D. J. C. YATES
'746
concentrations depends primarily on the charge type of the electrolytes and somewhat on their individual nature. For a rough idea of the quantities involved we can disregard the individual variations and use the theoretical limiting slope in dilute solutions which has been estimated by Redlich to be 2.8 for 1 : l electr01ytes.l~ For higher rharge types this must be multiplied by the valence i'act or11
where p is the number of different kinds of ions in an electrolyte (2 for those in Table I), Y is the number and-z th_e charge of the jth ion. Thus calculated, 1 ' - Vofor a 0.4 M solution is 1.8, 9.4, 14.4, 26.5 and 104 ml./mole, respectively, for electrolytes of the charge types 1: 1, 2 : 1, 2: 2, 3 : 1 and 3:2. The possible error from this source is, therefore, small for 1: 1 electrolytes but increases steeply for those of higher charges. Other possible errors may arise from formation of complexes or ion pairs which results in the formation of species of lower charge and may decrease the electrostriction. Thus the possible systematic errors all tend to give lower electrostriction values from the present method. Consistent with this all the results from the present methsd in Table I are lower than the estimates from V , data excepting that of S a O H (which is slightly higher). The agreements are expected to be best for electrolytes of low charge type and small intrinsic volume. For such ions, the agreement, considering the small systematic (13) 0 . Redlich. J . Phys. Chem., 44, 619 (1940).
Vol. 65
errors still present, can be considered to be within the uncertainty of the two methods. As the intrinsic volume and ionic charges become higher, the deviations increase more or less systematically, as expected, but still remain well within the bound5 of possible error. We can conclude, therefore, that the two methods, all things considered, are in good agreement for all the electrolytes in T_able I. The electrostriction values obtained from 1'0 data are considered more accurate. In detail, the cwrious low value for-Li+ and the large value of OH- deduced from Vo data are confirmed. The postulated similarity in behavior of monatomic and polyatomic ions such as Ba++ and Sod is also amply supported. For comparison with previous analyses the estimates of Couture and LaidleP are 52 ml./mole for all 1 : l electrolytes and 78 ml./mole for all 2 : 1 electrolytes, in sharp disagreement with the compression method as regards both magnitude and trend, particularly for 1: I electrolytes. Hepler's analysis15has been shown to be unsatisfactory for the Vo data of polyvalent ions (see paper I). For monovalent ions it gives E values oi the correct order of magnitude, but, as shown in Table I, gives consistently larger deviations from the compression values than our estimatss. We therefor, feel that our treatment of the Vodata is more consistent with the experimental data availahle. Acknowledgment.-I am grateful to Dr. H. 1, Friedman. I.B.M. Research Center, Yorktown Heights, New York, for suggesting the present approach. (1%) A. M. Couture and K J. Laidler, Can J . Chem , 34, 1209 (1956). (15) L. G. Hepler, J . Phus Chem , 61, 1426 (1957)
IKFRARED STUDIES OF THE SURFACE HYDROXYL GROUPS O S TITASIUlll DIOXIDE, AXD OF THE CHEAIISORPTION OF CARBON MOSOXIDE AND CARBON DIOXIDE BY D. J. C. YATES' School of Mines, Columbia University, New York 27, IT. Y. Received August 20, 1060
An infrared investigation has been made of the surface properties of anatase and rutile. Bfter evacuation a t 150" residual ater was detected, while after evacuation a t 350' only residual OH groups remained on the surface. These OH group^ show marked differences from those found on alumina and silica. In one case, a rutile sample, ammonia had been used in its preparation, and NH containing species remained on the surface even after evacuaticm at 350'. All the OH, and the XH, groups could be exchanged readily with deuterium at 350'. The chemisorption of carbon monoxide and carbon diovide has also been studied a t room temperature. Carbon monoxide is very weakly chemisorbed. In every case all of the adsorbed gas could be removed by evacuation for 30 seconds, and at lower initial pressures, in 5 seconds. The adsorbed carbon monoxide probably is held by its carbon atom to one of the surface oxygen atoms. Despite some slight reduction of one of the powders, shown by a color change after the deuterium treatment, no carbon monoxide molecules chemisorbed to metal sites were detected in any experiment. Carbon dioxide was much more strongly chemisorbed, probably as GO,species. Some evidence also has been found that carbonate-like species are present. Quite large variations in the surface properties of all the titanias used were found in all the above experiments. These variations most probably are due to differences in method of preparation, rather than to differences in bulk properties. 11
Introduction Many accurate and detailed calorimetric measTitanium dioxide has been used as an adsorbent urements of heats of adsorption Of rare gases on in a large number of investigations of physical titania have been reported by 4 s t 0 n , ~M ~ r r i s o n , ~ (2) W. D. Harkins, "The Physlcal Chemistry of Surface Films," adsorption, following its extensive use by Harkins.2 (1) Now at the National Physical Laboratory, Teddington, -Middles e x , England.
