R oy& i K . ^Jeuflon.
THE D A V I S O N CHEM ICAL C O R P O R A T IO N , BALTIMORE 3, M D .
A D S O R P T IO N m easurem ents for w ater on silica gel have been m ade, by b o th static an d dynam ic m ethods, over a range
* of relative hum idities in the neighborhood of room tem per
a tu re (5, 8). O ver th e lim ited range of tem p eratu re the adsorp
tio n a t a given P/P„ value appeared to be virtu ally independent o f tem perature. T he present w ork w as u ndertaken to obtain d a ta over a wider range of tem peratures; and it was hoped th a t some simple relation m ight be found to hold, a t least approx
im ately, betw een tem perature and pressure for a given ratio of w ater to gel.
As it has been considered th a t th e presence of perm anent gases is th e cause of adsorption-desorption hysteresis (7, 8, 11), it was decided to m ake m easurem ents by th e sta tic m ethod in the absence of air, using all-glass ap p aratu s w ith no stopcocks ex
posed to th e evacuated system (Figure 1). T h e sample was contained in bulb A , provided w ith a therm ocouple well. T he dead space above th e sam ple was small, th e volume between the to p of the sample and th e m ercury surface in th e connecting tube being 5 -1 0 % of th e (apparent) volume occupied b y th e gel.
Tw o tungsten contacts were sealed in to th e tu b e connecting A w ith th e rest of the apparatus, so th a t w hen m ercury completed a n electric circuit betw een them , its surface was in a reproducible position. T he sam ple tube was surrounded by a glass jacket, w ith asbestos packed over th e to p and in to th e b ottom ; th e jac
ket, in tu m , was surrounded by a tube furnace, which could be regulated to give the desired tem perature. D uring a ru n a slow stream of air was led downw ard through th e jack et to equalize th e tem perature.
Pressures were m easured on a closed-end m anom eter, B, pro
vided w ith a scale divided into millimeters. U pper lim it of the scale was about 900 mm. T h e m anom eter was so arranged th a t
th e space above th e m ercury could be freed of gas by raising the m ercury an d sweeping the gas through capillary tra p a t th e top.
A modified M cLeod gage, C, was used b oth for m easurem ent of low pressures of air or w ater vapor and for low concentrations of air in w ater v apor (9). T he comparison tu b e was closed, and the relative volumes were such th a t if w ater v apor was present in excess of about 20% of its vapor pressure, condensation occurred both in th e gage and in th e comparison tube when th e mercury was raised to th e reference m arks; accordingly, th e pressure of w ater vapor canceled out. T he reading of pressure of perm anent gas was tak en ju st as if w ater vapor were absent.
U n it D provided m easurem ent ofithe quantities of w ater tra n s
ferred to an d from th e sample. B y cooling, w ater could be con
densed in either of two calibrated tubes for m easurem ent; and in case a portion of w ater was being removed, it' m ight th en be dis
tilled in to one of the sm all bulbs and sealed off, for a gravim etric check.
T he u n it providing air-free w ater is shown in Figure 2 (1, 10).
I t was essentially a fractionating column; perm anent gases accum ulated a t th e top and were removed b y opening th e u n it periodically to vacuum. T he w ater in th e reservoir was circu
lated by a sm all flame beneath th e double-wall side tube. As the w ater in th e side tu b e w arm ed up , increasing vapor pressure forced it u p th e outlet tu b e; it could n o t re tu rn to th e reservoir through th e lower connection because of the floating spherical check valve. I t finally blew over, spreading in a th in film on the wall of th e reservoir; thereupon a fresh portion entered through th e check valve, and th e cycle was repeated. Passage of cool ta p w ater through th e condenser a t the top ensured th e continuous flow of w ater vapor up th e column.
In the absence of air, pressure-temperature measurements for silica Cox chart. In the absence of air, no hysteresis is detected for changes gel-water have been made, up to about 9 0 0 mm. pressure, with 1 of temperature or of composition. In the presence of 1 0 mm. partial to 3 0 % added water. The data give a straight-line relation on a pressure of air, no hysteresis is detected for changes of temperature.
