decaborane were found to dissolve readily and melt without leaving a residue. Samples which have been alloIT-ed to stand for some time develop a white opaque coating nhich is resistant to solution and is left after the bulk sample of decnborane has melted. The nature of this material has not been determined. The fact that recoveries on the order of only 98% were obtained may be clue to the presence of this material. The hydrolysis of dwaborane is writttm R S f0llOll s ( b ) :
BlaH14
+ 30 H20
-+
10 HPBOj
+ 22 H?
This would indicate t h a t 44 electrons are involved per mole of decaborane. This assumption, which has also been used b y other workers in decaborane chemistry, has been utilized in the present investigation. LITERATURE CITED
(1) Gallant, IT. K. A,, “Apparatus for
Differential Thermal and Therniogravimetric Analysis,” Division of Analytical Chemistry, 133rd Meeting, ACS, San Francisco, Calif., l p r i l 1958.
( 2 ) Hill, It‘. H., Levinakas, G. J., Sovick, IT. J., “Iodometric hIonitoring of Bo-
rane-Containing Atmospheres,” University of Pittsburgh Report CCC-1024TR-129 (-4ug. 12, 1955). (3) hlessner, A. E., .Isa~. CHEY.30, 547 (1958). (4) Simons, H. L., Balis, E. IT-.) Liebhafsky, H. A Ibid., 2 5 , 635 (19j3j. (5) Stock, ii., ”“Hydrides of Boron and Silicon.” uw. 80-5. Cornel1 Univrrsitv, ”, Ithaca; N: Y . , 1933.
RECEIVED for review ?\lay 4, 1959. .4ccepted September 10, 1959. Division of Analytical Chemistry, 135th Meeting, ACS, Boston, Mass., April 1959.
Photometric Indicator Titration of Weak Bases in Acetic Acid The Modified Type II Plot KENNETH A. CONNORS and TAKERU HlGUCHl School of Pharmacy, University of Wisconsin, Madison, Wis.
b A refinement of an earlier photometric titration method is based upon an equation which takes into account solvolysis of the salt formed during the titration. Quantitative recovery data and exchange constants are reported for such weak bases as acetamide, urea, dimethylpyrone, thiourea, caffeine, antipyrine, and triphenylguanidine, titrations being performed in acetic acid with Sudan 111, Nile Blue A, p-naphtholbenzein, and malachite green as indicators. Recoveries were 98 to 101% for all samples in the 20to 30-mg. range. Some special cases of mixture analysis have been developed and data are presented. The ready accessibility, through the exchange constant, of the salt formaation constant makes this quantity a convenient measure of basicity for structural work.
I
color changes associated with the titration of weak acids and bases are frequently too gradual t o permit accurate visual determination of the end point. The recent development of methods for the linear extrapolation of photometric measurements has greatly extendcsd the range of substances \\-liich can be determined by acid-base titration (3). These methods are based upon a description of the reaction as a competition between 1,he indicator and the sample for the titrant species. For example, the titration of a base NDICATOR
sample with a n acid is represented by Equation 1: BHA+I=IHA+B (1) where I is a n indicator base, B is the sample base, and HA is the acid. The indicator is present in such small concentration that the amount of acid bound as indicator salt is negligible; solvolysis effects are also neglected. The equilibrium constant, K,,, for Equation 1 is called the exchange constant. K e x = CIHACB/CICBHA = K:H4/Kp
(2)
K,BHA is the formation constant for the salt BHA; i t describes the equilibriuni B HA4= BHA. A similar constant is defined for the indicator salt, IHA. These equations are nritten n i t h the salt and acid in the undissociated form, which is the case in acetic acid; in aqueous solution they would be JT ritten in the dissociated form. The symbolism used here follows t h a t of Kolthoff and Bruckenstein ( 5 ) . The titration equilibrium in Equation 1 is described by Equation 3.