Reinhold Publ. Corp., New York. N. Y., 1952. (3) W. A. Steele and J. G. Aston, J . -4m. Chem. (1957), and earlier papers.
Soc
, 79, 2393
May, 1961
747
h F R r l K E D STUDIES OF THE SCIlF.iCE OF ‘ h ’ A S I . 4
100 80
60
40
a.
20
I
0-1
3800
I
I
I
I
I
I
I
2600 2200 1800 1400 Wave no., cm. - l . Fig. 1.-Spectra obtained on evacuation: anatase (a) evacuated a t 150’ ( l ) ,350’ (2), and exchanged with deuterium a t 350’ (3); rutile (b) with spectra 1, 2 and 3 obtained under the same conditions as for the anatase. 3400
3000
Pace5 and their co-workers. It seems that, one of t’he reasons for the popularity of this oxide as an adsorbent for physical adsorption has been the idea t>hatit, is chemically stable, and has a reproducible surface. Severtheless, the results of Reyereon and Honig6 studying nitrogen dioxide adsorption and the observations of color changes on evacuation at elevated t,empe:ratures,6-8 show that titania is quite reactive. Spectroscopic dataQ obtained on single crystals of rutile demonstrate that the crystal can be made oxygen deficient by hydrogen treatment. Recent infrared studies have shown that silicalo and aluminall retain hydroxyl groups on their surface even after long evacuation a t elevated temperatures. It is of int,erest to determine whether anat,ase and rut’ile i-etain similar groups on evacuation. These groups have been found, and have also been studied by replacing t,hem with OD groups ( 4 ) L. E. Drain and J. A. Morrison, Trans. Faraday SOC.,49, 654 (1953), and earlier papers. (5) E. L. Pace, w. T. Berg and -4. R. Siebert, J . Am. Chem. SOC..78, 1531 (19561, and earlier pa.pers. 16) L. €1. Reyerson and J. M. Honig, ibid., 76, 3917 (1953). (7) Y. L. Sandlrr, J . P h y s . Chem., 68, 54 (1954). ( 8 ) (a) J. Gebhardt and K. Herrington, ibid., 69, 120 (1958); (b) A. W. Caanderna and J. M. Honig, ibid., 63, 120 (1959). (9) D. Cronemeyer, Phys. Rev., 87, 876 (1952). (10) R. S. McDonald, J . Phgs. Chem., 68, 1168 (1958). (11) J. B. Peri, paper presented a t the Second International Congress on Catalysis, Paris, July 1960.
using deuterium. I n addition, it has been found that carbon monoxide and carbon dioxide are chemisorbed at room temperature. The carbon monoxide is extremely weakly chemisorbed; all t,he adsorbed gas can be removed by evacuat,ion for a few seconds. The carbon dioxide is much more strongly held and can only be removed slowly. TABLE I PROPERTIES OF TITANIA POWDERS Codeno. Crystal form Drying temp., O C . Surface area, m,p,g.
MP-1579 .4natase 110
(BET) 290 (Evacuated at drying temp.) 89.5 % lo,o
ignition,
Chlorides, % Sulfur, % Other impurities, total %
.... 0.41 0.24
AMP-1208 IMP-1608-1 Rutile -4natase 120 120
MP-1608-4 Rutile 225
185
100
89.6
.... .... 0.23
85 94.4 5.1 0.08
....
0.20
94.1 4.1
.... ....
0.34
The powders were pressed into self-supporting discs10 in a 1” diameter die, using pressures of 2000 to 4000 lb./in.a. As 0.10 g. of powder was used, the optical “thickness” of the samples is 20 mg./cm.*.
