Anal. Chem. 1981,
From eq 10 and 11, it then follows that the function dk,l has the properties of conditions I1 and IV, if the function is symmetric in the elements of [y&l)]. As one of such measures we propose
which is essentially the Euclidian distance between two points, Pk = (log r J k ) , j < i) and Pl = (log rij(l),j < i), in the ({)-dimensional Euclidian space, and con~equentlythis satisfies the conditions I11 and V.
RESULTS AND DISCUSSION In this section, we apply the dissimilarity measure dk,l defined by eq 21 to an example of Anders (I, Table I) to compare the values of dk,l with those of z * k , ~ ( M ) given by eq 13. A part of Anders’ data (1,Table I) is relproduced in Table
I. Table 11, shows the vadues of z * k , l ( M ) fior M = 1.5 and M = 1.3 (in parentheses) anld the correspondiing values of dk,l for the data in Table I. The correlation coeffiicient between z:#(1.5) and z*(1.3) calculated from the values in Table I1 is 0.882, while the coefficients are 0.796 and 0.542 between t i and z*(1.5), and d and z*(1.3), respectively. Thus the d values appear to be more correlated to z*(1,5) values than tlo z*(1.3), for this particular sample and the particular choices of M. The d values in the above table show that the sample, No. 29, is far different from the rest of the samples in its concentration-ratio structure, but this is not so apparent from the values of z*. The measure, eq 21, will be better thain the correlation number of Anders (1) in practical applications. Also, the statistical properties of the measure will be derived, in cases
1579
53, 1579-1582
where one can assume the log normal distribution model on the population distribution, from which the original data, eq 1, are drawn. As in Anders (I),one may settle a critical value for judging a pair of individuals to be “correlated” or “similar” in applying our d of eq 21: Given a “significance level” a , a pair of individuals, k and I , are judged to be “correlated” or “similar” if and only if dk.1 < c,, where c, stands for the sample lower a-quantile of the empirical distribution of all the d values for a large set of original data. Likewise, one may use the sample upper a-quantile c,’ as a critical value for uncorrelatedness or dissimilarity.
ACKNOWLEDGMENT The authors are grateful to the reviewers and the editor for their valuable comments through which the present article was revised.
LITERATURE CITED (1) Anders, 0. U. Anal. Chem. 1972, 44, 1930-1933. (2) Koba, T.; Masumoto, K. “Mechanism and Analysis of Water Poilutlon”; Japan Geochemical Society: 1978; pp 53-83. (3) Kodama, T.; Eba, H.; Oda, H. Alchl-ken KO@ Chosa Senta Shoho i.a- m . - ., 6.,. 121-124. .- . .- .. (4) Kodama, T.; Kadowaki, L.; Oda,H. Alchl-ken Kogal Chosa Sente Shoho 1978, 8 , 125-128. (5) Kawae, Y.; Takaoka, K. Bull. Fukul Sosen 1978, 11, 117. (6) Takagl, Y.; Chlkano, S.; Masuda, K. FukulNken. Rep. 1978, 14, 168. (7) Kodama, T.; Eba, H.; Okuda, H. Alchl-ken Kogel Chosa Senta Shoho 1917, 5, 40-47. (8) Seklguchi, K.; Salto, Y.; Hayashi, F. Jpn. J. Air Pollut. 1977, 12, 466-469. (9) Koba, T.; Terada, K.; Masumoto, K.; Ichinokl, S. Jpn. Proc. Anal. Chem. 1978, 37, 27. (10) Koba, T., Terada, K., Honjyo, T., Masumoto, K., Jpn. J. Anal. Chem. 1975, 2 4 , 18-25. (11) Watanuki, K.; Takano, T. Hot-Spring Scl. Res. 1974, 25, 26.
RECEIVED for review January 12,1981. Accepted April 6,1981.
