INFRARED TRANSMITTANCE OF FLUORINE-CONTAINING POROUS GLASS
Oct., 1963
t
and c2, and Fig. 4 shows a comparison of the experimental and theoretical results. It is seen that the agreement between the two is indeed excellent.
0.4
TABLE I ANISOTROPY AND h CALCULATED FOR VALUES OB ELECTRICAL BENTONITE ]'ARTICLES O F VARIOUS SIZES
0.3
Mean OLI ma for ai a a for bl for fraction 62 for fraction particle size, B. fraotion 1, om.$ fraction 2, 8111.3 I, cm./v. 2, om./v. 6064 4.93 X 10-18 2 4 X IO-'$ 1 47 X 10-3 0.11 X 4110 5.80 X 10-1a 1 32 X 10-13 2870 2.68 X 10-13 0 606 X 10-13 0 95 X 10-2 0.223 X
-
An(E) -
An8
- -24
1
=
-2[f@1
(bl, c1, E )
+
fiiP2
(bz, c2, E ) ]
(25) with 4 from eq. 12 to fit the birefringence data. For An8 the calculated value from a.c. measurements can be substituted. With this siniplification and previous knowledge of c1 and c2 from a.c. measurements, the d.c. birefringence measurements were fitted to eq. 25 to evaluate bl and bt with the help of another program on the IBM 7090. Again, a method of least squares was used. Table I lists the values of bl and b2 calculated in this manner. Also in Fig. 5 we have shown a comparison of experimental results with the birefringence values obtained by the theoretical expressions of eq. 25 and 12. It is seen that the agreement is not so good as would be desired. This is due to the fact that better estimates of bl and b2 could not be obtained due to numerical difficulties in the computer program. Furthermore, the assumption of bidisperse distribution is an approximation so that the estimates of b are accurate only t o the order of magnitudes given. Conclusions An extension of birefringence saturation theory, which considers the case where orientation torques due
/
L
/
0.2
-
The evaluation of bl and bl from the experimental measurements under d.c. fields is somewhat more complicated, since we have to use the equation
2219
v)
21: 0.1
L '
0 -0.1
L
'
0 000
100
200 300 400 500 FIELD STRENGTH, E .
Fig. 5.--Comparison of experimental and theoretical values of birefringence under mixed orientation (particle size 2970 A.).
to permanent and induced dipole moments are perpendicular to each other, shows that for certain ratios of these two dipole moments the birefringence undergoes a minimum and exhibits reversal of sign with increasing electric field. This theoretical analysis supports the model proposed to explain the anomalous birefringence behavior of bentonite in our earlier publications. A comparison of the experimental results for three particle sizes of bentonite (size measured with an electron microscope) with the theoretical analysis leads to evaluation of the electrical anisotropy of these particles as well as the value of b which is directly proportional to their permanent dipole moments. Acknowledgments.-The
author is grateful to Prof.
C. T. O'Konski for helpful suggestions, to Mr. W. N. Syn for carrying out the numerical computations, and to Mr. C. M. Hart for assistance in experimental measurements.
CHANGES IN INFRARED TRANSMITTANCE OF FLVORINE-CONTAINING POROUS GLASS ON HEATING BY THOMAS H. ELMER, IAND. CHAPMAN, AND MARTINE. NORDBERG Research & Development Laboratory, Corning Glass Works, Corning, N e w York Received April 16, 1963 The infrared transmittance of plates of porous glass impregnated with aqueous animonium fluoride solution was measured as they were heated in air and in uucuo from room temperature to 1000". The elimination of water is shovn t o be complete at 800". At this temperature, the untreated glass still contains appreciable water bonded as free OH groups at the surface. The complete removal of water is due to a replacement of OH groups by fluoride ions. Shrinkage of ammonium fluoride-treated porous glass was somewhat lower than for untreated glass. Changes in the infrared bands obaerved during rewetting and further heating are discussed.
