243
CHEMISORBED AND PHYSISORBED WATERON METALOXIDES of experimental investigations of specific substances containing two peptide groups in a variety of conformations, the accent in this paper has been placed on treating rotatory strengths as functions of two variables, in this case the angles cp and $. The applications given in the last section have taken advantage of the fact that the Cotton effects of the n,a* transition in polypeptides, as well as in dimers, are largely vicinal in character and depend principally on nearest neighbors. The general semiquantitative agreement which is obtained between experiment and calculation for these polymers as well as for specific model diamides indicates that the maps obtained give a t least a coarse-grained description of the rotatory strength over the cp-$ plane. These maps are presumably most accurate where strong rotatory strengths are predicted and least accurate near nodal lines which are produced by the cancellation of large contributions from several mechanisms. Continuation of this work with the rotatory strengths of polypeptides as functions of chain length and conformation is now well under way.
It should be emphasized that the methods of sections I11 and IV are strictly applicable only to two peptide units and that the ultimate purpose of their presentation here is to provide an interpretation to a number of experimental studies on diamide models which have been performed in our laboratory. Experimental and theoretical results on these compounds will be presented later. Polymeric systems (helical or disordered) may be treated with no difficulty by the methods of section I1 and a number of such calculations have already been p e r f ~ r m e d . ~The ~ discussion of polymers given in section IV serves only to illustrate that a useful qualitative picture of the dependence on cp and $ of the n,n* rotatory strength may be obtained from the dimer maps. No such correlations are possible with the n,n*transition because it is extensively delocalized and therefore sensitive to total polymer geometry.
(35) V. Madison, unpublished work.
The Relation between the Amounts of Chemisorbed and Physisorbed Water on Metal Oxides by Tetsuo Morimoto, Mahiko Nagao, and Fujiko Tokuda Department of Chemistry, Faculty of Science, Okayama University, Tsushima, Okayama, Japan
(Recdved Augud t8,1968)
The first and second water adsorption isotherms and the water content on the metal oxide surfaces have been measured for samples of TiOa (rutile) and a-FezOs(hematite). After the first adsorption isotherm had been torr at room temperature for 4 hr, and obtained, the sample was degassed under a reduced pressure of then the second adsorption isotherq was measured at the same temperature as before. The determination of water content was carried out by the successive ignition-loss method. The amounts of physisorbed water (V,) and chemisorbed water (V,) were estimated from the two isotherms. The total amount of chemisorbed water was obtained by adding the water content ( v h ) to the value of V,. The sum of V , and VI, decreases slightly with an increase in the temperature of pretreatment of the sample, which indicates that the extent of reversible rehydroxylation of the sites dehydrated at higher temperatures is somewhat diminished. The ratio of V,, expressed in the number of water molecules per unit area, to the sum of V , and Vh, both of which are expressed in the number of hydroxyl groups per unit area, was found to be about 1:2 for all samples used, which indicates that a water molecule is adsorbed on two hydroxyl groups through hydrogen bonding.
Introduction It has been made clear that the surfaces of most metal oxides have hydroxyl groups in the atmosphere which play important roles in such surface phenomena as catalytic action, selective adsorbability, hydrophilicity, surface electrification, etc. Recently, the properties of the surface hydroxyl
groups have been investigated from the following standpoints. Surface hydroxyl groups can generally be removed by condensation dehydration at higher temperature in OBCUO, but the removal temperature considerably differs depending on the nature of the solid substrate.' On the other hand, the rehydroxylation of the dehydrated surface can proceed in the Volume 73, Number 1 January 1060
244 saturated water vapor at a rate possible to follow experimentally, which may also bo different with a variety of solid substrates.2,a The energy of surface rehydroxylation has been determined by heat-ofimmersion calorimetry for several oxides and has been found to be different for each substance.4-6 Infrared spectroscopic studies have shown that water molecules are adsorbed on the surface of oxides through hydrogen bonding with surface hydroxyl groups.7-18 The purposes of the present paper are to investigate quantitatively the relation between the amount of water chemisorbed as surface hydroxyl groups and the amount of water physisorbed on these groups and to discern the structure of the adsorbed layer of water on metal oxides. Hollabaugh and Chessick14have discussed theratio of the number of physisorbed water molecules to the number of surface hydroxyl groups on metal oxides. However, a measurement of hydroxyl groups remaining on the oxide surfaces was not made in their study, so that the discussion seems insufficient. I n the present paper, the true relation between the amounts of physisorbed and chemisorbed water has been dealt with, the latter including the number of hydroxyl groups remaining on the surface in addition to the amount of chemisorbed water determined from the adsorption isotherm of water. Using these data, we can consider the detailed structure of the adsorbed layer of water on metal oxide surfaces, as will be described later.