Experimental Materials and Sample Preparation.-All the titanium dioxide used was the gift of the National Lead Company, Titanium Division, South Amboy, N. J. The majority of the work has been done on one anatase preparation (Code No. MP-1579) and with one rutile (MP-1208). Some meas-
D. J. C. YATES
748
Vol. 65
100
80
@ :I
60
9
-3
40
V
20
0
2200
2100
2200 2100 2000 Wave no.. em.-'. Fig. 2.-Carbon monoxide adsorbed on anatase (a and b) and rutile (c, d and e). Pressures after adsorption in cm.: (a) 0.10 0.27,0.64, 2.04,5.88; (b) 0.20,0.59; (c) 0.10,0.29,0.71, 2.14, 3.88, 11.84,35.10; (d) last dose on spectrum c after evacuadion for 5 seconds; (e) 0.71, 1.81, 4.05. urements have also been made on anatase (MP-1608-1) evacuated at 150" until values in the region of 10-6 mm. and rutile (MP-1608-4). Unless otherwise stated, all re- were obtained. The spectrum of the sample then was results with anatase and rutile in this paper were obtained recorded from 4000 to 1220 cm.-l. The sample now was with samples prepared from the MP-1519 and MP-1208 evacuated, in a similar way, at 350°, and its spectrum repowders. Chemical and physlcal properties of these pow- corded over the same region. In some cases the OH groups were exchanged with deuders, supplied by the manufacturer, are given in Table I . The anatase MP-1608-1 is identical with that used by terium, as follows. After recording the second spectra, Czanderna and Honigsb with the code number MP-980-1. the sample was heated while pumping to 350". Dried No details of the methods of preparation are known other deuterium at about 10 cm. pressure then was admitted, than all the samples were made from titanium tetrachloride left over the sample for times ranging up to 25 minutes, except MP-1579, which was made from titanium sulfate. and evacuated for 10 minutes at 350". The spectrum Carbon monoxide (C.P. grade) and oxygen were supplied was recorded again. It was found that all of the samples by the Matheson Company. Deuterium was supplied by could be completely deuterated with a total time of treatthe Stuart Oxygen Co., and was better than 99.5% pure. ment of 25 minutes, which is considerably less than needed These ases were dried by paFsage through traps cooled to for silica and alumina at the same temperature. 77°K. %efore use. Commercial carbon dioxide was puriTo these surfaces containing either OH or OD groups, the carbon monoxide or carbon dioxide was added in a series of fied by distillation. Apparatus.-A small (volume 48 em.*) externally heated doses, of known initial pressures. The spectrum then was glass cell (path length 3.4 cm.) with magnesium oxide recorded, and the pressure after adsorption noted. Volumes windows was used, described elsewhere.12 An all-glass of gas adsorbed could not be calculated readily as the vacuum system was used, with a mercury diffusion pump samples used were so small (0.03 9.) as to adsorb quite small and rotary backing pump, and i t gave a kinetic vacuum of quantities of gas. 5 X 1 0 4 mm. A Perkin-Elmer double beam model 21 Results spectrometer was used with a fluorite prism. I n examining Residual Water and Hydroxyl Gmnps.-Figure 1 highly scattering materials, it is an advantage if the sample is placed at a focus. An acceesible focus was provided by shows the spectra recorded on evacuating the an external mirror system placed in the sample beam of the anatase (a) and rutile (b) samples. Evacuation model 21. Wire gauzes were placed in the reference beam of the anatase a t 150° (Fig. l a , 1) shows strong to balance out the absorption due to the mirrors, cell and bands at 3675 and 1605 cm.-I, and a repion of zero sample, and were changed at 2800, 2000 and 2300 em.-'. T o maintain the energy arriving on the detector at normal transmission between 3350 and 3100 em.-'. levels, the spectrometer slits were widened. At 2200 em.-', After evacuation a t 350' (Fig. l a , 2), the 3675 em.-' the following slit widths (in mm.) were used: for the anatase band splits into two bands a t 3715 and 3675 em.-', 0.07, for rutile 0.20. These give corresponding spectral and the 1605 cm.-1 band and the broad band slit widths of 2.4 and 5.3 em.-'. Procedure.-After inserting in the cell, the samples were centered at about 3250 em.-' are removed. I n addition, a sharp band appeared a t 1360 cm.-'. On deuteration (Fig. la, 3), the OH bands are re(121 C. E. O'Neill and D.J. C. Yates, to be published.
May, 1961
INFRARED STUDIESOF THE SURFACE OF TITANIA
749
