Analysis of Phosphorus Compounds Automatic pH Titration of Soluble Phosphates and Their Mixtures J. R. VAN WAZER, E. 1. GRIFFITH, and J. F. MCCULLOUGH lnorganic Chemicals Division, Research Department, Monsanto Chemical Co., Dayton, O h i o
points after hydrolysis measures the total phosphorus (total phosphorus pentoxide). I n addition, the third hydrogen of the orthophosphate, which is normally not titratable, can be released for p H titration by use of a precipitating cation, as in the method of Gerber and Miles (6). Thus, three titrations (before and after hydrolysis and after addition of silver ion) can he used in defining a mixture of ortho- and polyphosphates.
The increasing commercial importance of condensed phosphates and the absence of really good analytical methods led to an investigation of the method of pII titrations as applied to mixtures of condensed phosphates. Titrations for total phosphorus pentoxide, as well as for end-group phosphorus pentoxide, are precise to better than 5 parts per thousand; but the orthophosphate titration, using precipitation by silver, is accurate only within about 5 parts per hundred. The method as applied to mixtures of ortho-, pyro-, and tripolyphosphates is simple and relatively rapid, and requires a minimum of attention from the analyst.
T
APPARATUS
HE earliest differentiation between the various types of phosphates was made a century ago by Graham ( 7 ) , who classified the phosphates into three groups, called ortho-, pyro-, and metaphosphates. Many authors following Graham employed this threefold scheme of classification (6). However, the existence of crystalline sodium tripolyphosphate and the obviously radically different varieties of sodium metaphosphatrs ledto a broadening of the analytical schemes ( 3 , 4,8). These later, more inclusive wet-chemical procedures were still not vrrv satisfactory and, in essence, consisted only of solubility fractionations, to each fiaction of which the name of a specific species is rather arbitrarily given ( I S ) . The main obstacle to devising schemes for the anal? sis of phosphate mixtures soluble in n-ater rests in the chemical similarity of the molecule ions present. Modern studies and interpretation indicate that three, and only three, classes of phosphate? are stable to any evtent in aqueous solution ( 1 2 ) : (1) the monophosphates usuallv called orthophosphates, (2) the straight-chain phosphates (usually called polyphosphates, which include thc well-known pyro- and tripolyphosphates), and (3) the simple single-ring phosphates (usually called ring metaphosphates, of which only two species, trimeta- and tetrametaphosphnte\, are now known) All of the phosphates are made up of unit 4.5 end point ( 7 )
or ( 4 h
% N a 2 0 = 3,10----
- E')-V for original pH it'
< 4.5 end point
Average chain lrngths of polyphosphates (not containing
By using the appropriate factor, ions other than sodium can be
9l
.4 stock solution of orthophosphoric acid was prepared containing 0.2600 millimole per gram. A sample to 1w titrated was pipetted
0
(3)
Percentage of total PZO; as middle groups = 100
Special Procedure for Long-chain Compounds (13). I n some cases the Method 2 titration shows no orthophosphate, and the Method 1 titration is considerablv smaller than the Method 3 titration. This indicates a long-chain and/ or a ring phosphate. The exact number-average-number of phosphorus atoms per chain in a mixture of polyphosphates (chain, but no ring phosphates) can be determined by dissolving 5 to 50 grams of phosphate in 100 to 250 ml. of water, h Waring Blendor is convenirnt for this operation, Most of the solution is measured into a beaker and titrated as in Method 1. A IO-ml. aliquot is hydrolyzed and analyzed for total phosphorus pentoxide. Because the solution is analyzed before and after hydrolysis, the exact weight of the original sample and the losses ill the Waring Blendor are of no importance. This general procedure can be used for chains of any size, assuming that rings and orthophosphate are a k e n t . For the short chains, only a few grams of phosphate are required, whereas for the longest chains nieasurahle by end-group IO titration, as much as 50 to 100 grams must br' titrated. DETERllIYATION O F TITROlIETER PRECISIOY
