results that porous bed salt filters will only be useful for treating waters of low origiiial salinity. It is interesting to note that in all these cases R is quite close to R,.
PARAMETERS O F POROCS 0.04 7' = 1jo,p =
ik2,
TABLE 1 BED8-4LT FILTERS FOR
WHICH iJ ==
1 G . / C 1 L 3 , t / e ~ = 80, A S D mzp) = 0.01 MOLE/KC.WATERACCORDIXG TO THE POREMODEL These results are based on a Donnan exclusion in d i c h the only departures from ideality considered are the electrostatic interactions of the point counterions. Y t w r
mole/kg.
0.025
.os 1
5
R,,
84 42 21 4 2
1.
R,
A.
83
41 20 3 2
These conclusions apply to lattice cells with a void fraction of about 36% or less if the relatioii
R' - --- R_ _ Ro (45) R Ro R is used between the equivalent capillary radius R', the cell radius R, aiid the charge radius Ro, and if the fixed charge in the lattice model arises from a surface density u on the cylinder of radius Ro. This can be shomii as follows: In the lattice model, mocr)pis given by the equation
mole/kg.
WLQ(~),
1 7 3 3 6 7 43
Coiisider next the energy IV coiisunied in viscous flow per unit volume of water traversing the filter, which can be estimated as follows. W is equal to the pressure drop P across the filter. I n the pore model, if we assume Poisseuille flow in the pores purely for the sake of argument, W caii be written (43) where 7 is viscosity of the water, d the filter thickness, and q the volume of water put through the filter in unit time. If we use R , in place of R, eq. 43 can be written (44)
the factor of pioportioiiality being a combiiiatioii of various physical constants. Thus for fixed nzz(,),the energy consumed per unit throughput varies as the fourth power of the mean activity of the salt in the feed water. (An integration over the parabolic velocity profile characteristic of Poisseuille flow shows the product water salinity to be (3/2)m2(r)when & >> 1).
Thus w ~ o ( ~is) pthe same as in the related capillary. In such a case, eq. 45 becomes identical with eq. 37 aiid the lattice cell and the equivalent capillary have the same value of (e-?. Thus when R'IR, calculated from eq. 45 and 41 is plotted against ( e - < ) for the lattice cell, the curve of Fig. 4 is again obtained. Secondly, the geometric relation (46) between ?TLO(~ ) p and the radius R' from eq. 45 i s identical with eq. 42. Thus to each capillary radius R' determined by a given u and yk2m~2/mZ(r) there corresponds a manifold of lattice cells (with void fractions not exceeding 36%) whose parameters R and ROmust satisfy eq. 45. This is a very satisfactory state of affairs, first because the lattice model is presumably a much better replica of a porous bed than the capillary model, and second because a good way to compare these tmo dissimilar models is through eq. 45 which says that the volumeto-surface ratio is the same in both models. It is plausible to expect that to every capillary radius R' detcrmined by a u and a ratio there corresponds a porous bed with a volume-to-surface ratio equal to R' '2. Acknowledgment.-We wish to extend our heartfelt gratitude to Prof. George Scatchard for a series of illuminating discussions.
THE STRUCTURE OF ACTIVE CENTERS IX NICKEL CATALYST. 11. BY ITURO UHARA, SHOZO KISHIMOTO, TADASHI HIKIXO, YOICHIKAGEYAMA, HIDEBUMI HAJIADA, BXD YOSHIHIKO xUX4T.4
Chemzstry Department, Faculty of Science, Kobe Unaversnty, Milzage, Kobe, Japan Received August 2'0, 1962' When slightly cold-worked nickel is annealed, the disappearance of vacancies and dislocations takes place a t different temperature ranges, T , and TO,respectively, and the influence of the degree of cold-working and the existence of impurities on TOis considerable. The change of the catalytic activities of cold-worked nickel due to annealing at various temperatures wa9 measured for the following reactions: ( A ) hydrogenation of cinnamic acid, (B) dehydrogenation of ethanol, (C) decomposition of hydrogen peroxide, (D) para-ortho conversion of hydrogen, and (E) electrolytic generation of hydrogen (overvoltage). The activities decreased in two steps a t temperature ranges, Tal and T k 2 , and these temperatures coincided approximately with T , and TO, respectively, when specimens of the same material were employed for the measurements of both physical properties and catalytic activities. It was concluded that the active centers annealed a t Tal are point defects at the surface which coexist and vanish together with vacancies in the bulk metal, and that those annealed a t TA?are the terminations of dislocations at the surface.