+
+
1/X = ( K e J S ) ( I B / I ~ ~1/S (3) n-here S = milliliters of standard acid required to react stoichiometrically with the base originally present, X = milliliters of standard acid added a t any point, and I B D . 4 , the experimentally determined indicator ratio, is n-iitten for the ratio CI/CIHA. (This equality is not generally true, but is valid for the
very n-eakly basic indicators employed in this work.) A plot of 1/X US. IB/14yields a straight line, from nhich values of the end roiiit and exchange constant can be determined. This graphical photonietric titration method has been termed the Type I1 titration. I t s validity has h e m established for several systems (3). THE MODIFIED TYPE II TITRATION
Recent applications of the Type I1 method t o the titration of weak bases in acetic acid have revealed significant and reproducible deviations from linearity when the sample size is very small. Recoveries in excess of 100% result when this curvature is observed. Similar deviations recently have led to the derivation of a more complete titration equation for the Type I11 photometric plot ( 1 ) ; this treatment, which was developed for aqueous systems, takes into account hydrolysis of the salt formed during the titration. The equation developed for the aqueous system also describes the equilibrium in acetic acid, although not so exactly. Let X’ be the total concentration of acid added to the system a t any time prior to the end point, and let S’ be the total concentration of base initially present. Then, if dissociation is neglected for the case of acetic acid titrations,
+ S’ = CB + C B H ~
X’ = CHA CBHA
(4) (5)
The concentrations X’ and 5” can be VOL. 32, NO. 1, JANUARY 1960
93
converted to the previously defined volumes X and S with the relations X ' = X N / V and S' = S S / V , S being the normality of the titrant and V the total volume of the sample solution, Combination of these relations with Equations 2, 4, and 5 leads to Equation 6, which describes the titration equilibrium. This expression was used by Bodin and Higuchi in their work with Type I11 titrations ( I ) .
X S/[1
S
+ (V/NK:HA)(I~/I~ -
+ (IA/IB)( 1 / K d l
(6) Equation 6 can be put into usable form in the following manner: Consider the titration of two samples identical except for size and let sample 1 be the smaller of the two. The titration of sample 1 is described by Equation 6, the volumes being denoted by XI and SI. A similar equation involving X p and SZ describes the titration of sample 2. Subtracting these tn.0 equations a t constant I B / I A gives, after rearrangement, l/(XZ
- SI)
= [K,,/(SZ (IB/IA)
- SI)] x
+ i / ( s z - xi (7)
as V/KfHA.V is practically identical in the two cases. The quantity l/(XaXI) is calculated at given values of IB/IA by reading points from t h e and smoothed plots of 1/X1 vs. 1/Xz vs. IB/I. 0, (8: - 8:) = 0
I n this case Equation 8 becomes Equation 7; the exchange constant and end point apply t o base B. This case is a n important one because it describes the titration of a base, B, in a solvent containing a basic impurity, C. The solvent purity should not be critical, therefore, when the modified Type I1 technique is used. KZx < < IA/IB
Because varies from about 0.5 to infinity in the region of interest, KEx should be less than about 0.02. Then Equation 8 becomes:
- X,)
1 - (8:
- 8:)
-
If the end point for the stronger base is known or can be found from a n independent determination, the modified Type I1 Flot yields the exchange constant and end point for the n-eaker base. EXPERIMENTAL
Apparatus. Spectronic 20
Bausch and Lomb colorimeter. Two-
Spectral Characteristics of Indicators in Acetic Acid Base Max., Acid Max., AbsorpAbbreviation Mpa Mp" tivitj-b
Indicator Malachite green MG 425, 610 450 Xaphtholbenzein PNB 453 465, 693 ile Blue A NBA 632 455 s-I11 513 613 Sudan I11 a Italicized wave lengths are those employed in photometric titrations. b A t the italicized wave length; PNB and S-I11 in 0.05N perchloric acid, glacial acetic acid. Units are milliliters/milligram centimeters ( 4 ) .
K
94
ANALYTICAL CHEMISTRY
(MC'I
1. 4 . 8 4 mg. of urea with 0 . 0 9 8 0 N perchloric acid 2. 24.20 mg. of urea Modifled. Difference titration of 1 9 . 3 6 mg. of urea
(XZ
This equation has not been made useful for the general case of any two bases, but some special cases are of interest.
