Direct Extraction-Pycnometer Method for Oil Content of Refinery

Determination of Oil in Refinery Effluent Waters. J. B. Rather , J. A. Grant , V. H. Gunther , C. E. Headington , J. H. Karchmer , S. S. Kurtz , A. R...
8 downloads 0 Views 3MB Size
Direct Extraction-Pycnometer Method for Oil Content of Refinery Effluents W. S. LEVINE, G. S. MAPES, AND M. J. RODDY Technical Seraice Department, Socony-Vacuum Laboratories, Brooklyn 22, N . Y . Existing methods for determining the oil content of waste water are not entirely satisfactory. If rapid, they are inaccurate; if accurate, they are time-consuming. This report describes a rapid and accurate method, whichis especially applicableto waste waters having appreciable quantities of volatile material. The oil is extracted with carbon tetrachloride. The extract, after filtration and separation from entrained water, is run into a two-armed pycnometer

T

HE disposal of waste from manufacturing and refining processes into waterways always has been a serious problem to industry and adjacent communities. Many industries have converted their wastes into marketable by-products. Despite the great progress made in this direction, the problem is of everincreasing interest to communities using water for recreation, transportation, and human consumption. In the petroleum industry, refineries have set up elaborate installations to minimize the amount of waste products getting into neighboring streams. The efficiency of such installations is checked a t regular intervals by analyzing the ivaste water leaving the refineries and the water in adjacent streams. An important phase of this program is to determine the oily material content in these waters. For this purpose, numerous methods have been suggested. These fall into four general categories: Methods in which the oil is adsorbed on a fresh precipitate, usually ferric hydroxide, extracted from the precipitate with a volatile solvent. and weighed after evauoration of the solvent (i,6, 9, IO, !I).' Methods in which the oil is extracted from the water with a volatile solvent and weighed after evaporating the solvent (3, 5, la). Methods in which volatile constituents are se.p?rated by boiling the water sample, after which the nonvolatile oil is extracted from the remaining sample with a volatile solvent. The oil is weighed 8). after evaporation of the solvent (1, I, Methods in which the oil is separated from the water by solvent extraction or gas stripping. The solution or gas stream is then analyzed with an instrument such as an infrared spectrophotometer (11) or a mass spectrometer ( 7 ) . Y

Of the methods reported, the American Petroleum Institute has recommended two procedures in its latest publication ( 2 ) . These are the flocculation-extraction U-tube method and reflux distillation-extraction method. The U-tube method uses ferric hydroxide to collect the oil and diethyl ether to extract it from the precipitate. The procedure is rapid, but if volatile material is present, it gives low results. In the reflux distillation-extraction method a sample is boiled to remove volatile material and the remainder is extracted with benzene. The solvent is evaporated-the last traces in a special flask-and the residue weighed. The method is more accurate than the U-tube method and determines both the volatile and nonvolatile oil present. However, the procedure requires about 2 hours for a determination and the use of bulky equipment such as 3-liter distillation flasks, and Pliter separatory funnels. Apparatus of this size requires considerable bench apace and is awkward to handle when i t has to be picked up and shaken. -4detailed and extensive discussion of the nonvolatile oil phase of this procedure can be found in a recent article by blusante (8).

and weighed. The oil solution will weigh less than an equal volume of pure solvent. From this difference in weight and an assumed value for the density of the oil (based on prior measurements) the oil content of the sample is computed. A nontechnical man using two extraction units can do two determinations in 1.5 hours. With all types of samples, including a gasoline with initial boiling point of 100' F., the average deviation from known values is less than 6%.

Thus, although many methods have been recommended for this analysis, no method is available that is rapid, accurate (whether the oil has little or a high percentage of volatile constituents), sensitive to about 1 p.p.m., and simple enough to be run by nontechnical personnel using inexpensive equipment. The purpose of this investigation was to develop a procedure that has these characteristics to a greater degree than other methods. However, this method is not sensitive to 1 p.p.m. of oil but only to 10 p.p.m. THEORY

