Adsorption of nitrogen and carbon monoxide on alumina up to 3000 atmospheres pressure

ADSORPTION OF NITROGEN

AND CARBON MONOXIDE ON ALUMINA

UP TO 3000 ATMOSPHERES PRESSURE

/

•

BIBcicHHEEK

D1:R

TECHNISCHE HOGESCHOOL

DELFT

PROEFSCHRIFT

TER VERKRIJGING VAN DE GRAAD VAN

DOCTOR IN DE TECHNISCHE WETENSCHAP

AAN DE TECHNISCHE HOGESCHOOL

TE DELFT OP GEZAG VAN DE RECTOR

MAGNIFICUS DR. R. KRONlG, HOOGLERAAR

IN DE AFDELING DER TE

C

HNISCHE

NATUURKUNDE, VOOR EEN COMMISSIE

UIT DE SENAAT TE VERDEDIGEN OP

WOENSDAG 25 OKTOBER 1 961

DES VOORMIDDAGS TE 11 UUR

DOOR

PARAKAT GOVINDANKUTTY MENON,

M.Sc., M.Tech.

GEBOREN TE CHERPU, KERALA, INDIA

1961

•

DiT PROEFSCHRiFT iS GOEDGEKEURD DOOR DE PROMOTOREN

PROF. DR. J.

H.

DE

BOER

EN PROF DR. Ä.

M. J.

F.

MICHELS

THE ENTI~E EXPERIMENT AL WO~K ON WHICH TH IS THESIS IS BASED WAS CA~RIED OUT UNDER THE DIRECTION OF PROF. A. M. J. F. MICHELS IN THE VAN DER WAALS LABORATORY, UNIVERSITY OF AMSTERDAM. THE THESIS IN lTS PRESENT FO~M WAS DEVELOPED AND WRITTEN UNDER THE GUIDANCE OF PROF. J. H. DE BOER, WH EN THE AUTHOR WAS WORKING IN THE CA TAL YSIS DIVISION OF THE CENT~AL LABORATORY, STAATSMIJNEN IN LIMBURG,

T 0 our joint family, especially the younger generation there

CONTENTS

Page

Introduction 9

Chapter I - Adsorption at high pressures

1. Importance of adsorption at high pressures 11 2. A survey of earlier work on adsorption at high pressures 13

3. Scope of the present work 14

Chapter 11 - Apparatus and experimental procedure A. Adsorption measurements at high pressures 1. Principle of the experiment

2. The glass capsule

3. Choice of the adsorbent and the adsorbate gas es 4. The high-pressure experiment

5. Calculation of l',.g 17 17 19 20

21

22 B. Low- and medium-pressure adsorption measurements 23

1. Measurement of adsorption at low pressures 23

2. Measurement of adsorption at low temperatures 24

3. Medium-pressure measurements 25

C. Preparation of the gases and their purity 26 Chapter m - Corrections to he considered in adsorption

measure-ments at high pressures

1. General corrections . 27

2. Thermal expansion and compressibility of alumina 27 3, Dimensional changes in rigid adsorbents on physical

adsorption of gases on them 28

4. The true density of an adsorbent . 30

5. The specific volume of alumina A and alumina B 31 6. Corrections for the dimensions of the N2 and CO

molecules 32

Chapter IV - Experimental results and preliminary discussioill

1. Experimental data 34

2. Adsorption of N2 on alumina A and alumina B 51 3. Adsorption.above the critical temperature of the adsorbate 52

4. Course of the adsorption isotherms at high pressures 53 5. The second ascending part of the high-pressure adsorption

Chapter V - Entropy and mobility of adsorbed molecules

1. Evaluation of entropy data 56

2. Comparison of physical adsorption of A,

N

2

and CO

on alumina, charcoal and graphite 59

3. lnduced dipole moment in adsorbed molecules 61

Chapter VI - Analysis of adsorption isotherms at high pressures

1. Correction for the decrease in dead space due to the

volume of adsorbed molecules

62

2

.

The adsorption layer on a surface

63

3.

Differential adsorption

66

4.

Absolute adsorption

66

5. Surface coverage at high gas den8ities

69

6

.

Reorientation of adsorbed molecules at high densities

72

7.

Temperature dependence of the second ascending part

of the nitrogen adsorption isothern s

₇₃

8. The absence of reorientation in thc case of adsorbed

CO molecules at high den si ties

74

Chapter VII - Some general considerations

1. Surface structure of partially dehydrated alumina 75

2. Reproducibility in adsorption-desorption measurements 77

3. Absolute accuracy in adscrption measurements at high

pressures

78

4. Relative accuracy in adsorption measurements at high

pressures 80

5. Some possible refinements 80

6. Suggestions for future work 82

Summary ₈₄

Samenvatting 88

References 92

Curriculum vitae 95

INTRODUCTION

The application of high pressures and the introduction of specific catalysts have been two of the most important factors which led to the rapid development of chemical industry during the last fifty years. The development of the Haber-Bosch process for the synthesis of ammonia in the Badische Anilin and Soda-Fabrik at Ludwigshafen in 1913 was in deed a significant landmark in the chemical industrial revolution. The experience gained in working at 300 - 400 atm. pressure at high temperatures and in the use of catalysts on an industrial scale soon proved to be inval ua bie : in the next two decades synthesis of methanol and higher alcohols, hydrogenation of coal and tar to produce synthetic fuels and other by-products, Fischer-Tropsch synthesis, iso-synthesis, Oxo, Arobin, Mersol and Synol processes, several cracking and polymerization processes in the petrochemical industry, etc., were all developed. This, in turn, was followed by the synthesis of polyethylene at 1500 - 3000 atm. pressure in the Imperial Chemical Industries in 1935, the development of the new industrial chemistry of

acetylene (Reppe Chemistry) at Ludwigshafen, the Claude process for the

synthesis of ammonia at 1000 atm. pressure, the Lürgi medium-pressure complete gasification process to produce water gas from low-grade coal, the water-gas shift reaction under medium pressure, etc. Many of these processes can be carried out only at high pressures and/or in the presence of specific catalysts.

Catalysis itself has in the meanwhile developed as a vast field of research and of technology, cutting across many other fields. Thirty years ago the ad hoc approach of Haber's time of trying every conceivable combination of metals or oxides to discover a suitable catalyst for a chemical process was still in vogue. However, it was alrcady known that the chemisorption of one or more of the reactants on the surface of the catalyst was a pre-requisite for most of the heterogeneous catalytic reactions, although a measurable amount of chemisorption was not always a sure guide to catalytic activity.

During the 1930's the application of quantum mechanics to the problems of the solid state led to a great improvement in the understanding of the physical chemistry of metals, semi-conductors and insulators.

In

the next decade these ideas were extended to the study of phenomena occuring at solid surf aces such as chemisorption, catalysis and corrosion, wh en the influence of the electron ic configuration of the solid on these phenomena was readily recognized.

