SORPTION OF MTROUS OXIDE AND SULPHUR DIOXIDE BY GLASS D . H. BANGHAM AND F. P. BURT
I n a previous paper1 experiments on the sorption of ammonia and carbon dioxide by a glass surface a t oo C. were described, and the time and pressure relationships were discussed. Similar experiments have been carried out with nitrous oxide a t oo C., using the same tube of glass wool as sorbent*; and the results confirm substantially the conclusions arrived a t from the carbon dioxide series. Nitrous oxide as supplied for anaesthetic purposes(that is, practically free from other oxides of nitrogen) was collected in a gas-holder over aqueous potash. After drying over calcium chloride and phosphoric oxide it was solidified at liquid air temperature and freed from permanent gas by means of the
FIG.I
mercury pump. After fractional distillation a product was obtained which had a melting-pressure of 662 mm. With this gas sorption experiments were carried out at a series of pressures ranging from 44 mm. to 640 mm. Every effort was made to obtain as large a time-range as possible, and in one case the observations extended from 5 2 seconds to 5 days. The sorption values and times together with the momentary pressures are given in Table I, and the log s, log t graphs derived therefrom are shown in Fig. I . For corresponding times and pressures the sorption values for nitrous oxide are invariably smaller than for carbon dioxide. Two 2
J. Phys. Chem. 29, 113 (1925) The surface of the sorbent was very nearly 3 square met,res.
SORPTION BY GLASS
TABLEI Nitrous oxide. Time-sorption measurements 24. July 1922 (2nd Exp.) Time (mins) 25
35 50 65 *
85 131 I95 1225
I510 2655
P (mm. of Hg)
1g.Oct. 1922 (5th EXP.1 S
(cc. @ N.T.P)
640.67 640.52 640.29 639.98 639.67 639.37 638.80 636.81 636.63 635.97
1.625 I .656 I .689 I . 722
1.752 I . 809
1.855 2.085 2.119 2.200
19.July 1922 (I st Exp)
1.5 4
7 I1
I4 I7 23
28 31 35 40
50 65 85 113 148 268 1290 I655 27-42
382.40 381.65 381.18 380.85 380.59 380.40 380.06 380.04 379.89 379.80 379.73 379.61 379.40 379.17 378.90 378.72 377.88 376.17 376.29 375.46
I.
117
1.196
Time (mins)
1.96
7 IO
I3 16 I9 24 29 34 39 49 59 79 119 I470 2819
P (mm. of Hg)
( c c . @ N.T.P)
386.74 385.77 385.52 385.24 385 ' I 4 384.95 384.80 384.62 384.54 384.39 384. I 3 384.02 383.73 383.32 380.63 379.87
1 . I54 1.254 I . 281 1.313 1.324 1.346 I ,362 1.378 1.389 I ,406 1.433 1.446 1.472 1.514 1 ' 792 1.872
5
1.251 I .287 1.314 1.332 1.371 1.372 1.389 I .401 I ,411 1.424 1 * 450 1.479 I . 506 1.525 1.587 I . 760 1.790 1.847
1I.Oct. 1922 (4th EXP) 0.86 2.65 4.7 8 I1
16 21
31 41 60 80 140 I73 I 400 2920 4225 7150
180.73 179.90 179.53 179.06 178.90 178.60 178.40 178.17 177.94 177.62 177.41 176.96 176.78 174.79 174.09 173.75 173.23
0.723 0.800 0.832 0.877 0.891 0.920 0.938 0.959 0.980 I .OIO I ,027 I . 069 I
.085
I . 262
1.336 1.370 1.419
D. H. BANGHAX AND F. P. BURT
542
TABLEI (Continued) Time (mins)
P
S
(mm. of Hg) (cc. @ ?rT.T.P)
Time (mins)
26. Sept. 1922
1.17
4.50 7.25
9.1 14.2 17.3 23 30 40 60 90 I 28 132 I325 2660
106.59 106.38 1 0 5 97 105.67 105.45 105.23 105.12 105.00
104.85 104.60 104.32 104.04 103.85 103.81 101.98 101.33
S
(cc. @ N.T.P)
30th Oct. 1922 (6th Exp.)