This method is based on the fact that carbon tetrachloride containing a small quantity of oil will weigh less than an equal volume of pure carbon tetrachloride. The difference in weight is the difference betmen the n-eight of the oil dissolved in the carbon tetrachloride and the n-eight of a volume of solvent equal to that of the oil. If x represents the grams of oil in a sample and d the assumed density of the oil, then x/d is equal to the volume of 1 oil and to the volume of nn carbon tetrachloride displaced by the dissolved oil. Since the density of the water-saturated carbon tetrachloride is 1.59, the weight of the carbon tetrachloride displaced by the oil is 1.59 x/d The net loss in weight represents the difference bet w e e n t h e w e i g h t of carbon tetrachloride displaced and the weight of the oil. Hence,

where

W,

Figure 1. Pycnometer for Determination of Oil in Refinery Effluents 1840

weight in grams of solution of oil in solvent WI = weight in grams of a n e q u a l volume of solvent =

V O L U M E 25, NO. 1 2 , D E C E M B E R 1953 This equation involves two assumptions: (a) formation of an ideal solution when the oil dissolves in the carbon tetrachloride, and (a) value for the density of the oil. While the latter assumption is a disadvantage of the method, the same assumption is made in calculating volatile oil contents in the API distill% tion-extraction method ( I ) and the API reflux distillation-extraction procedure (8). In these methods a value of 0.88 is assumed. In this procedure the value used is 0.87, the density of oil taken from an oil separator. ~

1841 Stirrer, any glass or metal stirrer that can tit into the globe sePamtorY funnel. A paddle type has been found suitable. It should extend to within 1.5 inches from the bottom of the funnel. A variable high-speed motor, clamped to a separate ring stand t o minimize vibrating the rest of the apparatus, drives the stirrer which should be rotating a t abaut 2400 r.p.m. Fritted-glass filtering fungi, a coarse porosity funnel h a v h g a

c a ~ ~ [ ~f { ~ ~~ $ ~ ~~ o y $& ~ ~ ~c ~h~ $ ~ ~~side ann is connected to the fritted-glass filtering funnel inlet tube with rubber tubing. The flask is tilled with glass wool and closed with a two-hole cork StODDer. A lone elass tube extendine almost to the bottom of the fladk'throuah on: Eole is connected bvmeans of rubber tubine to source of Gmpressed air. The otLer hole remains open a i d is closed with a tinger when i t is desired to apply pressure to the mixture in the fritted-elass funnel.

a

..

boik stopper. Phosphoric acid solution. Dilute 10 ml. of concentrated acid t o 100 ml. with distilled water. Paper pulp. Whatman ashless tablets are suitable. Preparation of Apparatus. A photograph of the auparatus D

I .

hole of a twb-hole cork stopper, grooved down its d e , which fits into the fritted-&sa funnel. A gli~ssair inlet tube, just passing

through a one-hole cork &per

also slotted down the side. Any

pycnomete r arms. A piece tubing in which a plug of qualitative filter . . ,ofm w Tygon i& mnnerta t h e fiinnel tn the nvmnmoter Tho paper 18 mu...".. "___ plug is made by rolling up, fairly tightly, a piece of tilter paper about 0.75 inch square and folding it once. Use a new plug for each determination. The paper plug removes any water droplets that may be suspended in the solution passing into the pyenometer. Pycnometer Calibration. Fill the pycnometer with potassium dichromabsulfuric acid cleaning solution and allow to stand for about an hour. Pour aut the ;leaning solution and wash the pycnometer thoroughly with distilled water. Rinse i t three t o four times with C.P. acetone and dry by suction, connecting the center arm of the pycnometer to the suction source. Dip the pycnometer in a beaker of acetone, dry carefully with a clean, lintless cloth and weigh to the nearest 0.1 mg. using a tare. (The bulb of a 50-ml. pipet sealed a t both ends and supported by insertine one end into a rubber-lined metal or wooden stand makes a suhb Add to the adjust Allow thi uhases t o szuarate. Close the cylindrical f t n n d stbucock pen-tbatof the l&ge separatory funnel and run solvent into the h i tted glass filter until i t is about two thirds full. Whc?n t,he carbon tetrachloride filters into the lower funnel. open t he stopcock and permit the solvent to flow into the pycnnml eter at, a rate of about 5 to 10 ml. per minute until about 0.5 ml. reman tinue manne top of the graduations. The &de arm of the pycnometer is always 1mnnected to the extraction apparatus when being filled to avoid trapping air. Dip the pycnometer in acetone, dry with a clean lintless cloth, and w eigh to the nearest 0.1 mg. using the tare. Place the pvcnoineter in the constant temperature bath. When the level bf the solvent in the pycnometer arms becomes constant, record the BUIrn of the heights of the solvent in the arms, reading to the nearetit 0.5 mm. Pou P a drop of solvent out of the center arm of the pycnometer and re]peat the weighing as described above. Continue to do this until t.hree or four sets of readings are obtained. Empty the l _ l l _ ~ _ ll _ _ l _ I