(2)
2
4
6
8
IO
12
MI. of bose added Figure 1. Typical W e a k Acid Titration Illustrating method employed to locate end points
14
1757
V O L U M E 2 6 , N O . 11, N O V E M B E R 1 9 5 4 accounted for. I n the case of mixtures of pure sodium pho+ phates, it is found that
+ %' S a l 0 + %' HsO = 100.0 i 0.5
% total P20a
d = inilliliters of base required t o titrate an original aliquot that ha3 been acidified t o a pH < 4.5 (if not already sufficiently acidic) fyom the end point near p H 4.5 t o end point near pH 9.5. Ah = milliliters of base required t o titrate a hydrolyzed sample from the end point near p H 4.5 t o the end point, near p H 9.5. D = milliliters of base required t o titrate from the original p H of a n aliquot t o the end point near p H 9.5. E = milliliters of base t o titrate from thc end point ca. p H 4.5 t~othe original p H of the sample when the original p H is greater than 4.5. If only standard base is to be employed, the original pH of the sample is noted and then the sample is aridified to a p H < 4.5. The sample is then titratcd with standard base and the milliliters required t o titrate from the p H 4.5 end point to the original p H are equal to E (see Figure 1). An alternativr method is to titrate directly from the original p H to the p H 4.5 end point with standard acid. E' = milliliters of base to titrate from the original pH of the, saniple to the p H 4.5 end point whcn the original pH is less than 4.5. 1V = normality of base employed. S = milliliters of base used to titrate orthophosphate in the silver titration. It7 = weight of saiiiplc per aliquot.
II I
1
I
1
2
4
6
8
1
I
IO MI
Of bas0
I2
I 14
I
I
I
16
18
20
Qdd6d
Figure 2. Effect of Sodium Chloride upon p€I of Phosphate Titrations
The method of pH titr:ttion is most convenient for the antilysis of mixtures of ortho-, pyro-, and tripolyphosphates. Three titrations arc required for. the annl>-ses of mixtures containing all three components. The ivcak :wid concentration of the niixture is obt:tined by treatment 1, tho orthophosphxtr content by treatmelit 2, and the strong acid concentration by treatment 3. The above titrations yield two equations which may I)r solved simultaneoiisly to yicld for tripol~yhosphate
1'
=
A*' - A '
where 1' is millimoles of tripolyphosphate, Ah' is the millirquivalents of strong acid, and A ' i- the milliequivalents of weak :tcid; and for p>-rophosphate
Svvc~r:tlmethods for ohtaiiiiiig ~ i i ( 1pc~intsfrom the snloot,h titr:ition curvcs were investigated. Th(, plot of ApH/nil. 2'snil. of titrant as well as several geoniet ricd construct,ioris was. considered ( 5 ) . It was finally decidrd that, simple bisection of the breaks was :is accurate as any other procedure and was c-onsider-ably niorc rapid. 