/ VOLUME
Typical titration curves
226 93 150 193
others in
milliliter microburet, graduated to 0.01 ml. and read with a magnifying glass. Photometric titration circulation unit (6). Indicators. Sudan 111, recrystallized from 1 to 1 benzene-absolute ethyl alcohol, melting point 203' C.: p-naphtholbenzein, recrystallized from glacial acetic acid. melting point, 236' t o 241' C.; S i l e Blue A and malachite green, used directly. The spectral characteristics of these indicators are compiled in Table I. Reagents. Glacial acetic acid, reagent grade. Purified acetic acid, dried b y refluxing with boron acetate and distilling (a)and analyzed for water by visual Karl Fischer titration. All compouncls titrated n ere from commercial sources and were recrystallized before use. Titration Procedure. Stock solutions of perchloric acid, indicator, and base were made in d r y acetic acid. T h e titrant was prepared by taking aliquots of t h e indicator and acid solutions and diluting them to give convenient concentrations of both. Standardization of the titrant was b y visual titration of potassiuni biphthalate using p-naphtholbenzein indicator; the indicator contained in the titrant did not interfere with this visual end point. Sample solutions were prepared from solutions of indicator and base diluted to a volume such that the indicator concentration x i s identical with that in the titrant. dl1 titrations n'ere performed at 25" f 2" C. Titrant was delivered in increments of such size that 15 to 20 determinations were made in the range 0 < I B & < 2. The indicator ratio was calculated from the spectral data according to the usual formula (3); the evaluation m s conveniently made with the use of nomographs constructed for this purpose. RESULTS
Titration of Single Bases. Glacial acetic acid was the solvent in all titrations. T h e titrant acid was 0.lX perchloric acid in acetic acid for
titrations with Nile Blue A, p-naphtholbenzein, and malachite green, and 0.45N perchloric acid for Sudan I11 titrations. I n most cases t h e sample weight in the initial titration was 5 t o 7 mg. and 25 to 35 mg. for the second; the effective sample titrated was therefore in the range 20 t; 30 mg. I n subsequent discussion sample size” will refer t o the effective or difference sample.
A typical titration is shown in Figure 1; the sample was urea and the indicator was Kile Blue A. The points obtained by plotting 1/(XQ - XI) against the indicator ratio fell on a straight line and the recovery \vas essentially quantitative. The same behavior was obcerved in all titrations. Some scatter of points occasionally occurred a t lower indieator ratios, probably because of the large volumes being subtracted; thP Sudan I11 systems exhibited this deviation to the great& degree. Reduction of the sample size to 10 to 15 mg. also increased the scatter and reducpd the accuracy. Cse of a sample of size zero for the initial titration (equivalent to a n indicator blank titration) was lcss satisfactory because of poor reproducibility. Such a system is completely unbuffered and is very sensitive t o trace impurities in the solvent. T h e quantitative results are summari z d in Table 11. The recoveries were satisfactory for every system, and only the titrations with Sudan I11 exhibited inconveniently large deviations from t h e mean. A considerable part of the deviation observed in all cases is thought t o be due t o variations incurred in plotting the data. The exchange constant, by its relation t o the salt formation constant, is a measure of basicity. The exchange constants of a series of bases with a single indicator and a single acid thus permit evaluation of t h e relative basicities of the compounds. Table I11 contains the exchange constants for the systems titrated. The relative precision of these values (excel t for Sudan I11 titrations) is about 5%. Titration of Mixtures. CASE I. -4s Kile Blue A levels antipyrine a n d triphenylguanidine t o t h e same base strength (Table 111), their mixture should yield total equivalents of base. Three determinations x e r e made of a miyture containing 0.1219 meq. of antipyrine and 0.0963 meq. of triphenylguanidine (total, 0.2254 meq.). T h e mean result was 0.2249 meq. of base recovered, or 99.8%. The mean deviation was 1.3%. Observed K e xwas 0.01 0.01. CASE 11. The system urea-Me Blue A was chosen t o provide a n example for this case. The impurity (base C) was n-ater, acetamide, or ammonia. T h e results, listed in Table IV, show
+
Table II.
Quantitative Results of Titration by the Modified Type
Base Acetamide iicetamide ilntipyrine Antipyrine Caffeine Caffeine 2,6-Dimethyl--ppyrone Thiourea Thiourea Triphenylguanidine Urea Urea Urea a
Mean deviation
Table 111.
=
Per Cent Recovery Mean Xean deviation0 99 7 3 7 97 9 1.7 97 9 0 8 100 9 0 7
Indicator s-I11 NBA
NBA
PKB NBA MG SBA SBA
100 4 99 0
99
NBA
PXB hIG
No. of Determinations m
6
4 5 9 a
0 9
1 0
0 3 0 5
1
100 5 99 4 98 4 100 6 98 0 98 0
PSB 1-B A
II Method
1 5 0 7 1 6 1 3 1 5
3
-,
a
3 21 7 (j
Z(Xi - 8 ) n-1 .
Exchange Constants for Some Base-Indicator Systems in Acetic Acid
Indicator Base .%retamide Urea Dimethylpyrone Thiourea Caffeine .htipyrine Triphenylguanidine Table IV.