m y iaooratory using mm metnoa mouia not assume mat tne oil in its waste water has this density. Wherever possible, each laboratory should determine its own density values. This may be done by skimming some oil from the surface of an oil separator, centrifuging, and measuring the density of a portion of the oil phase. The weight of carbon tetrachloride equal in volume to the solution of oil in carbon tetraohloride is measured with a two-armed pycnometer (Figure l), which is an adaptation of one described by Lipkin, Davison, Harvey, and Kurtz ( 6 ) . Actual volumes of pure solvent or solutions are not measured, but 8. C U N ~ relating weight of carbon tetrachloridein the pyonometer to the sum of the heights of solvent in the arms of the pycnometer is prepared. Thus, when the pycnometer is filled with B solution of oil in carbon tetrachloride, the weight of an equal volume of the pure solvent is that weight, read off the curve, that corresponds to the sum of the heights of the sample solution in the arms of the pycnometer. ANALYTICAL METHOD

Ap aratus. Constant temperature bath, a conventional bath used !or determining the viscosity of oils. This consists of a circular glass tank having a capacity of about 5 gallonsgan ASTM Saybolt viscosity thermometer (17 F-51), 66" to 80 F. range, electric stirrer, and suitable thermostatic controls and heaters t o maintain the bath temperature at 77' zt 0.08" F. The bath is equipped with cooling ooils through which cold water may be run when necessary. A copper or brass platform, with a surface 6 X 4 inches and a height of about 3.5 inches is placed in the bath. Tho water level in the bath is adiusted so that when the pycnom-

r_l

_i_

'1842

pycnometer. Repeat the filling, weighing, and bath immersion steps until a total of six to eight sets of readings are obtained, Calculate the weight of carbon tetrachloride in the pycnometer by subtracting the weight of the empty pycnometer. Plot the weight of solvent against the corresponding sum of the heights of the solvent in the pycnometer arms on Cartesian coordinate paper, Draw the best straight line through the points. Procedure. Collect the sample in a quart bottle according to API recommended procedure (1) until the bottle is nearly full. Make a crayon mark on the bottle to indicate sample volume. Add 25 nil. of carbon tetrachloride to the sample, stopper with a metal foil-covered cork, and shake the mixture well for about 2 minutes. Transfer the contents of the bottle to the 1-liter separatory funnel. This can be done without lifting out the stirrer by using a glass funnel having two right-angle bends in the stem (Figure 2). Add 3 ml. of the phosphoric acid solution and about 0.5 gram of paper pulp (one fourth of a compressed tablet broken into small pjeces) to the sample to minimize the formation of stable emulsions. Start stirrer and adjust motor speed so that the mixtureis uniformly opaque. While the mixture is being stirred, add 20 ml. of carbon tetrachloride to the sample bottle, stopper, and shake vigorously for 2 minutes. Stop the stirrer after 15 minutes, and allow the two phases to separate. Run the lower phase into the fritted-glass filter, making certain that the lower separatory funnel stopcock is closed, Allow any emulsion in the carbon tetrachloride phase as well as a little of the paper pulp to run into the filter. Close the stopcock. Add the 20 ml. of solvent in the sample bottle to the large separatory funnel and repeat the agitation of the mixture as described above. Meanwhile, let the first extract go through the fritted-glass filter into the lower separatory funnel. If filtration is slow, apply a little pressure to the solution. To do this, connect the air inlet tube in the fritted-glass filter stopper to the side arm of the filtering flask and close the open hole with a finger. For best results, minimize the time pressure used to push the solution through the filter. Although it is necessary to process the carbon tetrachloride solution out of the fritted-glass funnel before running in the next extract, it is not necessary to have all the water accompany the first extract going through. If pressure is used to aid in filtering the carbon tetrachloride solution, the pressure should be cut off when the water layer reaches the fritted-glass surface. Open the stopcock of the lower funnel. Adjust the flow rate so that 3 to 5 minutes are required for all but about 0.5 ml. of the lower phase to flow into the pycnometer, Repeat the separation of phases and add a third 20-ml. portion of carbon tetrachloride to the sample bottle, which is then processed through the apparatus in the same manner. When the third extract is in the lower funnel, run enough of it into the pycnometer to fill the latter to the desired height. The height depends on the room temperature. If room temperature is around 26" C., fill the pycnometer to about the middle of the arms. If around 20' C., fill it to the bottom of the graduations, and fill to the top of the graduations if room temperature is near 30" C. A little experience will enable an operator to fill the pycnometer to the proper height. When the pycnometer is filled to the desired height, there should be a few milliliters of solvent remaining in the lower separatory funnel. If more carbon tetrachloride is needed to fill the pycnometer, add 5 t o 10 ml. of the water-saturated solvent to the 1-liter separatory funnel and carry it down through the extracton apparatus. Bgitation is not needed. Dip the pycnometer in acetone, wipe with a lintless cloth, and weigh to the closest 0.1 mg. using the tare. Subtract the weight of the empty pycnometer t o get the weight of the solution. Place the pycnometer in the bath. When the level of the liquid in the arms becomes constant, note the height in millimeters in each arm to the nearest 0.5 mm. and add them. Read from the calibration curve the weight of carbon tetrachloride corresponding to this sum. Fill the sample bottle to the crayon mark with tap water. Pour the water into a 1-liter graduated cylinder. The volume of water in the graduated cylinder equals the volume of sample analyzed. The original sample volume can not be measured directly without risking loss of oil. Blanks. When a new batch of carbon tetrachloride is used, make a blank run exactly as described under pycnometer calibration, omitting the repeated weighings and taking the weight a t only one liquid level. If the weight of the reagent is the same for the sum of the liquid level heights as that from the calibration curve, the new batch of solvent has no blank. If the weight of new solvent in the pycnometer is greater than that read off the curve, the blank on the new solvent batch is positive and the difference between the two weights is added to all subsequent weights read off the curve. Conversely, if the weight