Developments in the study of physical adsorption, caused by van der Waals forces, were also taking place at a rapid pace. Langmuir's unimolecular adsorption theory, Polanyi's potential theory and the capillary condensation theory could not account for all the va ried characteristics of physical adsorption isotherms observed experimentally. Several other theoretical attempts have since been made of which, perhaps, the best known is the multimolecular adsorption theory of Brunauer, Emmett and Teller (BET)!. The BET theory has been particularly successful for the estimation of reasonably reliable surf ace are as of solid catalysts. A knowledge of the surface area in turn has had many practical applications as

in the calculation of promotor distribution on catalyst surfaces, in the determination of pore volume and po re si ze and in the study of specific activity, poisoning and sintering of industrial catalysts, notably, ammonia, methanol, Fischer-Tropsch and petroleum catalysts2_•

More rigorous theoretical approach to the problem of physical adsorption, in particular, multimolecular adsorption, is severely handicapped by mathematical difficulties. The anisotropy of the adsorbent-adsorbate system and the energetic heterogeneity of ordinary solid surfaces make the problem far more complicated than the theory of the liquid state3 . Experimental results, however, can of ten be interpreted in terms of simpier models~ and thermodynamic methods can be employed to derive useful quantities from the results.

Experimental investigations of surface phenomena like adsorption and catalysis have been intensified during the last fifteen years. Apart from volumetric and calorimetric studies of adsorption, a variety of new and powerful techniques have been employed such as measurement of surf ace potentials, chemical analysis of surface films, radioactive tracers, magnetic permeability, nuclear magnetic resonance, vacuum micro-gravimetry, gas chromatography, optical metallography, polarizing spectrometry, multiple interferometry, electron diffraction, electron microscopy and field emission microscopy. The results emerging from all these studies do not always lead to the same congruent picture; the phenomenon of adsorption itself is gradually being revealed as one of unexpected complexity. Nevertheless, many results of experimental studies, together with the theoretical knowledge derived from them, have had a great influence in many fields of science and have led to many applications in the practical fields of catalysis, corrosion, electron emission, adhesion, welding, mechanical wear, lubrication, etc. They mayalso find equally important long-range applications in several other branches of science lying between geology on thc one hand and biology on the other.

CHAPTER I

ADSORPTION AT HIGH PRESSURES 1. Importanee of adsorption at high pressures

Chemical reactions on an industrial scale in the 1000 - 3000 atm. pressure range seems to be a definite possibility in the near futuré. The high-pressure synthesis of polyethylene and the Claude process for the synthesis of ammonia serve already as a pointer in this direction. But one of the main hurdles in this development is our ignorance on the effect of pressure on catalytic activity. Freidlin, Vereshchagin and NumanovG _{have observed that compression of} sup-ported catalysts to 20,000 atm. improves in general their bulk density, mechanical strength, activity and useful life, but the energy of activation of dehydration reactions studied by them is not changed. Pressure did not change the active surfaces, crystallattices and po re characteristics of alumina and silica gel catalysts; this means that compression did also not affect or alter the nature of active centres in the catalysts, if such active centres are operative.

These results suggest another !ine of approach to investigate the gas-solid interface at high pressures. Since the adsorption of the reactant or reactants on the catalyst surface is a pre-requisite for heterogeneous catalytic reactions, how does th is adsorption vary under high pressure ? Even otherwise, adsorption studies at high pressures and relatively high temperatures may be a better approximation to actual conditions in industrial catalytic reactions than are the convent ion al low-pressure low-temperature measurements.

Though chemisorption has the dominant role in catalytic phenomena, it is known th at physical adsorption too can sometimes contribute to catalytic activity6. Some cases of pure "physical catalysis" have also been reported quite recently7, 8.

According to the Rideal mechanism of contact catalysis, a chemically adsorbed reactant molecule can, under favourable conditions of temperature and pressure, react with another reactant molecule in a physically adsorbed phase or in the gas phase. Recently, Ehrlich9 _{has suggested that a physically adsorbed layer on} a surface can serve as a reservoir of atoms or molecules, from which subsequent chemisorption can take place. This rather obvious assumption means physical adsorption to be the real forerunner to all heterogeneous catalytic reactions and also to the chemisorption which is hitherto considered as the basic and essential part of catalytic reactions (Emmett10_).

But our knowledge of physical adsorption under Ieaction conditions, i.e., above room temperature and very often under high pressures, is very scarce. Brunauerl l _{has repeatedly pointed out the necessity of testing experimentally} whether multimolecular adsorption above the critica 1 temperature of the adsorbate and necessarily under high pressures is pos si bIe or not. In theory, at least, this seems to be not impossible. In 1933 - 35, Maass and coworkers12 , 13 investigated the adsorption of propylene and dim ethyl ether on alumina neaI thc critical

temperatures and up to saturation pressures. They obtained S-shaped isotherms at even 90

above the critical temperature. Brunauerl l _{concludes therefore th at}

multimolecular adsorption is possible even above the critical temperature; th is conclusion, however, is questionable. In the literature of adsorption these results are considered as rather anomalous, but so far nobody has verified them.

At low pressures, adsorption is determined primarily by the properties of the adsorbent surface (chemical nature, energetic heterogeneity, porosity and acces si bic surface area, lattice defects, impurities, etc.). At higher pressures, no doubt, thc nature of the surf ace is still important, but other factors like interaction among adsorbed molecules, capillary condensation and multimolecular adsorption become relatively much more significant. Complications mayalso arise here due to both adsorption and absorption taking place simultaneously. lf during the process of activated adsorption the adsorbed gas slowly dissolves into the adsorbent lattice as has been argued by BeecP4, or if there is a slow diffusion

of the gas along the grain boundaries of the crystal lattice as proposed by

Bcnton15 _{and Morozov}1G_{, then these processes of dissolution and diffusion should} be much more pronounced at higher pressures. Perhaps, high-pressure adsorption measurements may thus provide a straight-forward evidence in solving this highly controversial problem in activated adsorption.

Another unsolved problem is the course of thc high-pressure adsorption isotherm vis-a-vis the prediction from Polanyi's potential theory. The potential theory predicts a maximum at some intermediate pressure. According to the ordinary definition of adsorption, the amount adsorbed is the excess material present in the pores and on the surf ace of the adsorbent over and above that corresponding to thc density of the gas in the bulk ph ase at th at temperature and pressure. Obviously, if the gas is subjected to such high compression th at its density becomes equal to that of the adsorbed phase, the amount adsorbed measured experimentally, and calculated according to the above definition, must become zero. Hence the high-pressure adsorption isotherms must exhibit a maximum even by elementary considerations. Such maxima have actually been observed by von Antropoff et al17 _{for adsorption of A and N2 on charcoal up}

to 400 atm. pressure, by Coolidge and Fornwalt18 _{for C02, N20 and SiF4 on}

charcoal up to 100 atm., and by others. But Krischevsky and Kalvarskaya19 _have observed th at the adsorption decrease af ter the maximum is less than that predicted by Polanyi's theory. They have attributed th is to the difference in molar volumes of adsorbed and ordinary liquids.

The sorption of methane by coal at high pressures has been investigated

by several workers (cf. next Section). This still continues to be an important

problem due to its direct bearing on explosion hazards in c~al mines.