Experimental Section Materials. The samples used in this study were Ti02 (rutile) and two kinds of cr-Fe203 (hematite). The original sample of rutile, obtained from Teikoku Rako Co., was a commercial one prepared from titanium sulfate solution. One of the hematite samples, which was obtained from Nippon Bengara Go., was made by the calcination of FeSOl.7Hz0 a t 800" for 7 hr (F€&(I)). These two samples were treated with 0.1 N nitric acid several times to remove basic impurities, were washed repeatedly with 0.1 N ammonia to remove acidic impurities, then were thoroughly washed with distilled water, and finally were dried a t 110" for 8 hr. Another sample of hematite was prepared by calcination of a-Fe00H (goethite) a t 800" in air for 5 hr, the latter having been precipitated by the oxidation of an aqueous solution of ferrous sulfate by hydrogen peroxide and washed with distilled water. The second sample of hematite was immersed in hot water at 80" for 3 days and then was dried a t 110" for 8 hr (Fe203(II)). Surface Area Measurement. The surface areas of samples were determined by applying the BET theory to the nitrogen adsorption data obtained at liquid nitrogen temperatuze, assuming the molecular area of nitrogen to be 16.2 A2. Water Vapor Adsorption Isotherm. The adsorption isotherm of water on metal oxides was determined The Journal of Physical Chemistry
T. MORIMOTO, M. NAGAO, AND F. TOKUDA
a: 0 0
5.0
-
4.0
-
3.0
-
\ cn
c C 0)
s L 48
g
2.0 -
1.0 -
0
200
400
600
000
IO00
Temperature , "C Figure 1. Water content of oxide surfaces a t various temperatures: 0, TiOz (rutile); 0,a-FezOa(I); @, a-FezOs(II).
volumetrically by using an adsorption apparatus which was equipped with an oil manometer to measure sensitively the equilibrium pressure. The sample taken for every run was about 10 m2. Prior to the adsorption measurement, the sample was treated a t a given temperature (250, 600, and 900" for Ti02and 250, 600, and 800" for cr-Fez 0,) under a reduced pressure of torr for 4 hr. The adsorption measurement was made at 18" for Ti02 and at 25" for cr-Fez03, and the adsorption equilibrium was usually established within 30 min for every addition of water vapor to the adsorption system. The desorption measurement carried out as a (1) T. Morimoto, M. Nagao, and F. Tokuda, Bull. Chem. SOC.Jap., 40, 2723 (1967).
(2) T. Morimoto and M. Nagao, Kolloid-Z. 2. Polym., 224, 62 (1968). (3) Unpublished data. (4) G. J. Young and T. P. Bursh, J. Colloid Sci., 15, 361 (1960). (5) T. Morimoto, K. Shiomi, and H. Tanalra, Bull. Chem. SO& Jap., 37, 392 (1964). (6) T. Morimoto, M. Nagao, and M. Hirata, Kolloid-Z. 2. Polym., 225, 29 (1968). (7) A. V. Kiselev and V. I. Lygin, Kolloid. Zhur., 21, 561 (1959). (8) J. H. Anderson, Jr., and K. A. Wickersheim, Proc. Int. Congr. Phys. Chem. Solid Surfaces, North Holland, Amsterdam, 252 (1964). (9) G. J. Young, J . Colloid Sci., 13, 67 (1958). (10) D. J. C. Yates, J. Phys. Chem., 65, 746 (1961). (11) K. E. Lewis and G. D. Parfitt, Trans. Faraday SOC.,62, 204 (1966). (12) G. Blyholder and E. A. Richardson, J . Phys. Chem., 66, 2697 (1962). (13) G. Blyholder and E. A. Richardson, ibid., 68, 3882 (1964). (14) C. M. Hollabaugh and J. J. Chessick, ibid., 65, 109 (1961).