650 I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y
The various units were connected to the m anifold through Y- trap s with floating spherical check valves (Figure 3). B y suitable m anipulation of the stopcocks, the m ercury was brought to and held at th e desired position.
METHOD O F O PER A TIO N
A weighed sample of silica gel, of known w ater content, was introduced into the sample tube and the la tte r was sealed.
Pum ping was then commenced and continued u ntil the partial pressure of air was low. As a general rule, in pum ping either the sample or the w ater unit, direct comm unication was not made betw een the u n it and th e pum p; rather, the m anifold and gage were opened alternately to the pum p and to the unit. Air was removed from the sample by alternate adsorption, and desorption a t elevated tem perature, of water, and required considerable time. In th e earlier stages the desorbed w ater was condensed in one of the m easuring tubes, the vapor phase being pum ped off w henever the p artial pressure of air built up enough to interfere w ith distillation. Later, when air removal was more nearly complete, th e w ater was distilled directly into the w ater u n it;
m ost of the air remained above the top plate and was transferred to the evacuated gage for m easurem ent alm ost completely by a single expansion.
D uring the course of air removal, it was observed th a t when w ater was removed from the sample by progressive increase of tem perature, the last portions a t the highest tem peratures carried the highest concentrations of air; the first fraction m ight show virtually none. T his is n ot in accord w ith the view th a t adsorbed gases should be displaced from the silica gel by the preferential adsorption of water.
W hen the air had been removed to the extent th a t a t no stage did its p a rtial pressure am ount to more th an a few thousandths of a millimeter, th e sample was brought to the chosen activation tem perature and held there for several hours while in communica
tion w ith the w ater unit, the pressure being noted. T his defined th e initial s ta te of the sample. An appropriate q u an tity of w ater w as th en condensed in a m easuring tube, and the sample was
cooled while open to th is m easured p ortion of w ater.
A fter it h ad been adsorbed, m ercury was b rought to and k ep t a t the reference contact, and pressure readings were tak en a t various tem pera
tures. As th e dead space be
tw een th e sam ple and the m ercury surface was small, th e graph obtained m ay be re
garded as an isostere. Suc
cessive m easured portions of w ater were added in th e same way, to obtain isosteres for v a r i o u s w a t e r c o n te n t s . W ater could also be removed from the sam ple; it was col
lected in a m easuring tube and th en distilled into one of the sm all bulbs an d sealed off for gravim etric determ ina
tion.
A t th e completion of tests on a given sam ple, dry air was ad m itted and th e sam ple tube was c u t from th e ap p aratu s to p erm it determ ination of the final w ater co n ten t of the sam ple; from this and from th e m easurem ents of quanti
ties of w ater added and subtracted, the ratio of w ater to sample corresponding to th e various isosteres was com puted. Values thus obtained could be com pared w ith those calculated from th e tem perature and pressure of a ctivation a t th e beginning of th e series.
T em peratures were tak en by therm ocouple, w hich was cali
b rated against a N ational B ureau of S tan d ard s certified therm om eter. A dditional check was obtained by tak in g pressure and tem perature readings w ith liquid w ater in th e sam ple bulb;
tem peratures from vapor pressure of w ater checked therm ocouple indication w ithin 1° C.
A t the tim e these experim ents were m ade, equipm ent for the precise autom atic control of tem p eratu re was n o t available, and m anual control was used. T em perature v ariatio n probably represents the m ajor source of error in th e d a ta ; low therm al con
d uctivity of the sam ple was a com plication as th e therm ocouple was a t the center and d id n o t im m ediately reflect tem perature changes of th e environm ent. Technique finally ad o p ted was to bring pressure to a chosen value by m anipulation of heating and to hold it there u n til d rift of therm ocouple reading ap p aren tly ceased.
In some of th e accom panying d a ta , pressures w ith ascending and descending tem peratures do n o t agree. A t first sight this m ight be tak en as evidence of hysteresis; b u t as th e higher pres
sures are observed w ith rising tem peratures and hysteresis would produce the opposite effect, th e explanation m u st be sought else
where. Incom plete a tta in m e n t of therm al equilibrium , in these cases, appears a plausible cause for the disagreem ent; w ith rising:
tem perature and tem p eratu re gradient from surroundings to- therm ocouple, th e therm ocouple a t the center of th e gel would show a reading lower th a n the average tem perature of th e sample.