placed by two sharp OD bands at 2740 and 2705 ern.-'. The 270,5 em.-* band has a shoulder at 2685 crn.-l. This may be due to a small proportion of a third t,ype of OD (and, of course, OH) 80 groups. A corresponding shoulder could not be detected on the peak a t 3675 cm.-I, as this OH peak is quite broa,d. 60 Evacuation of the rutile at 150' (Fig. lb, 1) gives a spectrum with bands at 3680, 1610 and 1420 40 cm.-l, and a broad blackout between 3450 and 3150 cm.-'. Evacuation a t 350' (Fig. lb, 2) decreases the intensity of the 3680 cm.-' band, but it did not split into two components. The 1610 20 and 1420 cm.-l bands were removed, while most of .sd the broad band (due to hydrogen bonding) is 0 removed, leaving a band a t 3320 ern.-'. The band at 3680 cm.-l is due to OH stretching vibrations, while that a t 1610 em.-' is caused by the OH bending vibration in adsorbed water. k? No definite assignment can be given to the 1420 cm.-' band, but it may be a result of some form of adsorbed carbonate complex. This could have been formed from carbon dioxide in the air while the rutile W : ~ Sin storage. Adsorption of Carbon Monoxide.-Figure 2a shows the two bands found when carbon monoxide is added to deuterium treated anatase. At the lowest pressure utsed (0.1 cm.) the main band is a t 2203 cm.-', with a shoulder at about 2190 crn.-'. The lower frequency band increases in intensity v much more than the other a t higher carbon mon01 ' 1 ' ' ' I 1 oxide pressures, t,he two bands being of equal in1700 1600 1500 1400 1300 tensity a t 0.64 cm. At the highest pressure used Wave no., cm.-l. (5.88 cm.) the 2188 ern.-' band dominates the Fig. 3.-Carbon dioxide adsorbed on anatase (a): presspectrum. The deuterium treatment mentioned sures after adsorption in cm.: 0.025, 0.14, 0.52. Evacuaabove was carried out a t 350', and the sample was tion (b) showing (lower curve) adsorbed gas at a pressure 0.52 cm. and evacuations for total times of 1.0, 8.5 and cooled down in vacuo before adding the carbon of 20.0 minutes. monoxide. As bulk titanium dioxide easily can be r e d u ~ e d the , ~ possibility was considered that the gate the reversibility of the adsorption, the sample chemisorption of carbon monoxide was affected by was then evacuated for a further 60 seconds, to the deuteration. The same sample was re-evacu- make certain of removing all the last traces of the ated a t 350", cooled in vacuo and 15 cm. of oxygen carbon monoxide. Carbon monoxide then was readded a t room temperature. The oxygen then was adsorbed without any other pretreatment. At a evacuated at the same temperature for five minutes pressure of 0.57 cm., a band of 76% transmission and then carbon monoxide re-adsorbed. The at 2190 em.-' was recorded, and a t a pressure of resulting spectra are shown in Fig. 2b. At a pres- 2.20 cm., a band of 57% transmission a t 2188 sure of 0.59 cm., two peaks were found with nearly cm.-' was found. Comparison with Fig. 2c shows the same position and intensity as those in Fig. 2a a band at 2188 em.-' with 57% transmission a t a at a pressurtl of 0.64 cm. pressure of 2.14 cm. Evidently the adsorption of When carbon monoxide is added to deuterated carbon monoxide is completely reversible, within rutile, only one band is observed (Fig. 2c). A t experimental error. After the second dose (pressure 2.20 em.), the sample was evacuated for five 0.10 cm. pressure, it has a frequency of 2195 cm.-', and shifts t o lower frequencies a t increasing pres- seconds. All the adsorbed carbon monoxide was sures. At the highest pressure used (dose 7, 35.10 removed. Adsorption of Carbon Dioxide.-Figure 3a shows cm.), the band is at 2182 cm.-'. This particular rutile. went very gray when deuterated. The that very strong bands a t 1580 and 1320 cm.-' are sample used, to obtain the spectra shown in Fig. formed on adding carbon dioxide to anatase. A 2c had been fully deuterated (350' with 10 em. weak shoulder can be detected on the low frequency Dz for 20 minutes), and then 10 cm. of oxygen was side of the former band, at about 1500 em.-'. added ai, 350O. The oxygen was evacuated a t Further growth of both bands took place on in3jOo, and the sample cooled in vacuo,before the creasing the carbon dioxide pressure; in particular carbon monoxide was adsorbed. Evacuation of the 1580 cm.-' band had a peak transmission of dose 7 for five seconds at room temperature re- 3% at the highest pressure used. On evacuation for times of up to 20 minutes moved nearly all of the adsorbed carbon monoxide as shown by Fig. 2d; a further five seconds pumping (Fig. 3b) both bands were considerably reduced removed the rest of the adsorbed gas. To investi- in intensity, but the 1580 cm.-' band remained
4
750
D. J. c. YATES
Vol. 65
If hydrogen bonding is present, the v2 vibration of water increases14 in frequency by 10-50 em.-'. Detailed studies of hydrogen bonding in a nitrogen matrixl6 show that monomeric water has its v 2 80 vibration a t 1600 ern.-', dimeric water at 1620 cm.-l and polymeric water a t 1633 em.-'. The value for residual water (1605 cm.-l) on this 60 surface is very close to that of monomeric water. This indicates that the majority of the water molecules left on the surface after evacuation at 40 150' are isolated from each other, and not held as aggregates. d The absence of the 1605 em.-' band after wacua20 tion a t 350' indicates that it is unlikely that any E water as such is present. ,4ny water held to a 100 surface after this treatment must be adsorbed by + forces considerably stronger than those usually k? operative in hydrogen bonding, and the bending 80 motion of this water would be strongly hindered by the surface forces. It is possible that this would cause v 2 to decrease in frequency. If such a band 60 is a t frequencies lower than 1220 em.