245
CHEMISORBED AND PHYSISORBED WATERON METALOXIDES
Pressure , cmHg Figure 2. Adsorption isotherms of water on TiOg (rutile) a t 18', pretreated a t 250°, a; 600', b; and 900°, 0,first adsorption; 0, second adsorption.
preliminary test for each adsorbent showed no hysteresis, indicating that the samples examined were nonporous. After the first adsorption isotherm of water had been obtained, the sample was placed) under a reduced pressure of 10-6 torr at room temperature for 4 hr, and then the second adsorption was carried ouB a t the same temperature as before. I n the case of rutile it has been reported that partial reduction of the sample occurs when it is treated in vacuo a t an elevated temperature.l4,'6 In the present case, the color of the sample changed slightly to light gray after treatment in vacuo a t temperatures higher than 600", but we utilized the sample in this state for the adsorption measurement. The BET plots'6 of the water adsorption data on every sample showed a linear relation in the range of relative pressure from 0.05 to 0.35, and the monolayer capacity of water was calculated from the linear part of the plot. Water Content Measurement. The surfaoe hydroxyl groups can be removed by condensation dehydration upon heat treatment of the sample in vacuo, but the removal temperature differs depending on the nature of the metal oxides.' By employing the apparatus for the measurement of water adsorption isotherms, the remaining water content of the samples treated a t a given temperature was measured by the successive ignition-loss method as described previously.6 When the powdered samples were treated a t higher temperature in DUCUO, not only water vapor but also other volatile gases were found to be expelled. Accordingly, during the measurement of water content, all the gases evolved at, a given temperature were trapped as a first step in a bulb kept a t liquid nitrogen tem-
perature and were determined volumetrically a,fter reevaporation a t room temperature. Then most of the reevaporated gases could be trapped again in the bulb a t -72", where the remaining gases were determined in the same way as described above. When only water vapor was present in the measuring system, the vapor was completely retrapped by this procedure, the pressure of the remaining gas being zero. When the sample was treated a t 300", the amount of these remaining gases was negligibly small and increased with an increase in the temperature of treatment of the samples. If the doetected values are expressed in molecules per 100 A2, they are, for example, 0.54 a t 900" for TiOz, and 0.76 and 0.29 a t 800" for Fe203(I) and Fe2o3(II), respectively. The water content values illustrated in Figure 1 of this paper are those free from these contaminations.
Results and Discussion The water content abtained at various temperatures is given in Figure 1; the value is expressed in OH groups/100 A2. This value means the number of hydroxyl groups remaining on the surface a t the temperature indicated. The data show that, with rising temperature of treatment, the water content decreases linearly up to about 400" and decreases exponentially a t higher temperature. At 800" the remaining number of hydroxyl groups is negligibly small in every sample. The water content of the two kinds of a-FeaOs is similar over the whole range of treatment temperature, in spite of their differences in origin and history. The (16) J. Gebhardt and K. Herrington, J . Phys. Chem., 62, 120 (1958). (le) S. Brunauer, D. H. Emmett, and E. Teller, J . Amer. Chem. SOC., 60, 309 (1938).
Volume 7 3 , Number 1 January 1969
246
T. MORIMOTO, M. NAGAO, AND F, TORUDA
Pressure , cmHg Figure 3. Adsorption isotherms of water on a-I?eaOa(I) a t 25O, pretreated a t 250°, a; 600°,b; and 800°,c: 0,5rst adsorption; e, second adsorption.
Pressure , cmHg Figure 4. Adsorption isotherms of water on or-FezOa(I1) a t 25", pretreated a t 250°, a: 000°, b; and 800°, c: 0,first adsorption; e, second adsorption.
water content of TiOz is considerably smaller than that of a-FezOa over the whole range of temperature, which may reasonably result from the fact that the removal of surface hydroxyl groups starts at different temperatures depending on the nature of the substrate; they can be removed from TiOz at a lower temperature than from a-FezOa. l7 I n Figure 2, the water adsorption isotherms obtained at 18" on Ti02 are shown, the pretreatment temperatures of which were 250, 600, and 900'. The adsorption isotherms of a-FezOa pretreated at 250, 600, and 800" are given in Figures 3 and 4. The adsorption The Journal of Physical Chemistry
curves of both substances show a similar tendency that represents multimolecular adsorption. I n all cases the second adsorption isotherm is lower than the first one by a definite amount specified by the condition of treatment, the difference being due to the irreversible adsorption of water. is the Calculated data are summarized in Table I. specific surface area of the sample obtained by nitrogen adsorption, The data of f,"*show that the heat treat(17) T . Morimoto, M. Nagao, and F. Tokuda, BUZZ.Chem. Xoc. Jap., 41, 1633 (1968).