A N A L Y T I C A L CHEMISTRY Table I. Type Oil SAE X o . 10

Results on Synthetic Samples

Oil Added, AIg.

Oil Found, hlg.

31.5 33.0 62.6 74.0 97.7 109.0 205.5

31.3 32.0 63.3

Recovery, %

74 .. n .

99 97 101

on

1 .. .

94.3 105.0 192.0

96 96 94 .4verage recovery 97.5

Separator oil

21.2 43.4 62.5 95.6 200.0 226.4

113 102 102 94 97 91 Average recovery 9 9 . 8

Kerosine (initial boiling point 330' F.)

29.0 57.0 84.0 92.0 110.2 140.2 200,s

23.4 59.0 84.0 81.2 113.0 118.0 184.7

Gasoline (initial boiling point 100' F.)

25.0 36.9 51.0 63.2 80.1

24.0 44.2 63.7 90.1 194,7 205.6

88 104 100 88 104 84 92 Average recovery 9 4 , 3 25.0 26.8 39.0 48.7 73.0

100 73 77 77 91 Average recovery 8 3 . 6

of the solvent in the pycnometer is less than that read off the curve, the blank is negative and is subtracted from subsequent weights read off the curve. However, with C.P. carbon tetrachloride, the blanks from one batch to another should be small. Calculations. The oil content of the sample is calculated by means of the following equation: Oil, p.p.m. = 106 X

( W , i B - W,) V (1.59/d - 1)

where d = assumed value for the density of the waste oil V = volume of sample taken for analysis, milliliters Ti71 = weight of carbon tetrachloride plus oil in the pycnometer, grams W z = weight of carbon tetrachloride from the calibration curve, grams B = blank correction, grams Report oil content of the sample to the nearest part per million. RESULTS OBTAINED

