHEMISTRY

788 used. Because different colorimeters may not give the same response even when they are of the same make, and the response of a given instrument may change significantly from month to month, any specification method for the measurement of color density should be designed to compensate for variations in colorimeter response. Prior to the work of Osborn and Kenyon (8),no established photoelectric color measurement method provided for this necessary requirement. Osborn and Kenyon compensated for instrument response variations bv use of a carefully selected set of color comparison standards. The same result was achieved independently in these laboratories by use of an easily prepared single comparison standard for a number of products, having the same general spectral characteristics and covering an approximately twentyfold color concentration range. Ausiliary comparison standards are used for relatively dark and relatively light liquids.

color density as the standard, and a value of 0.00 indicates “waterwhiteness.” The relative color density method is empirical, inasmuch as the value obtained is of significance only when expressed with relation to the color comparison standard used and, to a lesser extent, nith relation to the length of the light path used. I t does not indicate the hue of the sample but merely measures the total light absorption of the sample; thus, it is possible for samples visually differing in hue or saturation to have the same relative color density. The method is particularly useful for measuring color changes in a series of liquids exhibiting similar spectral characteristics. In fact, for such a series of liquids, an approximate correlation of relative color density values xvith I.C.I. tristimulus values could be made in a manner similar to that described by Diller and coworkers ( 4 ) . However, in the applications for which the relative color density method has been found useful, the results obtained from such a correlation would appear to be of little practical value. Results of similar magnitude and reproducibility would undoubtedly be found with an I.C.I. tristimulus green (luminosity) filter (9) in place of the chosen broad band filter but, for all practical purposes, the results would still be empirical and subject to the limitations of interpretation or correlation found for the relative color density values. The relative color density method is designed particularly as a specification or control method for lubricating oils and similar products, but is generally applicable to any nonaqueous or aqueous solution. I t is particularly applicable to materials, such as petroleum products, which show predominant absorption in the near-ultraviolet portion of the visible spectrum (see Figure 1). The method is most precise when applied to samples having color denqities close to that of the comparison standard (relative color density of unity), especially when the spectral characteristics of the test material are similar to that of the standard. The pracision decreases rapidly with increasing color density values greater than 2.

20

-

cell depth :13 mm. I

Oioo

Figure 1.

4b0

500

I

I

600 700 W a v e Length, mu

I

,

APPARATUS

BOO

.kny sensitive photoelectric colorimeter may be used, if it is provided with a tungsten-filament, incandescent light source and with one or more barrier-layer type photocells, and constructed to permit use of the parallel, planar-surface glass cells described below. The glass light “correction” filter must have the relative spectral characteristics shown in Figure 2. I t is desirable that the transmittance at any xave length be within 5Ti of the value shown; greater variations are permissible, provided they are proportional over the entire spectral range. Because the filter has a relatively high transmittance, it is sometimes necessary to reduce the intensity of the incident light in order to obtain proper adjustment of the colorimeter, especially n-hen adjusting the 100%

Spectral Characteristics of Typical Petroleum Products

The method given in this paper uses Gardner-type color comparison standards (6) composed of an aqueous solution of iron and cobalt chlorides acidified with hydrochloric acid. These standards are easily prepared, universally reproducible, and stable indefinitely, provided they are kept at a moderate room temperature (20” to 25” C.). They have been generally accepted by industry and tested over a period of years in various commercial visual color methods. In the method described here, the relative color density ,oo I I I I I I of a liquid is obtained by measuring the optical density of the liquid and of a given color comparison standard, using any of several commercially available photoelec80 tric colorimeters equipped with a specified broad band Net R e s p o n s e o f / light filter, a suitable light reduction mask, and light abf sorption cells of specified depth having parallel, planar faces. The relative color density of the test material is calculated by dividing its optical density by that of the color comparison standard. The result obtained is, therefore, a direct comparison of the color density of the test matelial to that of the standard. I t is of direct psychophysical significance, as the light filter is chosen so that the net response of the colorimeter approximates the response of the normal eye. Moreover, as a consequence of Beer’s laiv, relative color density is, to a fair approximation, proportional to the color concentration in the test material. For example, a relative color density of 2.00 simply means that the test material is Wave L e n q t h , m u twice as dark as the comparison standard, a value of 1.00 indicates that the test material has the same Figure 2. Effect of Color Filter upon Instrument Response

-

E

-

-

V O L U M E 2 1 , NO. 7, J U L Y 1 9 4 9

789

minute periods until a constant weight is obtained. From the weight of cobalt sulfate, calculate the cobalt concentration in the stock solution. Standards A, B, and C. Weigh a quantity of the iron stock solution calculated t o contain exactly the quantity of iron given in Table I ; transfer to a 100-ml. volumetric flask, using a minimum amount of 0.6 -Y hydrochloric acid to aid in the transfer. Introduce into the same volumetric flask, either from a buret or from a graduated pipet, a volume of the cobalt stock solution calculated to contain exactly the amount of cobalt given in Table I. Cool to room temperature, dilute to 100 ml. with 0.6 S hydrochloric acid, and mix thoroughly. S.A>\IPLE PREPARATION