-', it would not have been observed here, as 1220 ern.-' m s the lowest frequency scanned. 40 The band a t 1360 cm.-l (2, Fig. la) might be the OH bending vibration of strongly held water, as it only appeared after the high temperature treat20 ment. This seems very unlikely, however, as the 1360 em.-' band is not affected by the deuterium 1700 1600 1500 1400 1300 treatment which removed all the OH groups. S o Wave no., cm.-'. other band at 1360 cm.-l was found under similar Fig. 4.-Carbon dioxide adsorbed on rutile (a): pres- conditions, and seems peculiar to sample MP 1579. sures after adsorption in cm.: 0.04, 0.11 and 0.71. Evacua- S o definite assignment ran be given to this band tion (b) showing (lower curve) adsorbed gas a t a pressure of 0.71 cm. and evacuations for total times of 0.5, 1.5 and on the basis of these experiments, and further experimental work is needed to settle this point. 6.5 minutes. The presence of two OH stretching vibrations the most intense, as on adsorption. The shoulder after evacuation a t 350' is of interest. Of the a t 1500 em.-' was removed by the first evacuation other oxides which have been examined, silicalo (1 minute) so that the species producing this band shows an asymmetric band at 3740 cm.-', with a probably is held quite weakly. broad band due to hydrogen bonding extending Very considerable differences are found between to lower frequencies. When the hydrogen bondrutile (Fig. 4) and anatase (Fig. 3) on adsorbing ing system is removed by evacuation at higher carbon dioxide. On both spectra, a strong band temperatures, lo only a single, very sharp band at 1330 to 1320 em.-' occurs, but the very strong remains a t 3749 em.-'. On a y-alumina aerogel," 1580 ern.-' band found on anatase is extremely three bands at about 3790, 3730 and 3700 cm.-' weak on rutile. It forms a shoulder a t 1580 cm.-l have been reported. With a partly amorphous16 on the high frequency side of the band at 1485 y-alumina (Alon C), spectra \ - e have obtained cm.-' (Fig. 4a). Furthermore, the 1325 em.-' show a broad unresolved OH peak at about 3660 band was the weaker of the two strong bands on the cm.-l. The two OH peaks on the anatase sample anatase, but is the strongest on the rutile. used here may be due to the presence of two relaThe 1580 crn.-' shoulder on Fig. 4a is produced tively uniform regions of the surface, which could by a weakly held species, in a similar fashion to the plausibly be associated with two crystal faces. 1500 cm.-l shoulder on the anatase. Figure 4b Support for this suggestion is provided by the shows that the 1580 em.-' band is removed by parallel between the spectra obtained on adsorbing evacuation for 30 seconds. Both of the strong carbon monoxide. The latter show two absorption bands at 1485 and 1325 cm.-l are removed to the peaks on anatase, and one on rutile (1;ig. 2 ) . The same extent, by further evacuation. rutile has only one OH peak in the stretching region. Rutile.-Although there is only one OH stretchDiscussion Residual Water and Hydroxyl Groups. Ana- ing peak present after evacuation at 350" (2, Fig tase.-The bands in the 3700 cm.-l region are l b ) , the nature of the residual n-nter i. iimilar t o caused by OH stretching vibrations, these vibra(13) G. Heraberg, "Infrared and Raman Spectra," L3 T.m Tinstrand tions being present in water and in OH groups. Co.. Ino., Nea York, N. Y . , 1945. (14) G . C. Pimentel and A. L. hlcClellan, "The Hldrogen Bond The band a t 1605 cm.-l is due to the OH bending H. Freeman and Co., San Francisco, 1960. (vz) vibration of residual adsorbed water. In W.(15) hl. van Thiel, E. D. Becker and G. C. Pirnentel J . Chem I'hys , water vapor, where the molecules are freely rotat- 27, 486 (1957) (16) J).itn E i i p I i I i P c I I,\ tli? innnilf ing, this vibration has a frequency13 of 1595 cm.-l. 100
:p g
"
May, 1961
INFRARED
STUDIES OF
THE SURFACE OF
TITAYIA
751
that found with the anatase. The v2 vibration after evacuation a t 150° (1, Fig. lb) is at 1610 em. -l, which might indicate that more dimeric16 species are present than on the anatase. The broad band a t 3320 cm.-l (2, Fig. lb) left after evacuation a t 350°, is a t a frequency close to that of NH stretrhing vibrations. The only bands that occur in this region are those caused by hydrogen stretching vibrations. Acetylenic CH groups absorb" near 3300 cm.-', but such groups seem unlikely to be present on rutile. The possibility of some form of contaminant containing 1;H groups was confirmed by the suppliers of the material, who commented on enquiry being made that ammonia had been used in the preparation of rutile MP-1208. On deuteration (3, Fig. lb), the 3680 and 3320 em.-' bands are removed and OD and ND bands appear at 2725 and 2475 cm.