411 of the data reported herein were obtained by the bisection method, as shown in Figure I , It, has been demonstrated ( 1 4 ) th:it the addition of elec.trol!.tc$ s pH of the soluto .Colutions of condensed phosp1i:ttes l o w e ~ ~the tion. As can be seen froni Figure 2 , this lowering exists ;I,t d l pH vdues between 2 and 10; anti the end points occur at lower ilts th:tn in salt-free solutions. pH values in the presence of excc This means that the ovor-all c~oncwitration of electrolytes :tffccts p H titrations of phosphates when indicators are employed to obtain end points. [This is the basis for the use of sodium chloride in the AACC tleterminatioii of the neutralizing value of sodium x i t i pyrophosphate ( I ) . ] Hoivcvrr, t h e end poitits in Figure 2 are not displacd and thus t'he automitt,ic p H titrations are valid, Pither with or nithout t,hc 1)resenc't~of excess electrorq. Sodium (ahlorid? i* srlnict inirs :tdv:iiitageously eniployrd to ,sh:upeii thv crid points of tiii,:itions in which the change of pH 1wr. millilitr~rof titriiirt d d c d is not large at the end point.. This atltlit ioii rff(~c~tivrly iiicwtisrs tlic strrngth of a weak acid function. Table I is a compilation of the results of 16 titrations of a stock solution of ort,hophosphoric acid at various sample sizes, i.at,es of titrations, and scnsitivities of the Titronieter. The sensit,ivity rontrol autonistic:tlly turns the Titrometer off where there is a suddcn change in pH, when high scnsitivity is e m p l o j d . It is concluded from Table I that the precision of a phosphate t,itratioit is b u t .lightly greater when masimuni sensitivity is r i ~ e d roriip:iretl to no wnsitivity, t,he rate of addition does not infuciice the precirion provided the volume of solution is iiot large, a i d the larger the quantity of solute being titrated the greater the ion of the titration. For solutions requiring 25 nil. of titrant, using a minirnuni sensitivity and a maximum feed rate (ea. 3 ml. per minute), a mean deviation of 2 or 3 parts pcr thousand may be expected. The Precision-Dow Titrometcr is especially suited to titrations hetn-een end points at. ~vhicli ApW/inI. does not, become exceedingly great. Surh end points are found for the phosphates (14, IO), as coml):ti,ed to the sharp type of end point found for the tit,ration of a strong arid \)y a strong base. Tahle I1 s h o w the titration of tlir weak acid function of pure phosphates at maximum f e d ratr anti a minimum sensit,ivity setting, as well as the valuer ol)taiiied 1)y titrating solutions of the same phosphates after they have hcen hydrolyzed completely t,o orthopho$iate hy refluxing thein in aqiieou.: zolutioris near -~ -
~~
Table I.
=
:3;2.1' - .t,,' - SI12
where P is millimoles of pyrophosphate, and S' is milliinoles of orthophosphate oht,ained from treatment 2. RESULTS
Weak Acid 'Titrations of a Standard Orthophosphoric Acid
Feed Rate
\'oIrrint~
((I =
Llnximriin Rate) R R R
3 9 3 4 3 9 9 3 3 (I
The data prese~~trci here are representative of the analyses performed and the following remarks are based upon total experience with the method as well as upon the data presented.
~~
( 0 . 2 I ; O O inilliniole p ~ pram r
(I
J'
~~
3 9
Senpiti\-ity 3Iax.
Max. Off Off
off
Off 3Inx. Max.
Off Off
Max.
Off 51n x . Xlax. Max. Max.
I r a n , Parts) 100
0 2600 n 2606 0 258.5 0 2602 0 2687 0 2606 0 260~ 0 2611 0 2597 0 2598 n 2619 0 2580 0 2601 0 2603 0 2597 0 2600 .4baolute average
0
-0 -0 +0 -0 +0
+n
2 3
1 4 2
2 1-0 4 -0 1 -0 1
+0
-0
An
a
6 I
-0 1 -0 1 0 0 2
1758
ANALYTICAL CHEMISTRY
Table 11.
Titration of Pure Phosphates before and after Boiling in Acid Overnight (18 Hours)
Phosphate KH~POI
SaiPlOi
KaaPa0la.6HtO
Calcd. Weak Acid t o Be Titrated, N e q , Theory Before After boiling boiling 7 347 7 347
7 519
6 303
7.519
9 4j4
Weak Acid Found. Deviation from Meq. Theory, P a r t / 1 0 0 Before After Before After boiling boiling boiling boiling 7 326 7 297 -0 3 -0 7 7 272 7 306 -1 0 -0 6 7 319 -0 4
7 538 7 528 7541
7 501 7 510
4-0 3 +O 1 +03
-0.3 -0 1
6 355 6 334 6 324
9 403
+O
E
9 466 9 431
+O fO
5 3
-0 7 +O 1 -0 3
0 4
0.4
...
.ibsolute averages
-~
,..
