Adsorption isotherms by a rapid flow method

Adsorption isotherms by a rapid flow method

JOURNAL OF COLLOID SCIENCE 18, 65-72 (1963) ADSORPTION ISOTHERMS BY A RAPID FLOW METHOD Klaus Robert Lange The Philadelphia Quartz Company, Research ...

337KB Sizes 3 Downloads 57 Views

JOURNAL OF COLLOID SCIENCE 18, 65-72 (1963)

ADSORPTION ISOTHERMS BY A RAPID FLOW METHOD Klaus Robert Lange The Philadelphia Quartz Company, Research and Development Department, Primos, Pennsylvania Received April 30, 1962; revised May 28, 1962 ABSTRACT A simple flow m e t h o d has been developed for the d e t e r m i n a t i o n of adsorption isot h e r m s of high surface area solids w i t h nitrogen. The gas flow is controlled b y the pressure drop across a fine capillary in such a way t h a t t h e flow rate is m e a s u r a b l y indep e n d e n t of b a c k pressure. Comparison of surface areas using reference silica-alumina samples shows excellent a g r e e m e n t ; areas from p ore size d i s t r i b u t i o n s agree well w i t h those o b t a i n e d b y B E T calculations. A p p r o a c h to equilibrium is essential a n d has been shown to be v e r y close b y v a r y i n g t h e sample size of a 700 m.2/g, silica gel sample. Results are shown with Types I, II, and IV isotherms using silica gel, finely di-

vided silica, silica-alumina, and alumina samples. INTRODUCTION

In recent years there has been an evolution of flow methods to facilitate the determination of surface areas, pore size distributions, and pore volumes. The incentive for this lies in the length of time needed for the determination of the vapor-solid adsorption isotherm by the conventional volumetric method. Innes (1, 2) in 1951 devised a rapid method suitable for area and pore volume determinations, the latter in a semicontinuous way. The principle involved is that of a regulated gas flow into the sample holder--an automatic delivery burette, as it were. Because of the relatively high flow rates involved in that method (greater than 8 ml./min.) it has not been widely applied to the determination of the complete isotherm from relative pressure, p / p o = 0 to 1. Our method is designed so that lower flow rates (4-5 ml./min.) can be attained to allow the determination of complete isotherms so that accurate BET areas may be obtained. For pore size distributions we use the calculations presented by Cranston and Inkley (3). The particular calculation method used for area determination is that of Joyner, Weinberger, and 5/~ontgomery (4); this method is rapid and accurate once the necessary preliminary calculations have been done. The samples have been chosen to illustrate the kind of data obtained with isotherms of Types I (silica gel), and II (finely divided silica), and IV (silica-alumina and eta alumina) according to the BDDT classification 65

66

LANGE

(5). In the latter two cases we obtained adsorption isotherms independently by arrangement with the Atlantic Refining Co. (eta alumina) and Lukens Laboratory, Newton, Massachusetts (finely divided silica and silicaalumina). EXPERIMENTAL

Apparatus The schematic diagram in Fig. 1 shows the essential features of the apparatus. The nitrogen flows through the capillary at 25-40 p.s.i.g, forepressure. The capillary is constructed by flattening and crimping a 1/~" o.d. copper tube until a delivery rate of less than 6 ml./min, is obtained. We operate at about 4.3 ml./min. S.T.P. ordinarily, against 1 atmosphere back pressure. It is found that when this flow rate is established at the high fore-pressure, the delivery rate does not change with back pressure varying from 0.5 cm. to atmospheric pressure, in a measurable way. The gas enters the secondary manifold and sample bulb and is adsorbed by the sample. Pressure in the system is measured with a double-bellows recording pressure gauge as a function of time. The gas flow rate itself may be measured with a wet test meter or soap film meter, conveniently. The constancy of the flow rate is ascertained by allowing the gas to flow into an evacuated, known volume (2 1. is sufficient). The rise in pressure is then plotted as a function of time as a check. The measured flow rate is then used to calibrate the meter. The dead space is obtained by the Innes technique. The system is evacuated, the empty sample bulb is cooled down, and the gas is allowed to flow into the secondary manifold and bulb. The volume of gas is plotted against partial pressure and must be subtracted from the values obtained with a sample present. We also correct for the sample volume. TO '~tE'T "TEST

N~130-40PS, i ~ '

,--

X

;

CAPILLARY

(5

1~4E'TER

/ I

,+

I,

,k

~..TO MCLvD ~AUGE~

X~

To VACUUM

SAI~PLF-.