Introduction The structure and the temperature of disappearauce of lattice defects in metals -which were subjected to cold-Jvorking, etc., have been studied in recent years by measuring the rate of release of defect energy (AP),'
changes of density (D), extra-resistivity (Ap), and hardness ( H ) 011 annealing. ' Typical behavior of twisted (1) L. M. Clarebrough, M. E. Haryreaves, and G . W.\Test, I'roe. k!oii. sot. (London), A232, 252 (1955); 'wag., 1, ,528 (1956); TT-. B ~ "Defects in Crystalline Solids," The Physical Society, 1 ~ 5 5p. , 212.
~
~
.
STRUCTURE OF ACTIVECEXTERS I N SICKEL CATALYST
May, 1963
nickel (only the AP' curve is that of a ground specimen) is shown in Fig. 1. Changes on annealing take place in two (or three) temperature ranges, Le., T , (ea. 200-330') and T D (-~400-750'), differing sample by sample) and the range centered a t loo', which was suggested to be due to the disappearance of interstitial^.^-^ Nothing is known of the relation between these defects and catalysis. h gradual decrease in the breadth of X-ray diffraction line is observed between room temperature-300' and a sudden change a t 515-550', and the hardness of cold-worked nickel was reported to decrease a t 550-750°.6 The change a t T , is believed to be due to the disappearance of vacancies.' At T Drecrystallization and a sudden decrease in hardness are noted and these changes have been attributed to the disappearance of dislocations. As the concentration of defects increases with the degree of cold-working, stronger interactions among them are naturally expected, Le., the formation of divacancies, clusters, vacancy-dislocation complexes and dislocation networks. These give rise to the depression of TDas follows : Cu.-Discussed in a previous paper.' Pt:- Z'D found by Rishimoto* from the measurement of the thermoelectric force is centered at 530 and 450' for 68 and 88% compressed specimens, respectively. Ni.-The rate of release of defect energy, AP for a specimen deformed in torsion to ndll = 1.87, where n = number of turns, d = diameter of wire or cylinder, and 1 = gage length, and AP' for a powder specimen of equal purity prepared by grinding, and recovery of density of specimens subjected to various degrees of working are shown in Fig. 1. On the other hand, 7'0 is elevated remarkably by the existence of impurities as shown in Table I.
997
6
4
.
2
1
s X
5
0
z
3
s
v . 0
d
2 1
0 100 50 0 0
200
600
400
Annealing temp.,
BOO
O C .
Fig. 1.-The rate of release of strain energy ( A P ) , changes of density (Dj, extra-resistivity ( A p ) , and hardness ( H ) as functions of annealing temp. for Ni (99.67,) deformed in torsion to nd/Z = AP = the rate of release of strain 1.87 (Clarebrough, et dl). energy for Ni powder (Mitchell, et aZ.z). APo = that of twisted Ni (99.85%, nd/l = 2.01) (Clarebrough, et a1.3).
TABLE I THE TEMPERATURE CORRESPONDING T O THE MAXIMUM OF AP AS THE FUNCTION OF THE DEGREE OF COLD-WORKING AND THE CONTENT OF IMPCRITIES Purity (%) Cold-working
99.6
Pulverized: 500' (AI'') Twisted n d / l = 2.34: 610'
99.85
99.96
70% Compressed:
70% Compressed: 3250
520' Twisted n d / l = 2.01: 450' (APo)
n d / l = 1.87: 630-660' ( A P )
T , remains nearly constant irrespective of the method of preparation or the degree of working.6,8 Depression of T D results in an unavoidable overlapping of T , and T Das seen in Table 1 and Fig. 1,especially in the density recovery curve. (In the lower part of T D ,rearrangement of dislocations may also occur.) I n order to establish unquestionably the relation between lattice defects and active centers in metal catalysts, therefore, it is necessary to employ the ideii(2) D . Mitchell and F. D. Haig, PhiZ. Mag., 2, 15 (1937). (3) L. M. Clarebrough, M. E. Hargreaves, M. H. Loretto, and G. W. w e s t , Acta M e t . , 8, 797 (1960). (4) A. Sosin and J. A. Brinkman, ibid., 7 , 478 (1959). ( 5 ) J. E. Wilson and L. Thomassen, Trans. A I M E , 22, 769 (1934). (6) "Vacancies and Other Point Defects in Metals and Alloys," The Institute of Metals, London, 1958. (7) I. Uhara, S. Panagimoto, K. Tani, G. Adachi, and S . Teratani, J . Phys. Chem., 66, 2691 (1962). (8) 5.Kishimoto, to be published.