The sample must be essentially free from turbidity or suspended solids. If it appears turbid when viewed through a thin layer-e.g., the 2-mm. absorption cellfilter through a fine filter paper or clarify in a centrifuge before measuring the color. 400

I

500

I

600

I

700

I

800

1 900

I

1000

PROCEDURE

Select a pair of matched 10-mm. colorimeter cells. Fill one cell with distilled water and the second with the Figure 3. Spectral Distribution of Photocell, Net Instrument sample to be tested. Polish the cell surfaces Lvith a Response, and Energy of Light Source clean, lint-free cloth or tissue and place the water cell in position in the colorimeter. Adjust the colorimeter to a scale reading of zero optical density (100% transmittance). Replace the water cell with that containing the sainple and record the resulting scale reading. Reinsert the water Table I. Composition of Standard Color Solutions cell and check the instrument adjustment. If any readjustment is Standard Iron Cobalt made, obtain a new reading for the sample. Discard the sample. G./100 nl. G./lOO ml. Clean and dry the cell and fill with color standard B. Pro.I 9.70 1.61 ceeding as directed for the sample, obtain a colorimeter reading B 2 . 7 7 0 , 5 3 7 for the standard. (This measurement of the standard need be 0.0333 0.0134 C made only sufficiently often t,o ensure constancy. In most cases, one reading per xeek will suffice.) Calculate the relative color density as directed below. If the value obtained is greater than 2.0, repeat the color density meastransmittance setting. Satisfactory reduction of the amount of urement using standard and matched 2-mm. colorimeter cells. light entering the sample cell can be accomplished by use of a If the value obtained is less than 0.1, repeat the measurement screen, or a mask, prepared from black photographic paper having using standard C and matched 50-mm. cells. I n borderline small, uniformly distributed holes or slits of appropriate dimencases, make and report measurements with both combinations of sions. standard and cell depth. Note. This filter is usually referred to as the Korth Sky fi!ter; it is made of Aklo glass obtainable from the Corning Glass Works. CALCULATION It has been chosen t o transmit a somewhat wider band than a filter accurately duplicating the luminosity response of the average eye to If the colorimeter reads directly in optical density, or in units daylight because such a filter spreads the density scale a t the upper proportiona~to optical density, calculate the relative color denand lower ends in a useful manner without materially affecting the of the follolTing equation: sity by correlation with other colorimetric measures such as A.S.T.M. union R color numbers. Relative color density = Matched pairs of glass cells are necessary t o permit measurement of the light transmittance of 50 * 10, 10 * 2 , and 2 * 0.5 R = colorimeter reading for the sample, and = colornim. layers of test material. These cells should have parallel, imeter reading for the standard. planar faces and be of proper dimensions to Permit use in the If the colorimeter scale is graduated in terms of transmittance, colorimeter employed. A 2-mm. cell depth may conveniently be calculate the relative color density by means of the folloming obtained by inserting an 8-mm. glass block (of clear optical glass) equation: in a 10-nim. cell. 2 - log R Relative color density = 2 - logs S T A S D A R D COLOR SOLUTIONS W a v e Length,

mu

s

s

Iron Stock Solution. Dissolve 500 grams of C . P . ferric chloride hexahydrate in 120 ml. of 0.6 *V hydrochloric acid and filter through glass viool. Determine the iron content of the solution as follows: Accurately weigh 0.8 to 1.2 grams of the test solution into a glass-stoppered weighing bottle and transfer to a 250-ml. Erlenmeyer flask containing 30 to 40 ml. of 0.6 .V hydrochloric acid. Heat the solution to boiling and add 10TGstannous chloride solution dropwise until the solution is colorless, being careful not to add more than 2 drops in excess. Cool to room temperature, dilute to about 80 ml., add 10 ml. of 5% mercuric chloride solution, and allow to stand for 4 to 6 minutes. Add 3 ml. of 85Yc phosphoric acid and 6 drops of 0.2y0 barium diphenylamine sulfonate indicator solution, and titrate with standard 0.1 N potassium dichromate solution until 1 drop produces a deep purple color which does not fade within 5 seconds. Calculate the iron concentration in the stock solution. Cobalt Stock Solution. Dissolve 100 grams of C.P. cobalt chloride hexahydrate in 300 ml. of 0.6 S hydrochloric acid. Determine the cobalt content of the solution as follows: Pipet exactly 1 ml. of the test solution into a weighed platinum dish, add 5 ml. of 18 3 sulfuric acid, and caut;fously evaporate to dryness. Ignite in a muffle furnace a t 500 * 25' C. for 5-

where R and S are expressed in per cent transmittance. I n reporting the relative color density, specify the standard used and the depth of the absorption cell. If the sample has been clarified in any way, report this fact. DISCUSSIOY

I n any photoelectric color measurement, there are four fundamental factors involved: (1) the spectral energy distribution of the light source, (2) the spectral response of the photocell, (3) the spectral characteristics of the light filter used, and (4)the depth of test material through which the light passes. The effects of the first three upon instrument response are illustrated in Figures 2 and 3. Common photoelectric colorimeters are generally equipped with a tungsten filament light source and a barrierlayer, photovoltaic type of photocell. For these colorimeters, consideration of the response relationships involved (Figures 2 and 3) makes it clear that only a broad band filter such as the North Sky filter will lead to a net response approximating that of the normal eye.