Every year hundreds of papers are being published àn the adsorption of gases at pressures below atmospheric, but the data at high

p~essures

are still very

scaTce. The main difficulty, in addition to the experimental technique, is "the

rapidly growing inaccuracy and uncertainty of measurements where sorption is

increasing only slowly with pressure, and the amount of gas or vapour in the poorly defined dead space within the apparatus is rapidly becoming the dominant quantity observed." 20

Thus, accurate adsorption measurements at high pressures are not only of direct importance to the chemical industry, they mayalso be of help in solving

2. A survey of earlier work on adsorption at high pressures

As mentioned in the last Section, adsorption at high pressures is still a scarcely investigated field. The earliest attempts have been to measure the sorption of methane on coal under pressure. In recent years, Palvelev21 and Khodot22 have obtained methane sorption isotherms for Russian coals up to 1000 atm. The apparent sorption they found either increases only very slightly with pressure above 200 atm, or declines from a maximum near this pressure to a limiting value which is weil above zero at 1000 atm. The maximum in methane sorption on coal has also been observed by van der Sommen, Zwietering, Eillebrecht and van Krevelen23 and by Moffat and Weale24. The isotherm actually levels off to a saturation value, if the amount adsorbed, measured experimentally, is corrected for the increasing density of the gas phase23. The general conclusion is th at methane is physically adsorbed on the large internal surface of coal, and is not chemisorbed or held in the type of solid solution formed by compressed gases in rubber and linear high polymers. The adsorbed am ou nt at 1000 atm is higher than what is to be expected. This may be due to the compressibility of co al increasing the pore volume at high pressure and hence changing the dead space in the apparatus. A more probable cause is the penetration of methane molecules at high pressures into spaces between the coal lamellae, which are not included in the dead-space determination with helium at 1 atm pressure. This interstitial adsorption is accompanied by a slight expansion of the coal structure which has been measured by electrical resistance strain gauges attached to suitably cut blocks of coa12_{4 . No expansion takes place in}

compressed helium, nor is it seen wh en the coal sample is encased in metal which mechanically prevents expansion. The sorptive capacity of co al itself is dependent on its ultrafjne structure, which has not yet been determined with certainty.

The earliest out-standing high-pressure adsorption work, however, is the investigation of McBain and Britton20 _{on the sorption of N20, C2H4 and N2}

on steam-activated sugar charcoal up to 50 atm pressure. The sorption balance developed by them has been further refined in the studies of Maass and co-workers12. 13 and Coolidge and Fornwalt18 (see the previous Section). But,

mounted as it is in glass, it cannot be used for pressures above 100 atm;

uncertainties also creep in into the results duc to the large buoyancy correction to be applied to the results at high densities. Adsorption units with an all-steel side to handle the gas under pressure and a glass side to measure the volume of gas on expansion to 1 atm pressure have been used by von Antropoff et aP7 (see p. 12), Frolich and White25 (H2 and CH4 on charcoal up to 150 atm), Moffat and Weale24 (CH4 on coal up to 1000 atm) and others. In most of these measurements the Bourdon-type pressure gauge has been used for measuring the pressure. The de ad space in the adsorption vessel has been determined in very unsatisfactory ways. For example, von Antropoff et al and Frolich and White have taken copper shot of the same volume as the coal sample and determined the so-called expansion isotherms ; the former have determined the density of coal using liquid nitrogen as the pyknometer liquid.

Ray and Box2G have measured the adsorption of H2, N2, CO, C02, CH~,

C2H2, C2H4, C2Hû, propylene, propane and n-butane on a single sample of cocoanut charcoal at 100 - 450°C in the pres su re range 0 - 14 atm. Adsorption equilibrium data for binary hydrocarbon mixtures on silica gel and several types of activated carbon up to 20 atm pressure have been published by Lewis, Gilliland, Chertow and Cadogan27. Quite recently, Jones, Isaac and Phillips28 have studied the adsorption of C02 and N2 on porous plugs of lampblack at 19, 30 and 32°C up to 80 atm, and Czaplinski and Zielinski29, of He, Ne and H2 on colloidal 13

silica and alumina at liquid nitrogen temperature up to 30 atm pressure.

Only very Iittle has been done on the chemisorption of gases on solids at high pressures. Emmett and Brunauer30 have studied the chemisorption of N2 on a doubly promoted iron synthetic ammonia catalyst up to 50 atm pressure. Their method consists in chilling the adsorption equilibrium from high pressure and high temperature to 1 atm and room temperature, reduction of the chemi-sorbed N2 by H2 and subsequent estimation of the ammonia so formed. Here desorption of N2 may occur during the chilling process. Furthermore, one cannot be sure that the whole of the chemisorbed N2 will be readily converted to ammonia by the hydrogen stream. Sastri and Srikant31_{• 32 have used an improved} vers ion of the direct volumetric static technique of Frolich and White25; the adsorption of N2 and H2 and of a 3H2 : 1N2 mixture has been measured on a commercial synthetic ammonia catalyst (Aerocat) at 97 - 400°C and up to 50 atm pressure. A similar volumetric method has recently been employed by Vaska and Selwood33 for simultaneous measurement of hydrogen chemisorption and specific magnetization up to 140 atm pressure on a supported nickel catalyst; chemi-sorption of H2 on nickel at room temperature is found to be essentially complete at ab out 100 atm pressure. .

The glass piezometer type of apparatus, used by van der Sommen et al23 for the study of the sorption of methane on coal up to 500 atm pressure represents a significant advance in technique. The calibration of the volumes of the piezometer bulbs with mercury and the use of the Michels pressure balance for measuring pressure have ensured much higher accuracy than in any previous work in this field. This method is, in a way, the forerunner to the new experi-mental technique des cri bed in this thesis.

An elegant method, which requires no compressor or oil press to raise the pressure, nor a gauge for measuring the high pressure, has been used by Vasil'ev34 _{for the adsorption of C02 on two types of silica at -85 to 40°C up} to 80 atm pressure. In this method therm al compression is employed to obtain high pressures. A known amount of gas, measured at 1 atm pressure, is trans-ferred into a calibrated steel U-tube, cooled in liquid nitrogen. The coolant is removed and the U-tube is now opened to the evacuated adsorption vessel, the helium dead space of which is already known. At equilibrium, the U-tube is isolated from the vessel and the gas in it is measured af ter expansion to 1 atm pressure. From these data, the amount of C02 adsorbed as also the equilibrium pressure are calculated using the accurate C02 compressibility data of Michels and Michels35.3_{6 .}

3. Scope of the present work

During the last four decades highly specialized techniques for the accurate measurement of

P -

V -T relations of gases up to 3000 atm pressure have been developed in the van der Waals Laboratory in Amsterdam. Taking advantage of these techniques, a new method for accurate study of the adsorption of gases on solids has now been developed. Two distinctive features of this method over the earlier high-pressure adsorption measurements are :

1. the adsorbent and the adsorbate gas are enclosed in glass and not in metal even at a pressure of 3000 atm;

2. the use of the free-piston type pressure balance ensures not only an accuracy of 1 in 10,000 in pressure measurements, it also maintains the pressure constant even if the amount of gas adsorbed varies with time.