0
200
400 600 Annealing temp., " C .
800
Fig. 2.-Catalytic activities (scales being arbitrary) of toldworked Ni( I j as functions of annealing temps. A = hydrogenation of cinnamic acid by twisted (e) and compressed (0) S i . F, = dehydrogenation of ethanol (A). W = hydrogenation of CzH4 by Ni foil (Eckell)l*(@). K l and KP = temp. ranges for sudden decrease of thermoelectric force of Ni(I), corresponding to T , and T D ,respective1y.g
tical specimens or a t least specimens of the same material slightly cold-worked for the measurements of both physical properties and catalytic activities. The thermoelectric force, S , also is sensitive to the presence and disappearance of lattice defects. Since the measurement of S can be performed easily with small quantities of specimens, it offers a eonvenient and effective method for studits of this field es shown by Kishimoto.8 In Fig. 2 and 3, the temperature ranges corresponding to T , and T D for sudden changes (9) 8. Kishimoto, J . Phys. Chcm., 66, 2694 (1962).
I. UHARA,S.KISHIXOTO, T. HIKISO,Y. KAGEYAMA, H. HAMADA, ASD Y. KUMATA
998
Vol. 67
Experimental
I
\
'..
200 300 400 300 Annealing temp., O C .
600
Fig. 3.--Catalytic activities (scales being arbitrary) of S i for p-o conversion of Hz as functions of annealing temps. D = slightly cold-worked Ni(1) ( 0 0 means two measurements). Z = heavily cold-worked foil (O).?O X28 and YZQ = ordinary reduced catalysts (V and A, respectively). K 1 = T,, K Z = T D .
.
200
?
2 A
Y
c
0.20
180 on
!i
I 0.22
140
0.24
100
0.26
I
100
1
200
I
300
400
500
I
1
600
700
2
i
60
800
Annealing temp., " C .
Fig. 4.-Catalytic activity for the decomposition of HzOz (C, A), hydrogetl overvoltage (TH(E),0),and hardness (H, 0 ) of cold-worked Xi( 11)as functions of annealing temps.
in A' on annealing cold-worked nickel (sample I) determined by Kishimoto are represented with the horizontal lines, K1 and Ks. For sample I1 which is less pure than (I), no reliable data on S can be obtained because of large fluctuatioiis probably owing to noliuniform distribution of impurities. It is, therefore, necessary to employ samples of high purity for this method. Instead, T D of sample I1 was determined from the chaiige of hardness 011 annealing (measured by K. Iwasaki with a micro-Vickers hardness tester) as shown in Fig. 4, H. Uhara and co-workers1° have studied the relation between lattice defects and active centers in metallic catalysts and established experimentally that the ends of dislocations at the surface are the active centers of copper catalyst for the decomposition of diazoilium salt? lo and the dehydrogenation of ethanol.' They also reported preliminarily that in the case of nickel catalysts the active centers are surface point defects for some reactions and these are both point defects and ends of dislocations at the surface for some other reactions." I n this paper, details of experimental evidence for the conclusion concerning the structure of active centers in nickel are given. (IO) I. Uhara, S. Yanagimoto, K. Tani. and G. Adachi, Xature, 192, 867 (1961). (11) I. Uhara, T. Hikcno, P. Kumata, H. Hamada, and Y. Kaaeyama, J . Phys. Chen., 66, 1374 (1962).