Table 111. Titration of Graham’s Salt after Hydrolyzation 0.1492.V Base Required __ t o Titrate a n Aliquot Sample I1 fi~= Sample I (n = 150) IIours MI. Hours 2 1 24 90 7 I 25 01 25 26 25 07 100 100 25 06 ~
Table IV. PzOs, AOAC Molybdate,
%
63.38 52.47 52.51
Av.
52.45
Av.
57.42 57.31 57.44 57.39
100) 111.
15.01 15.06 15 16 15.17
Analyses of Commercial Sodium Tripolyphosphate PgOa, p H Titration,
70
Deviation from RIolybdate AI.,, Part,’100
1,570 1,705
52,59 52.21
4- 0 . 3 - 0,s
1.028 1.028
57.15 5E.31
-
Weight Uspd in P H Titration
The pII titration method for orthophosphate described by Gerber and Miles (6) is much simpler when performed upon the automatic Titrometer (see Figure 3) than u-hen indicators are used. Table V indicates that t’he base used in the silver orthophosphate analysis should be standardized under the conditions of the orthophosphate analysis in addition to the standardization employed in general use. The results of the work reported here were intentionally obtained with base that was standardized with potassium acid phthalate and could be improved by a standardization under the conditions of the silver orthophosphate titrat’ion, as shown by the last column of Table V. The values given in Tables V and VI are averages of three or more individual determinations. Table VI was prepared to determine whether an excess of silver nitrate would improve the method by more completely precipitating the orthophosphate as trisilver orthophosphate. It may be concluded that a large excess of silver nitrate is of little aid in improving t’he analysis. It was reasoned that the escess of precipitate in the titrated solution might absorb hydrogen ions and thus make them unavailable. Attempts to use benzene or methanol in the solution to decrease the adsorption of ions by the silver precipitate not only failed to iniprove the analyses but made them poorer than without the niixed solvents. The inherent errors in the silver precipitation method prompted a search for other cations capahlc of precipitating phosphate and releasing the third hydrogen from orthophosphate. Cadmium, cobalt, lead, mercury(II), and thorium were tried. The cadiiiiuni ion was the only ion that was promising, and as it showed no great advantage over silver ion, this work was discontinued. A recent puhlication (IO)indicates that cerous ions may be suhsti-
0.5 0.2
zero pH for 10 hours or more. This titration is equivalent to the strongly acidic functions of the phosphates and measure.: the total phosphorus peiitoside. The samples may be analyzed after hydrolysis with the same precision as was obtained before they were hydrolyzed. Although the nionopotassiuni ortho) be hydrolyzed, it was included iu phosphate ( K H ~ P O Icannot Table I1 to show that the process of boiling in acid did not alter its titration value. The fact that the deviations in Table I1 are consistently either positive or negative for a given sample inHydrochloric acid added here. 01 dicates that the absolute accuracy is about as precise as the average deviation of 4 parts per thousand. II’ Although the well known molybdate methods-e.g., (2)-for I I I I I I 0 2 4 6 8 10 I2 14 I( total phosphorus pentoxide content are very satisfact,ory, it MI of basadded has been found simpler in many mqes to employ the p H titration after hydrolysis for this determination. As there is one