Using this method the adsorption of N2 on a sample of alumina (B) has first been measured at 0, 50, 75 and 100°C up to 500 atm as a test case. This has proved the success of the technique as also the perfect reversibility of the adsorption.desorption cycle for the N2-alumina system in this temperature range. Hence it is foUowed by more systematic measurements of the adsorption of N2 up to 3000 atm pressure at ·7.6, 0, 25, 50, 75 and 100°C on an alumina sample (A) of larger surf ace area. A new and unexpected phenomenon obserYed in the N2 adsorption isotherms at high densities (in the pressure region 1500 - 3000 atm) has led to a further examination of high·pressure adsorption, in this case of CO on alumina A at 0, 25 and 50°C.

To cover the pressure range 1 - 6.5 atm, a medium.pressure adsorption unit has been set up; using this the adsorption of N2 on alumina A at O°C has been measured. An apparatus for low·pressure adsorption measurements has also been made during the above work; the adsorption of N2 at ·195.6, ·150, -125, -100, -78, 0 and 20°C and of A and CO at -78, 0 and 20°C on alumina A are also reported here. These medium and low·pressure measurements, however, are meant only as subsidiary to the main high·pressure work. Moreover, the literature of adsorption abounds with data for low·pressure and low-temperature studies. In particular, such studies for nitrogen adsorption and desorption have been extensively employed by de Boer and coworkers in Delft" for investigating the structure and texture of aluminium hydroxides and aluminas, obtained by different methods of preparation and extents of dehydration. Hence the results of the present low.pressure adsorption measurements have been discussed only when they become relevant in the context of the high·pressure adsorption resuIts; the experimental results and conclusions of the investigations in Delft have also come very handy for this purpose.

Molecular interactions in N2 and CO at high densities have earlier been the subject of experimental and theoretical studies in the van der Waals Laboratory. The present adsorption studies may in a way be looked upon as an attempt to know more about the behaviour of these two gases at high densities on the surface of an adsorbent, as se en from the stand point of molecular physics. The discussion in this thesis, however, is from the standpoint of adsorption and catalysis, seen in the context of present ideas in surface chemistry.

Classical treatments of adsorption like the Langmuir adsorption isotherm and the potential theory of Polanyi have been applied to the present high.pressure adsorption results. But these theories have only very limited success and they do not contribute any new insight into the present results. Hence they are not included in this thesis.

Simple thermodynamic methods have been employed to estimate entropy changes in the present adsorption results; from entropy data inferences can be drawn regarding the mobility of adsorbed molecules on the surface and also the changes taking pi ace in the surf ace layer when the surface coverage progressively increases due to more and more adsorption. The nature of physically adsorbed A, N2 and CO molecules on alumina has been compared with the behaviour of these gases when adsorbed on charcoal and on graphite under comparable conditions of temperature and surface coverage.

The importance of accuracy in determining pressure, volume and temperature in high.pressure adsorption studies can never be overemphasized. A detailed examination has been made here of the various corrections to be considered in

.) For instance, see the Doctorate theses (Delft) of Houben (1951), Fortuin (1955), Steggcrda (1955), Meijs (1961) nnd Lippens (1961).

studies of this type and of the possible approximations where the parameters involved are not known with certainty. Chapter III is completely devoted to this topic, still additional discussions about it have become necessary in parts of two other Chapters (VI and VII).

The concepts of differential and absolute adsorption in high·pressure adsorption studies have been discussed in detail and the data for both are given in Tables. These concepts have led to a better understanding of the high.pressure adsorption results and a consistent explanation for the second ascending and endothermic part of the

N2

isotherms and for the absence of it in the case of CO adsorption.

The nature of the complex and heterogeneous surface of partially dehydráted alumina has been briefly discusscd. Finally, some suggestions are given for further research in the field of high-pressure adsorption.

CHAPTER 11

APPARA TUS AND EXPERIMENT AL PROCEDURE A. ADSORPTION MEASUREMENTS AT HIGH PRESSURES 1. Principle of the experiment

A rigorous definition of the amount adsorbed, .6g', as obtained from

measurements at high pressures, is given below in order to indicate clearly the assumptions involved in calculating it.

First, for the dehamination of the helium density of the alumina used, a known weight of alumina is taken in a vessel of known volume. From th is volume and the amount of helium required to fill the vessel to a pressure not exceeding 1 atm at a given temperature, a volume apparently inaccessible to helium is defined by the ideal gas equation, PV = constant. The experiment is carried

out at various low pressures at 0, 25 and 50°C; within the limits of the

experimental accuracy, the values of the inacccssibie volume were constant. The

weight of the alumina sample divided by this apparently inaccessible volume is

called the helium density of the specimen·. The difference between the total volume of the vessel and the apparently inaccessible volume is referred to as the free volume.

For the investigation of nitrogen adsorption under high pressure a twin sample of the above alumina is placed in a piezometer of known total volume vp. Then a volume v 0 is calculated by dividing the weight of the alumina sample by the helium density as determined above. (In the procedure adopted it was impossible to measure the helium density of the actual sample used; this density was checked af ter all the high-pressure measurements were finished; it was found to be the same as th at determined earlier for the sample mentioned in the last paragraph). A known amount g of nitrogen is brought into contact with the . alumina and the pressure vs volume relationship is studied at constant temperature. The density p of the gas is taken from the known compressibility isotherms. Then the quantity .6 g' is defined as

(1) wh ere v is the total free volume of the gas phase in the piezometer.

For further elucidation of the principle the experimental set-up has to be

examined. The glass-piezometer technique for determining the compressibility

isotherms of gases has been described previously35, 36. For the present purpose

.) In Chapter IIl, a correction will be applied to th is helium density value to account for the dimension of the helium atom.

Fig. 1. The glass piezometer mounted in the high-pressure vesseI. Inset: the glass capsule containing alumina in the top bulb. The piezo-meter used for the present adsorption studies has 9 platinum contacts instead of 8 shown in this Figure.

some details of the glass piezometer had to be altered. The fjnal shape is shown In Fig. 1. It consists of two large bulbs at the lower end instead of one in the norm al high-pressure piezometer (cf. reference 36) while the top bulb contains a capsule filled with the dehydrated adsorbent and closed at the top with a thin

diaphragm. The diaphragm is made to break at an extern al pressure of 20 - 30 atm. F or the final measurement it is necessary to know the total volume outside the capsule and the free volume inside. These two volumes are determined separately. The determination of the free volume inside the capsule is described in the next Section. The volumes between the platinum contacts 1, 2, 3 ... 9 can be calibrated with mercury in the usual way3G, but this is not possible for the top bulb. This difficulty is overcome as follows : the piezometer is filled with nitrogen to obtain a pressure of about 5 atm at contact 1 at 25

°

C. The pressures are measured when the mercury reaches contacts 1 and 2; from these data, the known volumes below contact 9 and the known P·V·T relation of the gas, the volume of the top bulb (outside the capsule) as weil as the amount of gas can be readily calculated.

The pressure is th en raised til! the diaphragm of the capsule breaks and the adsorbent becomes accessible to the gas. The real experiment can now start, i.e., the determination of the pressure when the mercury reaches the different contacts. From these pressures p is calculated, and from it l>g'.

