A Surface Area-Hydroxyl Correlation Employing Trimethylsiloxy-Treated Ammonium Silicates Gene E. Kelluml and James R. Hahn Dow Corning Corp., Midland, Mich.
A direct relationship between hydroxyl and surface area was obtained employing trimethylsiloxy-treated ammonium silicates. Six separate sets of samples were prepared from sols having surface areas from 500,to 1300 M2/g. Agglomeration at a pH of 5 to 7 produced lower surface areas as a function of reflux time. The surface area of the samples was measured by alkali titration just prior to treatment with trimethylsiloxy groups. Determination of carbon gave the treatment efficiency and total hydroxyl minus adsorbed water gave surface hydroxyl concentrations. After correction for inconsistent surface treatment and errors in surface area measurement, the hydroxyt data showed an excellent correlation with surface area from 300 to 1300 M2/g. The effect of surface treatment upon hydroxyl appeared to change slightly over the wide surface area range obtained. The average hydroxyl group population per m$ at 300 M2/g was 6.34 and 6.80 at about 1000 MZ/g after corrections were applied.
THE SURFACE CHEMISTRY of silica has generated volumes of experimental data attesting to the widespread interest in this subject. Much confusion exists in the literature for one reason or another. One area where disagreement is found is in the role of the hydroxyl group in some of the surface phenomenon of silicas. At least one good review is available which indirectly treats this problem (I). It has been generally accepted that no absolute function of surface area with hydroxyl exists. The BET method for surface area has not been a perfect solution to the problem of surface area definition, particularly with treated materials. In some instances the surface area of a silicate under a surface coating is desired, but cannot be obtained from BET data. A titration method for determination of surface area has been in use for several years (2). This method must be calibrated for the specific materials being used, often employing BET data for the standard. The titration procedure is not applicable to hydrophobic, treated silicas. Methods proposed for quantitative analysis of hydroxyl apparently have been limited in application for determination of surface area, even over short ranges. No method has been presented to adequately measure the surface area of trimethylsiloxy treated silicas and correlate this to the original surface of the materials. The work reported here was motivated by the desire to apply recently developed analytical techniques for determination of water and silanol to a systematic series of silicas that could be studied before and after surface treatment. Objectives were to obtain a surface area-hydroxyl correlation for a treated silica, and observe the reproducibility of sample preparation and analytical methods. The work described below provides unique techniques for the systhesis of conPresent address, Gulf Research and Development Co., Kansas City Laboratory, 9009 W 67th St., Merriam, Kan. 66202
(1) J. A. Hockey, Chem. Ind. (London), 1965, p. 57. (2) G. W. Sears, ANAL.CHEM., 28, 1981 (1956).
952
ANALYTICAL CHEMISTRY
sistent, fresh silicas, and applies recently published methods to analysis of the substituents necessary to obtain the desired correlation. The data allowed comparison of treatment efficiency and reproducibility, surface areas before and after treatment, and concentrations of hydroxyl based upon the original surface. EXPERIMENTAL
Silica Materials. The silicas used in this study were prepared by known methods of the art. The ammonium silicate sols were prepared as described by Frederick J . Walter (3). These sols were aged for various lengths of time such that silicas of a variety of surface areas were obtained. A total of six silica sols of surface area ranging from 500 to 1300 M2/g were prepared. Each sol was deionized by passing through a column containing a strong cation exchange resin. The deionized sols were then refluxed at a controlled pH (pH between 5 to 7) for 24 hours. Under these conditions the silica particles agglomerate accompanied by a reduction of surface area with reflux time. Aliquot samples were taken at various times during the reflux such that a variety of surface area silica hydrogels were obtained from each sol. A total of 32 silica samples with surface areas ranging from 320 to 1300 M2jg were obtained. A small portion of each aliquot sample was used for the surface area measurement and the silica from the remainder of each sample was trimethylsiloxy [MesSiOl,2] coated by methods described by Leslie J. Tyler (4). The Me,SiOl,2 coated silicas were recovered as powders and were all dried for 16 hours at 150" C . Surface Area Measurement. The silica sols were titrated with alkali employing the method of Sears ( 2 ) to give the surface