Two samples of nickel wire containing the following impurities (in per cent) were employed. Xi(1): 0.009 C, 0.011 Si, 0.001 S, 0.092 hln, 0.022 Cu, and 0.033 Fe. X ( I 1 ) : 0.023 C, 0.07 Si, and 0.006 P, other elements being unknown. Wire was annealed a t 800' in hydrogen atmosphere for 1 hr. and twisted or rolled in air slowly so as to prevent temperature elevation; the degree of working being represented by nd/l or by the degree of compression defined by 100(d-6)/d, where 6 = the thickness of the resulting plate. The change of catalytic activity of coldworked metals due to annealing a t various temperatures12may be studied by the following methods: (i) pieces of specimens coldworked under conditions as identical as possible are divided into several groups and each group is annealed a t different temperatures. (ii) a group of pieces is employed throughout a series of experiments by repeating annealing a t increasing temperatures followed by the activity measurement. Although the latter is excellent in avoiding fluctuations of data (see below) and in keeping the experimental conditions constant, it is applicable only when poisoning is absent or a t least pxovery from it is easy on repeated catalytic reactions. B large number of pieces of cold-worked catalysts must be employed to avoid the fluctuation of activity due t o non-uniformity of cold-wTorking and of diutribution of impurity, since their influence on the generation and disappearance of lattice defects cannot be neglected. The activity of catalysts employed in our research was much lem than that of the usual catalyst; therefore, special devices are frequently necessary for determining the yield and for the selection of experimental conditions. The reproducibility of results was ascertained for reactions (A), (B), and ( C ) . (A) Hydrogenation of trans-Cinnamic acid.-N(1) was ( a ) twisted (nd/l = 0.21) or ( b ) 787, compressed (the apparent surface area = 3.65 cnx.z) and then polished with emery paper in water. One cc. of 0.45 N alcoholic solution of cinnamic acid (extra pure) was put in Tarburg's apparatus and the hydrogen uptake was measured a t 25'. (B) Dehydrogenation of Ethanol.-Ni(1) wire (total surface area = 15.7 cm.2) was twisted (nd/l = 0.25) without previous annealing, washed with pure alcohol, dried and put in hydrogen a t 200" over 1 hr. to remove the oxide layer a t the surface. Pure nitrogen saturated with ethanol vapor free from aldehyde a t 30' was sent on the catalyst a t 200' a t the rate of 1 l./hr. for 2 hr. Aldehyde produced was dissolved in cold water and the vield was determined colorimetrically with m-phenylenediamine . l a After the reaction the catalyst was annealed in a flow of nitrogen at a higher temperature for 1 hr., put in hydrogen overnight and then the activity measured again, etc. The catalyst was not exposed t o air throughout the experiment. (C) Decomposition of Hydrogen Peroxide .-Xickel(I1) wire (the surface area = 3.22 sq. cm.) was tw-isted ( n d / l = 0.41) and thrown into 30% hydrogen " -peroxide solution in Warburg's apparatus a t 20". [D 1 Para-ortho Conversion of Hvdroaen - - .-Seventv-four per cent compressed nickel(1) (the apparent surface area = 15 cm.l) was polished with emery paper in acetone, washed with alcohol and acetone and dried by evacuation to mm. a t 150". Hydrogen purified by passing through a palladium tube was adsorbed a t liquid nitrogen temperature on active carbon which had been degassed a t 300" t o 10-6 mm. The resulting hydrogen contains 50.4% of the para-form. The rate of conversion t o the ortho-form was measured a t 150', where the equilibrium content of para-hydrogen is 25.0%, with a Pirani gage with platinum wire in a bath maintained a t - 110'. The catalyst was annealed in the reaction vessel without contact with air. Hydrogen (15-19 mm.) was circulated constantly in the vessel with a mercury pump. (E) Electrolytic Generation of Hydrogen.-Hydrogen overvoltage ( T E ) was measured a t 25" in 0.12 N HC1 solution and compared a t a current density of 1 0 - ~amp./cm.z. A twisted specimen of nickel(II)(nd/l = 0.217) was employed as the cathode. Even when i t was brought in contact with air a constant value of K H was obtained after flowing a weak current by which adsorbed oxygen was probably removed.
Results The catalytic activity of nickel (I or 11) cold-worked (12) Cooling after annealing must be slow to avoid the generation of lattice defects due t o quenching. (13) B. Richard, Anal. Chem., 20, 922 (1948); J. Bailey, Ind. Eng. Chem., Anal. Ed.. 13, 834 (1941).
STRUCTURE OF ACTIVECEKTERS IN KICKEL CATALYST
May, 1963
and annealed a t various temperatures is shown in Fig. 2, 3, and 4 on arbitrary scales together with K1 and K z for S measured with sample I and H for sample 11. (A) CijHbCH-CHCOZH H, = C6H,CHzCHzCChH, (Fig. 2, A). About 0.6 pl. of hydrogen was absorbed in min. when the catalyst was not annealed. (B) CzH60H = CHsCHO Hz, (Fig. 2 , B).-At 200' the formation of ethylene and ether is negligible and the decomposition of aldehyde to carbon monoxide was also not detected. The reproducibility of the activity was shown as follows :
+
+
Experiment
Relative activii y
Treatment after catalytic reacn.