2. The glass capsule

The glass capsule is made to conform to the following requirements:

a) The dehydration of the adsorbent has to be carried out in the capsule at 300 and 600°C under vacuum (Super-max glass for 600° and soft glass for 300°C). b) To allow the pressure measurements at contacts 1 and 2 for determining the top·

bulb volume, the di::phragm should not break bel ow 20 atm pressure.

A B ) ( 0 hi9h YG<uum

_

Î

thin diophro9m O.04mm

Fig. 2. The glass capsule in its three stages: a) during pressure testing; b) during tbe dehydration of

alumina; c) in the final form.

c) To restrict the pressure drop on breaking, the diaphragm should break at a pressure not higher than 30 atm, and the volume of the capsule should not be large.

To determine the thickness of the diaphragm, the capsule is immersed in a .slightly coloured liquid of the same refractive index as the glass. The thickness can then be measured with a microscope. By a few trials, 0.04 mm at the thinnest part of the diaphragm has been found to be the optimum.

The capsule consists initially of two parts, A and B, separated by the spherical diaphragm (Fig. 2). The section A, later to be filled with alumina, is evacuated and sealed oH. The capsule is tested at an extern al pressure of 20 atm.

To determine the free volume inside the capsule, it is necessary to know (a) the internal volume of the empty capsule and (b) the volume, vo, of the alumina as defined earlier. The volume (a) is obtained fr om the extern al volume of the capsule and the weight and density of the glass, whereas V_{o is caleulated from the weight and helium} density of the dehydrated alumina. In addition, the loss of weight on dehydration is also determined. The sequence of steps involved in these determinations is the following:

(cf. Fig. 2) : .

1. B is sealed oH undei vacuum to prevent the collapse of the diaphragm during the subsequent evacuation of section A at high temperature.

2. The tip of A is opened and a glass tube is sealed to it; the capsule with th is extension is weighed along with some glass wool.

3. Through the extension tube, A is filled with alumina and the glass wool is placcd at C as shown in Fig. 2b. The who Ie is th en weighed. It is connected

to a mercury diHusion pump. '

4. With the diHusion pump working, the alumina "is dehydrated for 6 hours at the desired temperature. The vacuum obtained was of the order of 10-3 _{mm Hg;} the pressure inside the capsule, ho wever, can be higher than this value due to the liberation of water vapour from the alum;na and the diHusion limitations within the alumina column. The capsule is sealed oH at D while still under vacuum. 5. The capsule and the tube containing the glass wool are weighed together. Care

is taken that no alumina has been deposited on the glass wool. 6. Section B is cut oH at E, leaving the capsule in its final form. 7. The weight and extern al volume of the capsule are determined.

8. The density of the glass is determined for a specimen subjected to the same heat treatment as the capsule during dehydration of the alumi na.

9. The helium density of the alumina is determined on a conventional Iow-pressure adsorption apparatus using a twin sample of the alumina dehydrated simultane-ously and under the same pumping conditions as for the one in the capsule. 3. Choice of the adsorbent and the adsorbate gases

In the present study alumina has been chosen as the adsorbent not only because of its technical importance as a widely used catalyst by its own merit, as a carrier for metallic and oxide catalysts and as a promotor for many other catalysts (e.g., iron catalyst for ammonia synthesis, in which Iess than 1

% of alumina content can cover

up to 35 % of the total surface of the catalyst37), but because it has a highly porous structure and high adsorptive capacity. It has also the distinctive advantage of having very hard granules, which wil! not be easily deformed under pressure. Be,sides, the dehydration and low-pressure adsorption characteristics of alumina trihydrates have been studied in detail by de Boer and coworkers3_{8 •} 39; hence it wil! be possible to correlate their results with the high-pressure adsorption results, especially since the same type of alumina is used for this work also.

Two samples of Peter Spence Type H alumina have been used. The alumina has already been heated to about 600°C by the manufac:urers. For the present work, one sample of the alumina is dehydrated under vacuum for 6 hours at 300°C, the other at 600°C. The BET surface area has been determined using nitrogen at the boiling point of nitrogen. The samples have the following characteristics :

TABLE I

Characteristics of the Alumina Samples A and B

Alumina A

Dehydration temperature, °C

Loss of weight on dehydration (based on the dry sample), %

Helium density, g/cm3 _{(not corrected; see Chapter lIl, 5)}

BET surf ace area, m2_/g

Weight of alumina in lhe capsule for N!! adsorption, g

Weight of alumina in the capsule for CO adsorption, g 300 2.1" 3.3589 120.1 3.9657 4.7881 Alumina B 600 2.9 3.4273 94.2 3.2570

X-ray analysis has shown that both aluminu A and alumi na B consist of a mixture of y-and x-alumina .

• Further dehydration at 1200°C rcmoved another 1.29% by weight of water.

Nitrogen has been taken up to study pure physical adsorption on alumina in the temperature range 0-100·C, without any complications due to chemisorption. Subse-quently, the adsorption of CO on alumina A has also been measured at high pressures to compare it with the N2-alumina system. The physical properties of these two gases

are VEIy nearly alike, but there is a significant difference in the arrangement of their nuclear charges_ This difference must be reflected in the electron orbitals and, to

some extent, in the interactions between molecules, especially at high densities 40,41_

4. The high-pressure experiment

The piezometer consists of two bulbs with volumes of about 32 and 38 cc at the lower end, one bulb (10 cm long nnd 1.3 cm outside diameter) at the top and 5 or 6 bulbs varying in volume from 0.9 to 0.35 cc in between, interconnected by short sections of capillary tubing. The capsule containing the dehydrated and evacuated

alumina is enclosed in the top bulb (Fig. 1). The experimental set-up for P-V-T measurements is shown in Fig. 3. The piezometer is mounted in a high-pressure vessel,

evacuated and filled with nitrogen as described earlier3n The amount of the gas is to be such that it gives a pressure of about 5 atm at contact 1. The oil thermostat in which the steel container is placed is regulated at 25 ± 0.005°C. Dil is pumped in into the steel vessel from the press till mercury inside the piezometer reaches contact 1. The pressure is measured with a pressure balance and the temperature of the ther-mostat with a platinum resistance thermometer. The pressure is then raised till the mercury reaches contact 2 at about 9 atm. FIOm these measurements the top-bulb

volume (outside the capsule) can be calculated as mentioned in Section A 1 of this

chapter. The procedure is repeated with different fillings of the gas. Each set giv~s an independent value for the volume of thc bulb; the deviation from the average

amounts to 1 in 5,000. For the last filling (abaut 20 atm at contact 2) the amount of gas taken in is also calculated. During all these measurements the diaphragm of the capsule is still unbroken.

On compressing the gas from contact 2 to contact 3 the pressure increases about

fourfold. In this pressure range the diaphragm of the capsule collaps es and the

adsorbent becomes accessible to the gas.