weakly Figure 3. Tj-pica1 Orthophosphate Titration by Silver Precipitation Method acidic hydrogen per phosphorus atom in both pyro- and orthouhosuhate, it is not necessary to . I ~hydrolyze beyond the pyrophosphate stage; and the relTable Y. Ihterinination of Sodium Orthophosphate in RIixtures with Pyrophosphate atively slolver rate of hydrolysis arid Tripolyphosphate of the pyrophosphate ion ( 1 5 ) 6Deviation % SasHPO, p H of Titration, percentage Taken Sa2&O4 from U e a n . Found X is thus not a source of error. AgSO3 .\I& of 1 02 .iddition PsOa NasPsOlo Nail’?O; XasHPOi Found PartsI100 h s an example of t,he time 73 0 0 0 100.0 98.7 -1 3 100 7 0 0 needed for hydrolysis, some e 50 2 -1 6 49.2 30.0 50.0 37.4 0 0 8 16 10 0 10.0 7 J 80 0 J -16 0 8.32 data are presented in Table 1.5.0 60.0 20.0 20.0 18.2 -9 0 3 18 6 111 f o r t w o s a m p l e s of .kbsolute average 7 5 Graham’s salt (long-chain polv0 0 100 0 96 7 -1 3 100 7 130 0 0 0 10 phosphates). A comparison -0 6 50 7 49 7 50 0 50 0 0 0 10 75 0 10 6 10 0 10 0 10 4 +4 0 of t h e AOAC v o l u m e t r i c 10 15 0 EO 0 -1 0 20 2 19 E 10 30 0 60 0 20 0 20 0 molykidate method (2) with the pH titration procedure is given Absolute average 1.7 in Table IV. ~
V O L U M E 2 6 , NO. 1 1 , N O V E M B E R 1 9 5 4
1759
Table VI. Effect of Excess Silver Nitrate, Sample Size, and pH on Determination of Pure Orthophosphate PtOs, 1Ig. 100 0 100 0 100 0 100 0 100 0 100 0 149 7 59 88 60 09 250 0 250 0
Excess AgNOa,
yo
50 50 100 100 200 200 280 290 290 100 100
pH at Additiona 11 0 6 5 11 0 6 5 11 0 6 5 11 0 11 0 11 0 11 0 6 5
Theoretical NnOH Required, All. 10 35 10 35 10 35 10 35 10 35 10 35 15 50 6 20 6 22 25 58 25 58
NaOH
Actually Used, Ml. 9 94 10 08 9 80 9 89 9 95 9 79 14 57 5 95 5 85 24 68 24 89
Deviation from Theory P a r t - ’ion
Absolute average a
Table YII.
sample So.
1
I
- 5 ?>
I1
-4 --o
-4 6
-3 9 -5 6 2 0
-6 -4 -6 -3 -2
tuted for silver ions in the orthophosphate analysis. It is not known, hoxever, how well the method norlts with mixed phosphates. Table VI1 shows typical results for the analyses of mist~uiwof ortho-, pyro-, and tripolyphosphates. If rings or long (*hailis are present’ in the mixtures to be analyzed, they must be det,ermined by independent analyses and their presence redures the effectiveness of the method. Even in the absence of long chains or rings, the analysis is very dependent upon the correct determination of the total phosphorus pentoside. Considerable care must be exercised while preparing an aliquot for hydrolysis and in the subsequent titration. The tripolyphosphate analysis is dependent only upon the strong- and weak-acid titration and is not influenced by t,he analysis for orthophosphate. The analysis for pyrophosphate is dependent upon the orthophosphate analysis, however. I n general, the constituent present in the largest quantity may lie determined x i t h the greatest accuracy. REFERENCES
( I ) ;Im. ;Issoc. of Cereal Chem., “Cereal Laboratory Methods,” 5th ed.. DD. 2 0 2 4 . 1947. (2) hssoc. 0 6 : Agr. Chemists, “Official Methods of Analysis,” 7th ed., p. 334, 1950. (3) Bell, It. S . ,AXAL.CHEY..19, 97 (1947).
30.9 24.5 23.8 26.7 88.9 88.7 90 0 94.3 94.3
Per C e n t PtOa Pyrophosphate Taken Found 20.0 19.6 25.0 27.5 24.1
20.0 25.0 50.0 5.0
1. VI
25.0
VI1 VI11
90.0 95.0
IX
5 0
:;
950
2 0
3.1 5 0
980
8
4.3
30.0 25.0
IV
6
Alilirouirllate.