(3rd EXP.) 2.17
P (mm. of Hg)
0.590 0.608 0.644 0.670 0.685 0.709 0.717 0.728 0.741 0,762 0.786 0.810 0.827 0.830 0.987 I ,042
I .67 4.25 7.60
10.75 I4 I9 24 29 39 49 59 79 119 186 264 I 400 4205 5640
43.93 43.63 43.35 43 , 2 3 43.15 42.96 42.90 42.84 42.69 42 ’ 54 42.44 42.37 42.22 41.96 41.76 40.86 40.21 40.04
.
0.368 0.392 0.416 0 * 425 0.432 0.447 0.452 0.458 0.469 0.482 0.490 0.495 0.507 0.529 0.544 0.617 0.671 c .685
experiments were .performed at about 3 -atmosphere pressure; namely, the first experiment of the series where the initial pressure was 382.4 mm., and the fifth experiment where the initial pressure was 386.7 mm. The excellent agreement between these two precludes the possibility of any serious fatigue effects having occurred in the interval. I n the graph, a single log s, log t straight line is drawn to represent the course of sorption a t an average pressure.
As in the case of carbon dioxide the slopes of the log s, log t lines vary with the pressure. A table of “experimental” reciprocal slopes is given below (Table 11.) together with the “corrected” values deduced for sorption under constant pressure. These last were arrived a t by applying the same methods of correction for falling pressure as were used and described under carbon dioxidel. For such “corrected” times and sorption values the symbols t, and s, are used. The Pressure-Sorption-Time Relationships for Nitrous Oxide The pressure-sorption relations for nitrous oxide are represented in Fig. 2 . The heavily-drawn curves are derived from the experimental sorption- and pressure-values for t = I min. and t = 3000 mins., respectively (black circles). The corrected sorption values for t, = IO, 100,1000 and 3000 mins. are del
J. Phys. Chem. 29, r r g (1925)
5 43
SORPTION BY GLASS
TABLE I1 Values of m
=
6 log t 6 log s
m =-6 log s 6 log t from exoerimental log s, log t graphs
641.9 384.6 180.7 106.6 43.9
61og t from corrected log s, log t graphs.
15.35
15.35
1j.I
1j.I
13.7 13.4 13.05
13.4 12.95
12.4
FIG.2
noted by white circles through which fine lines are drawn to indicate the course of the “corrected” log s,, log p curves. For t = I min., corrected and experimental curves coincide, and the differences are very small even for t = 3000 mins. It was realized when too late to repeat it that the experiment at 642 mm. had yielded results somewhat a t variance with the others: thus, in Fig. 2 , the highest-pressure points fall considerably below any reasonably-shaped curves drawn through the remainder. Since the neighbouring % -atmosphere points are the results of two separate experiments in excellent agreement, it seems probable that a gross error of some 90 cmm. (such as would have resulted from misreading a millimetre division on the scale) was made in the measurement of the gas introduced, in this experiment. The value of the index m (15.35) also points to an error having been made, since it represents a disproportionately small increase on the %-atmosphere value (I 5 . I). The remainder of the points in Fig. 2 lie very satisfactorily. Table 111 gives data derived from the log s,, log p curves: in compiling it the pressure region beyond the well-substantiated -atmosphere points has been neglected. In column (I) are given selected ordinate values, and in columns (2) to (6)
D. H. BAXGHAM A S D F. P. BURT
544
TABLE I11 Nitrous oxide.
Data from corrected log s,, log p curves.