1 2 3
0 50 .48 .50
4
.10 (poisoning by aldehyde I ) 0.52 .72
Put in hydrogen flow overnight Put in hydrogen flow overnight Immediately after 3rd experiment without sending hydrogen P u t in hydrogen flow overnight
5 6
Put in hydrogen flow for 2 days
The activity observed after putting in hydrogen flow overnight was compared. By the catalyst not anmole of aldehyde was formed, nealed 2.4 X about half a t the point defects and another half at the ends of dislocations a t the surface as discussed below. (C) 2Hz02 = 2H20 02,(Fig. 4,C). About 10 ~ 1 of. oxygen was evolved in 40 min. by the catalyst not annealed (D) Para-H2 = ortho-Hz, (Fig. 3, D)- This reaction proceeded according to a first-order rate equation and the rate constant was 0.13 when the catalyst annealed a t 200' was employed. (E) 2H+aq+Hs, (Fig. 4, E).-Lowering of T H due to mechanical working such as grinding has been explained in terms of increabe of the surface area, and consequently, of depression' of the current density. But we know that even the surface area of reduced nickel (powder) does not change on annealing over T v , I 4 while T H of cold-worked metal with far less roughness of the surface increases suddenly on annealing a t this temperature. Siiiice the change a t the cathode is also a sort of catalytic reaction, it may be most rational to explain it in an analogous manner as in cases (A) (D).
+
-
TABLE I1 ( T A AND ~ T.42) O F NICKEL(SAMPLES I AND 11)
T H E SI?rTERIXG TEMPERATCRE
Reaction
Sample
.4 R
1 I I1 I I1
c
L)
E Physical properties
S H
TAi ( " c . )
150-270 200-300 300-400 -200-300 220-300
TAZ
200-250 (Ki)
("c.)
400-600 460-540 630-720 400-500 570-CU.700
T"
I I1
COLD-WORKED
TD
400-CU. 600 (K2) 500-CU.800
Discussion The comparison of these results with Fig. 1 indicates that the decrease of catalytic activities generally ~ takes place in two temperature ranges, i.e., T A and (14) S. Iijima, Rev. Phys. Chem. Japan, 14, 128 (1940).
999
T A ~in, concurrent with the disappearance of vacancies and dislocations a t T, and T D ,respectively, in the bulk metal. The coincideiice of Tal with T,, and TAZwith T D is fairly good when they are measured with an identical specimen or a t least specimens of the same material, as seen in the Table I1 and Fig. 2 , 3, and 4. It has frequently been assumed that the surface of nickel catalyst is perfect and the only heterogeneity comes from the variety of appearing crystal faces, and that sintering is explained only in terms of the decrease of the surface area. But the change of the area due to aiinealiiig over T , and in some cases a t T D was shown to be quite small compared with the decrease of activities even for ordinary catalysts with the highly developed surface so long as they are non-porous, e.g., the van der Waals adsorption area of reduced nickel does not decrease on annealing at 380°,14and that of deposited copper is almost invariable on annealing over TD.~O The concurrent change of the catalytic activity with various physical properties of the bulk metal indicates that the sintering of catalysts is not a mere shrinkage of the area but a qualitative and structural change given rise by the disappearance of (surface) lattice defects. The stepwise decrease of the activities demonstrates this most clearly. Accordingly, we may conclude that the active centers annealed at T A ~ are the terminations of dislocations a t the surface. The approximate coincidence of T Aand ~ T, indicates that Tal is substantially concerned with the disappearance of point defects. Since the active centers of the catalyst exist, of course, at the surface they must be surface point defects(P), although nothing is known of the physical nature of them so far. Of P in the catalyst with a slightly deformed structure and with low concentration of the lattice defects as studied in our research, the simplest type, ie., a surface vacancy (Sv) with one atomic size, and a projecting atom (Pr) may predominate over other complex types of defects. It has been considered that at T, vacancies begin to migrate to dislocations, to grain boundaries or to the surface, where they are absorbed or disappear. When a vacancy comes to a plane surface it will form Sv. If a considerable part of the surface is covered with Sv, the remaining part of the surface will form P r in a dispersed state, hence vacancies in the bulk and P a t the surface are not independent and various types of interaction including mutual annihilation are conceivable. At the temperature where the surface migration of P become possible, 1: 1 annihilation of Sv and P r resulting in the formation of a plane surface, or agglomeration of Sv (or Pr) only or combination with dislocations may occur. This temperature may probably correspond to T A ~ .It may be supposed that the surface migratioii is easier than the bulk ~ somewhat migration of vacancies, hence that T A is lower than T,. If so, catalysts must lose all of their activity originating from P when they are annealed a t the temperature range between TR1and T,, then the (15) Crystalline nickel powder prepared b y complete evaporation of amalgam is perfectly inactive for the hydrogenation of benzene, indicating the essential importance of the presence of surface defects.10 Heterogeneitv ot the surface of nickel catalyst as demonstrated also fiom the analysis of the d a t a of hydrogen adsorption 17 (16) P Zemsch and F. Lihl, Z Elektrochern., 56, 985 (1952). (17) (a) H. 6 Taylor, Advan. Catalyszs, I , 1 (1948): A . Eucken and W. Hunsmann. 2. p b y s z k . Chern., B44, 163 (1939); (b) T. Takaishi. zbzd., 14, 3/4, 164 (1958).