$111
NB.4
0.12
3.66 0.46 0.28
0.25 0.16 0.00
b
JIG
1.29
2.12
0.73 0.03
0.00
0.75
Effect of Solvent Impurity on Titration of Urea-Nile Blue A System
Impurity Concn., \lole/Liter
a
PSB
Inlpurity Water 0 01 Water 0 10 Acetamideb 0 005 .Immoniab 0 002 Mean of three determinations. lilso 0.0811.1with respect to Rater.
that no significant interference was encountered in the urea determinations. Because acetic acid always contains water, and may frequently absorb traces of volatile bases from laboratory air, this insensitivity to solvent impurities is a n important advantage. CASE111. With Nile Blue A as the indicator, antipyrine and urea satisfy the requirements of this special case. About 25 mg. of each base n as included in the sample. The end point for the antipyrine was calculated from knon 1edge of the sample size and titrant normality. The mean recovery for the titration of three samples n-as 97.9% of urea taken and the exchange constant observed was 0.46. Evidently, Equation 9 can be applied t o such a mixture. An attempt t o titrate aminopyrine as a monoacidic base b y use of Equation 7 yielded plots with reproducible curvature, indicating t h a t a second basic function n-as consuming acid. Aminopyrine n a s consequently treated as a mixture of two bases, the exchange constant for the first group being estimated a s 0.00. Equation 9 was applied, yielding recovery values which were semi-
Urea, M g .
Taken 20 5 19 0 21 8 21 4
Observed
Found5 20 5 19 1 22 2 21 5
Kexa
0 0 0 0
47 45
49 41
quantitative only, probably as a result of the extremely weak character of the second basic group. Nile Blue A was the indicator in these titrations. The best estimate of the exchange constant, for the second group is 5.0. DISCUSSION
The slope of the modified Type I1 plot is controlled by the magnitude of the exchange constant, and hence by the indicator, for a given base. The end point is most accurately located n hen the exchange constant is very small. When a measure of basicity is desired (and this will probably be the most useful application of the methodj, it is necessary to employ a differentiating indicator or one n hich gives an exchange constant greater than zero. Sile Blue A and p-naphtholbenzein are the most useful of the indicators in the basicity range studied. Sudan I11 is not recommended for Type I1 titrations. If the salt formation constant of an indicator is known, then the formation constants for bases titrated against this indicator can be calculated. Studies in VOL. 32, NO. 1, JANUARY 1960
95
this 1:tborntory girvt- vduc of 1.1 x 106 for t h o perchlorate formatioll coilstant of pnnphtholbcnzcin. With this vrilue the iridividual constants of thc other u c:ik b:ws niid indicators in T:ible 111 c:in be c:ilcuIntcd. Such constlints provide n convenient qu:mtitlltive nleaSIIrc of basicity dctcrIlijncd by j V i t h I! rcfercncc noid. Roliablc constants mny be obtriined
quickly by thc modified Type I1 plot; sevcral points in the indicator ratio r:iiipc 0.5 to 1.5 will s d f h for such detrrniinitions. LITERATURE CITED
( 1 )of nodill, J. 1 5 , Ph.D. thegi% University \\‘isconsin, 1958.
( 2 ) Eictielbcrger, W. C., LsJIer, V. K., J . .I m. Chert. SOC.55, 3633 (1933).
(3 karnstein, H b c h i , Charlcs, ‘rakCW ANAL r t e hCrrest. c- R.1 28, lx6 (lo56). (4) HJlghes, H, K., et a/., [bid., 24, 1349 ( 1902 ). ( 5 ) Kolthoff, 1. hi., hiekenstein, s., J. Ani. Ckem. SOC.78, 1 (1SS6). (6) Rehm C. It:, Bodin, J. I., coiiriors, K. A,, kiguctii, Takeru, X S A L CHEM. 31, 483 (1959). RECEIVEDfor review August, 6, 1939. Accepted Novcmber 2, 1959.
It i L.
L. LEWIS
and W. A. STRAUB
A p p h d Research laboratory,
United
Stales Steel Corp., Monroeville, Pa.
A nisthod has been developed for the routine determination of large amounts of nickel and cobalt in complex high-alloy and stainless steels, Nickel is isolated from other elements by dimethylgiyoxime precipitation with or without a preliminary ion exchange separation, and cobalt is isolated by ion exchange. The separated metals are then determined by titrafian with fetfiylgnedinitrilo)tp?traaceticacid. The method has the advantage of speed, yet retains the accuracy of the commonly used methods far determining these elements.