-l. The rutile (MP-1208) samples all turned grey after the deuterium treatment. The color change is due to a slighi reduction of the samples. This mas confirmed by adding oxygen to the sample (at 350') aft ei the deuterium had been evacuated. The samples immediately returned to their original white color. KO color vhunge took place when the oxygen was evacuated at 3.jOo. If desired all the MP-1208 samples rould he turned grey or white at wll by adding alternately deuterium (or hydrogen) and then oxygen at 330'. This effect was not observed with the other powders used, and it seems veiy probable that the color changes observed hcre and elsewhere6-* are specific to the particular titania samples used and are not necessarily dependent on the presence of burface contamination.3a More work with a range of titania powders examined under comparable conditions is needed to settle this point. The anatase l\I1>-1608-1 and rutile SIP-1608-4 were evacuated at 150 and 350'. For both samples, rery broad OH bands indicating considerable hydrogen bonding TI ere present after the 150' treatment. Little change took place on evacuating at 3.50'. These broad bands indicate that the residual water on these hvo samples is held in a radically different fashion from the samples shown in Fig. 1. The difference may result from differences in the method of preparation of the powders, but this is not certain. Wade and Hackerman have recently pointed out,'* with regard to heats of immersion, that very little can be inferred from the bulk structural and thermodynamic properties of solids. The same comment applies to other surface properties. Certainly much more work is needed before a given OH spectrum after evacuation at 350" can be regarded as characteristic of rutile and another, possibly different, OH spectrum characteristic of anatase. With surh wide variations in the OH spectra, it cannot be assumed that one OH peak is assoriated with rutile and two with anata1e. The surface characteristics of titanium dioxide sleeni sensitive to method of preparation. This does not seem to occur with silica,
where most samples have an OH peak at 3750 cm.-'. with a shoulder extending to lower frequencies. Chemisorption of Carbon Monoxide. Anatase. --When carbon monoxide is oxidized over nickel oxide, Eischens and Pliskin19 observed a band at 2190 cm.-l which they considered to be an oxidation intermediate. This band was assigned to a carbon monoxide molecule held by its carbon end to an oxygen atom of nickel oxide. This carbon monoxide and oxygen complex is expected to be linear and bears some resemblance in configuration to carbon dioxide. This hand has also been observed both on nickel and nickel oxide12 under conditions where no carbon dioxide was being formed from carbon monoxide. This species can sometimes be an oxidation intermediate, but its presence does not necessarily indicate that carbon monoxide is undergoing oxidation. The same configuration which has been found with nickel oxide probably is present here, the oxygen atom to which the carbon monoxide is attached being part of the titanium dioxide. Similar results (Fig 2%) are found on adding carbon monoxide to deuterium treated anatase, and on adding it to oxygen treated anatase (Fig. 2b). It is concluded that the sites active in carbon monoxide chemisorption are largely unaffected by reduction with deuterium at 350'. KO bands appeared in any experiment at frequencies lower than 2100 cm.-'. Bands between 2100 and 1800 cm.-l are characteristic of carbon monoxide held to metal sites. In view of the strong interaction between many polar molecules and the OH groups on silica (reviewed elsewhere2O-*l), the OD region also was scanned while the carbon monoxide was cheniisorbed. KO weakening of either of the sharp OD bands a t 2740 and 2703 crn.-l was observed, neither were they shifted to lower frequencies. Both these effects are characteristic of the formation of hydrogen bonds between the surface OH (and OD) groups on silica and adsorbed molecules. This clearly indicates that the sites for the cheniisorption of carbon monoxide on this anatase are not the residual OD (or OH) groups. The carbon monoxide is weakly chemisorbed ; all of the adsorbed gas giving the spectrum showii in Fig. 2a was removed by evacuation at I oom temperature for 30 seconds. Rutile.-The spectra shown in Fig. 2c were obtained, as mentioned earlier, on a deuterated surface which then was treated with oxygen at 350". In view of the marked color changes noted with this rutile, the chemisorption of carbon monoxide might have been affected by this oxygeii treatment. The same sample was re-reduced with deuterium a t 350' for 30 minutes, and the deuicrium evacuated for 10 minutes at the same teinpei attire. After cooling in 2rucu0, carbon monoxide wa,i added to this greyish sample. Hands at siinilar frequencies to those shown in Fig. 2c were found. At pressures of 0.12, 0.30, 0.70 and 2.1 e m , tho
1 lie I n f i a l e d Sptxtrdi of ( ornple\ ?Ii,lw IIIC? J Bell i n i j , M v t h u r n C ' o I t d , London, 1958 118) 11 H 18 adp and N Hackerinan I Phvs Chem. 64, 1196
(19) R P Eisrlirnq and W i Pllsl,in, 4diancrT ~n Catalysis 9 , 6R2 (1'3 57) (10) a) H P Li-~liensand W. 4 I'liqhin idid 10,1 ( l Y ) i 8 ) ill) 2 RliPppard, S p w t r o c h i m A c t a . 1 4 , 249 (1959) (21) I ) J c I n t p ' , t d i n n r c ~17, ratrli71~l~ 12, 28s (igfioj.