Micellar Effect on Indicators in a Perfluorochemical Emulsion and Nonionic Surfactant Solution Alaln Berthod, * Joseph1 Georges, and MylOne Breant Laboratolre de Chimie Analytique III, Professeu,r M. Porthault, Universlt6 Claude Bernard-Lyon 69622 Villeurbanne Cedex, France
Acld-base properties In a1 perfluorochemlcai emuislon used as a blood substitute were investlgatedspecftrophotometrlcally and potentlometrically. The pH measurements, performed with a cailbrated glass tslectrode, show that dissoclatlon constants of acids and bases are analogous to the correspondlng values In water. A micellar effect Is observed for lndlcator dyes. The surfactant’s chlef part In this micellar effect Is shown studying mater-surfactant solutlons and correlating with the oil-In-waiter emulsion results. The study is made with bromophenol Miie, phenolphtaleln, and neutral red. I n surfactant solutlons,the micellar effect, expressed by shifts of the indicators acidic dirssoclation constant, Is Interpreted as proceeding from variatlon of surfactant mlcelies size.
In order to establish pH measurements by means of indicator dyes in emulsion of perfluorochemical compounds used 0003-2700/8 110353-1579$01.25/0
I, 43 boulevard do 11 Novembre 19 18,
as blood substitutes, we made a study of behavior of some acid-base indicators in Fluosol 43. This medium is an oilin-water emulsion of perfluorotributylamine (25% w/v), the surfactant being a polyoxypropylene-polyoxyethylene nonionic copolymer (Pluronic F.68, mol wt 8750,3.2% w/v). The aim of the present work was to study the interaction between the nonionic surfactant used to prepare the emulsion and a number of acid-base indicators in a wide pH range by measuring their acidic dissociation constant. These interactions between the indicators and surfactants have been studied by several authors (1-4). Some of them ( I , 2) suggest a weak interaction by adsorption of dye on the surface of the surfactant micelles, while others ( 3 , 4 )suggest a stronger interaction even complex formation (3). Given the nonionic character of the surfactant, our study was based on the first hypothesis. This supposes a correlation between the micellar effect in aqueous surfactant solution and in oil-in-water emulsion expressed by a shift of the acidic dissociation con0 1981 American Chemical Society
1580
ANALYTICAL CHEMISTRY, VOL. 53, NO. 11, SEPTEMBER 1981
stant. The mechanism of the micellar effects on chemical reactions, as studied by Martinek et al. (5-8) supposes the existence of a distribution of the reagents in a bulk phase and a micellar phase. If the studied reagent is an indicator with an acid form HI and a basic form I-, it appears (HI),
bulk phase
[HIlm/[HIlb
(2)
If an acid-base reaction proceeds in the bulk phase
HI e I- + H+
PKA in water product
micelles
with the partition coefficient PHI
Table I. Dissociation Constant of Some Acids and Bases in Water and in Emulsion Fluosol43 (Confidence Limits ltO.1)
(3)
The dissociation constant is (4)
but in the presence of micelles, the measured value is the apparent dissociation constant which can be determined as (8)
(5) where C is the surfactant concentration from which the critical micellar concentration (cmc) is substracted, and Vis the molar volume of surfactant. The useful value will be
thymol blue I chloracetic acid bromophenol blue acetic acid methyl red maleic acid I1 neutral red carbonic acid I bromocresol purple 3-nitrophenol phenol phenolphtalein n-butylamine thymol blue I1 piperidine thymolphtalein tetramethylguanidin
(12,1 6 ) pKAa
1.65 2.86 3.85
2.2 2.7 5.0
4.75 5.0 6.07 7.4 6.40 6.12
4.6 5.4 5.9 6.0 6.4
8.38 10.0 9.1 10.61 8.90 11.12 9.7
8.5 9.9 10.2 10.3 10.0 11.0 12.0
13.6
13.4
PKAb molwt 2.2 467 94.5 2.6 4.9 67 0 4.6 5.4 5.8 6.0 6.4 6.9
60 269 116 289 61 540
8.5 9.9 10.2 10.3 10.2
139 94 318 73 46 7 85 430
7.0
11.1
a Measured by spectrophotometry. tentiometr-v.