activities will be partly recovered by the migration of decomposition of formic acid by nickel Duell and vacancies to the surface on heating to T,. ,4s a matter Robertsonz4found the generation of very liigh activities of fact, however, no example of catalysts mliich behave after flashing thein at high t,emperature and attributed in such a way when they are sintered is known. I n it to surface vacancies acting as active catalytic sites. our research, also, no distinct difference between T A ~ It is well known that vacancies are generated during and T , is observable. Consequently, we may conclude quenching (including radiation quenching), hence the that TAXis practically equal to T,, and that the active formation of P (Pr and Sv) also is quite conceivable. centers annealed a t T.41 are the surface point defects (v) Purely Chemical Method (Ordinary Catalysts).which coexist and vanish together with vacancies in Ordinary catalysts are prepared chemically and display the bulk metal. A method of assigning Sv or P r to much higher activity than the cold-worked metals. the active center for individual reaction may be They are considered to have highly distorted structures shortly reported. analogous to that of the extremely cold-worked speciAccording to Eckell, rolled nickel (foil) lost most of mens discussed abore, in view of the diffuseness of the activity for the hydrogenation of ethylene on anX-ray diffractlion pattern, abnormally high chemical nealing a t 275" and completely at 300" (Fig. 2, W). activity (pyrophoricity, etc.), energy content and heat This may be interpreted to mean that the active center of adsorption, and low density. For example, density for this reaction is some kind of P,contrary to Cratty of copper twist,ed to nd/Z = 1.8 is less than t,hat of the and Granato's p o ~ t u l a t e which '~ attributed it to a disloiiorrnal one (8.93) by 0.025%,1 whereas that of a sample cation. of reduced copper is less by 2.5%,z6and even a value, Contribution of the normal surface to the activities as low as 7.6 was report'ed.z6 The density of nickel of nickel is almost negligible or a t most only a few twisted (ndll = 1.41) is less than that of the normal per cent in every case so far studied. metal (8.90) by 0.045'f%0 (Fig. l), whereas that of reWith the increasing degree of cold-working, separaduced catalyst is sornet'imes less than 8.0, aIthough the tion of T , and T D (hence that of T A and ~ T A ~becomes ) existence of macroscopic voids may be suspected, too.27 diffuse resulting in only continuous decrease of activity Clustering of lattice defects and strong interactions with increasing temperature of annealing, as observed among them result in overlapping of T, and TD, or in the case of nickel foil, which is, of course, prepared T.41 and Taz, and consequently a continuous sintering by heavy rolling, for para-ortho conversion of hydrogen(at T,) curve as shown by curves X (nickel catalyst (ea. 200-500°) zo (Fig. 3, Z), reduced a t 300°)28and I' (catalyst reduced a t 340°)z9 in Fig. 3 for para-ortho Conversion of hydrogen. Their Generation of Active Centers active ceiit'ers may substantially be the same as those Various procedures other than the mechanical one generated att t'he surface of metals by cold-working, are known to generate active centers in metals. considering the approximate coincidence of T s with (i) Irradiation.-When annealed nickel is irradiated Tal (or T,) and T A (or ~ T D ) ,although the possibility with thermal neutrons (1.5 X 1016/cm.z) it gained cannot, be denied that two or more latt,ice defects comcatalytic activity of nearly the same magnitude as that bine to form complex active centers with new functions of the cold-worked one for the decomposition of hydrowhen the conceiitration of defects is high. Estimation gen peroxide. If the annealing experiment is perof t,he structure of active centers is possible from T , formed the sort of the active center generated may value for some reaction, e.g., since nickel catalyst be determined. loses most of the activity for the hydrogenation of (ii) Electrodeposition.