PR E ~ U R E GAUGE N i t r o c j e n f l o w reLte - 3 to G cc/mirL.

Fro. 1. Schematic representation of the flow apparatus.

ADSORPTION ISOTHERMS BY RAPID FLOW METHOD

67

Procedure The samples reported on were weighed as received, degassed by pumping at 10-5 ram. ttg for 16 hours and then heated at 300°C. until 10-5 mm. Hg pressure was re-established. During the pretreatment of the sample, the gas flow rate may be measured. After the pretreatment the sample bulb is cooled down with liquid nitrogen and shut off from the secondary manifold. The gas is then introduced through the secondary manifold into the fore-pump. An initial pressure surge is observed owing to the turning of the 3-way stopcock; a steady pressure of less than 5 ram. Hg is finally developed in the manifold in about 300 see. At this point the secondary manifold is shut off from the primary and the stopcock to the sample bulb opened. No further attention is needed until the system pressure reaches the desired final pressure, except to maintain a constant level of liquid nitrogen about the sample bulb. We generally allow the pressure to reach 0.95 p/po and vent the sample bulb to the atmosphere when the run is over.

Materials The silica xerogel, silica-alumina MSF-1, eta alumina are all available commercially, manufactured by Davison Chemical Co. (Div. of Grace Corp.). The 60 D&L silica-alumina was steamed MSF-1, obtained from the Atlantic Refining Co. The finely divided silicas are also obtainable from the manufacturers. The nitrogen used is Mathieson Co. Prepurified (99.996% N2 min.) and does not require further purification. ]DATA AND RESULTS

Areas Table I shows a comparison between multipoint BET areas obtained with our samples in comparison with those determined by the suppliers. The conventional method used was the volumetric technique. The values in brackets show the number of determinations made; no error TABLE I Sample

D a v i s o n MSF-1 SIO2-A1203 S t e a m e d MSF-1 Quso F-20 F i n e l y divided silica

Surface area (m.2/g.) Flow method

Conventional

496 (4) 161 (3)

496 (100) 161 (2)

292 (2)

292 (1)

68

LANGE

limits have been shown because of the low number of runs we made with each sample. It can be mentioned that, with the flow system, the spread was never more than 4 m.2/g, and that the steamed catalyst was measured with both the volumetric and Sorptometer with a spread of 2 m.~/g, between the two. The conventional data on the fresh MSF-1 were obtained by the manufacturer, those on the steamed catalyst at the Atlantic Refining Co., and those of the Quso by Lukens Laboratory, Newton, Mass.

Isotherms In order to establish reproducibility, two isotherms were run on separate samples of Alcoa F-10 alumina four months apart. Figure 2 shows the data obtained indicating individual calculated points on the isotherms. The two runs also involved the use of two sample bulbs of different volumes, so that the dead-space corrections were different. In order to establish accuracy, a sample of Davison eta-alumina was run on our apparatus and the data compared to unpublished data obtained by use of a quartz torsion balance (6), a static technique. The agreement is excellent below 0.4 pipe and above 0.85 piPe (see Fig. 3). At intermediate values the flow method gives a slightly higher adsorption. Since the flow is a nonequilibrium method, the deviation ought to be in the opposite sense. This means that the flow method is very close to equilibrium with this

14.0,

,~ t?.O.

ALCOA

J

F-tO A L U M I N A

x

/

o

x

x" x*

tJ e~::), XI

,n

~ 6o

xe x eX

o 0 0

o

x* oXO Xo o X e

5~ 2O

oX

X °X°~X

oO Xe

0 0 P-.ELATIVF_. P~ESSUIP--E P/Po F i e . 2. Reproducibility of t h e flow m e t h o d and comparison w i t h gravimetrie. X • Flow method, separate runs, sample wt. = 0.9333 g. and 1.3576 g.; N2 = 4.47 m l . / m i n , and 4.43 ml./min., respectively. © Torsion balance data.

ADSORPTION

ISOTHERMS

BY RAPID

FLOW

69

METHOD

IGO

c~

0

,DAVISON

140

ETA

~,LUI'IIt',IA...

X X X o

I~,0. d d r~ i,i

X X X

IoO

X 0

~n

X

<

×

× 0

GO_

XO

Z X

I,i

X

Z 40.

0 X

D .J

>o

X ~.0.