247
CHEMISORBED AND PHYSISORBED WATERON METALOXIDES Table I : The Relation between the Amounts of Water Chemisorbed and Physisorbed on TiOz (Rutile) and a-FezOa (Hematite)
Sample
Treatrnent temp,
Adsorption temp,
SNI,
OC
O C
mz/g
VP 9
vm,
rnl (STP)/ ma
H20 molecules/
v cI
Vhg
OH groups/
OH groups/
100 A2
100 A2
100 A 2
vc
+
(vo Ph,
OH groups/ 100 A 2
V d -k Vh),
Ha0 molecules/ OH groups
TiOz
250 600 900
18 18 18
9.94 9.38 4.60
0 213 0.233 0.231
4.30 4.14 4.51
5.91 6.35 6.76
1.65 0.10 0.01
7.56 6.45 6.77
0.57 0.64 0.67
a-FezOs(I)
250 600 800
25 25 25
14.5 14.0 6.41
0.242 0.253 0.237
4.71 4.25 4.45
5.70 7.59 6.73
4.53 0.17 0.06
10.23 7.76 6.79
0 46 0.55 0.66
a-Fe203 (11)
250 600 800
25 25 25
21.2 20.0 14.9
0.194 0.254 0.247
3.81 4.04 3.75
4.57 7.91 7.91
3.95 0.22 0.03
8.52 8.13 7.94
0.45 0.50 0.47
I
ment of samples a t higher temperatures leads to the large depression of specific surface areas, which indicates the occurrence of appreciable sintering. V , is the monolayer capacity of water obtained by applying the BET theory to the first adsorption isotherms. This value is found to vary slightly with the treatment temperature of the samples, and the value of the 600" treated sample is the largest of the three in every sample. The V , value calculated from the application of the BET theory should contain the amount of chemisorbed water in addition to that of the first layer of physisorbed water. In order to estimate the aniounts of the two kinds of adsorbed water separately, Jurinak'* determined the first and second adsorption isotherms of water on oc-FezOa, 0-FeOOH, and TiOz (anatase), taking the difference of the two resulting V , values to be the chemisorbed amount of water. Hollabaugh and Chessiclc,14on the other hand, obtained the irreversibly adsorbed amount of water on Ti02 (rutile) by subtracting the value of the second adsorption isotherm from that of the first adsorption isotherm at the relative pressure of 0.2. These authors assumed that the second adsorption isotherm, when measured after degassing the sample at room temperature, contains only physical adsorption of water. Indeed, this is true of a-Fe203and Ti02 (rutile).17 Also in the present investigation the monolayer capacity of physically adsorbed water, V,, was calculated by applying the BET theory to the second adsorption isotherm, and is expressed in H 2 0 molecules/100 as shown in Table I. The value of V , is almost the same on each sample, independent of the temperature of pretreatment. This suggests that the number of sites for physical adsorption of water does not depend on the temperature of pretreatment. If it i s assumed that the area of a water molecule physisorbed on the oxide surfaces is 10.8 A2, the number of water molecules which can possibly exist geometrically on th,e surface can be estimated as 9.2 H,O molecules/ 100 A2. The values obtained experimentally for a-
B2,
Fe203and TjOz are only 41-4801, of this number, which shows the water molecules in the first physically adsorbed layer to be in a loosely packed state. Jurinak'* and Healey, et aZ.,l9 reported that only two-thirds of the hematite surfaces were effective for the physical adsorption of water, a figure larger than the present results. As described above, it seems that there is a definite interval between the first and second adsorption isotherms on each sample, which can be considered the amount of irreversibly adsorbed water, Le., the chemisorbed water. We have regarded the difference of each pair of isotherms at the relative pressure of 0.2 as the amount of newly chemisorbed water (V,) during the adsorption process, as Hollabaugh and Chessick14 did in their work. Since a water molecule is considered to be chemisorbed on the oxide surfaces to form two hydroxyl group:, the value of V , is expressed in OH groups/100 A2. The V , values for the 250" treated samples are smaller than those of the samples treated a t higher temperatures. The amount of the remaining surface hydroxyl groups can be read from Figure 1, and the results are listed in column 8 of Table I. The sum of V , and Vh should be the total number of surface hydroxyl groups at the end of the adsorption process. The total amount of chemisorbed water on TiOz, generally, seems to be different from that on o-Fe20s: the latter is slightly larger than the former. Taking into account the lattice constants, we can estimate the number of hydroxyl groups which are able t o exist crystallographically on the oxide surfaces. In this estimation, we assume that only one hydroxyl group is formed on each surface metal atom. Th,e calculated value amounts to about 8 OH groups/100 A2 for the average of the (100) and (110) planes of Ti02 (rutile), whose cleavage is (18) J. J. Jurinak, J . Colloid Sci., 19, 477 (1964). (19) F. H. Healey, J. J. Chessick, and A. V. Fraioli, 60, 1001 (1956).