pairs of curves)
0 . 2 6 -0.16 0.06 2.620 I . 9 6 2.418 I .86 2 . 2 2 5 I .76 2.035 I . 6 6 I . 850 _I . 5 6 I .670 -
I-
-
-
--
2.625 2.390 2.173 1.965 1.770
547 2 , 4 7 5 2.325 2.270 2.115 2.077 1.922 1.887 1.735 1.702
2'
2.553 2.320 2.098 1.890 1.700
-
- -
- 0.157 0 . I45 0.148 0.148 0.148 0.148
0.072 0.070
0.150
0.152
0.155
0.150
0.075 0.075
0.155 0.152
0.070
0.152
-
0.153
0.153
0.072
6.5
6.5
6.6
-
MeanAlogpl 0.147 Alog t, whence(-) A h p 8, 6.8
=I
~
For over-all range from t,
= I
to t,
= 3000,
corresponding abscissa values, for log t, = 0.0, I .o, 2.0, 3.0 and 3.477 respectively, while columns ( 7 ) to ( I O ) show the values of the horizontal intercepts between adjacent curves. These intercepts are sensibly constant in each vertical column, and, as in the case of carbon dioxide, are almost exactly proportional to the corresponding differences in log t,. For the over-all range, the value of the constant
(i $ ~
:)sc,
which may be regarded as the quotient is 6.6.
As in the case of carbon dioxide, this relationship makes it possible to express log s (derived from the different sorption experiments) as a single-valued function of the complex variable 5.6 log pt
+ log l jp.dt.
The continuous
line obtained on plotting these variables shows only slight curvature over a large range of sorption values.
SORPTION BY GLASS
545
Desorption Experiment with Nitrous Oxide Some of the results of this experiment (performed by reducing the pressure in successive stages, following protracted sorption at a higher pressure) have already been published'. It was found that if sufficient time2 was allowed for desorption at each stage, the residual sorption values were related to the corresponding pressures by the Freundlich equation sn = k'p, n having the value of 3.2. Time-desorption measurements with nitrous oxide confirmed the conclusion previously arrived at3 that sorption and desorption follow similar time equations. In Fig. 3 the logarithms of the quantities of gas disengaged are plotted against the logarithms of the times reckoned from the moment of each pressure reduction. Early readings were taken in only one or two instances but, where the observations were few, the slopes could be gauged from those of the neighbouring lines. The coefficient r of the equation r log p = log t constant, (where j- is the quantity of gas desorbed in time t ) appears to decrease considerably in the later stages of desorption. All the gases so far investigated have yielded s, p values obeying Freundlich's equation under the conditions of our 115 desorption experiment. It still remains doubtful, however, whether this fact is indicative of a near approach to equi0 IO 20 3.0 librium conditions, or whether me are FIG.3 dealing with a fictitious regularity conditioned by the procedure adopted. While the rate of desorption was practically undetectable after 24 hours, it is noteworthy that a t the end of the first desorption step the s value passed through a minimum and began to increase at a very small but measurable rate.
+
With the hope of throwing some light on this question, the pressure, which at the end of the desorption experiment had been reduced to 9.86 mm. was suddenly raised to 38.7 mm. by the introduction of more gas, and the progress of the renewed sorption mas followed ovei a period of 3 days. The results are shown in Table IV. 'Proc. Roy. Soc. 105A, 487 (1924). The usual time interval allowed between successive pressure reductions was 24 hours, but on two occasions the system was allowed to stand over the week-end. This treatment had no apparent effect on the regularity of the results. 8 J. Phya. Chem. 29, 113 (1925).
D. H . BANGHAM AXD F. P. BURT
546
In Fig. 4 the values of log (sorption increment) are plotted against log t (time being measured from the moment of increasing the pressure). Not only are the early points seen to be closely linear, but the line drawn through them has a reciprocal slope of 13.7, in fair agreement with the value 13.05, found for the 43.9 mm. sorption experiment, starting with gas-free wool. I n contrast, however, to the latter experiment, the line becomes practically parallel to the log t axis after one day's sorption. The fact that the sorption appears to be tending to a limit in the immediate neighbourhood of the value assessed on the basis of the desorption results, using the Freundlich equation as an interpolation formula, (0.938 cc.), suggests that the applicability of the equation is not a fortuitous concomitant of the conditions imposed.