Introduction Porous glass has attracted attention as a rigid substrate for the study by infrared techniques of adsorbed molecules. The infrared spectrum of porous glass it-. self between 1.2 and 8.0 p shows a number of absorption peaks due t o surface OR groups, boron oxide, and silica. The principal band due to OH groups is that between
2.68 and 3.2 p, the position of the peak and the width of the band being dependent upon the degree of hydrogen bonding between the surface silanol groups and the presence of physically adsorbed water. The presence of these surface silanol groups has complicated the infrared sgectra of physically adsorbed molecules containing C-H, O-H,and N-El bands who% fundamental
THONAS H. ELMER, IAX11. Cii-irmx, . i x ATARTIN ~ E. NORDBERG
3220
90
1
These silanol groups in porous glass cannot be removed completely by heating even to temperaturcs where the structurc is consolidated and the high surface area is lost, which occurs a t 1000-1100". Folinan' recently attempted an exchange reaction by chlorinatioii with SOCh but found tliat only an 80% exchange was possible. Elmer2 noted that if porous glass way trrated with a dilute ammonium fluoride solution before being heated to temperatures above lOOO', which causes the porous siliceous structure to collapse and form a nonporous glass, the fundamental 011 vibrational frcquency peak in the infrared bpcctriun of the silica glass was coinpleteIy removed. Tlicreforc, it seemed that a porous glass codd he made that would contain no silanol groups and could be used far, among other things, the study of adsorption on silica of water, on sites other than surface 011groups. Experimental
loo 80
70 '
60
% 2
50
ap
40
30
a
I-
20
IO 0 WAVELENGTH,
p
.
Fig. 1.-Infrared spectra a t room temperature of ammonium fluoride-treated porous glass, 0.25 mm. thick, after heating in air to 300" and in uucuo from 300 to 800": 1, empty cell: 2, 110"; 3 . 300'; 4,500'; 5, 600"; 6,700'; 7, 800". The spectra at 900 and 1 0 0 0 O were the same aa in curve 7 .
loo 90
oo
' '
'O 60
f
40
L-
t t
2.69
A 2 loo
so 90
t
WAVELENGTH,
p
.
t
. 70
'
60
0
3
Vol. 67
50
t ;4 0 z 30 20
10 0 WAVELEYQTH,
p ,
Fig. %-Infrared spectra at room temperature of porous glass plates, 0.25 mm. thick, heated to 800" and exposed to water vapor a t 4.6 mm. pressure for 18 hr.: A, after heating in vacuo; B, after exposure to water vapor; curve 1, ammonium fluoridetreated glass; curve 2, nonfluoride treated glass.
vibrational frtqucncies all lie within the fundamental vibrational frtyuency region observed for isolated surface silanol and bonded silanol groups.
Samples.-Porous glass, an intermediate gl:iss obt:iinctf by heat treating and leaching a special soft alk;ili-borosilic:it(. glass,* was used in this study in the form of plates, 3 x 1 cm. X 0.23 and I-nim. thickness. Samples used in comparative stucfics between standard and ammonium fluoride-ti ratctf porous glass were cut from the same shret to eliminate anv posyiblc variations in glass properties due to manufacture. Ammonium Fluoride Treatment and Baking Procedure.-The ammonium fluoride treatment consisted of irnprrgna ting an nirdried porous platc in a 30yc XH,F solution :it room tempcrnturr for 20 min. followed by rapid rinscs in distilled miter to wash ofT excess fluoride salts from the surfaces. The gl:iss then was tiricd in an oven a t 110' for 10 min., c~)oledto room tcmpeiatiirc', tinti immersed in 1 r\i HNOa solution to fix the fluorides niorc coinpletely in the porous glass. After again baking at llOo for 15 min., the glass was placed in a cell and hcated from 100 to 300' in air in 30 min. and on to 800' or higher in vacuo in 10:)' steps for 0.5 hr. a t each successive tcrnperature (Fig. I ). The plates, for which infrared spectra arc shown in Fig. 2, deviated from the above heating schcdule in that thev n ere heated in about 3 hr. from 100 to S00" in vmuo :idthen hchntcd for 0.5 hr. a t 800". Other 0.25 and 1-mm. thick platea which had not been impregnated in ammonium fluoride solution were given thermd trratments identical with those given the fluorine-coittaining plates. The surface areas of 800" baked 1-mm. thick sat1q)les were determined by thr I3.E.T. method using Kea t its boiling point. The ammoniuru fluoride-trcnted gl:iss anti thc untretttd glass had surface areas of 105 a n d 190 in.2/g., rcspcc.tivcly. Length changes ocwirrlng during hrating of thc, porous plates were observed a t room tciiiperature hy niettns of a traveling microscope. Infrared Measurements.-"I'hr ( ~ 1 1for the jnfrnrrd studies was made of fused quartz and '36' : slliva glass, the porous sarnplr being held between the two p:ir;illd fzicrs of the Euscti qurirtz windows. The infrared spec-trti 'i\c'rr rwordcd u t room ternpcrature with a Perkin-Elmer llodrl 2 I infr:ircd s1)ectrol)hotoineter with sodium chloride optics. 'The cell also served as a rcfmctory vessel for vacuum baking the porous ghss plates.