ANALYTICAL CHEMISTRY

790 For reasons previously stated, the authors have chosen Gardtier-type iron-cobalt color comparison standards to compensate for variations in response from colorimeter to colorimeter. Comparison of Figure 4 and Figure 1 shows that the spectral characteristics of these standards closely approximate those of typical petroleum products except for the spectra1 region above 700 millimicrons. However, this difference can be neglected, as the net response of the colorimeter is comparatively small i n this region (Figure 2). 100

-

I

..__

Table 11. sa.rllplr

NO.

3-50 5-51

S-54

81:;

S-57 S-58 S-59 S-60

Stondard 80

-

C

Stondord B

I

Y-61

5-62 Standord A

S-63

c

0

8-64

0

8-65

c

c 2

Description of Color Test Samples

40

Description of Sample Lubricating oil SAE 10 Lubricating SAE 20

,L;,bz+z!; $ ~ ( ~ ~ ~ l~~ s~than s o l 1)o r Lubricating oil SAE 50 2;: i g (abnormal color) Lubricating oil, highly fluorescent ~~~~~~~~~~

Light kerosene (ordinary water-white kerosene) Dark kerosene (sample 5-58 containing 2% by volume of SAF. 10 oil) Aqueous inorganic standard 4 (contains 9.70 grams of iron and 1.61 grams of cobalt diluted t o exactly 100 ml. with 0.6 A' hydrochloric acid) .4queous inorganic standard B (contains 2.77 grams of iron and 0.537 gram of cobalt diluted t o exactly 100 m.1. with 0.6 .A' hydrochloric acid) hqueous inorganic standard C (contains 0.0333 gram of iron anc 0.0134 gram of cobalt diluted to exactly 100 ml. with 0.6 h hydrochloric acid) Aqueous inorganic solution X (contains 120 grams of nickelouc chloride hexahydrate i n 14 liters of 0.2 N hydrochloric acid solution) Aqueous inorganic solution Y (contains 0.0514 gram of iron ano 0,0214 gram of cobalt diluted t o 2000 ml. with 0.6 .V h y d r n chloric acid) Aqueous inorganic solution z (contains 700 grams of couper sulfate pentahydrate in 1 4 liters of approximately 0.1 V ~111. furic acid solution) ___.

Cell depih

20

I

11, Wove Length,

Figure 4.

Spectral'Characteristics Solutions

I

I

Table 111. Kelative Color L)ensity Yalues

* 13 m m

Sample

I ,

S-50 S-51 5-52 s-53 8-54 5-55 S-56 5-57 S-60 54-61 5-62 a-63 R-6.5

mu

of Standard (:olor

(Obtained with five makes of photoelectric colorimeters) Relative Color Density, Standard B, 10-Mm. Cell Fisher Lumetron CencoModel SheardHelligc Klettelectro0.76 1.03 1.79 0.05 Too dark Too dark 2.85 1.60 2.47 1 .oo 0.06 0.07 0.27

0.73 1.02 1.90 0.04 5.31 3.77 3.07 1.78 2.64 1.00 0.04 0.05 0.29

0.71 1.04 1.94 0.01 5.04 3.87 3.09 1.74 2.70 1.00 0.03 0.05 0.43

0.79 1.06 1.86 0.05 4.47 3.52 2.93 1.72 2.48 1.00 0.02 0.04 0.31

0.77 1.0;. 1.83 0.03 5.14 3.86 2.91 1.69 2.44

With regard to the stability of the chosen standards -1,B, and C described above, repeated optical density measurements over a i.oc period of 4 months indicated that they are stable when kept a t a 0.02 0.03 moderate room temperature (20"to 25' '2.). However, when not 0.31 in use, they should be kept in tight glass-stoppered bottles, as ~ _ -. _ . ~ __ - __ standards A and R , particularly, have a tendency to rthsorh water from the air. 'rahle 17. Kelative Color Density \-slues of Light Sample* Although the authors have chosen the particular color stand(Obtained with four makes of photoelectric colorimeters) Relative Color Density, Standard C, 50-hfm. Cell ards described for use in their work, it is intended that other suita.^ " * 4u-ivim ble st,andards should be used for specific applications, particuFisher T,umetron CencoCell. electroModel SheardKlettlarly t,hose involving specification testing. The standard selected ~ aX ~ i i p l r 402E Sanford Summerson O. photometer for a particular application should approximate &s closely as s-53 0.98 1.00 1.18 1.12 Too lighr Too light Too light 5-58 'roo light possible the color density and spec.tral characteristics of a sample 0 .64 0 . 4 9 0 . 5 8 01.00 .63 of the test material which just passes the test or has polor den1.00 1.00 1.00 2 , l i 1.74 I .3F; 1.67 5-63 iity in the most critical range. 0.11 'roo light 0.09 S-64 0.14 In order to check the reproducibility and general .... usefulness of the proposed method, sixteen samples Table V. Precision of Relative Color Density Results Obtained in of petroleum products and colored aqueous soluDifferent Laboratories tions, designated as 5-50 to S-65, inclusive, were Relative Color Density for Indicated Standard and Cell Deptha .. prepared (Table 11). Their relative color density Standard A, 2-Mm. B, IO-Mrn. Standard C , 50-Mm. values were determined on five commercially avail.._ Cell Depth -Standard Cell Depth Cell Depth able photoelectric c o l o r i m e t e r s : t h e K l e t t dumnierson photoelectric colorimeter (Model 900-3), the AC Model Fisher electrophotometer, the Hellige-Diller photoelectric colorimeter (Model IOj-A), the Cenco-Sheard-Sanford photoelectric colorimeter (Catalog KO. 12,340),Table and the Lumetron colorimeter (Model 402-E). 111 gives the results of the measurements made in 10-mm. cells relative to standard B. Table I\- gives those made in 40- and 50-mm. cells relative t,o Jtandard C. The results in Table 111 indicate that for those samples in which the relative color density values are between Oeland 2.0, good a@eement is Obtained by the five different instruments; in this range, the average deviation from the mean of the relative