area just prior to trimethylsiloxy treatment. After treatment the surface area was determined utilizing nitrogen sorption techniques. The apparatus and methods were similar to those given by Emmett and Brunauer (5) and Nelson and Eggertsen (6). Measurement of Surface Treatment. The effectiveness of surface treatment was determined by analysis employing standard micro train-combustion techniques. Determination of Hydroxyl and Water. Silanol group concentration was obtained using methods of Kellum and Smith (7, 8). The exact procedures are available from the above references. Surface water concentrations were determined with the modified Karl Fischer reagent titration method reported by Smith and Kellum (7-9). The specific procedures may be readily obtained from their work. ~~
~
(3) Frederick J. Walter, U. S. Patent 2,671,056; Chem. Abstr. 48,
5402e (1954). (4) Leslie J. Tyler, U. S. Patent 3,015,645 (1957); Chem. Absrr., 52,
33981 (1958). (5) P. H.Emmett and S. Brunauer, J. Amer. Chem. Soc., 59, 1553 ( 1937). (6) F. M. Nelson and F. T. Eggertsen, ANAL.CHEM., 30, 1387 (1958). (7) R. C. Smith and G. E. Kellum, ibid.,39, 339 (1967). (8) G. E. Kellum and R. C. Smith, ibid.,39, p. 341. (9) R. C. Smith and G. E. Kellum, ibid.,38, 67 (1966).
130C
/ A
IlOC
9oc
400
SAMPLE SET :
C 70C
m 200
I
I
I
I
I
I
I A l 1
I
I
I
I
I
m:
50C
RESULTS AND DISCUSSION
The silicas were prepared in six separate sets containing five or six samples each. The primary particles, before agglomeration, varied in surface area from about 500 to 1300 M2/g. The values for the various sets of materials were 495, 655, 795, 800, 1100, and 1300 M2/g, The agglomeration procedure then produced surface areas to about 300 M2/g. Since silicas of the type reported in this work are extremely difficult to isolate as dry powders, in the same form as in solution, it is necessary to coat the particles with trimethylsiloxy groups to prevent agglomeration when they are removed from the water solution. To evaluate the proposed hydroxyl-surface area relationship, certain variables in this system had to be defined and evaluated. Some of the possible parameters present were:
(1) Total hydroxyl group concentration on the original surface. (2) Surface treatment with trimethylsiloxy groups. (3) Surface area measurements. (4) Carbon analysis measurements. (5) Hydroxyl analysis measurements. The following assumptions were made from the possible variables in order that the desired relationship be more effectively evaluated :
(1) Total hydroxyl groups/mp2 on the surface of all samples was nearly constant. (2) Trimethylsiloxy and hydroxyl groups/mp2 on the surface were equal to this constant: e.g., each bonded trimethylsiloxy group affected about the same number of hydroxyl groups. Variable surface treatment could be detected graphically and the number of bonded groups corrected for it. (3) Surface area measurements were generally correct, but when errors occur, they should be obvious. (4) Carbon analysis was quantitative. (5) Hydroxyl analysis was not biased by degree of agglomeration or basic particle size.
0
UNCORRECTED
0
CORRECTED
0
UNCORRECTED
I
CORRECTED
A
UNCORRECTED
A
CORRECTED
30C 10
14
18
22
26
30
34
38
WT. % Me3Si01,2
Figure 2. Selected experimental data of trimethylsiloxy us. surface area The above assumptions allowed observation of the hydroxyl parameter essentially free of other variables. On a routine basis these parameters left unevaluated could cause somewhat variable analysis data. The following calculations were considered for evaluation of the data. The actual weight of silica in a sample was the observed weight minus the weight due to the trimethylsiloxy treatment. All sample weights were corrected. The trimethylsiloxy surface treatment was nearly constant, therefore all points on the weight per cent trimethylsiloxy versus surface area plots must fit a line considered to be representative for any set of data. Inconsistent surface treatment or surface area was corrected as shown in Figure 1. Plots of surface area versus weight per cent trimethylsiloxy and hydroxyl measurements are given. A representative surface area line has been drawn through each set of data. If an error in a surface area measurement is present, both the weight per cent trimethylsiloxy and hydroxyl points will be on either the right or left of the lines, like a 4 and c-d. The point from the trimethylsiloxy curve is positioned on the line by vertical
Table I. Recovery of Hydroxyl after Trimethylsiloxy Surface Treatment OH/M2 lost in treatment Sample Predicted" Found' Cabosil S-17 1.70 1.75 Ludox AS 1.06 1.24 Calculated assuming that two hydroxyl groups were lost by analysis for each trimethyldoxy group bonded. 0
VOL 40, NO. 6, MAY 1968
953
DATA SET: 0
UNCORRECTED
I CORRECTED
1100
x
0 UNCORRECTED
CORRECTED
900
-
700
-
so0
-
300 0
A
UNCORRECTED
A
CORRECTED
.5
I 2.0
1.0
3.0
5.0
4.0
WT.