Pressures are then measured at contacts 1 to 9 and in the reverse order at 0,

25 and 50 ° C. The pressures during the adsorption and desorption series agree in

general within 1 in 10,000 and hence no hysteresis is involved in this case. More gas is now filled in to go to higher pressures. For every new filling, pressures at contacts 1

and 2 have to he in the range already covercd by the higher contacts for the earlier filling; this overlap enables the calculation of the amount of nitrogen added. In this

way the pressure range is extended to 3000 atm. For the last filling of gas, measure-ments at - 7.6°C have also been taken. Af ter finishing the work at lower temperatures,

p

A

Fig. 3. The experimental set-up for P-V -T measuremenls of gases up to 3000

atm. pressure. A, Glass piezometer (mounted in the high-pressure vessel; for details see Fig. 1. The oil thermostat is not shown); I, valve for isolating the piezometer from the gas-filling side; L, oil press; M, steel capillary; Q, gas cylinder (pure gas); R, pressure balance; S, mercury-in-glass differential mlnometer for gas-fillings up to

10 atm. pressure; U, pressure gauge (gas, 0-100 atm); W, pressure gauge (oil, 0-4000 atm); V, connection to h·gh vacuum.

measurements at 75 and 100°C have been made. This Eequence of measurements is followed to avoid endangcring thc low-temperature measurements by the possible poisoning of the alumina due to thc enhanced vapour pressure of mercury at higher temperatures. However, a redeterm·nation of tbc adsorption at 25

oe

at the end of all olher measurements does not indicate any alteration in tbe activity of the alumina due to exposure to mercury vapours at 100"e for a few days.

Further details of the experimental tcchnique along wilh the exact figures for the volumes of the piezometer bulbs, thc total gas in tbe piezomcter at each gas filling, the corr;ctions applied for the thermal expansion nnd compressibility of the piezometer glass, et~., in the case of adsorption of N2 on alumina A and alumino 8, have been published elsewhere by Michels, Mcnon and ten Seldam 107.

5. Calculation of L'> 9

In the following, Pi denotes tbc pressure :md pi the density of the gas in thc piezometer wh en the mercury reaches the ith contact; v'i denotes the corresponding volume when thc capsule is unbroken.

As mentioned in the preceding Sec:ion, the pressures PI and P2 have been measured before breaking thc capsule, in order 10 determine the total amounl g of gas and to calibrate the top volume. The following rplations have been used :

g = P1!(PVh g = P2!(PVh L'>v = V'1

-

V'2

22

V'1 PI.

,

V2 P2 v,! V'2 (2) (3) (4)

where the quantity (PV) denotes the PV product of nitrogen in amagat units at tbe experimental pressure and temperature. Sin cc I:::. v is known from the calibration of the piezometer with mercury, v' 1, v' 2 and gare readily calculated.

For any measurement carried out af ter breaking the capsule the quantity

(Pi / (PV)i) (5)

is calculated where v f is the free volume inside the capsule as defined in Section 1. The method for determining this is given in Scction 2.

Since v'i

+

vf equals vp - V

o of eq. (1)

I:::. g' g - Pi (v'i

+

vf) = g - Pi vi (6)

where the densities Pi are calculated from the pressures Pi using the isotherm data of nitrogen determined earlier in the van der Waals Laboratory 43, 44, and vi

=

v'i

+

vf·

Then, I:::.g

=

I:::.g'/w, where w is the weight of alumina in the capsule. The quantity I:::.g is the apparent adsorption, cxpressed in cm3_{STP/g alumina.}

F or the second nnd higher fillings of gas, the pressures Pl and P2 must fall within the range already covered by the previous fillings. The I:::.g vs P plot for these preceding fillings is used for interpolation purposes. For example, the amount of gas g2 of the second filling is calculated from

1, 2 (7)

where I:::.g' Pi denotes the value of I:::.g' for the pressure Pi, taken from the

I:::.g vs P plot. By measurements at thc h;gher contacts the pressure range is extended furthtr; for instance, for the second filling I:::.g' is again calculated from

I:::.g' Pi Vi, i

=

3 - 9 (8)

The procedure for CO adsorption measurcmen(s is cxactiy the same as above. Isotherm data for CO have also been measured earlier in the van der Waals Laboratory40,41,

B. LOW- AND MEDIUM-PRESSURE ADSORPTION MEASUREMENTS

1. Measurement of adsorption at low pressures

A conventional low-prdsure adsorption apparatus (Fig. 4) is set up for thts purpose. A glass tube of Ulijform cross-scction of exactly 1 cm2, serving as a gJl; burette and provided with a manometric limb, is connected to a mercury manometer and a catalyst tube containing the adsorbent on one side. On the other side it is connected to a gas-filling manifold and a high-vacuum system. The gas burette is surrounded with a water jacket. The whole apparatus is set up in a room, maintained at a constant temperature of 20 ± 0.10

C. Mercury levels in the burette and its manometer limb are read with a cathetometer against a standard invar metre scale.

Tbe dead space in the catalyst tube and in the manometric space at different heights of mercury in the c10sed Iimb are calibrated using helium at TOom temperature Wh en the whole system is evacuated and a gas is taken in the burette, the amount of gas can be readily determined by measu:ing thc mercury levels at two different prcssurcs of the gas. On letting in gas to the cvacuated catalyst tube, tbe amount transferred is known from the volume and p:essure of the gas remaining in the burette, whereas the unadsorbed gas lefl ovcr in thc catalyst tube and in tbe manometric space is known from the pressure indicated on the manometer and the previous

calibrations. The difference between these two gives the adsorbed amount at that pressure and temperature.

For measurements at -78, 0, and 20°C, a large amount of alumina (38.72 g.)

is taken in the catalyst tube 50 as to make thc amount adsorbed high enough to measure accurately.

b

~

\0 high

vocuum

Room temp. 20 ! O,l·C

pure

H.

2. Measurement of adSorption at low temperatures

Fig. 4. Low-pressure

adsorption unit. The catalyst tube b is used for taking a larger quantity

of the adsorbent as in low-pressure adsorption measurements. The helium inlet of b serves to transfer a dehydrated ad-sorbent into b in an

at-mosphere of helium_

Low-pressure low-temperature adsOIption measurements have been carried out using the apparatus set-up for medium-pressure measurements. In tlIis, the gas burette is connected to a catalyst tube which is kept in a cryostat (Fig. 5). The temperature of the copper b~ock in the middle of the cryostat can be kept constant anywhere .between -188 and +50°C. The glass catalyst tube is connected through

copper capillaries (2 mm 1.D.) to a short mercury manometer which is provided with a sharp platinum contact in one limb, b. The mercury level in this tube can be kept exactly at the platinum contact (as checked with a Wheatstone bridge circuit) by manipulating the connecting valves to vacuum or to a nitrogen gas cylinder. In this way the volume of the manometric space in limb a is always kept constant. The pressure in limb b (same as in a, since the m:rcury levels in both limbs are equal) is measured on a mercury manometer, M, 310 ems long. This m3nometer can be used for measuring pressure directly up to 4 atm. Counter-pressure applied by means of the vapour pressure of methyl chloride at O°C (191.5 ems, measured experimentally) enables

Pt ,e9"lolor PI lhermometer

To gos buretto

r>---and hign

.

vocuum

I 5

,

3 4 2 ~ ...

•

b 'Lmm ,

~

To Wheetston. bridge Alumina To vacuum

-6 ~ M N2 M To vo cuum

~

~

Ic.