Saniplr No.

Av. ralue

Deviation from Average MaxiMean mum

...

Deviation from Average Av. Maxivalue Mean mum 01 ..704 3 0 0.03 0.06 .03 0 .11 1 . 9 0 0.05 0.15

Av. value

Deviatiun--~ from Average MaxiMean mum

.. .. .. .. .. ..

... S-50 . . 8-51 . . . . ., " ... 5-52 0.56 0.03 0.09 ,. , 1',23 0,'30 0.8: ,.. ... ... . . . . . . 6-53 . . . . . . . . . . . . 5-54 0.18 ... .. ...... 6-55 11 .. 26 54 00 .. 00 75 0.12 .,. ... .. ... S-56 0.82 0.05 0.17 5-57 0.60 0.02 0.04 f,73 0:Ok 0.16 . . . .0 . .1 3. 0.45 0.05 . . . . . ...... S-58 0.25 0,..8 0 0.66 . . . . . . s-59 . . . S-60 1.00 01004 6.04 ...... 5-61 . . . . . . l'.CO 0:005 0.05 0.'04 1',00 ... ... 8-62 . . . . . . 1.80 00.'004 .30 1.47 ... 8-63 . . . . . . 0.210.13 0.50 ... S-64 0.31 0.06 0.17 S-65 5 Based on unpublished results collected b y Subcommittee V I of A.S.T.M. Committee laboratories utilizing seven different makes of commercial pho. j3-2 from 12 toelectric colorimeters.

...

. . . . . .

... ,..

......

......

... ...

... ... ...

... ... ... ... ... ...

...

......

..

V O L U M E 21, NO. 7, J U L Y 1 9 4 9

_-

791

--__

.__-

precision found by the authors (Tables 111 and IV). Inasmuch as the color sam(Instrument: Fisher eleotrophotometer) ples were being tested over a Firat Week Second Week Third Week Fourth Week long period of time, their relaStand- Stand- Stand- Stand- Stand- Stand- Stand- Stand- Stand- Stand- Stand- Standtive color densities were measard C ard A ard B ard C ard B ard C ard A ard A ard B a r d A ard B a r d C Jam2.410502.410502.410502.41050ured on the Fisher electrophopje mm. mm. mm. mm. mm. mm. mm. mm. mm. mm. mm. rnin \o. cell cell oell cell cell cell cell cell cell cell cell rell tometer once a week over a pe0 21 0 77 riod of 4 weeks in order to deter0 21 0.7Y 0.21 0.75 . S-50 0.21 0.76 0 27 1 05 0.28 1.03 0.27 1.03 . 8-51 0.26 1.03 mine whether the color density of n 59 I 83 0.60 1.m 0.59 I 78 , . 8-59 0.60 1.75 ... ,, . 1.12 1.14 ... 9-53 1 02 I Ok any of the sampleschangedwitl~ 1 61 1.58 .., , . I.'60 1.62 . . 9-54 1 25 time. The results, given in 1.38 ... I .26 1.27 ,.. 3-55 0.85 2.85 0.86 2.81 0.86 2.88 5-56 0 86 2 91 Table VI, indicate that little or n 63 I 69 0.63 857 1.66 l1,6:> 1 .6.5 0.64 1.63 .. ... .., ... no change took place. The data 9-58 0 52 I1 h,i 9-59 , , 0' ii:i 0'61 , . . ... 1 00 2 41 also indicate the high degiw of i.00 2.47 1.0C~ 2.48 1.01 2.,% 3-60 9-61 11.37 1.00 o 3; I on O.RR i.nn 0 35 I no repeatabilit) obtainable 1% hen I 00 I ('(J 1 54 I 6+ the same colorimeter 15 useti for I? 0 08 n series of color nieaquremr.nts i 42 I 1 71 I I4 Color dtaiisities rc4ativca to _-___ _ _ _Ytandards A and H were obtained for sixty saniplrs of lubricating 011s 1 e p esentlng ~ LUelve commercial brand. iTable w1o1 ileiisity value fot a single ~ r i i p l eranges i ~ o n i0 01 t o 0.04 VII) These color measurements were made on t t r t a Fiiher For all the samples in Table I11 the average deviation from thr rlectrophotometer and the Klett-SummeisoIl photoelrctric ~ n e a nof the relative color den;iitie5 for a single sample ranges from c~olnrimeter. I n addition, the A.S.T.M. union color numbers 1.0 to 12.5%, with an aveiage relative error of 4.4O:. The r d a 1'1) of these samples were determined. I n most ~ t l . w ~ the , tlve color density data obtained with itaridard C (Tahlr IT'i 4 t o ~ final 1.S.T.M. union color numbers reported were obtained b\ sn avei age relative error of I 1 .6Lc . averaging the readings obtained by three different ope1at 01s These sixteen samples nere also cooperatively tested in a pi(^The results in Table VI1 demonstrate the good instrument -togram hponsored by Subcommittee VI of A.S.T.Lf. Committrr instrument reproducibility obtained by the propowd method and D-2 Relative color drn4ty measurements were made in accoidits applicability to lubricating oils and materials of .iniilai color wire with the described method in twelve different laboratorie. c,haracteristics. Table VI11 shows the reproduc.ibilit1 of the relausing a total of seven different InakeEi of photoelectric coloitive c d o r density and A.S.T.M. union color ( 1 ) methods and meters. The data obtained, given in Table V, show that the Show that the precision of the relative color density values is reproducibility of the method in different laboratories using corii( onsiderahly better than that of the visual method xhen the demini photoelectric colorimeters i s approximately the iame a\ tlir I'ahle VI.