e /'
6.0
I
8.0
7.0
OH
Figure 3. Selected experimental data of hydroxyl
US.
surface
area
1.5
Set I
I1
500
I11
455 425 380 320 665
BET S.A. 687 630 628 594 459 287
540
515 405
IV
V
350 850 615 610 495 370 795 560
VI
954
535 445 346 1270 1115 1025 810 760 540
Wt.
MeaSiOliz 33.1 26.5 21.6 19.1 16.1 11.2 14.1 13.1 13.1 12.3 10.1 17.9 15.4 15.4 12.8 10.2 21.8 17.3 16.7 13.8 11.0 18.9 15.7 15.3 13.1 10.5
578 800 678 550 503 356
ANALYTICAL CHEMISTRY
32.4 26.2 25.0 20.4 17.4 13.1
Ip,
I 2.5 OH/muZ
I A. I 3.0'6.0
I 6.5
7.0
TOT. gpr/rnu2
Figure 4. Trimethylsiloxy, hydroxyl, and total groups per mp us. surface area
Table 11. Experimental Data Measured surface area 1070 890 742 642 495 343
d*.l
I 2.0 2.5'2.0 Me3SiOh/muz
z
Wt. OH 6.49 5.21 3.96 3.41 2.78 1.64 3.53 3.25 3.10 2.56 2.00 4.68 3.41 3.32 2.89 2.11 5.66 4.12 3.88 3.37 2.59 5.72 3.45 3.64 2.89 2.30 8.02 6.76 5.94 6.41 5.42 4.18
Wt.
z
HzO
0.30 0.41 0.45 0.71 0.59 0.31 0.89 0.49 0.47 0.53 0.83 0.39 0.37 0.91 1.02 0.79 1.53 1.85 1.61 1.53 1.22 0.13 0.71 0.70 0.65 0.16
0.13 0.17 0.14 0.18 0.14 0.12
movement along the surface area axis and that amount of correction applied to the point from the hydroxyl curve. If the assumption made was correct both points should lie on the lines. An inconsistent surface treatment will be evident from the occurrence of the hydroxyl and trimethylsiloxy points on opposite sides of their respective curves. In this instance the trimethylsiloxy point is moved along the weight per cent axis to the line. The corresponding weight per cent correction is then applied negatively to the hydroxyl point. A correction of 0.1 % trimethylsiloxy theoretically equals a correction of 0.021% hydroxyl. Table I shows that each trimethylsiloxy group reacts so that analysis indicates a loss of two hydroxyl groups. Both silicas mentioned in this table were available before and after surface treatment. This phenomenon has not been completely explained but may arise from steric effects of the trimethylsiloxy group. Corrections for inconsistent treatment applied to hydroxyl results must then be doubled to be correct. Each 0.1 trimethylsiloxy correction is then equal to 0.042% hydroxyl. If errors were as assumed, then both points should lie on the lines. Each set of results was subjected to this data treatment. Both original and corrected weight per cent data were plotted on the same graphs. The trimethylsiloxy groups/mp2, hydroxyl groups measured/mpz, and calculated total groups/mp2 were plotted versus surface area. The original data adjusted for sample weight errors are presented in Table 11. Nitrogen sorption surface area data are included for the two high surface area samples to indicate how ineffective it was on treated silicas. Figures 2 and 3 give three sets of results in weight per cents before and after data corrections were ap-
plied. The correlation is reproducible and the materials used appear fairly consistent. Figure 4 shows the observed and calculated groups of hydroxyl, trimethylsiloxy, and total groups per mp2 of surface area, All curves show the same general trends indicating that assumptions made were generally reasonable. The slopes of group/mp 2. the various lines were calculated as The slopes M 2/g of the hydroxyl plots varied from +9.8 X to -7.0 X 10-4 as shown in Table 111. Slopes of the trimethylsiloxy curves varied from f4.2 X to -23 X The greatest variation in results obviously occurred in the trimethylsiloxy data. Figure 5 presents the average groups per mp2 of surface as calculated from the six separate sets of data. Slopes were -0.27 X -7.0 X lov4,and -12.0 X 10-4 for hydroxyl, trimethylsiloxy, and total groups, respectively. One might conclude that the effect of trimethylsiloxy upon hydroxyl shifts slightly over a large surface area range due to the net negative slope of the averaged groups per mp2. When overall treatment efficiency varies, the