The method has been tried on samples prepared by shaking known amounts of oil with tap n-ater and on a large number of refinery effluent samples. Table I shows typical results obtained with an ShE 10 oil, oil reclaimed from separators, a kerosine, and a gasoline. The recovery of the SAE 10 and separator oil is good-9T.5 and 99.8%, respectively. Kerosine recovery is satisfactory, while that for gasoline is someFvhat low. However, considering the volatility of this product, the recovery is fair. A factor tending to lower recovery of the gasoline is the difficulty met in weighing out milligram amounts of the gasoline and transferring them to the sample bottle. About 84% is the recovery achieved on a regular gasoline. In waste water, the bulk of the lo^ boiling material weathers away, so that such samples nil1 never contain the quantities of light material put into these synthetic samples. Three oils, blended for hmerican Petroleum Institute cooperative work on oil-in-water methods, were also used to prepare synthetic samples. Table I1 shows the results obtained with these oils. As is the case with other synthetic oil-in-water samples, those made with the SAE 70 oil and the mixture of this oil with kerosine showed good recovery. The percentage recovery on sample 3 runs were not quite as good as for those made with samples 1 and 2, but considering the volatility of the gasoline they are satisfactory. However, if one recalls that the percentage recovery on a straight-run gasoline u-as 83.6% (Table I), better recovery on this sample is believed possible. This was verified by the coop-

V O L U M E 2 5 , NO, 12, D E C E M B E R 1 9 5 3

1843

erative test results obtained with this method and summarized by Headington (4). His data show that with six independent laboratories reporting the average recovery on sample 3 11 as 101%. Table 11. Analysis of Cooperative Samples Oil .4dded, hlg.

Type Oil S.4E 70 (No. 1)

Y.4E 70 kerosine ( N o . 2)

l/s

18.6 49.5 55.7 110.0 110.0 112.1

+

1 / ~

S.4E 70 kerosine gasoline ( S o . 3)

I/J 1/a

++

56.3 59.6 60.7 61.8 103.0 105.4

Oil Found, 32g. Recovery, 16.1 87 47.4 96 53.6 96 99.5 90 102.5 92 115.8 102 Bverage recovery 94 54.5 97 56.4 93 59.1 98 61.8 100 Average recovery 98 86.4 84 93.0 88

r0

1/a

Average recovery

86

A large number of refinery effluent samples were run to test the method in routine operation. The results were compared with values obtained on duplicate samples by the -4PI U-tube method (2). The work was done by a nontechnical operator and Table I11 shows some of the comparison values obtained. On the average, results by the direct extraction method are about twice those by the U-tube method. Of course, the more volatile material in the sample, the greater sill be the difference. Another factor that must enter into such a comparison is eampling. The samples for this comparative work were taken a t the same time and every effort made to get duplicate samples. Yet, as shown in Table IV, some results can only be due to poor sampling. Such data emphasize the fact that sampling must be done with great care to obtain significant results.

Tests showed that three extractions, each of 15 minutes’ duration] are enough to extract the oil quantitatively from a water sample. The first extraction removes about 75% of the oil, and the second one removes all but traces of the rest. The third extraction picks up traces of oil that may still remain and washes down any oil-bearing solvent on the sides of the fritted-glass filter and lower separatory funnel. The ability of carbon tetrachloride to extract quantitatively the oil from water samples is also proved by Headington and coir-orkers in their infrared method (11). While this investigation waq concerned with a method for determining the total oil content of refinery effluents, others may wish to determine the volatile and the nonvolatile oil content. This method can easily be adapted for this purpose. hfter determining the total oil, the pycnometer contents are poured into a weighed beaker. The solvent is evaporated on a steam plate and the residue dried t o constant weight. The weight of oil in the beaker is the nonvolatile oil. The difference betn-een this value and the total oil content is the volatile oil present. Of course, the nonvolatile oil is not all oil, but may include other material suspended or dissolved in the water that is partially or wholly extractable with carbon tetrachloride. Some refineries are intereqted in an estimation of the acidic content of the extracted oil. I n the -4PI distillation-extraction method ( 1 ) and the reflux distillation-extraction method (8)] the oil remaining after evaporating the solvent is diqsolved in a suitable solvent and titrated with standard alkali. In a similar fashion, the nonvolatile oil obtained above may be titrated for its acidic content. For best results, a few precautions must be taken in cleaning the apparatus after a determination and before starting another. The large separatory funnel needs only rinsing with water. The lower separatory funnel, after remaining solvent and water have been poured out, should be rinsed twice with carbon tetrachloride. It is not necessary to r i p e it dry and the use of acetone to clean it is not advisable.