0 0

d,

d~

ds

de

~ ELAWIVE

o:s

o:~

P~SUI~.E

o~7

o~

do

,.b

P/Po

FIG. 3. Comparison of gravimetric and flow methods. X Flow; © torsion balance. Type IV isotherm and that the differences noted are probably those involved in testing samples in different laboratories and by different methods. Davison's silica xerogel was used in order to investigate further the question of approach to equilibrium in the flow system. Figure 4 shows the data obtained with this sample which has an area of 710 m.2/g. Table II summarizes the results. The data go in the expected direction for a nonequilibrium method. It is quite certain that equilibrium is attained up to the condition of 700 m. 2 total area present. The upper portion of the isotherm is more sensitive to sample size, as expected, but quite acceptable when one considers that the total difference between runs 1 and 3 represents only 6 % of the total amount of gas finally adsorbed. We were particularly interested in evaluating Type II isotherms and had a volumetric determination made by Lukens Laboratory. This comparison is shown in Fig. 5. The two methods agree up to p/po = 0.45. After this value the data tend away from each other increasingly. On the other hand, in Fig. 2 the open circles showing torsion balance data on F-10 alumina indicate a deviation in the opposite sense on the part of the static data. Our only conclusion is that Type II isotherms present difficulties when it comes to attainment of equilibrium, no matter which method is used, because of the asymptotic nature of the high pressure region.

70

LA~OE Z50

i,

EO0

150 O ~o

|00.

D J

>0

50.

O

,, ,,

o

,[

o'z

d4

RELATIVE

d~

o:~

PRESSURE

~.o P/?o

FIG. 4. The effect of sample size on flow data: adsorption of N~ on SiO2 gel at - 196°C. TABLE II Run 1 2 3

Sample Wt. (g.) N2rate(ml./min.) 0.400 1.000 2.000

4.41 4.28 4.26

V~(ml.)

Pore vol.(ml.g.)

163 163 154

0.38 0.37 0.36

Pore Size Distributions

We have used the d a t a of Cranston and Inkley (3) to calculate pore size distributions using a digital computer. The samples used for this calculation all give T y p e I I isotherms and most of the pores are between agglomerated ultimate particles, rather t h a n in the ultimate particles themselves. Table I I I shows the areas obtained b y the multi-point B E T calculation as compared to the pore size distribution area. The agreement between the calculated areas b y the two different methods, as shown in Fig. 3, gives us confidence t h a t this method does attain equilibrium with T y p e I I isotherms over the whole range from p/po = 0.94, which is the range covered b y the P S D calculation.

ADSORPTION ISOTHERMS BY RAPID FLOW METHOD

QU

~:%5o-

50 F-~0)F'II,4[I..Y D|VII,,C~/:51LICA vED

X

71

/

d 0 a

~ loo. o

z

I~ 5 o _.1 0

.~,

o

.~,

~LATIVE

.8, I='I~.ESSUI~

P/Po

FIG. 5. Comparison of volumetric and flow methods. •

.6

,.o

Flow; X volumetric.

TABLE III Area (m?/g.) *

Sample

Cab-O-SilM5 HiSilX-303 Syloid 308 HiSil 233 Quso F-20 Quso G-30

BET

PSD

194 169 256 119 292 281

210 186 252 107 291 278

* Approx. 0.5 g. sample used, N2 rate 4.2-4.6 ml./min, for all samples.

CONCLUSIONS The data indicate that for isotherms of Types I, II, and IV a flow apparatus, in which the nitrogen flow is regulated by a fine capillary, gives isotherms of sufficient accuracy to be of general utility. Surface areas may be determined with an accuracy equal to conventional methods while the higher relative pressure portions of the isotherm are within the deviations to be expected from conventional volmnetric or gravimetric techniques.

72

LANGE

The savings in time and experimental manipulations are great--a typical determination taking from ~ to 11/~ hours instead of several days. ACKNOWLEDGMENTS I would like to acknowledge the support for this work by the Management of the Philadelphia Quartz Company, who kindly allowed these data to be published. My thanks go also to Dr. Hans L. Gruber of The Atlantic Refining Company for allowing the use of the torsion balance data. REFERENCES 1. 2. 3. 4.

INNES, W. B., U. S. 2, 729, 969, Jan. 10, 1956. INNES, W. B., Anal. Chem. 23, 759-763 (1951). CRANSTON,R. W., AND INKLEY, F. A., Advances in Catalysis 9, 143-153 (1957). JOYNER, L. G., WEIN:BERGER,E. B., ANDMONTGOMERY,C. W., J. Am. Chem. Soc. 67, 2182-2188 (1945). 5. BRUNAUER, S., DEMING, L. S., DEMING, W. E., AND TELLER, E., J. Am. Chem. Soc. 62, 1723 (1940). 6. GRvnER, It. L., Personal communication, Atlantic Refining Co., Research & Development Department.