Relative Color Density Values of Color Test Samples, Showing Their Stability over a Period of Four Weeks

, .

,

,

Table VII.

Description of Oil Sample Brand A 9 A E 10 S.4E 20 3AE 30 3AE 40 9 A E 50 S A E 60 Brand B SAE 10 S 9 E 20 9AE 30 9AE 40 3AE 50 98E 60 Brand C 3AE 1 0 SAE 20 3AE 30 J A E 40 3AE 50 3AE 60 YAE 70 Rrand D 3.4E 10 3 A E 20 J A E 30 SAE 40 Brand E (green) 3.4E 10 qAE 20 3AE 30 S A E 40 3AE 50 3AE 60 4 4 70~

Relative Color Densities and Union Colors of Sixty Commercial Lubricating Oils Representing Twelve Brands Standard A, 2.4-h'Im. Cell Fisher Kletteleotrophotom- Summerson eter

(Obtained with t w o colorimetern) Standard B. 10-Mm. Cell Fisher Klett electroDescription of Union photom- SummerOil Sample son Color eter Brand F (pale) S A E 10 1.08 0.99 3112' SAE 20 1.43 1.47 4. SAE 30 1.94 1.88 2SAE 40 1.70 1.80 0SAE 50 2.08 5 1.97 Brand G 2.98 3.20 7+ %E 10 SAE 20 1.20 1.21 4 SAE 30 1.40 1.37 y/1 SAE 40 1.72 1.76 09AE 50 1.47 1.53 4'/2 SAE 60 1.47 1.51 hl/l SAE 70 1.90 2.01 5+ Brand H SAE 10 41.02 1.01 S 4 E 20 1.16 1.17 4+ S.4E 30 5 1.73 1.88 SAE 40 5 1.89 2.00 SAE 50 1.84 1.98 5 SAE 60 2.07 2.21 6S 4 E 70 3.12 3.39 7+

0.26 0.39 0.56 0.52 0.61 0.93

0.27 0.40 0.53 0.46 0.58 0.88

0.34 0.36 0.51 0.42 0.45 0.57

0.31 0.39 0.49 0.43 0.41 0.53

0.28 0.34 0.56 0.62 0.61 0.70 8.99

0.26 0.33 0.54 0.61 0.59 0.69 0.99

0.31 0.29 0.44 0.35

0.32 0.31 0.45 0.33

1.07 0.99 1.38 1.11

1.06 1.00 1.43 1.13

0.67 0.73 0.67 0.65 0.73 0.89 0.97

0.65 0.73 0.65 0.62 0.70 0.86 0.97

2.18 2.28 2.18 2.13 2.23 2.75 3.08

2.31 2.46 2.32 2.27 2.38 2.98 3.3.5

43'/2+

4+ 4-

5+ 6 5+ 5+ 6

7 7+

Brand I (Penn.) S.4E 10 SAE 20 SAE 30 SAE 40 SAE 50 Brand J S.4E 10 SAE 20 SAE 30 Brand K SAE 30 SAE 40 Brand L (Hi-Duty) SAE 70

Standard .4, 2.4-Mm. Cell Fisher electroKlettphotom- Summereter Ron

Standard B. 10-Mm. Cell Riqhpr - .-....

eleotrophotometer

Klett Summerson

Union Color

0.15 0.16 0.24 0.26 0.44

0.14 0.15 0.22 0.25 0.42

0.67 0.75 0.98 1.02 1.46

0.66 0.74 0.96 1.02 1.51

21/z 21,'s 3 3 4.