Pressure m easurem ents are subject to correction for vapor pressure an d for lower density of m ercury in the m anom eter leg- connected to th e sam ple when it is a t elevated tem p e ratu re.
These corrections were n o t applied; th ey are opposite in sign and in m ost instances, probably of no greater m agnitude th a n v a ria tions caused b y tem p eratu re fluctuations. D a ta obtained are- presented in T able I.
July, 1945 I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y 651
Table I. Water Adsorption Measurements T ,
° C. P ,
M m . H g
- 1 % H o O
P / P o T P
i r , n
P / P o T P
0(W TJ ,A
,
P / P o T nor w„n_P P / P o T _ 1 c\0/ PT T - AP / P o
A /o o / o n j u ¿ /c JtiaU- O /o *12' - / “ - i u /o r i 2 U
-9 4 .5 4 . 0 0 .0 0 6 4 69 10 .5 0 .0 4 6 9 114 60 0 .0 4 8 9 50 7 . 8 0 .0 8 6 5 53 18 0 .1 6 8
121 1 9 .3 0 .0 1 2 6 97 48 0 .0 7 0 4 149 2 1 8 .5 0 .0 6 3 0 59 14.3 0 .0 9 9 0 60 28 0 .1 8 7
149 67 0 .0 1 9 3 111 9 2 . 5 • 0.0 831 168 387 0 .0 6 8 4 6 7 .6 24 0 .1 1 4 8 67 4 1 .6 0 .2 0 3
167 1 34.3 0 .0 2 4 3 126 165 0 .0 8 3 2 191 713 0 .0741 7 5 .5 3 6 .8 0 .1 2 4 8 6 8 .5 4 5 .4 0 .2 0 7
181 213 0 .0 2 7 7 138 2 4 3 .5 0 .0 9 5 0 84 54 0 .1 3 1 0 75 6 4 .2 0 .2 2 2
197 353 0 .0 3 2 2 149 3 4 6 .5 0 .0 9 9 8 - 4 % H sO ---. 94 86 0 .1387 86 108.4 0 .2 4 0
207 467 0.0347 157 465 0 .1082 71 19 0 .0779 105.5 140 0 .1518 9 4 .5 148 0 .2 3 3
221 650 0 .0366 162 560 0 .1 1 4 9 100 7 5 .5 0 .0994 117 216 0.1 5 9 5 104 230 0.263
232 862 0 .0 3 9 6 170 687 0 .1 1 5 8 126 2 1 3 .5 0 .1189 128 334 0 .1762 115 346 0.273
146 431 0 .1 3 4 6 138 447 0 .1744 132 602 0 .2 8 0
- 1 5 % H , 0 - - 2 0 % H iO — 167.5 643 0 .1525 146 587 0 .1832 137 699 0.281
4 0 .5 15.4 0 .2 7 2 41 20 .3 0 .3 4 8 168 825 0 .1457 154.5 756 0 .1906 143 869 0.294
57 3 7 .8 0 .291 47 2 9 .5 0.371 139 751 0 .2 8 5
62 5 0 .3 0 .3 0 7 56 4 8 .4 0.391 --- 2 5 % H , 0 --- 138 725 0 .2 8 3
71 84 0 .3 4 4 64 7 4 .8 0 .4 1 6 29 13.1 0 .4 3 6 45 3 2 .6 0 .4 5 2 131 570 0.273
81 128 0 .3 4 6 73 120 0 .4 5 1 29 13.3 0 .4 4 4 59 7 0 .7 0 .4 9 6 124 455 0 .2 7 0
89 184 0 .3 6 4 83 180 0 .4 4 9 39 2 3 .3 0 .4 4 5 71 132 0 .541
1 00.5 302 0 .3 9 1 9 4 .5 284 0 .4 5 6 47 3 7 .9 0 .4 7 6 81 207 0 .5 6 0
1 21.5 612 0 .3 9 2 103 407 0 .4 8 2 54 5 6 .5 0 .501 86 263 0 .5 8 2
125 697 0 .4 0 0 110 528 0 .491 60 7 7 .6 0 .5 2 0 100 427 0 .5 6 2
132.5 837 0 .3 8 3 111 675 0 .5 1 7 67 111 0 .5 4 2 110 630 0 .5 8 6
127 731 0 .3 9 5 113 613 0 .5 1 6 73 151 0 .5 6 8
119 538 0 .3 7 2 122 809 0 .6 1 0 84 236 0 .5 6 6 --- 30 % HsO
117 509 0 .3 7 6 125 905 0 .5 1 9 01 316 0 .5 7 9 24 1 2 .9 0 .5 7 5 28 14.9 0.526
116 491 0 .3 7 5 126 895 0 .4 9 9 9 4 .5 353 0 .5 6 8 37 2 8 .7 0 .6 0 9 64 119 0 .6 6 5
114 468 0 .3 8 2 122 772 0 .4 8 7 95 367 0 .5 8 0 42 3 9 .3 0 .6 3 9 79 238 0.699
111 396 0 .3 5 6 100 335 0 .4 4 1 103 514 0 .6 0 8 49 58 0 .6 6 0 94 435 0.711
1 0 5 .5 306 0 .3 3 2 108 637 0 .6 3 4 51 68 0 .7 0 0 26 13.7 0 .5 4 5