Resclts Infrared Measurements.- -The infrared sprctra of ammonium fluoride-treated glass obtained a f t t r successive heating in uuczm at tempcraturrs from 300 to 1000' are shown in Fig. 1 . The strong absorption band observed at around 2.7 p due to OH groups together with the weaker bands at 1.4 and 2.2 decrease with temperature as shown. llfter 800" baking, all these bands are absent, indicating that the glass no longer contains silanol groups in its structure. The bands a t 1.4, 2.2, and 2.7 p were also absent when an ammonium fluoride-treated porous plate was (1) M. Folman, Tram. Fa?aduf/Soc., 67, 2000 (1961). (2) T.H. Elmer, U.S. Patent 2,982,053, (Iltry 2. 1961). (3) h l . E. Nordbprg, J . A m Ccram. Soc., 27, 299 (1944).
Oct., 1963
INFRARED TRANSMITTANCE OF FLUORINE-CONTAINING POROUS GLASS
heated in vacuo from 100 to 800" at furnace rate in 3 hr., but these bands were still present in an untreated porous plate heated to 800" in a similar manner (see Fig. 2A). The treated and untreated porous plates after heating to 800" were exposed for 18 hr. a t room temperature to water vapor a t a pressure of 4.6 mm. The spectra of the two plates were then recorded (see Fig. 2B). The spectrum of the ammonium fluoride-treated glass shows only a broad asymmetric band between 2.6 and 3.4 p with a peak at 2.76 p. The spectrum of the untreated glass shows two bands, a sharp one a t 2.68 p and a broad one a t 2.74 p. The former remains distinct a t 2.68 p , which indicates that there is no perturbation of free silanol groups by the adsorbed water. Another ammonium fluoride-treated porous glass plate which had been exposed to water vapor a t a pressure of 4.6 mm. after heating to 800" was degassed in vacuo a t temperatures ranging from 25 to 1000" to determine whether the glass gives off its sorbed water reversibly. The absorption band decreased on heating as shown in Fig. 3 . After heating a t 800" or higher, only the background of the fused quartz window remained, indicating that the glass was completely dehydrated again.
Discussion The spectra in Fig. 1 show that ammonium fluoride treatment of porous glass removed all the OH grmps, the small band at 2.73 fi being due to the fused quartz cell. Furthermore, the small bands a t 1.4 and 2.2 p observed in the porous glass containing no fluorine (Fig. 2) also have been removed, thus showing that they are due t3 overtones and combination overtones of the fundamental silanol stretching and bending frequencies. Their absence can prove to be a useful criterion of whether or not there are silanol groups in the glass, even in the presence of adsorbed mater whose overtones and combination overtones were not observed in this experiment. The fact that the OH absorption band at 2.7 p can be eliminated by the introduction of ammonium fluoride in porous glass indicates that fluoride ions substitute for hydroxyl groups in the glass. The infrared spectra, however, lead one to conclude that the replacement of hydroxyl groups by fluoride ions in porous glass does not become appreciable until it is heated to elevated temperatures. Results of successive heating in 100" steps for 0.5 hr. at each temperature in vacuo show that the replacement of hydroxyl groups by fluoride ions is complete at 800". This effect of fluoride ions is not surprising, since from the point of view of crystal chemistry, fluoride ions and hydroxyl ions are similar. The ions have not only approximately the same size but are isoelectric. Evidence for fluoride ion substitution for hydroxyl ion can be found not only in nature (e.g., in topaz, ,41~(F',09)&3i04, the percentages of fluorine and hydroxyl vary greatly) but also in synthesized minerals. For instance, Eite14has shown that the replacement of hydroxyl in mica by fluoride ions was successful. Further evidence for the interaction of fluoride ions was reported by Specht,6who showed that these ions in aqueous solution are either adsorbed on (4) W. Eitel, "The Physical Chemiatry of the Silicates," The IJniversity of Chicago Press. Chicago, Ill., 1954, p. 1123. ( 5 ) R. C. Specht, Bnal. Chhsnz., 28, 1015 (1956).