DISCUSSION

rl basic advantage of the method is its applicability to samples with volatile constituents. Seperal factors make this possible. 1. S o heating steps are involved. 2. S o suction is required; this is important because it is the step involving the drying of the ferric hydroxide precipitate by suction that promotes the inaccuracy of the U-tube method with samples containing volatile constituents. It has been found that even with relatively nonvolatile oil, some is lost if suction is applied for 10 to 15 minutes to the precipitate on which the oil is adsorbed. 3. In the use of pressure to force the carbon tetrachloride through the fritted-glass filter, the compressed air strikes a thin layer of water and not the carbon tetrachloride. This minimizes the loss of volatile matter by the use of the compressed air.

One factor in this method is the need to assume a value for the density of the oil; other oil-in-water methods make a similar assumption and an average value is not hard to obtain. The largest error that could result from an incorrect assumption, if 0.87 were assumed and the density was really 0.80, would be results that are 10% too high. If 0.87 17ei-e assumed and the density was 1.00, results would be 18% lo^. However, these are extremes. In practice the range of density is much less. -4more realistic range would be 0.81 to 0.89. In special cases it might be desirable to eliminate altogether the need for making this assumption. One way to accomplish this is to run duplicate samples by this method using carbon tetrachloride on one sample and carbon disulfide as the solvent on the duplicate. Thus, two equations are obtained in which the unknowns are the density and the oil content of the sample. By solving the two equations simultaneously, the density and oil content are obtained. Since this scheme requires two determinations, it is not practical for routine work. The method assumes, of courqe, that the two solvents extract the same kinds and quantit? of suhstances out of the sample.

Table 111. Comparison of Results by Direct ExtractionP) cnometer 3Iethod and .4PI U-Tube Method Oil, P.P.M. Direct extractionpycnometer method API U-tube method

Table IT.

49 38 14 9

46 28

57 14

52 28

64 42

43 23

49 42

28 9

31 19

148 150

Comparison of Results on Apparently Nonrepresentative Samples

Direct extraction method U-tube method

2379 42

Oil, P.P.M. 83 152 140 1874-

950 103

The fritted-glass filter is cleaned by removing the layer of paper pulp, rinsing twice with carbon tetrachloride, and wiping dry. Then strong suction is applied to its stem until bubbles of air are no longer visible on the underside of the fritted glass. The filter is then wet with a little carbon tetrachloride. This treatment is important in maintaining the rapid filtering characteristics of the fritted glass. If filtration is slow, i t usually means the fritted glass is wet with water. Drying with suction will improve its operation. Acetone should not be used to clean the filter, as it appears to be difficult to get all the acetone out of the fritted glass with suction. The pycnometer is cleaned by pouring out its contents, rinsing twice with C.P. acetone, and applying suction to the center arm for about 2 minutes. The acetone is added from a buret or separatory funnel connected to the side arm of the pycnometer by means of Tygon tubing. It must be perfectly clean for the next run. ,4n important feature of any method for determining oil in refinery effluents is the ease with which i t can be handled on a routine basis. -4 nontechnical operator working with two ex-

1844

ANALYTICAL CHEMISTRY

traction units can analyze two samples in 1.5 hours of over-all time. This figure includes the time required to clean up the apparatus for the next determination. The operator time required is about 40 minutes, which also includes cleaning time. Nevertheless, the method is not as rapid as the U-tube method, since the same nontechnical operator can do six determinations in an over-all time of 2.5 hours, of which only 50 minutes is operator time. However, results by the direct extraction method are much more accurate. Some samples, after standing for several days, required the use of compressed air to force the bulk of the carbon tetrachloride solution through the fritted glass. Duplicate samples analyzed within a few hours of receipt gave no difficulty a t all. It is therefore recommended that samples be analyzed as soon as possible after they are received. LITERATURE CITED

American Petroleum Institute, “Disposal of Refinery Wastes,” 4th ed., Sect. 1, Appendix 5, 1949.