101 271 0 .3 4 4 26 6 . 6 0 .2 6 2 110 647 0 .601 62 111 0 .6 7 7 105 606 0 .6 7 0
96 220 0 .3 3 4 49 3 1 .7 0 .3 6 0 118 869 0 .6 2 2 70 170 0 .7 2 6 108 711 0.708
86 139 0 .3 0 8 73 115 0 .4 3 2 1 16.5 834 0 .6 2 5 82 276 0 .7 1 7
93 269 0 .4 5 7 103 468 0 .5 5 4 94 434 0 .7 1 0
109 530 0 .5 1 0 100 416 0 .5 4 8 106 662 0 .7 0 6
119 735 0 .5 0 9
ISOSTERES
P lo tted on a Cox c h art (Figure 4), a graph of log P vs. modified 1 / T (£, S, 4), th e points lie reasonably close to straig h t lines;
greatest deviations are a t th e lower pressures, where pressure readings are proportionately less accurate. A sim ilar graph is shown elsewhere (6), b u t w ith no inform ation as to source of d a ta or experim ental details.
Figure 2. Unit for Pro
viding Air-Free Water Figure 3. Y-Trap with Spher
ical Floating Check Valve
T he sam e d a ta are presented in another form (Figure 5), as isosteres on a graph of relative hum idity vs. tem perature. This plot gives som ew hat greater separation of th e isosteres. I t is very sensitive to errors in tem perature readings. Over the range of m easurem ents, th e relation appears to be linear; b u t linearity m u st obviously fail a t sufficiently high or sufficiently low tem peratures and pressures. I t m ay be speculated th a t th e approach to the 0 and 1 values of P /P o would be asym ptotic.
T he d a ta are typical of those obtained w ith this apparatus.
T he concentrations of w ater given are subject to some uncer
tain ty ; accidental admission of air to the sam ple on one or two occasions necessitated reactivation, and it was found late in the investigation th a t apparently th e activation procedure was not sufficiently prolonged to bring the w ater content to th e same value in all cases.
M ost of th e d a ta presented here are self- consistent in th a t they are from th e same series of m easurem ents w ithout reactivation, and the differences in w ater content are known from direct m easurem ent. In some of the runs after reactivation, the w ater content m ay have been different by as much as 2% from the supposed value. I n all cases th e d a ta for given iso
steres gave straig h t lines on the Cox chart, b u t th e straig h t lines were no t always coin
cident w ith others for supposedly the same w ater content. F or th is reason some of the late r d a ta are om itted, although the P -T values are probably b etter from improved technique and give sm oother graphs. This u n fortunate lack of precise definition of w ater content, however, does not in any way alter th e general p a tte rn of converging straig h t lines on th e Cox chart. The point of convergence is approxim ately 3.5 X 10‘ mm., 1000° C. Silica gel of another type was tested in the sam e way and gave similar re
sults; spacing between the isosteres and the point of convergence were slightly different from those shown here.