115
ApK 0.6 -0.2 1.0
-0.1
0.4 -0.2
-1.4 0
0.8 0.1
-0.1 1.1
-0.3 1.2 -0.1
2.3 -0.2
Measured by po-
T a w e l glass electrode. All additions of reagents were performed with a 200-pL microsyringe Gilmont.
EXPERIMENTAL SECTION The emulsion Flumol43 was prepared by the Green Crow Corp. (Gamou-Cho,Joto-Ku, Japan) and supplied by MEDAC GmbH (Hamburg 36, B.R.D.). The perfluorotributylamine (mol wt 631, density 1.88, molar volume 0.357 L/mol) is fully substituted and therefore highly inert (no basic properties). The nonionic surfactant is a block copolymer type A-B-A (polyoxyethylene-polyoxypropylene-polyoxyethylene). The molecular weight is 8750 with 1750 for polyoxypropylene block and the molar volume is 7.5 L/mol. Pluronic F 68 is supplied by Produits Chimiques Ugine Kulhman France. The emulsion presents two phases: an organic one composed of perfluorotributylamine droplets, stabilized by Pluronic F 68 (weight average particle diameter: 0.09 pm) (9) and an aqueous phase. Owing to the proportions of the different constituents, the total surfactant present stabilizes oil particles and practically no part remains free. The organic phase corresponh to 16% and the aqueous phase to 84% of the total volume of the emulsion, but it is assumed that only the surfactant at the droplets surface plays a role in the indicator interaction. In surfactant's aqueous solution, beyond the cmc, the lipophilic parts (polyoxypropylene)of molecules of Pluronic F 68 are joined together in micelles. In both mediums, surfactant solutions and emulsion, indicators are in the aqueous phase. Interactions with surfactant's hydrophilic part (polyoxyethylene) at micelles interface, are compared with interactions at the droplets surface. Indicators and colorless acids and bases are from Schuchardt (Munchen, B.R.D.) and Prolabo (Rhone-Poulenc, France). All reagents (analytical grade) were used as received. To avoid variation in the absorbance rate of Fluosol43, we made titrations with concentrated reagents (NaOH 9.6 mol/L, HClOd 11.36 mol/L). In the beginning, the concentrations of acid or base are roughly mol/L and the concentration of the indicator about 10" mol/L. The ionic strength was regulated at 0.1 by lithium perchlorate. All concentrations are expressed as mole per liter of emulsion. The spectral measurements were carried out on a Beckman 25 spectrophotometer with cells having an optical length of 1 mm, since Fluosol43 is a powerful light absorber. The potentiometric measurements were made with a Minisis millivoltmeter (SoleaTacussel, Villeurbanne, France). In a previous work (IO) it was shown that saturated calomel electrode (SCE) prepared in Fluosol 43 acted as a reference electrode. The working electrode was a
PROCEDURE Determination of Acidity Constants of Indicators in Emulsion. The p K d values of indicators studied are measured at the same time potentiometrically and spectrophotometrically. The principle of the method, outlined in a publication (11),will be summarized here. An indicator dye HI (concentration 10" mol/L) is introduced into a solution containing an equimolecular quantity mol/L of a colorless acid-base pair HA/A- with an acidity constant pKA. The pH of the bulk equals the PKAvalues. The indicator dye is split in both acidic and basic form, HI/Iwhich each absorbs light at different wavelength. The acidity constant of the indicator pKhd is correlated to the pH of the bulk by eq 7 . If PKAis known, this equation gives p K d and vice versa. Both the concentration[HI] and [I-] are measured spectrophotometrically. First a determination is made with a strong acid (HC10,) and with an indicator whose acidic dissociation constant must lie between 1and 2. The pH values, computed as -log [HClO,], give the pK,d value of the first indicator (thymol blue I). The latter allows the determination of the dissociation constant of the chloracetic acid-chloracetate pair. Further values are obtained spectrophotometricallyby measuring successively and alternately the p K d of an indicator and the pKAof a colorless pair following the order shown in Table I. In practice, not a single measurement in the buffered solution with [HA] = [A-] is made but a series of measurements by means of titration. Acids HA are titrated with a strong base (sodium hydroxide) and bases A- with a strong acid (perchloric acid) and an indicator HI. The titration is recorded potentiometrically to locate the equivalent point by a potential jump (Figure la) and spectrophotometrically in order to obtain a correct isosbestic point (Figure lb). The half neutralization point of the potentiometric curve gives the potential corresponding to the titrated reagent pKA. The glass electrode vs. SCE is calibrated according to this procedure. Its potential obeys the Nernst law in this medium and pH range (slope 57 mV/pH). The mean of a set of three measurements is listed in Table I, the precision is &O.l PKAunit. Study of the Shift of pKhd Values. This study consists of determining the pKa variation of an indicator with the surfactant
ANALYTICAL CHEMISTRY, VOL. 53, NO. 11, SEPTEMBER 1981 .o.3
J
r’
APK
i
1581
1
-.0.4 O a k _
C
.