-By means of electrolysis benzene at ea. 350°,30the active center may be P. under high current density (>O. 1 amp./cm.2) catalytiIf the distortion of the crystal lattice is extreme as in cally active metals are obtained. They contain imthe case of Raney nickel, the stxuct'ure becomes nearly purities as oxide, hydroxide, etc., and distortion of the amorphous and hence the terms, vacancy and dislocastructure is shown by means of the X-ray tion, which have been defined for defects in a nearly The existence of dislocations is estimated from their perfect' crystal, lose their strict sense, and their properhigh hardness and was ascertained in the case of copper ties may be different from those of isolated ones. In from the catalytic activity characteristic of dislocasuch a case, assignment of active centers to individual tion~.~ The generation of point defects during eleclattice defects is difficult. Although it was also trolysis can be demonstrated by the catalytic activity found in some cases that the activity remains after of electro-deposited nickel for the hydrogenation of annealing a t temperatures higher than T , or TD, ethyleiie observed by Umemura. 2z especially when the cat,alyst is supported, their physical (iii) Reduction of Salt with Base Metals.-Copper conditions are too complex to be explained now. powder containing many dislocations can be obtained At the present stage of our knowledge, we must, by means of the reaction of copper sulfate solution first of all, assign active centers to individual surface prepared a nickel with zinc dust.7 Urushibara, et defects for each reaction employing slightly distort'ed catalyst with a high activity by means of the reaction crystals, though experimental difficulties owing to of nickel chloride solution with zinc dust. It can be utilized for many catalytic reactions, hence it is certain (23) Y.Urushibara and S. Nishimura, Bull. Chem. Soc. J a p a n , 27, 480 (1954); Y.Urushibara, 8. Nishimura, and H. Uehara, ibid., 28, 446 (1955). that it contains both P and dislocations. (24) M. J. Duell a n d A. J. Robertson, Trans. Faraday A'oc., 66, 1416 (iv) Quenching.-In the study of the catalytic (1960). (18) J. Eokell, Z.Elektrochem., 39, 433 (1933). 26,96 (1957). (19) L. E C r a t t y , J r , and A V Gianato, J . Chem PILUS., (20) E. Cremer a n d R. Kerber, Adban Catalyszs, VII, 82 (1955) (21) W. Blum and C . Xasper, Dzscusszo?i~Fa,aday Soc , 31, 1203 (1935); C. H. Desoh, tbzd., 31, 1043 (1935); D J Maonaughtan and 4. R . Hotliersall, abzd.. 31, 1168 (1935). (22) K. Umemura, Bull. Tnzz. Osaka PTB/.,A9, 91 (1960)
( 2 5 ) R. S . Pease, J . Am. Chem. Soc., 46, 2296 (1923). (26) fi. Audibert and A. Raineau, Compt. Tend., 197,596 (1933). (27) ,J. IT. bfellor, "A Comprehensive Treatise on Inornanic and Theoretical Chemistry," Vol. XV, London, 1936, pp. 52, 53. (28) E. Fajans, Z.p h y s i k . Chem., B28, 252 (1935). 129) G. Tammann, 2. anorg. allgem. Chem., 224, 25 (1936). (30) E'. Lihl and P. Zemsoh. 2. Elektrochem., 66, 979 (1952).
May, 1963
SURFACE COKDCCTANCE OF SODIUM CHLORIDE CRYSTALS
their low catalytic activity are sometimes unavoidable. Our present circumotaiice is quite analogous to the study of the solution, in which a dilute solution was studied at first with success because of the simplicity of the physical conditio11of the state treated.
1001
Acknowledgment.-MTe wish to express hearty thanks to Dr. H. Takegoshi, Mr. M. Hasegawa, Mr. S. Taniguchi, Mr. K. Iwasaki, and Mr. N. Murakami for their helpful assistance, arid t o Prof. T . Iida and Dr. A. Saika for their kind advice.
THE SURFACE CONDUCTAKCE OF SODIUM CHLORIDE CRYSTALS AS A FUKCTION OF WATER VAPOR PARTIAL PRESSURE BY GEORGESIMKOVICH Xax-Planck-Instit ut f u r physikalische Chenaie, Gottingen, Germany Received August 21, 1962 Surface conduction of a S a c 1 cryetal in H20 vapor may be due t o (1)individual adsorbed semi-hydrated ions, and/or ( 2 ) saturated NaCl solution formed by virtue of capillary condensations in grooves on the surface of the crystal. In the latter case, the conductance is supposed to be decreased by the presence of a detergent, e.g., 1-hexanol. Samples having a low conductance show no appreciable effect of the presence of hexanol. So conduction through adsorbed semi-hydrated ions seems to prevail. Samples having a higher conductance show an effect of the presence of hexanol supposedly due to conduction in grooves. Formulating the law of maes action for the formation of adsorbed semi-hydrated ions, one may calculate the average number of HzO molecules of adsorbed semi-hydrated cations and anions. A value of 3 is found a t low humidities, whereas a value of about 10 corresponding to a second hydration shell is found a t high humidities.