0.28 0.47 0.60 0.77 1.03 1.36 1.24

0.27 0.46 0.57 0.75 1.03 1.34 1.23

0.91 1.50 1.87 2.39 2.95 Toodark Too dark

0.92 1.59 2.01 2.59 3.25 4.52 4.08

4'/1 5 6 7 847+

0.19 0.22 0.26 0.28 0.65 1.02 1.68

0.19 0.21 0.24 0.26 0.59 0.95 1.66

0.68 0.75 0.85 0.94 2.03 3.21 Too dark

0.69 0.74 0.82 0.93 2.15 3.50 Too dark

0.21 0.78 0.98 1 ,2.5 1.57

0.17 0.78 0.94 1.23 1.57

0.61 2.12 2.72 Toodark Toodark

0.61 2.31 2.99 3.88 5.31

0.28 0.44 0.67

0 25 0.41 0.61

0.93 1.39 2.03

0.91 1.43 2,15

3'/z 44-

0.43 0.59

0.42 0.56

1.37 1.89

1.42 1.99

44-

4112

1.69

1.71

Toodark

5.58

Too dark

s1/z

3 3I/z

3I/i 3'/? 5

+

7 Too darb 3

5t 6+ 8 8+

5+

792

ANALYTICAL CHEMISTRY

viations are considered on a relative percentage basis. This is true even for samples S-55, 5-56, and S-60, which are really too dark for accurate measurement relative to standard B. Color data gathered in the past have often been expressed in A.S.T.M. union color numbers and, therefore, it is useful to know the relationship between any new color system and union color. Using data similar to those given in Table VIII, the relative color density values relative to standards A and B (measured in 2.4and 10-mm. cells, respectively), were plotted against the corresponding A.S.T.M. union color numbers. The resulting curves (Figure 5 ) give an approximate correlation of relative color density and the -4.S.T.M.union color scales.

Table VIII. Reproducibility of Relative Color Density and A.S.T.M. Union Color Number Results Relative Color Density A.S.T.>\I. Union Color Standard B, IO-Rfm. CellQ Nurnberb Sample Maximum R I aximu m No. Av. deviation Av. deviation 0.75 0.04 3 0.5 1.04 0.02 3.5$ 0.5+ 1.86 0.50.08 a3.75 0.23 7.5 0.5 2.97 1 0.12 7 1.72 0.5 0.06 2.55 0.15 J. J 0.5 a Rleasurements made using colorimeters of five different makes. b Based on unpublished results collected by Subcommittee VI of A.S.T.M. Committee D-2 from 12 cooperating laboratories.

et

In the practical application of the method, a discontinuity of calculated values occurs in progressively darker oils when it becomes necessary to change from one specified combination of cell depth and standard to another. From a reproducibility standpoint, this discontinuity is serious only in borderline cases. Table I S demonstrates the degree of constancy of the ratio of relative color densities obtained with standard B (10-mm. cell) to those obtained with standard -4(2-mm. cell) for various oils and for three different makes of colorimeters. hpparently, a definite relationship exists between the two sets of relative color density values but it is not sufficiently constant from one material to another to permit accurate conversion from one scale to another. 10

[

I

I

I

I

I

1

Table IX.