500
400
W4VELENGTH
WO
nm
Flgure 1. Titration of acetic acid (1.74 X to-’ mol/L) by sodlum hydroxide (9.7 mol/L) with bromophenol blue ( e - iod mol/L) in Fluosol 43. (a) Potentiometric cuirve, potential of glass electrode vs. SCE plotted with x , the rate of progress of the reaction. At x = 0.5 the pH value is equal to the acetic acid PKA. The1 potential value is 117 mV. (b) Absorption spectral of bromophenol blue. The acetic acid PKA calculated by eq 7 is 4.6. The pH values are (A)2.8, (0)3.7, (*) 4.3, (m) 4.9, (*) 6.9, (0)12.6.
concentration in water. ‘Toensure that the observed variations are not caused by ionic strength or pH changes, all solutions are well buffered at a known pH and the ionic strength is fixed at 0.1 by adding a dissociated salt (lithium perchlorate). The absorbance variations of a chosen form of the indicator are recorded spectrophotometrically. Taking into account the dilution, the absorbance variation is correlated by the e!q 7 to the apparent acidic dissociation constant pKhd,a of the indicator. In order to compare several indicators, we pjotted the difference ApK = P&d,app - PKhd vs. the surfactant concentration.
RESULTS AND DISCUSSION As the results listed in Table I show, no significant change is observed in pKA values for colorless I-[A/A- couples that are highly soluble in water. However, the values of acidity constants in the emulsion differ from that in water by a ApK of about one unit for the indicators that are sparingly soluble in water. In general, the indicators have a greater PKhd value, except neutral red, whose p&d decreases. It appears that these observations cohere with the fact that the former have a neutral acid form HI and a ionic basic form I- whereas neutral red has a ionic acid form HI+ and ti neutral basic form I. This suggests that an interaction occurs between the nonionic droplets of the emulsion and the neutral form of the indicator. In addition, no change in wavelength of absorption maximum of dye and dye micelles is observed; i.e., there is only a weak interaction. The role of the surfactant is demonstrated in Figure 2 where the difference in the apparent p&d value in aqueous solutions of surfactant and the pKhc1 value in water is plotted vs. the Pluronic I? 68 concentration. In order to make a comparison, one indicator out of each series was chosen: phenolphtalein representing the phtalein series, bromophenol blue the siilfonphtalein series, methyl red for the azoindicator series, and neutral red whose acid form is ionic. The methyl red value of ApK is not presented in Figure 2 because it is equal in water and in surfacitant solution (ApK = 0 ) whatever the surfactant concentration. That bears out the hypothesis of an interaction between the nonionic micelles or droplets and the neutral form of the indicator: the acid form of methyl red is a “zwitterion” (12) wlhile the basic form bears a negative charge. The apparent values of P&d of bromophenol blue and phenolphtalein increase (ApK > 0) while that of neutral red decreases up to a surfactant concentration of about 2 X mol/L (2% ur/v). Beyond this level, the apparent value of PKhd is almost stabilized (Figure 2).