2221
90
I
2.73 50
Y
2
30
2 76
t 800. 70
72.73
Fig. 3.-Infrared spectra a t room temperature between 2.5 and 3 p of porous glass, 0.25 mm. thick, after exposure to water vapor a t 4.6 mm. pressure and after successive degassing in vucuo a t temperatures from 25 to 800". The sample was heated a t 25' for 2.5 hr. and for 1hr. a t all the other temperatures. N.D. = not degassed.
the surface of borosilicate or soft glass or bound on the lattice. Sonders, Enright, and WeyP gave similar evidence as early as in 1950 in connection with the importance of hydroxyl groups for the wettability of a soda-lime silicate glass. Finally, in the preparation of the "water-free" glass mentioned earlier it was found that the fluoride ion concentration in the glass, as determined by chemical analysis, is approximately equal to the hydroxyl concentratim in the glass with no ammonium fluoride treatment. The hydroxyl concentration was determined in a manner similar to that used by Scholze.' During the course of heating the ammonium Auoridetreated sample to 800", irregular contractions and expansions mere noted. Although these effects are small they seem to be larger than can be explained by inaccuracies in measurement. During heating, ammonium fluoride and any other fluoride, such as ammonium silicofluoride, which may have formed in the pores will dissociate and may react further with the glass as the temperature is raised to account for the irregular shrinkage. Shrinkage was slightly less for the ammonium fluoride-treated glass than the untreated glass, over-all linear shrinkage based on original sample length after heating a t furnace rate to 800" being 1.6 and 2.0%, respectively. The decrease in shrinkage can be explained by the fact that fewer silanol groups on the surface are available to form Si-0-Si bridges when fluoride ions are present in the glass. The nitrogen adsorption measurements show that the combined ammonium fluoride and heat treatment of porous glass causes a greater reduction in internal surface area than the thermal treatment of the glass to 800" alone. Silica attack resulting from the ammonium fluoride treatment would reduce the measured surface area either by direct removal of silica to eliminate surface or by the precipitation of siliceous reaction products within the pores to block them. The question that remains is the state of the water adsorbed on the 800 " baked, fluorine-containing porous glass. It causes the broad asymmetric band shown in Fig. 2B. Considerable evidence has been accumulated to suggest that most of the water adsorbed upon (6) L. R. Sondsra, D. P. Enright, and W.A. Weyl, J . A p p l . Phys., 21,338 (1950).
(7) H. Scholse, G'lastecchn. Ber., 83, 81 (1959).
2222
L. E. TOPOL
ordinary porous glass is hydrogen bonded to silanol groups. The substitution of fluoride ions for hydroxyl groups in the case of the ammonium fluoride-treated glass has changed the interfacial forces between the glass and water; the glass has become more hydrophobic. Unpublished work in these Laboratories shows that there is a reduction by a factor of 10 in the amount of water adsorbed a t room temperature and low per cent relative humidities by ammonium fluoride-treated over untreated porous glass under the same water vapor pressure. The considerable decrease in water adsorbed cannot be explained solely on the basis of loss of surface area following the ammonium fluoride treatment. This suggests that water adsorption upon a fluoride ion via hydrogen bonding with one of the hydrogen atoms of the water molecule is not likely. The absence of the peak at 2.68 p in the case of the fluorinecontaining glass (curve 1 in Fig. 2B) suggests that no free silanol groups are formed as mater is adsorbed a t room temperature. Furthermore, the overtone bands at 1.4 and 2.2 p which are due to silanol groups were not observed. Evidence of a silanol peak at 2.68 p for the above glass was found on reheating it after exposure to water vapor as shown in Fig. 3, presumably due to the replacement of fluorine atoms by hydroxyl groups. The activation energy necessary to produce this hydrolysis reaction calls for temperatures higher than 300'.