(19fi0)
752
D. J. C. YATES
T‘ol. 65
corresponding peak percentage transmissions were on going from the gas to the liquid (and solid) state. A non-spectroscopic point is that a t room tem92.5, 86.5, 76.5 and 59.0. Inspection of Fig. 2c shows that, a t similar pressures, the bands on the perature the “vapor pressure” of carbon monoxide oxygen treated surface had about 2 to 4y0 less peak is so high (of the order of lo6 atmospheres) that transmission. The reduced surface seems to ad- the relative pressures used here would be extremely sorb a little less carbon monoxide over a pressure small. Under such conditions, only a very small range from 0.1 to 2.0 cm., than does the oxidized amount of physical adsorption would take place. surface. Evacuation for five seconds of the carbon Another indication is that no bands due to adsorbed monoxide on the reduced surface (pressure before molecules have been observed on adding carbon evacuation 2.1 cm.) removed about 95% of the monoxide at room temperature to silica and aluadsorbed gas, and a further evacuation of five mina.12 As physical adsorption is relatively nonseconds removed the rest. As with the anatase, the specific, bands with similar intensities would have sites for the carbon monoxide adsorption on rutile been observed if this process was occurring on silica seem relatively unaffected by deuterium treatment and alumina. a t 350’. No bands were detected on rutile a t Both the occurrence of a band, and its frequency, frequencies lower than 2100 cm.-l. when carbon monoxide is added to titania a t room The carbon monoxide is adsorbed in the same temperature indicate that chemisorption forces are configuration on the rutile as it is on the anatase. involved. However, the adsorption of the carbon The 2725 cm.-l OD band was scanned while monoxide is completely reversible. This does not the carbon monoxide was adsorbed on the rutile. occur with strong chemisorption, and it is concluded No change in frequency or intensity of this OD that carbon monoxide is weakly chemisorbed on band was observed. titanium dioxide. On rutile MP-16084, the addition of carbon No other reference to the adsorption of carbon monoxide to a sample evacuated a t 350“ gave a monoxide on titanium dioxide has been found band a t 2180 cm.-’ (Fig. 2e). Over the pressure except for the remark by Gray, et U L . , ~that ~ carbon range used (up t o 4.0 cm.) all the bands were monoxide shows a small reversible adsorption a t weaker than those found under similar conditions room temperature. No other details were given with MP-1208, but this may be due to the lower except that this adsorption did not change the surface area of MP-1608-4. In common with both magnetic susceptibility of the titanium dioxide. the other samples, the carbon monoxide on MP- From this it was concludedz5 that the carbon 1608-4 was weakly held, evacuation for 30 seconds monoxide was physically adsorbed. removing all the adsorbed gas. Chemisorption of Carbon Dioxide. Anatase.I n view of the current interest in the nature of It is very noticeable that the strong bands found weak chemisorption,zz the desorption of carbon on adsorbing carbon dioxide (Fig. 3) are extremely monoxide was investigated in more detail on rutile. broad, with half widths of some 40 cm.-’ or more. After evacuation of dose 7 (Fig. 2c, pressure 35.10 The resolving power of the prism-slit combination cm.) for five seconds, the weak band shown in in this region is much smaller (in cm.-l) than the Fig. 2d was obtained. Afurther five seconds pump- widths of the bands. Indeed, weak water vapor ing removed all the remaining adsorbed gas. The vibration-rotation bands, due to a slight unbalance ease with which carbon monoxide can be removed by of the optical system, were resolved in the 1950 to isothermal evacuation from titanium dioxide indi- 1450 cm.-l region, superimposed on the peaks due cates that some consideration has to be given to the to the carbon dioxide. The very marked width possibility of physical adsorption being present. of these bands may be because the surface is heteroTwo factors indicate that this is very unlikely geneous for carbon dioxide chemisorption; or, to be the case. The similar absorption frequencies what is almost the same thing, the carbon dioxide of the bands observed here and those ~ b s e r v e d ’ ~ Jmolecules ~ are all held with generally similar orienon nickel and nickel oxide on adding carbon mon- tations with respect to the surface, but quite large oxide indicate that similar surface configurations variations exist in the configurations of individual are likely to be present. In the latter experiments, adsorbed molecules. the carbon monoxide is certainly chemisorbed There are several possible species which might in the usual meaning of the term. Furthermore, be responsible for these bands. If carbonntes are the frequency of the bands found on titania is present, Miller and Wilkins26 have shown (using higher by about 45 to 62 wave numbers (about 2 8 carbonates) that a very strong band in the region to 3%) than the fundamental vibration frequency23 1450-1410 cm.-l would be found. If some analog of gaseous carbon monoxide (2143.2 cm. -I). of bicarbonate (COa-) were formed, no band in the Tn physical adsorption, decreases in frequency of about 0.2 to l.5yohave been found24relative 1450-1410 cm.-’ range would be present, but it to the gas phase. Similar shifts have been re- would be replaced by two strong bands in the 1632ported on change of state; Herzhergl3 shows that 1600 and 1410-1300 cm.-l regions.26 The latter for a range of small molecules decreases of from assignments are tentative” as only a small number 0 to 5Y0 in the fundamental frequencies occur of bicarbonates have been studied. If the carbon dioxide is negatively charged by electron transfer (22) “Chemisorption,” Discussion edited by W. E. Garner, Butteron adsorption, one might expect spectra characterworths, London, 1957. istic of a COZ- group. These groups occur in (23) G. Heraberg, “Spectra of Diatomic Molecules,” D. Van Nostrand, Inc., New York, N. y., 1950. (24) N. Sheppard and D. J. C. Yates, Proc. Roy. Soc. (London), U S E , 69 (1965).
(25) T. J. Gray, C. C. McCain and N. G. Masse, J . Phys. Cham., 63, 472 (1959). ( 2 5 ) F. A. Miller and C. H. Wilkins, Anal. Chem., 24, 1253 (1952).