~

r

-

1

~M'lhYI

llhl)'"'"''

Cryostat ' -Room temp. 20 ! 0.1 °C (a) (b)

Fig. 5. a) Apparatus for adsorption measuremen:s at low temperatures and at

medium pressures (up to 6.5 atm). b) Details of the cryostat

measurements up to 6.5 atm. The whole apparaius is kept in a constant-temperature room at 20± 0.1°C.

The catalyst tube, containing about 6.5 g. alumina, and all the connecting glass and copper capillaries are first thoroughly evacuated. The dead space in the catalyst tuhe as also in the limh a and the copper capillaries connecting this to the valves 1 and 2 is calihrated with helium. During volume calihrations valves 2 and 3 are kept open at particular positions of the vake spindles; these positions are maintained for

suhsequent adsorption measurements also. Af ter the calihrations, the system is again evacuated. The cryostat is then cooled to the desireel temperature. Known amounts

of gas are introduced into the system from the gas burette. The level in limh b is kept

at the platinum contact and the pressure of gas over the adsorhent is noted from the manometer M. In this way, adsorption of N2 on alumina A has heen measured

at -150, -125 and _100°C anel 0-100 cm. Hg pressure. 3. Medium~pressure measurements

Into the low-temperature set-up descrihed in the last Section, a requisite amount

of nitrogen is transferred from the gas burettc when the alumina is at low temperature,

50 that on dosing valve 1 and raising the temperature of the cryostat to O°C, the desorbed gas in the catalyst tuhe can exert a pressure up to ahout 6.5 atm. This pressure can still he measured on the mercury manometer jf the methyl chloride

counter-prcssurc is applicd. Thcreafter known amounts of the gas from the catalyst tube are withdrawn into tbe gas burette through valve 1, noting the pressure of the remaining gas as before. Tb: s sct of measurements actuaJly constitutes a desorption

series. In practice, the low-pressure adsorption at -150°C is first measured and, on raising the temperature to 0 ° C, tbe desorption series at medium pressures is foJlowed in the same cycJe.

C.

PREPARATION OF THE CASES AND THEIR PURITY

a) Nitrogen Pure nitrogen is preparcd by the thermal dccomposition of dry sodium

azide in an evacuated high-pressurc bomb4~. The gas evolved is passed through a trap cooled in solid carbon dioxide and condenscd in a receiver bomb cooled in Iiquid air. A mass-spectrometric check does not show any trace of impurities.

b) Carbon Monoxide CO is preparcd by the thermal decomposition of nickel

carbonyl in the apparatus shown in Fig. 6. Because of the toxie nature of CO and still more of nickel carbonyl, the entirc apparatus is set up in a fume chambcr provided with a powerful exhaust fan. The apparatus is thoroughly evacuated before letting in the carbonyl into the distiJlation flask and subsequently f1ushed out a few times with CO. Nickel in the form of fine powder is caught in the first trap. Unreacted carbonyl (or that freshly formed from the nick el powder) and possibly traces of moisture and carbon dioxide are trapped in a condenser cooled by solid carbon dioxide and two others in series cooled in Iiquid air. Carbon monoxide is condensed in a stainless steel

.bomb cooled in Iiquid nitrogen and finaJly distiIled over into a brass bomb. Here also a mass-spectrometric check has shown no trace of impurities in the CO prepared. c) Argon The argon used is obtaincd from Philips Laboratory in Eindhoven. It is specified by the suppliers as spectroscopicaJly pure.

d) Helium The helium used for helium density and dead space determinations is the very pure gas obtained from Kamerlingh Onnes Laboratory, Leyden_

Gos somplo (

tor mass

SpoclrOCJroph

Fig. 6. Apparatus for the preparation of pure carbon monoxide.

-1

10 high vocuum

CHAPTER

111

CORRECTIONS TO BE CONSIDERED IN ADSORPTION MEASUREMENTS AT HIGH PRESSURES I . Ge.n.eral corrections

In normal P -V -T measurements at high pressures corrections are always made for the change in glass-piezometer volumes due to the difference between the volume calibration temperature and the temperature of the experiment, and the contraction of the piezometer volume due to the applied pressure. The pressure balance readings are corrected for the deviation of room temperature from the calibration temperature of the cylinder-piston assembly, and of atmespheric pressure from 76 cm mercury. Since the pressure balance does net directly register the pressure of the gas in the piezometer, a correction has to be applied to the observed pressure to account for the difference between the oi! level under the piston of the balance and the mercury head inside the piezometer; this cor-rection varies for the different contacts of the piezometer and slightly with temperature for the same contact, but it is practically independent of the pressure applied at any particular contact. The capillary depression of mercury in the piezometer capillaries has been taken into account as also the vapour pressure of mercury at higher temperatures (75°C and higher). But the enhanced vapour pressure of mercury at high pressures due to intermolecular forces (cf. Rowlinson and Richardson45 ) has not been considered in the P-V-T data determined in the van der Waals Laboratory.

In the high-pressure adsorption work an additional volume correction has been applied for the change with temperature of the volume of the glass of the capsule enclosed in the top bulb of the piezometer. Taking the coefficient of cubical expansion of soft glass (used for the capsule for alumina A) as 2.55 X 10-G cm3

_/O

_{C, the volume of the glass of the capsule is 3.4003 cm}3 _{at 25 ° C}

and 3.4068 cm3 _{at 100°C. The compressibility correction for this volume of glass} is quite negligible. The thermal expansion and compressibility of alumina have also been neglected; their order of magnitude is discussed in the next Section.

2. Thermal expansion and compressibility of alumina

The coefficients of therm al expansion and compressibility are not known for the particular alumina samples used in the present studies and hence it is not possible to apply any correction to the dead-space value to account for these two factors. However, the order of magnitude of these corrections can still be roughly calculated.

The volume expansion for alumina (corundurn) from 20°C to 100°C is 0.14%46. In the study of the high-pressure adsorption of N2 on alumina A, the volume expansion for the temperature range 20 - 100°C for 3.9657 g of alumina

of helium density 3.359 g/cm3 _{will be 0.00165 cm}3_{, if the above value for} corundum is applicable to partially dehydrated alumina. At 572 amagats (the highest density reached in the measurements at 100°C), this volume expansion can cause an error of about 0.9 cm3 _{in the observed adsorption of 23.5 cm3 , i.e.,}

an error of 3.8

%.

This may have some influence on the exact course of the

isotherm in its second ascending part at high densities. But th is 3.8

%

is the maximum limit, at lower densities and at lower temperaturcs the error will be much less.

For alumina A, helium densities have actually been measured at 0, 25 and 50°C. The values are 3.3471, 3.3589 and 3.3546 g/cm3 _{respectively; the} dis-crepancies among these values lie within the limits of experimental accuracy.

The compressibility value for alumina has not yei: been reported in the literature. For synthetic saphire at 30°C, in the general relation for volume change per atmosphere pressure,

-6V

V

_o

aP -

bp2,

a = 3.36 X 10- 7 atm-1 , but the value of b is not reported46_.