Theoretical It is well known that the resistance of a solid insulator, e . g . , glass, decreases markedly with increasing humidity of the surrounding atmosphere.l-’ This is usually ascribed to the occurrence of conduction along the surface of the solid. It is the objective of this paper to try t o clarify the nature of this kind of surface conduction. To this end, it seems expedient to investigate a simple salt, e.g., SaC1, rather than glass whose structure is rather iiivolvecl and where hysteresis phenomena prevail. In the absence of HzO, surface conduction may occur by the movement of individual adsorbed ions, or ion vacancies in the outermost lattice plane. At room temperature, such defects are supposedly very rare because of their high energy contei2 t and accordingly surface conductance in a dry atmosphere is very low. I n the presence of H20vapor, two limiting mechanisms of surface conduction may be anticipated, vix., (1) conduction due to the formation of individual adsorbed semi-hydrated ions schematically shown in Fig. 1 with an energy content much lower than that of adsorbed nonhydrated ions and (2) coiiduction through a saturated NaCl solution formed by virtue of capillary condensation of water vapor in sufficieiitly narrow grooves shown schematically in Fig. 2 . I n what follows the characteristics of each limiting case are discussed. Intermediate conditions are considered below. (1) The formation of adsorbed semi-hydrated ions may be described by the equation SaCl(s)
+ (ml + m2)HdXg)
=
+
n’a(OH2)m,+(ad) C1(HzO),,-(ad) ( 1) where ml and me, respectively, are the numbers of H20 molecules associated with individual cations and anions (1) J. 8.Dryden and P. T. Wilson, Austral. J . A p p l . S a . , 1 , 97 (1950). ( 2 ) KI. Kantzer, B u l l . znst. zeme, No. 5, 11 (1946). (3) P.Le Clem Szlzeates Ind., 19, 237 (1954). (4) K. Kawasaki, K. Kanou, and Y . Bekita, J . Phyb. Soc. J a p a n , 2, 222 (1Yi8).
(5) N. Chirkov, Russ. J . Phys. Chem., 21, 1303 (1947). ( 6 ) U‘. A. Yagcr and S.0.Morgan, J . Phys. Chem., 35, 2026 (1031). (7) A. Ya. I h a n e t s o v , Z h . Fzz Khzm., 27, 657 (1953).
sitting on the surface of the bulk crystal shown schematically in Fig. 1. In general, one has to expect simultaneous formation of adsorbed ions involving a variety of hydration numbers, i.e., nal = 1,2,3. . .and m2 = 1,2,3. . .For the sake of simplicity, one may tentatively assume that ions with particular hydration numbers ml* and m2* prevail. Then nearly equal numbers of adsorbed cations with ml*H20 molecules and adsorbed anions with m2*H20 molecules are formed in accord with eq. 1. Denoting their concentrations in mole per unit surface area by I‘, assuming low surface coverage of the adsorbed ions, and applying the ideal law of mass action to the reaction stated in eq. 1,one has ~z/Pm’”+mz* = K
(2)
where p is the H2O partial pressure and K is a constant. I n the case of low surface coverage, one inay assume constant mobilities of the adsorbed ions so that the conductance becomes essentially proportional to the surface concentration r. Thus, under conditions where the volume conductance can be neglected, the conductance of a sample of given dimensions, i.e., the reciprocal of the resistance R is expected t o be proportional to I?, whereupon it follows with the help of eq. 2 that 1/R
=
const p”
(3)
where
m
’/dml*
+ m2*j
(41 is the average number of mater molecules found in the hydration shells of adsorbed cations and anions. Taking logarithms of both sides of eq. 3, one has =
log (l/R)
= m
log p
+ const
(5) Thus, measuring the resistance of a NaCl crystal a t different water vapor partial pressures and plotting log l/R us. log p , one may expect a straight line. Actually, deviations from a straight line may occur because of the occurrence of a variety of adsorbed ionic species with different numbers of H,O molecules in the hydration shell, in particular because of the occurrence of a second