J. P h y s . Chem.,
Volume 73, Number 1 Januarg 1969
248
T. MOIEIMOTO, M. NAGAO, AND F. TOKUDA
known to be perfect, and to about 9 OH groups/100 bz for the (001) plane of a-FeZ03.20As a whole, these values are nearly equal to the maximum value of the total chemisorbed water in Table I, which means that the surfaces are completely hydroxylated after the adsorption proccss. If we examine the data in Table I in detail, it is found that the sum of V , and V h , or the total amount of chemisorbed water, decreases slightly with the rising temperature of pretreatment of the sample. In the previous paper," alternate repetition of the adsorption measurement and degassing treatment showed that the reversible chemisorption occurs on dehydrated sites of the oxide surfaces, when the degassing treatment of oxide samples is carried out at temperatures lowcr than 500". The present result implies that the rehydroxylation of the oxide surfaces dehydrated at higher temperatures than GOO" in vacuo is not perfect. It is known that most metal oxides reveal the maximum in the plot of heat of immersion in water against Pretreatment temperature a t a temperature of 400-G00°.4-6 This fact has been explained in terms of the stabilization of oxide structure, which would occur a t higher temperature of treatment; the stabilization results in the retardation of rehydroxylation upon immersion of the oxide surfaces in water. The present results seem to support the above explanation for the phenomcnon-of-immersion anomaly. The change in the total amount of chemiaorbed water by the pretreatment temperature is larger with a-Fe203(I)than with a-FezO&3), as seen in Table I, which indicates that the extent of rehydroxylation may depend upon the origin and/or the history of the aample. The ratio of the first-layer capacity of physical adsorption (V,) to the total number of underlying hyv h ) is nearly equal to 'I :2, as droxyl groups ( V , shown in Table I, except in the case of TiOz pretreated a t extremely high temperature. This suggests that a water molecule is adsorbed on two surface hydroxyl groups in most cases. IGselev and Lygin,' who studied the adsorbed state of water molecules on silica by means of infrared spectroscopy, concluded that a water molecule is adsorbed on two surface hydroxyl groups through the formation of a hydrogen bond between the oxygen atom of the water molecule and two hydrogen atoms of
+
Xhe Journal of Physical Chemistry
the neighboring hydroxyl groups. The quantitative demonstration for their conclusion has not been given. Also, on the surfaces of Ti02and Fe203it has been found by infrared spectroscopic investigati~nl~-'~ that the physical adsorption of water occurs through hydrogen bonding to the surface hydroxyl groups, while the quantitative relation between the two kinds of surface species has not become clear. The fact that the cross-sectional area of the mater molecule in the Fonolayer on TiOz (rutile) was extremely large, 23.5 A2/Hz0 molecule, made Hollabaugh and C h e s s i c l ~infer ~ ~ that the physical adsorption of water would occur through the localization of a molecule on two surface hydroxyl groups. In t)he case of TiOz pretreated at an extremely high temperature, the ratio V , : V , Vh mas found to be nearly 1 : 2 as in other cases, despite the possibility of partial reduction and the resultant nonstoichiometric structure of the surface. Thus we can, generally, conclude that the adsorption mechanism of water on the surfaces of Ti02 and aFe203is
+
b
C
d
At the first stage of adsorption, a water molecule will be physically adsorbed on an activated site (a) to form an adsorption complex (b), which will subsequently transfer to surface hydroxyl groups (c). Next, another water molecule will come to be adsorbed through hydrogen bonding on the two neighboring hydroxyl groups as shown in d. (20) E. S. Dana and W. E. Ford, "A Textbook of Mineralogy," John Wiley & Sons, Inc., New York, N. Y., 1960,pp 483,498.