Vol. 67
Folnian and Yates8 have shown that the water initially adsorbed by ordinary porous glass is not hydrogen bonded to silanol groups, but must be taken up by other sites. They suggest that bonding via the oxygen atom of the water molecule to a silicon atom occurs, producing a quite stable bond. This mould give rise to a band around 2.8 p and could be the source of the band noted for the ammonium fluoride-treated glass. I n this case, there is also the possibility that the water is bonded to surface atoms other than silica. I n the manufacture of porous glass, the heat treatment that the alkali borosilicate base glass receives causes phase separation into a silica-rich phase and a boronalkali-rich phase. This latter phase is removed by subsequent leaching in acid solution, leaving the porous skeletal structure. Some boron and aluminum oxides remain within this structure, although normally there would be few boron or aluminum ions a t the surface. However, if the porous glass is heated above 500' (as was the case in this experiment) boron is known to migrate to the surface since boric acid "whiskers" are observable if the glass is placed in a humid atmosphere. Here then some of the water could be bonded to surface boron atoms, producing the broad band a t 2.8 p which has been observed in water-containing boric oxide g l a s ~ e s . ~ (8) M.Folman and D. J. C. Yates, Trans. Faraday Soc., 64, 1684 (1958); also, A. N. Sidorov, Buss. J . Phys. Chem., 80, 9115 (1956). (9) A. J. Harrison, J . Am. Ceram. Soc., 30, 362 (1947).
ELECTROMOTIVE FORCE MEASUREMENTS IS MOLTEN DIVALENT METAL-METAL HALIDE SOLUTIONS' BY L. E. TOPOL Atomics International, A Division of North American Aviation, Canoga Park, California Received April WR, 196s Potentiometric studies on concentration cells of the type C, AT ( N L ) ,MXZ(1 - XI) /I MXZ(1 - SZ), M (A'i), C, where M represents Hg, Cd, Pb, and Zn, MX7the halide, and 1Y denotes the metal mole fraction, were made a t various temperatures. Slopes of plots of the log metal concentration 21s.e.m.f. yielded apparent Nernst n-values of 2 for Hg-HgCl? at 300" and for Cd-CdClz at 580°. For Pb-PbI? apparent Nernst n-values of about 2.2-2.3 were found at 5% and 693"; for Pb-PbClz the low metal solubility made detection of an n-value difficult but a value of 2 appeared probable. No ronstant e.m.f. values were obtained in Zn-ZnClz melts at 658 and 695". The above n-values of 2 indicate a two-electron reaction at the electrode and are consistent with the existence of metal atoms M or cation dimers hI2 +*. No evidence for the monomer ion M + was observed.
Introduction Electromotive force measurements of concentration cells with molten bismuth-bismuth halide solutions2z3 h a w yielded information as to the species present in these systems. These results ha\-e shown that more than one lower-valent metal species exists in these melts. Since a similar study in divalent metal-metal halide melts would be expected to yield information as to the entities present, the potentiometric procedure has been extended to cells with mercury, cadmium, and zinc dissolved in their respective molten chlorides and with lead in its fused chloride and iodide. (1) This work was carried out under the auspice8 of the Research Division U. 8. Atomic Energy Commlssion. (2) L. E. Topol, 8. J. Yosim, and R. A. Osteryoung, J . Phys. Chem., 65, 1511 (1951). (8, L. E. Topol and R. A. Osteryoung, ibid., 66, 1687 (1952).
o t the
Experimental Materials.-The chemicals used in this study were all reagent grade. HgC12 was dissolved in anhydrous methanol and filtered to remove any HgZC12 that might be present. The solution then was evaporated to dryness, washed with distilled water, and dried again. The salt was finally sublimed in a chlorine atmosphere a t 300". The system was purged of chlorine by repeatedly flushing it with dry nitrogen a t room temperature and then evacuating it. The CdCl,, ZnClz, and PbClz salts all were melted under an HC1 atmosphere; after several hours, the HC1 flow through the fused salt was stopped and Clz introduced. The molten salt was finally purged with nitrogen and filtered through a fine fritted disk. The CdC1, and ZnClz, after purification, were handled in a helium-filled drybox. PbIz was mixed with iodine and heated overnight in an evacuated, sealed vessel enclosed in an oven a t 150". The iodine was then removed from the mixture by sublimation a t 200" and the PbIz melted and filtered under nitrogen. -4gC1 and NaCl were dried under vacuum at 208". The metals used were melted under helium