THEKIXETICS OF THE HYDROLYSIS OF CHLORISE
May, 1961
ionized carboxyl Lc acids," giving strong bands in the regions 1610-1550 ern.-' (symmetrical vibrations) and 1420-1 300 cm.-' (asymmetrical). On the anatase, the GOz- species is predominant as the two strong bands a t 1580 and 1320 cm.-l fall well within the usual range for such species. The 1580 em.-' band is stronger than the 1320 em.-' band, however. The weak shoulder a t about 1500 em.-' is more difficult to assign; it does not fall into either the COZ- or GO3- region, but is only just outside the region normally assigned to carbonates (1450-1410 cm.-l). It is tentatively suggested that this shoulder is caused by a small proportion of carbonate-like species on the surface. Rutile.--The width of the bands (Fig. 4) is very similar to rhat found with anatase, a n d ' t h e significance of this has been discussed in the previous section. The strong band a t 1325 cm.-' and the weak shoulder a t 1580 ern.-' may be due to some type of GOz- species. However, if sufficient of these species are present t o form a strong band due to asymmetric stretching vibrations a t 1325 em.-', it is difficult to account for the weakness of the band a t 1580 em.-', presumably due to the symmetrical stretching vibrations. It seems very unlikely that the extinction coefficients of these two modes of vibration of the COz- species could be drastically affected by the surface on which they are adsorbed. The assignment of the bands on the rutile surface is rather uncertain for this reason. The strlong band at 1485 ern.-' shown in Fig. 4a may be caused by a carbonate-like species. Although outside the range normally assigned to
753
carbonates (1450-1410 cm.-l) this may be because these surface species do not have any strictly comparable analogs in bulk solids. Possibly configurations intermediate between a carboxylate and a carbonate group are present on the surface. More work is needed before such species, if they exist, can be assigned with certainty. What is certain is that carbon dioxide is strongly adsorbed on the samples of anatase and rutile used here, probably both as GO2- and GOa species. Nevertheless, wide variations in intensity of the 1580 and 1485 cm.-' bands occur when anatase and rutile are compared. I n common with the spectra of the residual hydroxyl groups and carbon monoxide, the spectra of the chemisorbed carbon dioxide show clearly that there are quite wide differences in surface properties between the two samples. While this may be due to the different bulk crystal structure of the two oxides, this is unlikely, as very large differences in the spectra of the residual OH groups were found between the anatase and rutile mainly used, and the other samples also studied. The marked differences in surface characteristics shown in all the figures for anatase (MP-1579) and rutile (MP-1208) are likely to be due to variations in their method of manufacture. With titania, a t least, it is evident that a great deal of care must be taken in sample selection before the experimental data obtained can bear any relation to surface properties which may be calculated from bulk values of lattice constants, crystal structure, polarizability and so on.
THE KINETICS OF THE HYDROLYSIS OF CHLORINE. 11. THE HYDROLYSIS I N THE PRESENCE OF ACETATE BY ASSALIFSHITZ AND B. PERLMUTTER-HAYMAN Department of Physical Chemistry, Hebrew University, Jerusalem, Israel Receined August 31, 1060
The hydrolyEis of chlorine in the presence of acetate ions has been investigated. The results can be explained by the assumption that the rlzaction proceeds along three parallel paths: one the same as in pure water, the second involving a reaction between molecular chlorine and acetate ions, and the third involving a reaction between molecular chlorine and hydroxyl ions. I n the presence of acetate and acetic acid the second reaction predominates and its rate constant can be determined ( k = 5.5a x 102 mole-' 1. sec.-l). In the resence of acetate alone, the second and third reactions make comparable contributions. 'The rate constant for the thirfreaction can only be estimated approximately. In order to contnbute to the observed rate, this reaction must be diffusion-controlled.
Introduction We have reinvestigated recently' the kinetics of the hydrolysis of chlorine in pure water, using the continuous flow method. We concluded that the reaction takes place according to
of the hydrolysis of bromine2 in phosphate buffer solution we found that the observed rate was compatible with the mechanism HzO
+ X1 + A -
HOX
+ X- + HA
(3)
where A- represents the buffer anion. The next xn+ H ~ O HOX + X- + H + (1) step in our program of investigating the kinetics of (where X represents the halogen) with no ap- the hydrolysis of the halogens is the study of the
preciable contribution from a reaction involving the hydroxyl ion On the other hand, when investigating the kinetics
hydrolysis of chlorine in buffer solution in order to verify whether mechanism 3 is operative for chlorine. I n this paper, the results in the presence of acetate are reported; the possible correla-
(1) A. Lifshitz and B.Perlrnutter-Hayman, J . Phya. Chem., 64, 1663 (1960).
(2) A. Lifshitz and B. Perlmutter-Hayman, Bull. Reasarch Counci2 ramal, 8 8 , 166 (1959).
x2+ OH-
HOX
+ X-
(2)