Figure 2. Values of ApK = pKkxl,epp- pK,, vs. surfactant concentration. For phenolphtalein (PP) mol/L buffered by phenol 0.1 moVL with sodium hydroxide 0.05 mol/L, pH 9.7). For bromophenol blue (BPB) (IO-‘ mol/L buffered by succinic acid 0.05 mol/L with sodium hydroxide 0.05 mol/L, pH 4.4). For neutral red (NR) (2 X mol/L buffered by 0.025 moVL disodium hydrogen phosphate and 0.025 mol/L sodium dihydrogen phosphate, pH 6.9).
Table 11. Values of the Partition Coefficient Pa bromophenol phenolIApK 1 blue phtalein neutral red 0.01
0.05 0.10 0.15
18 25
310
24
302
9
0.30
5020 4130 5300 5400
311 285
325 365 140
0.60 1.00
5820
171
a Calculated with eq 9, for some experimental points of Figure 2. (Solution conditions are stated in legend of Figure 2.)
Making the assumption that the ionic compounds have no affinity to nonionic micelles, from eq 5 it can be obtained ApK = log (1
+
gv)
and also
(10lAPKl- 1)(1-
=
cv
cv)
(9)
Table I1 shows the PHIvalues calculated by eq 9 for some experimental values with a Pluronic critical micellar concentration of 2 X 10” mol/L (13). The partition coefficient is relatively constant for smaller values of ApK, from which an average value of P is estimated: P = 24 for the bromophenol blue, P =: 350 for phenolphtalein, and P = 5000 for neutral red. Figure 3 shows the experimental and theoretical curves as calculated with eq 8 substituting the average values of P. For the three indicators examined the variation of the dissociation constant is no longer represented by eq 8 at higher concentrations. These results can be compared to previous studies on the same surfactant (13) or on nonionic surfactants with similar properties (14). The values of cmc reported diverge widely and vary, according to the method used, from 2 X 10“ mol/L to 1.2 X lov4mol/L. Some authors (13, 15) explain these variations by the hypothesis that at low concentrations, the Pluronic F.68 solution seems to contain monomolecular micelles whereas at higher concentrations polymolecular micelles could occur. The variations of the partition coefficient P in high surfactant concentration can be explained assuming that the interaction between the dye and the sur-
1582
ANALYTICAL CHEMISTRY, VOL. 53, NO. 11, SEPTEMBER 1981
.
. A
NR
P.5000
PP
Ps350
BFe P: 2 4
-
0.7
4
os
-
a4
-
03
.
0.2
.
01
.
L
0 log l v r f s s t a n t
consentratlo"
Flgure 3. Experlmental values of l A p K l vs. logarithm of surfactant concentrationin full line, compared with theoretical values according to eq 9 in dots.
Table 111. Values of the Partition Coefficient P Calculated with Equation 9 indicator methyl red bromophenol blue
phenolphtalein neutral red
PHIa
54 320 410 860
0 24 350 5000
a In Fluosol 43 ApK is listed in Table I, the product CV In surfactant solution. is (32/8750)7.5 = 0.027.