hydroxyl varies by twice that amount. This is responsible for some of the scatter between the various sets of data, as evidenced in sets I and VI which do show a difference in treatment efficiency. If the trimethylsiloxy data were moved to the average line, the hydroxyl would also follow the average relationship. The data indicate about 6.34 hydroxyl groups per mp2 at 1000 M2/g and 6.82 hydroxyl groups per mp2 at 300 M2/g. The range of results is well within the values expected for these type materials. Variations noted in the slope of plotted data are considered typical for this type of silica system. It might also be conjectured that the measured hydroxyl group concentration is correct and that the HCl from the trimethylsiloxy treatment condensed some hydroxyl groups. This is something that is difficult to prove, however. Results from all six sets of samples were plotted as a single set of data with the points evaluated as previously described based upon a representative line drawn through the measured values. The slopes obtained from this calculation were -4.00 X +0.500 X and -6.33 X for tri-
I1 I11 IV V VI
+9.8 x +2.0 x -1.2 x -3.7 x -7.0 x Av. Slope = -0.27 X
Sample set no
S.A. range
I
350-1 100
I11
60C
40C
20c
I An. I 3.0 5.0
I
1.0
2.0
370-850
10-4 10-4
10-4 10-4 10-4
4.53-11.62 = 0.28-0.61 % OHT = 2.13-5.58 Z C = 4.64-8.16 Z HzO = 1.14-1.61 OHT = 4.54-7.22 C
=
I
7.0
TOTAL GPS/lOOAz
MEASURED GPS/IOOAz
methylsiloxy, hydroxyl, and total groups per mp2. The total groups varied from 6.34 groups/mp2 at 300 M2/g to 6.78 groups/mp2 at 1000 M2/g. The slope of the average total groups per mp2 was less negative than that obtained from the first evaluation but not significantly so. If it was assumed that the hydroxyl measured was correct, then the total groups would be between 4 and 5 per mp2 and would indicate less difference between the sample sets. Precision of the various analytical procedures was evaluated and is given in Table IV.
-2.3 -9.0 -6.0 -9.0
+0.58 Av. Slope = -7.0
6 5
Slope of total data
X 10-3 x 10-4 X 10-4
x 10-4 x 10-4 x 10-4
Av. Slope
Table IV. Typical Precision of Experimental Methods Number Duplicate average of Average range of results samples range
Z Hz0
I
I 6.0
Figure 5. Average surface groups per mp us. surface area
Table 111. Slopes Obtained from Plotted Data Groups/mp 2 Slopes were calculated as M 2/g Slope of Slope of hydroxyl data trimethylsiloxy data -1.5 x 10-4 + 4 . 2 x 10-4
Data set no I
800
0.0780 0.0230 0.190 0.0420 0.118 0.212
=
+2.2 x -2.0 x -1.4 X -1.4 x -2.2 x -5.6 x -1 . 2 x
10-4 10-8 10-3 10-3 10-8 10-4 10-3
Average re1 range
Re1 std dev, Z
1.05
0.286 2.26 1.72 0.341 4.35 1.87
5.60 4.29 0.688 8.72 3.67
VOL 40, NO. 6, MAY 1968
955
Sample analysis data are given for two sets of the silicas. Each result shown in the table was a duplicate average from which was determined as absolute and relative range. The standard deviation and relative standard deviation (RSD) were calculated from the average relative ranges. The carbon analyses yielded RSD values of 0.286% and 0.341 %. Adsorbed water determination data indicated RSD values of 2.26% and 4.35%. The RSD figures for total hydroxyl analyses were 1.72 and 1.87 %. The precision of analyses in all instances was considered excellent and reflected the effectiveness of the new methods for this type of application. The
speed, precision, and applicability of these methods are generally not possible with other methods. The silicas employed in this study were very consistently prepared, effectively treated, and reasonably uniform in surface character. The correlation obtained from this work probably does not hold for wet process or precipitated silicas in general. A correlation might exist for any silica of specific type but no one relationship should be completely effective. RECEIVED for review September 13, 1967. Accepted March 8, 1968.