The order of magnitude of the compressibility of metallic elements and ionic lattices can be calculated (cf. Bridgman47) from the relation (per atmosphere)

- 6V

= 8.6 X 10-14 _(M/d)4/3

V

_o

where M is the atomic or molecular weight of the substance and d is the density. For alumina with a helium density of 3.359 g/cm3

- 6V

- V-

=

8.137

X

10-8

o

For 3.9657 g of alumina used in the high-pressure adsorption measurements the change in volume at 3000 atm will be

- 6V

=

3.9657

3.359

X 3000 X 8.137 X 10-8

and hence quite negligible.

2.88 X 10-4 cm3

3. Dimensional changes in rigid adsorbents on physical adsorption of gases on

them48

Chemisorption of a gas on asolid seriously alters the adsorbent surface. But, in physical adsorption, the concept of inert adsorbents is generally taken for granted mainly for simplifying any theoretical treatment into that of a one-component system. As Brunauer49 _{has pointed out in 1956, "the theoretical}

arguments advanced in favour of this assumption (of inert adsorbent) are inadequate, and experimental data wholly lacking" (See also Brunauer50 _and

Yates and Sheppard51_{). Volume changes in rigid adsorbents on adsorption have}

been studied by Meehan5_{2, Bangham and coworkers}53_{, Briggs and Sinha54 and} McBain et a155_{. More refined measurements have recently been carried out by}

Cam-bridge. The dimensional changes measured for rigid solids like active carbon rod or porous glass when rare gases are adsorbed on them provide direct evidence

that the assumption of inert adsorbents in physical adsorption is invalid.

From surface potential measurements Mignolet62 _{has shown in 1950 that} van der Waals films of non-polar gases like Xe, N2 and C2HG on nickel films may exhibit considerable surface potentials. In 1959, from studies of adsorption on an atomic scale rendered possible in the field emission microscope, Ehrlich and Hudda63 _{and Gomer}64 _{have measured the lowering of the work function of} tungsten on adsorption of A, Kr and Xe at temperatures bel ow 800

K. The work function diminishes monotonously with Xe coverage even af ter more than a single adsorbed layer is formed. The lowering of the work function caused by adsorption is interpreted as due to a polarization of the gas atoms by the dipole layer of the metal surface.

lnfrared spectra of adsorbed molecules also show that the symmetry of a molecule undergoes a drastic change on adsorption because of the asymmetric nature of the surface forces48 . The presence of new bands in the spectra of adsorbed molecules at frequencies similar to those found in the Raman spectra confirms the existence of induced dipoles. All these results conclusively show that the adsorbent is not totally inert in physical adsorption.

Any rigorous treatment of physical adsorption has to take into account the perturbations produced both in the adsorbed gas and on the adsorbent surface. In this context the excellent review on "Molecular Specificity in Physical Adsorption" by Yates48 _{deserves special mention. The present discussion, however,} is confined only to the possible change in the dead-space value caused by any dimensional change of the alumina on adsorption of gases on it.

Two examples may be cited here to show the order of magnitude of dimensional changes on physical adsorption. YatesG1c has shown that when a monolayer of argon is adsorbed on aporous glass tube, it expanded 6.85 . 10-cm from its original leng th of 5.1 10-cm. For the same change to occur by thermal expansion, the sample wil! have to be heated by 250°C.

As another example for the expansion of the adsorbent, the surface tension changes in a small spherical adsorbent particle may be considered. The surface tension y of an isolated solid must be balanced byelastic strains induced in the solid. For asolid sphere of radius r, Shuttleworth!l5 has shown th at the pres su re difference P across the interface is given by

2y

P -

-

-

.

r

For high-area solids with small "effective" radii, this pressure can be quite high, e.g., if r = 80 A and y = 780 dynes/cm (the value for silica),

P = 1.95. 109 _dynes/cm2

~ 2,000 atm.

Thus the small particle of silica in vacuum may be conceived of as being under a hydrostatic pressure of 2,000 atm caused by its own surface tension. This stress in a sphere of this si ze with a bulk modulus of 1.0 . 1011 _dynes/cm2 _{produces a} decrease in radius of about 1.2

%,

relative to its radius with zero surface tension. When a gas is adsorbed on such a highly strained surface, the reduction in y

brings about a relief in the stress and naturally the solid has to ex pand . Yates48 concludes : "the absolute length change due to adsorption is smal!, but in rigid solids it is not negligible in comparison with other processes affecting the size of the solid".

The alumina sample used in the present high-pressure adsorption studie5 has an average effective radius of 200 A for the particles39 . The surface tension value for alumina is not known. If it is of the same order of magnitude as that for silica, then the considerably larger size of the alumina particles must make the interfacial pressure in vacuum and hence the expansion on adsorption smaller. Contraction of the solid on adsorption has been found to occur in cases where hydrogen bonding can take place between the adsorbate and the OH groups on the adsorbent surface. The stronger the hydrogen bond formed, the larger is /the contraction. Since the surf ace of the alumi na used in the presem study is only partially dehydrated and contains 1.29 weight

%

of residu al water, presumably in the form of OH groups on the surface, there would be enough OH groups to cause such a bonding. But, neither N2 nor CO forms such hydrogen bonds.

A third type of dimensional change in an adsorbent on adsorption can be caused by the penetration of adsorbed molecules into the interlamellar space in the adsorbent, as in the case of sorption of methane on coal at high pressures discussed in Chapter I, Section 2. This is not to be expected in the case of a crystalline and extremely hard solid like alumina. In fact, this has been one of the main considerations in specifical!y choosing alumina as the adsorbent for the

present high-pressure studies.

4. The true density of an adsorbent

A serious critici sm of the present work arises from the use of the helium density of alumina to determine the dead space in the capsule and using this dead-space value without any correction to nitrogen adsorption measurements. The concept of helium density of an adsorbent is itself open to question. As Coolidge18 _{has pointed out, no experiment on the dis placement of any fluid by}

an adsorbent wil! give the density unless some assumption is made concerning the amount of adsorption that occurs, and a definition based on such an assumption cannot logically be used to determine the adsorption itself. Since the adsorption of helium on ordinary solid surf aces is in general negligibly smal! at room temperature and even up to 100 atm pressure66 , helium has been extensively used for dead space calibrations in volumetric adsorption measurements. There does not seem to be any better method for this purpose at present. Many microporous adsorbents, also perhaps alumina, may contain several tiny pores into which even helium cannot penetrate. Such inaccessibly smal! pores, crevices or cavities are of no significance for adsorption and related surf ace phenomena. In addition, there mayalso be pores which are closed at both ends. The solid density measured by X-rays, however, does not take into account all these pores and also not similar vacant spaces and vacancies of molecular dimensions inside the lattices of real cyrstals. Hence the X-ray density wil! be prohibitively high to be used for the calculation of the tme specific volume of solids in adsorption studies.

Recently, de Boer and Steggerda67 _{have discussed the corrections to the}

helium density of micro-porous substances arising from a) the dimension of the helium atom itself and b) the slight adsorption of helium.

These corrections become appreciable for adsorbents of high surface areas.

Corrections for these two factors can be accurately estimated only if the absolute surface area, S, of the adsorbent, the distance r between the cent re of the