652 I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y
Figure 4 . Cox Chart
Per cent water is amount added to activated ( 5 2 5 ° F.) gel, based on weight of activated gel.
The u tility of th e Cox ch art lies in th e possibility of locat
ing an isostere w ith only two experim ental points or with only one if th e point of convergence is known.
Considerable atten tio n has been given to th e phenomenon of hysteresis in adsorption. Commonly, however, th e isotherm s have been involved; and it should be pointed ou t th a t the following observations involving tem perature changes are not directly applicable to isotherm s followed in th e usual way.
N o consistent tren d was detected betw een readings tak en with ascending and those tak en with descending tem peratures.
F urtherm ore, after a series of isosteres had been ru n w ith suc
cessive additions of w ater, the same q uantities were successively removed and check runs were m ade a t 30, 25, and 20% w ater;
the second set of isosteres was coincident w ith th e first set.
P oints from b o th sets are p lo tted together on th e isosteres.
W ith 5% an d again w ith 30% added w ater, air was adm itted to th e ex ten t of ab o u t 10 m m. p a rtia l pressure, and P - T m easure
m ents were taken. In b o th cases d a ta for ascending and descend
ing tem peratures were coincident; a t th e higher tem peratures, w here th e 10 m m. of air pressure was a sm all fraction of th e to ta l pressure, th e d a ta checked those tak en in th e absence of air.
T h u s in th is w ork no evidence of hysteresis in th e absence of air was found; an d th e presence of 10 m m. p artial pressure of air did n o t cause it, in changes of pressure w ith tem perature.
Aside from questions of precision and accuracy of th e d a ta here subm itted , an explicit sta te m en t should be m ade as to th e range of ap p licability. Silica gel is no t a substance of absolutely fixed
Figure 5. Isosteres of Relative Humidity vs. Temperature
composition and properties; ra th e r, by su itab le changes in m ethods of p reparation, corresponding v ariatio n s over an appreciable range can be produced in such p roperties as adsorp
tion characteristics and a p p aren t density. Accordingly, the above d ata, or adsorption isotherm s obtain ab le from them by cross plotting, are represen tativ e only of silica gel of th e specific ty p e em ployed; and th e sam e is necessarily tru e of an y pressure- tem perature-com position d a ta on such system s. I t is, th en , only th e ty p e of relationship (such as th e converging straig h t-lin e p a t
tern on th e Cox chart) th a t possesses an y degree of generality.
A C K N O W L E D G M E N T
T hanks are due E arl K . S eybert for th e p rep ara tio n of draw ings.
LITERATURE CITED
(1) B ruun, J. H ., In d. En o. C hem ., An a l. Ed., X, 212 (1929).
(2) Calingaert, G., an d D avis, D . S., In d. En o. Ch e m., 17, 1287 (1925).
(3) Cox, E. R „ Ibid., 15, 592 (1923).
(4) D avis, D . S., Ibid., 17, 735 (1925).
(5) D avison Chem ical C orp., R esearch D ep t., unpublished d a ta . (6) H ougen, O. A., an d W atson, K . M ., “ C hem ical P rocess P rin c i
ples” , P a r t 1, pp. 159, 160, N ew Y ork, Jo h n W iley & Sons, 1943.
(7) M cG avack, J . M ., Jr., and P atric k , W. A „ S m . Chem Soc.
42, 946 (1920).
(8) P a tric k , W . A., Colloid Sym posium Monograph, 7, 192 (1930).
(9) T ay lo r, R . K ., J . A m . Chem. Soc., 50. 2937 (1928) (10) Ibid., 52, 3576 (1930).
(11) Zsigm ondy, R ., Z. anorg. allgem. Chem.. 71, 350 (.1911).
Ph e b e n t e d before th e E l e v e n t h A n n u a l C h em ica l E n g in eerin g S y m p o s iu m (A d so rp tio n a n d I o n E x c h an g e) h eld u n d e r t h e au sp ices of th e D iv is io n of I n d u s tr ia l a n d E n g in eerin g C h e m is tr y , Am e b i c a n Ch e m i c a l So c i e t t a t C o lu m b ia U n iv e r s ity , N ew Y o rk , N . Y.