Reproducibility of Relative Color Density Values

[Ratio of relative color density with standard B (10-mm. cell) t o relative color density with standard A (2-mm. cell)] Fisher KlettHelligeElectrophotometer Summerson Diller Description of De viaDeviaDeviaOil Sample Ratio tion Ratio tion Ratio tion Brand .4 SAE 10 5.25 1.43 5.41 1.51 5.38 1.58 SAE 20 4.54 4.56 4.73 0.93 0.72 0.66 SAE 30 3.86 4.08 0.30 0.04 0 . 1 8 4.10 SAE 40 4.08 0.26 4.04 4.19 0.39 0.14 SAE 50 3.72 3.87 0.07 -0.10 3.89 -0.01 Brand B SAE 10 4.35 0.53 4 60 0.70 4.44 0.64 0.73 SAE 20 4.26 0.44 0.55 4.53 4 .45 S h E 30 4 04 0.22 0.42 4.20 0.40 4.32 SAE 40 4.40 0.50 0.55 0.60 4.32 4.45 SAE 50 0 . 2 0 4.16 0.36 3.90 4.10 0.08 0.34 . SAE60 3 96 4.14 0 . 1 4 4.19 0.29 Brand C 0.52 SAE 10 0.89 0.92 4.32 4.71 4.82 0.17 0.41 0 . 1 3 3 97 SAE 20 4.03 4 23 o n-L 3 (5.5 - 0 . 1 5 -0.29 SAE 40 3.94 3.53 -0.37 SAE 50 -0.22 3.43 3.68 3.50 -0.32 -0.47 SAE 60 -0 35 3,55 3.36 -0.46 3.33 -0.31 -0.23 SAE 70 3.67 3.37 -0.45 3.49 Brand D 0.43 ShE 10 . 0.45 0.40 4.23 4.27 4.30 0.16 0.53 3.96 4.49 SAE 20 0.59 4.35 -0.08 0.02 3.72 0.07 SAE 30 3.84 3.9! 0.04 0.07 3.84 0.27 4.1, SAE 40 3.89 Brand E (green) 0 .14 3.66 -0.33 3.77 -0.13 SAE 10 3.49 -n 29 3.51 -0.53 3.48 -0.42 3.29 SAE 20 -0 12 3.68 -0.37 3,53 -0.37 3.45 SAE 30 -0.18 3.62 -0.36 3.71 -0.16 3.46 SAE 40 0 . 2 3 3.57 -0.40 -0 44 3.38 3.50 SAE 50 -0 36 -0.1.; 3.44 -0 34 3.48 3.75 SAE GO -0.20 3.60 -0.35 -0.03 3.47 3.87 SAE 70 Brand F (pale) 0.85 1.12 0.49 4.65 4 94 4.39 SAE 10 0.25 4.05 0.29 4.19 A40 0.68 SAE 20 0.38 4.18 0.07 0.40 3 97 S 4 E 30 4.22 -0.12 3 68 -0.04 4 03 0 . 2 1 3.86 SAE 40 -0.09 3.71 -0.13 3 77 3 47 SAE 5 0 -0.35 Brand G SAE 10 3.80 -0.02 3.52 -0.38 3.33 -0.47 SAE 20 3.44 -0.38 3.49 -0.41 3.44 -0.36 SAE 30 3.27 -0.53 3,47 - 0 43 3.12 -0.68 SAE 40 3.27 -0.55 3.53 -0.33 3.38 -0.42 SAE 50 3.00 -0.82 3.35 -n.s 3 19 -0.61 -0.4 3 06 S.4E 60 ... ., . ,. 3 . 65 40 3 . 0 28 --0.72 0.78 SAE 70 .,. Brand H 0.81 4.61 0.42 0.61 4.32 4.43 SAE 10 -0.59 -0.10 3.21 0 . 3 4 3 xo 4.16 SAE 20 0.20 4 . 0 0 0 . 0 5 0.09 3.95 3.91 SAE 30 0.52 4.32 0.27 0 . 5 3 4.17 4.35 SAE 40 3 . 6 9 0 . 1 5 -0.11 -0.26 3.75 3.56 S.4E 50 -0.08 0 . 0 0 3 72 -0.34 3.90 3.48 SAE 60 2.62 -0.17 ... ... ... SAE 70 ... Brand I (Penn.) SAE 10 3.56 -0.26 4.00 0.10 4.15 0.35 SAE 20 2.99 -0.83 3.17 -0.73 3.23 -0.57 S.4E 30 2.81 -1.01 3.15 -0.75 3.11 -0.69 SAE 40 ... ,.. 3.15 -0.75 2.94 -0.86 SAE 50 ... ,.. 3.29 -0.61 2.96 -0.84 Brand J 0.99 0 18 4 . 7 9 0.26 1.08 4.08 S.4E 10 0.23 4.03 0 02 3.92 -0.14 3.68 SAE 20 0 .10 3 . 7 0 -0 15 3 . 7 5 -0.39 3.43 SAE 30 Brand K 0 .02 3 82 -0.08 3 . 8 4 3 . 6 5 0 . 1 7 SAE 30 0.18 3.98 0.10 4.00 3.54 -0.28 S 4 E 40 Brand L (Hi-Duty) -1.07 2.73 -0.70 3.20 SfiE 70 Average 3.82 .., 3.90 3.80 ... Mean deviation . ,. 0.42 ,.. 0 31 0.43 hlaximum deviation ,, , 1.43 ,. . 1.51 1.58

APPLICATIONS

Figure 5.

Correlation of Relative Color Density with

A.S.T.M. Union Color Numbers

This variation is probably attributable to differences in spectral transmission properties of some of the samples. However, neither reflectance nor fluorescence appears to bear any relationship to this variation. The most logical use of the average ratios would be for approximate conversion of a series of values from one systemto another when a transition of systems is necessary because of color changes during the progress of a series of experiments.

During the past two years, the relative color density method has given satisfactory results for a variety of materials, especially those in the color density range of standards A and B. - I t has been particularly useful in following the color changes occurring in experiments involving oxygen degradation of materials such as lubricating oils, gasolines, motor fuel blends, solvents, and polymers. It has also given considerable aid in determining the degree of refinement or purity of many chemical preparations. ACKNOWLEDGMENT

The authors wish to thank F. D . Tuemniler for his interest and suggestions during the course of this work, and D. R. McCormick and S. M. Woogerd for their assistance in the experimental work.

V O L U M E 2 1 , NO. 7, J U L Y 1 9 b 9 LITERATURE CITED

(1) Ani. SOC.Testing Materials, “A.S.T.M. Standards on Petroleum

Products and Lubricants,” Method D155-45T, November

1948. (2) Zhid., Method Dl56-38. ( 3 ) -in,. yoc, ~~~~i~~ Llarerials, ~ ~ ~ y m p o s i u non l Color,vslLlarch 1941. (4) Diller. I. M..Dean, J. C., DeGray, R. J., and Wilson. J. W., I X D . E X G . C H E Y . . . i S A L . ED.,15,367-73 (1943). ( 5 ) Diller. I. M.. DeGrq.. 11. J.. and Wilson, J. W,, Ibid., 14, 60i-14 (1942’.