factant is adsorption of the noncharged indicator form on the surface of Pluronic micelles. If the micelle size increases, the volume of the micellar phase increases more rapidly than the surface area. The apparent volume concentration in the micellar phase and the partition coefficient P = [HI],/[HI]I, tend to decrease equally with increasing surfactant concentration. The neutral red dissociation constant starts to change at 2 X lo4 mol/L which corresponds to the value of critical monomolecular micelles concentration as presented by Prasad et al. (13). The bromophenol blue constant starts to be modified at 10" mol/L and at 4 X 10" mol/L, the theoretical neutral red curve (with P = 5000) deviates from the experimental curve. Experiments with benzopurpurine and iodine show inflections at the level of concentration which is considered (13) as the critical polymolecular micelles concentration. At 2 X lov3mol/L, the theoretical curves with P = 350 for phenolphthalein and P = 24 for bromophenol blue no longer correspond to the experimental curves. At this high concentration level (18 g/L) polymolecular aggregates may be found in the surfactant solution. Figures 2 and 3 show no
abrupt change, which seems to indicate a gradual evolution of the Pluronic F 68 solutions structure. In order to correlate the indicator interaction in emulsion and in surfactant solution, the partition coefficients in Fluosol 43 were evaluated according to eq 9 with the ApK values as listed in Table I. These values of P are less scattered than those obtained for the Pluronic F 68 solution (Table 111). The monodisperse droplets in emulsion (average diameter 0.09 bm (9)) do not interact with the indicators the same way as the polydisperse aggregates in the surfactant solution. In addition the interaction is rather different in the emulsion since the lipophilic part (polyoxypropylene) of Pluronic F 68 is bound to the perfluorocarbone oil. Nevertheless the indicators interacting more strongly with the emulsion droplets are those that show the highest interaction with the surfactant micelles and vice versa. That suggests that the chief part in indicator adsorption is played by the surfactant with micelles in water solutions and at droplet interfaces in emulsion. Nevertheless it is possible to perform pH measurements by means of indicators in emulsion Fluosol 43, if shifts of indicators' p& are taken into account. For instance bromocresol purple is more convenient than neutral red to check pH in cases of medical applications of the emulsion.
LITERATURE CITED Mukerjee, P.; Banerjee, K. J. Phys. Chem. 1964, 68, 3567. Funasakl, Norlakl Nlppon Kagaku Kalshl, 1978, 5 , 722. Chem. Abstr. 1978, 85, 26006m. Nemoto, Y.; Funahashi. H. J. ColloM. Interface Sc/. 1977, 62, 95. Mackay, R. A.; Jacobson, K.; Tourlan, J. J. ColloM Interface Scl. 1980, 76, 515. Martinek, K.; Oslpov, A. P.; Yatslmlrskl, A. K.; Berezln, I.V. Tetrahedron 1971, 27, 2855. Martlnek, K.; Oslpov, A. P.; Yatslmlrskl, A. K.; Berezln, I.V. Tetrahedron 1973, 29, 963. Martlnek, K.; Oslpov, A. P.; Yatslmlrskl, A. K.; Berezln, I. V. Tetrahedron 1975, 31, 709. Martlnek, K.; Yatslmlrskl, A. K.; Levashov, A. V.; Berezln, 1. V. "Mlcelllzatlon, Solublllzatlon and Mlcroemulslons"; Mlttal, K. L., Ed.; Plenum Press: New York, 1977; Vol. 2; p 489. Technlcal Information; The Green Cross Corp. 1/3, Gamou-cho, Jot+ Ku, Osaka, Japan. Ser. No. 3, 1976. Breant, MylBne; Georges, Joseph; Mermet, Monlque AM/. Chlm. Acfa 1980, 115, 43. Breant, M y h e ; Auroux, Allne; Lavergne, Madeleine Anal. Chlm. Acfa 1978, 83, 49. Banyal, Eva; "Indicators"; Blschop, Edmund, Ed.; Pergamon Press: New York, 1972; pp 76-135. Prasad, K. N.; Tran Thuy Luong; Florence, A. T.; Paris, J.; Vautbn, C.; Selller, M.; Pulsleux, F. J. Colloid. Interface Sci. 1979, 69, 225-232. Kanellopoulos, A. G.; Owen, M. J. J. ColloM Interface Scl. 1971, 35,
120.
Tuzar, 2.; Kratochvll, P. Advan. Col/o/d Interface Scl. 1978, 6 , 201. "Handbook of Tables for Organic Compound Identlflcatlon", 3rd ed.; C.R.C. Press: Boca Raton, FL, 1979; pp 429-439.
RECE~VED for review January 30,1981. Accepted May 18,1981. This work was supported by the Centre National de la Recherche Scientifique, Equipe de Recherche Associ6e 474.