Ion Pair Dissociation of Solvated Bases and Base Perchlorates in Anhydrous Acetic Acid Orland W. Kolling Chemistry Department, Southwestern College, Winfield, Kan. 67156
A simple potentiometric method was devised for the determination of ion pair dissociation constants for representative solvated bases and their perchlorate salts in anhydrous acetic acid as the solvent. Sodium acetate and sodium perchlorate were the reference solutes used in this procedure requiring emf measurements on separate solutions of the base and the halfneutralized base. Trends in the perchlorate salt dissociation constant as a function of the basicity constant were correlated with the BruckensteinKolthoff theory of acid-base neutralization in anhydrous acetic acid. It was found that the value of the base perchlorate dissociation constant is always larger than the basicity constant, and their ratio ranges from 10: 1 for strong bases to 106: 1for very weak bases.
THE PRECISE INTERPRETATION of potentiometric and spectrophotometric titration curves for acid-base reactions in anhydrous acetic acid requires a knowledge of four equilibrium constants; the ionization constants for the solvent and the protonic acid; the basicity constant for the solvated base; and the ion pair separation constant for the salt containing the protonated base. Although the complete quantitative treatment of neutralization equilibria in glacial acetic acid was developed a number of years ago by Bruckenstein and Kolthoff ( I , 2), the experimental verification of that theory, and particularly its simplifying assumptions, has rested upon a very restricted set of reliable constants. Also, more accurate ion pair dissociation constants for salts are needed in the examination of ionic strength effects in acetic acid as the solvent (3). The present investigation was limited to the determination of basicity constants and ion pair dissociation constants for perchlorate salts, because perchloric acid is the most common titrant for bases and reasonably precise constants for the acid and the solvent have been established. A comparative (1) S. Bruckenstein and I. M. Kolthoff, J. Am. Chem. Soc., 78, 2974 (1956). (2) I. M. Kolthoff and S. Bruckenstein, in “Treatise on Analytical Chemistry,” Pt. I, Vol. 1, Interscience, New York, 1959, pp 499-511 and 524-31. (3) 0. W. Kolling, Trans. Kansas Acad. Sci., 70, 9 (1967).
956
ANALYTICAL CHEMISTRY
method based upon the sodium acetate-sodium perchlorate equilibrium as a standard system and the half-neutralization potential measured with the glass-calomel electrode pair was used to determine constants for a representative range of strong to very weak bases in anhydrous acetic acid. General trends in the dissociation constants for nitrogen bases and their perchlorate salts were observed. THEORY
The defining equilibria for the dissociation of the solvated base and its perchlorate salt, together with the corresponding constants, are shown by Equations 1 and 2. B*HOAce BHf
+ OAC-
KB =
[BHf] [OAc-] [BHOAc]
(1)
Bruckenstein-Kolthoff notation is used throughout and the additional defined symbols required are Ks for the autoprotolysis constant of the solvent and Ct for the stoichiometric concentration (M/L)for the particular solute, i. Because the response of the glass electrode in acetic acid is identical to that of the chloranil electrode used by Bruckenstein and Kolthoff, their derived equations based upon the Nernst relationship can be applied without alteration ( I ) . For a solution of base alone, Equation 3 gives the potential for the glass-calomel electrode pair. EB
=
+ E1 + RT In Ks - RTln KBCB F 2F
EoOc
(3)
(E, is the liquid junction potential for the reference electrode and EOcc is the total standard potential for the cell.) This equation forms the basis for determining the pKB of a base by direct comparison to a reference base having an identical concentration, provided CB is at a sufficiently low level that