Tripolwhosphate Taken Found 50.0 52.8 x o 49.7 62.7
I11
2
Determination of Mixtures of Ortho-, PJro-, and Tripolyphosphate“
90.0
10 0 5.0
1
s
22.3 24.7 26.0 53.2 4.3 7.1 10.0 5.7 5,7 947 94 3 96 9 950
Orthophosphate Taken Found 30.0 27.3 25.0 22.8 23.2
50.0 50.0 25.0 5.0
47.2 50 8 50.2 19.9 6.8
0 0 0.0
0.0
0 0 0 0
0 0
0 0 0 0
0 0 0 0
Total PzO, per titration, 0 2660 gram.
(4)
Bell, It. N., Wreath. .1.I t . , and Curless, W. T.. Ibid., 24, 1997 (1952).
(5) Dole, l l . , ”Principles of Experimental and Theoretical Electrochemistry,” Xew York, AIcGraw-Hill Book Co., 1935. (6) Gerber, A. B., arid lliles, F.T.. ISD. Esc,. C H m f . , AN.AL.ED., 13, 406-12 (1941). ( 7 ) Graham, Thomas, Phil. Trans., 123, 253-84 (1833). (8) .Jones. L. T., .LN.~L.CHEM.,14, 537 (1942). (9) Kolthoff. I. AI., and Saridell, E. B.. “Texthook of Quantitative Inorganic Analyseb,” p . 551, Sew York. Macmillan Co., 1948.
(10) Madsen, E. It.. and .Jaergard. T.K., dcta Chon. Scand., 7, 735 (1953). (11) Robinson, H. b.,T i a m . Electrockem. SOC.,92, 4 4 5 (1947). (12) Van Waser, J . R., “Encyclopedia of Chemical Technology,” Kirk and Othiner, eds., Vol. X, pp. 403-510, S e w York, Interscience Publishers, 1953. (13) Van Waser, J . R., J . Am.. ( ‘ h e n . SOC.,72, 644-55 (1950). (14) Van Waeer, J. R., and Campanella, D. A , , Ibid., 72, 659 (1950).
(15) Van Waaer, J. R . , Griffith, E. J., and McCullough, J. F., Ibid., 74, 4977 (1952). (16) Van Waaer, J. R., and Holst, K. A , , I b i d . , 72, 639 (1950). RECEIVED for review 1larcli 4, 1954.
Accepted August 21, 1954
Spectrochemical Determination of Copper in Crankcase Drainings C. R. HODGKINS and JOHN HANSEN Fsso Laboratories, Research Division, Standard O i l Development Co., Linden, The copper content of cranlicase drainings can be rapidly determined by an emission spectrograph method. The addition of hydrocarbon-soluble calcium and lead soaps to the sanlple before ashing serves two purposes: to provide a carrier for the sample ash of nonmetalloadditive lubricants, and to serve as a common matrix to reduce the interelement effect. The method cobers a range of 5 to 500 p.p.m. of copper in the oil and requires about 0.5 man-hour per determination. The results are on the average within +loyo of the amount found by chemical analysis.
T
HE standard or extended Chevrolet L-4 test (CRC L-4-949)
requires that two of the connecting rods be fitted with copperlead bearing inserts, while the other four are high tin Babbitt bearings. The determination of the copper-lead bearing weight loss is one of the principal factors of the test. I n some instances,
N. 1.
it is desirable to ascertain the intt:iTinrdiatevaiue of the bearing weight loss during the progress of a test. h correlation between the bearing weight loss of t h e copper content of the crankcase lubricant eliminates the necessity of removing the bearings from the engine for t,he intermediate neighings. The determination of the copper content of the lubricant a t fixed intervals during the test makes i t possible t,o follow the hearing R-eight loss and est,ablish the break point in the loss increase curve, thus enhanring the value of the test in evaluating the lubricant under study. Previous to development of the procedure described, spectrochemical copper determinations in this laboratory were made by a lithium carbonate-graphite coninion matrix technique (4). T h a t method involved quantitatively ashing the sample and determining the per cent copper in the ash, then calculating the copper content of the original blend. The number of weighings required in the ashing and subsequent sample preparation, as well as the calculations necessary t.o obtain a result, made the method