(2) American Petroleum Institute, “Manual on Disposal of Refinery Wastes,” Vol. IV, 1st ed., pp. 81-8, 1952. (3) American Public Health Association, “Standard Methods for the Examination of Water and Sewage,” 9th ed., p. 42-3,1946. (4) Headington, C. E., ANAL.CHEM.,25, 1681 (1953). (5) Kirschman, H. D., and Pomeroy, R., Ibid., 21, 793 (1949). ( 6 ) Lipkin, M. R., Davison, J. A,, Harvey, W. T., and Kurte, S. S., Jr., IND.ENG.CHEM.,AKAL.ED., 16,55 (1944). (7) Melpolder, F. AI., Warfield, C. W.,and Headington, C. E , ASIL. CHEM.,25, 1453 (1953). (8) llusante, A. F. S., Ibzd., 23, 1374 (1951). (9) Noll. C. A. and Tomlinson. W. S.. IND.ENG.CHEM, L hED.. ~ ~ . 15,629 (1943). (10) Scott, W. W,, “Standard Methods of Chemical Analysis,” Vol. 1 1 , 5th ed., p. 2078,New York, D.Van Nostrand Co., 1939. (11) Simard, R. G., Hasegawa, I., Bandaruk, W., and Headington, C. E., ANAL.CHEW,23, 1384 (1951). (12) Snell, F. D., and Biffen, F. M., “Commercial Methods of dnalysis,” p. 253,New York, McGraw-Hill Book Co., 19+4. RECEIVED for review May 6, 1953. Accepted August 28, 1953. Presented before the 18th Midyear Meeting of the American Petroleum Institute, Group Session on Analytical Research, New Yorh, May 1953.

Uncombined Calcium Oxide or Hydroxide in Lime and Silicate Products Volumetric Determination GUNNAR 0. ASSARSSON AND JEAN M. BOKSTRORI Chemical Laboratory, Geological Survey of Sweden, Stockholm 50, Sweden The methods for the volumetric determination of uncombined calcium oxide and hydroxide extracted from lime and silicate products by glycerol, ethylene glycol, and acetoacetic ester have been studied. The best results are obtained by conductometric titration with a strong acid. Reasonable values can also be obtained using an indicator: methyl red for glycerol and ethylene glycol extracts and bromophenol blue for the acetoacetic extracts. When an accelerator such as strontium nitrate is used in the glycerol extraction, the only convenient method is titration with an indicator.

0

NE of the most important determinations in the lime and

cement industry concerns the percentage of lime occurring as calcium oxide or hydroxide in the products, the so-called free or uncombined lime. Much effort has been expended in attempts to find suitable methods for this determination. h summary is found in modern handbooks on the lime industry (3). In connection with some investigations on pozzuolanas the authors required a tolerably accurate method for the determination of uncombined lime, but after a scrutiny of the methods described in the literature and after some experiments it was found that some of them are unsuitable and most of them need improvement. Thus the first step in this investigation was to find the conditions under which the determinations were sufficiently accurate and the analytical work could be best adapted. There are two chief groups of methods. The first involves the measurement of ignition loss (determination of calcium hydroxide) or of the heat of hydration of an anhydrous or an ignited sample a t well-defined humidification conditions. The second group includes methods based on an extraction of the uncombined lime with a suitable solvent. The former methods cannot be used in the presence of kinds of pozzuolanas usually encountered and are not discussed here. The extraction methods are suitable only when the solvent is sufficiently selective, completely dissolving calcium oxide or hydroxide without decomposing hydrated silicates or aluminates. Several solvents have been proposed and investigated. The

most acceptable are glycerol, acetoacetic ester, and ethylene glycol. The extracted lime can be determined volumetrically or gravimetrically, but in view of the difficulties encountered during the precipitation of calcium from organic solvents, volumetric methods have been exclusively employed. The problems involved are of such embracing nature that it was necessary to limit the scope of this investigation. An examination of the volumetric determination of calcium oxide or hydroxide dissolved in the solvents has now been undertaken, while the actual extraction methods for pozzuolanas must be treated separately. REAGENTS AND APPARATUS

The titration solutions used in the experiments described were: Hydrochloric acid (aqueous solution), 1 ml. equivalent to 10 mg. of calcium oxide. Sulfuric acid (ethyl alcohol solution), 1 ml. equivalent to 10 mg. of calcium oxide. Benzoic acid (ethyl alcohol solution), 1 mg. equivalent to 1 mg. of calcium oxide. Ammonium acetate (ethyl alcohol solution), 1 ml. equivalent to 2 mg. of calcium oxide. These solutions were calibrated in aqueous solution according to standard methods-that is, they were not standardized by carrying out a control titration of a known amount of calcium oxide under the conditions of the determinations. Usually a microburet was used for the acids.