793 (6) Gardner, H. A . , and Sward, G. G., “Physical and Chemical Examination of Paints, Varnishes, Lacquers, and Colors,” 10th ed., Washington, D. C., Institute of Paint and Varnish Re-

search, 1946. (7) Haren, Allen,Am. Chem. J . , 14, 300-10 (1892). (8) Osborn, R. H., and Kenyon, W.C., IND.ESG. CHEM.,;IN.~L.

ED., 18, 523-30 (1946), (9) Story, B. W., and Kalichevsky, V. rl., Ibid.,5 , 214-17 (1933). (10) U‘hyte, L. K., J . Am. Oil Chem. Soc.. 24, 137-40 (1947). RECEIVED A u g u s t 23, 1948.

Determination of Oli in Oil-Field Waste Waters H. DARWIN KIRSCHRI.4N, Lnicersity of California, Los Angeles, Calif.,

4hD

RICHARD POTIEROY,

.Montgomery and Pomeroy, Pasadena 1 , Calif. In anal5ses of oil-field waste waters “oil” is considered as the relativelj nonvolatile liquid component that contributes to the formation of oil films and deposits. The nonsaponifiable hexane-soluble fraction most closelg represents “oil content” in this sense. In anj analjtical procedure that uses a solvent, renioial of this so1,ent prior to weighing or iolumetric measurement must be accomplished in a manner to minimize evaporation of the oil. Drging the oil at room temperature after boiling off the solient has been found satisfactorj. In the wet-extraction method, samples of low oil content (below- 25 p.p.m.) jield about 9 5 q ~of the oil in two extractions and 999’~ in three; higher percentage recoiery is attained at higher oil contents. Naph-

T

HE determination of oil in oil-field waste waters is of considerable importance where pollution control agencies have set up standards of quality for wastes discharged into bodies of natural water, or where the efficiency of facilities for waste water treatment must he evaluated. In general, procedures previously recommended have not been investigated with enough thoroughness to provide an adequate basis for judging their accuracy and suitability: the results of such investigations, if any have been made, are not available in published form. This paper reports the results of an investigation undertaken to develop procedures that will yield valid results in terms of a reasonable understanding of the objectives of the “oil” determination. Two basic plans, the “wet extraction” procedure and the “flocculation” procedure, have been elaborated and compared. C O S S T I T U E N T S CONSIDERED A S OIL

Before evaluating analytical procedures for determination of the oil content of waste water, a decision should he made as to what materials are to be included in the category of oil. Oilfield waste waters contain all’the hydrocarbons of natural gas and oil from nietharie up, plus asphaltic and resinous materials, fatty acids, and other organic matter. The hydrocarbons having boiling points above normal atmospheric temperature-i.e., isopentane (2-methylbutane) and heavier-could he considered as constituting the oil component of the sample. \Then the oil is separated by solvent extraction, either with or without the aid of flocculation, there is no practical \yay of separating the solvent from the extracted oil without the loss of highly volatile fractions. In one published procedure for determination of oil in waste waters an attempt is made to overcome this difficulty by partially distilling the sample and separating the distilled oil from the water by gravity, with a subsequent volumetric measurement

thenic acids are extracted with the oil and must be separated by saponification. -4new procedure gives results differing by an average of less than 1 p.p.m. on duplicate samples. Ferric chloride is unsuitable as a flocculating agent, because it produces sulfur in the extract if the waste w-atercontains sulfide. Zinc salts are satisfactxy. Results by the flocculation method are reproducible with an average difference below 1 p.p.m. on duplicates, but the results are about 2 p.p.m. lower than those of the wet-extraction method, because of incomplete trapping of the oil by the floc and incomplete removal of the oil froni the floc. Because the wet-extraction method gives results of superior accuracy in less elapsed time, the authors consider it generally preferahle.

( 1 ) . But even in this procedure, a small amount of ethyl ether is used to collect all the oil in a capillary tube. Because of the necessity of evaporating this ether, all the pentanes and a t least part of the hexanes are lost. Some type of distillation procedure may he the only way to get accurate results as far as highly volatile components are concerned, hut any such procedure, if exact, is certain to he complicated and too lengthy to he of value for routine testing. The best way out of this difficulty is to ignore hydrocarbons that evaporate rapidly a t room temperature, and to consider that “oil content,’’ for practical purposes, means the oil that remains as such after moderate exposure to the air. This is a logical view, in so far as the pollutional character of the water is concerned, for the highly volatile oils do not contribute to oily films on the surface of polluted streams. The volatile hydrocarbons may contribute to explosive atmospheres in sewers, hut this condition may result from the presence of hydrocarbon gases as well as vaporizable oils. From an analytical standpoint this is a problem entirely separate from oil content. The final delimitation of the category designated as “oil” rvill depend in part upon the practicability of laboratory procedures. The objective chosen in this research Mas the separationof the components that are ielatively nonvolatile liquidsat room temperature and contribute to the formation of oil films and deposits. With this in mind a solvent should be chosen which will dissolve the oils effectively, hut ~villhave a minimum solvent action on nonliquid materials. In drying the separated oil prior to weighing, a procedure should he followed in which it will he possible t o evaporate the solvent used in the analysis, together with fractions of the oil similar to the solvent in volatility, hut without too great evaporation of heavier fractions. Solvent and drying technique must he chosen before comparisons of extraction procedures can be made.