HE conipiesity and thc number of high-alloy and stainless steels being produccd both commercially itnd experirnentttlly hnvc incrcswl as performance requirements have beconie more stringent. Demands for more rapid and reliable methods of chemical analysis for large amounts of nickel and cobalt in these steels have accompanied the increase. The dimethylglyoxime sepnration and gravimetric determination of nickel, although time-consuming, are still widely used. Several published methods for determining nickel in the dimethylglyosime prccipitnte include the direct titration of nickel with eyanide following the chemical dccomposition of the precipitate (4),and a redox titration based on the hydrosylamine liberated when the precipitate is hydrolyzed ( 1 ) . The first method is undesirable bccnuse of the toxicity of the cyanide reagent and the indistinctness of the end point. The other method is indirect and is subject to error if the precipitate is incompletely washed or incompletely hydrolyzed. The determination of large amounts of cobialt in steels has been complicated
I)y limitations i n mcthods for sepnrntion and dcterminntion. Intcrferenccs must be removed before the final precipitation of coLnlt with 1-nitroso-2-naphthol or the clcctrodeposition of cobalt ns the metal. In addition, whrn cobalt is ignitcld nnd n.sighccl as Co304,it will bo nonstoichiometric in composition unless ignited propcrly ( 4 ) ; and, in the clcctrolltic method, the cob:& mny not be depositcd completcly ( 2 ) . T h e , ferriryiriidv titrrition of cobalt in coniplcs stecls is not completely rcliablc vithout seprntions bccause of interfcrcnces from sonie co-alloying elements (4.). In recent years, both nickcl and cobatt have been titratcd successfully with (cthylcncdinitrilo)tctnacetic acid (EDTA) (8). Direct application of these titrations to the analysis of highalloy and stainless steels, however, is impossible because EDTA is nonspecific. A method was nceded for rapidly separating these elcments from each other and from other alloying elements prior to the EDTA titration. Diniethylglyosime, a specific reagent for nickel, was considered best for isolating nickel; however, no specific reagent is known for cobalt. A &isfactory method for isolating cobalt was not avsihble before the pioneering work of Kraus and Nelson (6). Their data indicate that cobalt could be separated from other elements in steels by adsorbing the sample on an anion exchange resin column and eluting selectively with hydrochloric acid. Analytical applications of :Inion exchange sepnrations in chloride media to heat-resistant nlloys nnd Alnico-type magnetic alloys have been reported (2, 5, 9). These procedures can be highly accurate, but they lack speed. A procedure has been developed that is based on isolating nickel and cobalt
(by dimcthylglyosime prccipitation and ion exchange separation, rcspectively) and then titrating them with EDTA. Large amounts of nickel and cobalt in high-alloy nnd st:iinlrss stecls with wide ranges in composition can be cletermined rapidly and reliably, and as accurately as by convcntiond nicthods. PREPARATION
OF
REAGENTS
ION ExcHANaE RESIN. Remove fiys
from Dowex I-X8 anion cschangc resin (J, T. Bnker Chemical Co.) by mising the resin as a n aqueous slurry and decanting three times. Remove traces of iron and aluminum remaining from the resin synthesis by washing alterrintely with concentrated hydrochloric acid and water until washings are colorless. For samples t h a t contain less than 0,3% copper, use 100- to W-mesh, and use 200- to 400-mesh for snmplcs with more than 0.37, coppcr. ION EXCHANGE COLUMIK.Obtain a column approximately 2 cm. in dinmeter and 18 cm. long, tapered at one end, and provided with a stopcock. Place a tuft of glass wool in the bottom of the column, add a slurry of cleaned resin, allow i t to settle, and add more resin until the packed bed. is about 14 cm. high. Cap the bed with a perforated ceramic disk from a Gooch crucible to prevent resin disturbance when the sample is added. Into the column insert a one-hole rubber stopper fitted with a funnel. Pretreat the column with 50 ml. of concentrated hydrochloric acid. Reagent-grade chemicals were used, unless otherwise noted. STLNDARD0,031ci EDTA. Dissolve 22.3 grams of disodium dihydrogen (ethylenedinitrilo)tetraacetic acid dihydrate in 2 litera of distilled water. Store in a polyethylene bottle. Standardize with standard 0.03M nickel solution at € 9I by using Eriochrome Black T ind!cator and manganese a8 a back-titrant (8). I n t h h and other