FIG.4
FIG. j
TABLE IT' Time from moment of increase of pressure (mins) 0
I
4 9 I2
I5 I9 24 30 48 81 142 217
1310 43 7 0
P
(mm. of Hg)
9.87 38.72 38.50 38.27 38.26 38.21 38.17 38. I3 38.07 37.96 37.82 37.77 37.64 37.30 37.11
S
S
(total) (cc. @ N. T. P)
(increment) (cc. @ N.T. P) -
0.627 0.808
0.825 0.844 0,845 0.849 0.853 0.855 0.860 0.869 0.881 0.885 0.892 0.922 0.924
(Freundlich equation interpolation value = 0.938 for p = 37.1)
0.181 0.198 0.217
0.218 0.222
0.226 0.228 0,233 0.242 0.254 0.258 0.265 0.295 0.297
547
SORPTION BY GLASS
The results of this experiment stand in contrast to one performed with carbon dioxide and discussed in our previous paper. In the latter the pressure was increased a t a moment when sorption was still progressing a t a measureable rate. Under these conditions the values of log (sorption increment) and log t yielded a graph initially very straight but afterwards becoming convex to the log t axis. This is shown in Fig. 5 . The direction of the departure of the later points from the linear appears, therefore, to depend on whether the rate of sorption, ds/dt, was positive or negative at the moment of pressure increase. An analogous conclusion was reached in the case of desorption1.
Experiments with Sulphur Dioxide This gas was obtained from a syphon of liquid sulphur dioxide in which the chief impurity is air. After passage over phosphoric oxide the gas was solidified in a fractionating vessel cooled in liquid air. The accumulating permanent gas was removed at intervals by a mercury pump. The solidified gas was then sublimed into a second vessel, a further quantity of permanent gas being
FIG.6
FIG.7
evolved during the process. Continued repetition of this treatment was required to remove the last traces of permanent gas which are very obstinately retained. The solid was then allowed to liquefy and further purified by fractional distillation. A middle fraction was introduced into an evacuated storage bulb. The melting-pressure of this was gas 13.2 mm. The first experiment with sulphhr dioxide, performed at an initial pressure of 376 mm., gave results quite in keeping with those for other gases, the log s, log t graph being strictly linear over the whole time range from t = %-minute to t = 4 days. This graph is represented by the upper line in Fig. 6, while the corresponding s, t values, together with the momentary pressures are given in Table 17. (experiment I ) . The reciprocal slope of the log s, log t graph is 16.7, a somewhat higher value than that found for the gases previously examined. -4t the end of four days, the pressure was suddenly reduced by about one half, and very careful time-desorption observations were taken. The graph in Fig. 7, where log ( ( { = quantity of gas desorbed) is plotted against log t (t = time from moment of reducing pressure), illustrates clearly the passage of the sorption value throuch a minimum (over-night). The straight line joining the three early points has a reciprocal slope of 2 2 : (compare values ranging from 29 to 16 in the case of other gases). Proc. Roy. SOC.105A, 487 (1924).
D. H. BANGHAM AND F. P. BURT
548
Further pressure reductions were subsequently made but no more time observations were taken. The final s, p values obtained for each step have already been published'. They show close agreement with the Freundlich equation, the value of the index n being 10.7. A repetition of the &atmosphere sorption experiment brought to light a conspicuous difference in the behaviour of sulphur dioxide from that of any gas previously tested. The results showed that the routine treatment adopted for
TABLE V Sulphur Dioxide Time-sorption measurements. Exp.
Exp. 2.
I.
Time P (mins) (mm. of Hg) (cc. @
0.50 1.37 2.27
3.55 5.47 8.95 12.5 16 21
26 31 41 56 76 I 06 167 255 2884 5795
376.35 374.99 374.21 373.70 373.28 372.63 372.17 371.89 371 * 5 0 371.21 370.94 370.53 370.00 369.54 368'94 368.29 367.62 363.06 361.76
S
K.T.P)
Time imins)
2.032 2,177 2.264 2.319 2.370 2.443 2,491 2,526 2 ' 565 2,595 2.627 2.667
16 26 36 76 136 I 82
2.722
1450
I.IO
.67 2.67 4.09 6.17 I
I1
P S (mm. of Hg) (cc. @ N. T. P)
377.21 376.60 375.96 375.39 374.92 374.21 373.78 373. I 7 372 ' 78 371.75 371 . o o 370.55 367.88
I
.830
I . 898 I . 968 2.032 2.083 2.164 2.211
2.279 2.317 2.423 2.498 2 ' 539 2.850
2.766 2.827 2,904 2,977 3.456 3.605
out-gassing the glass wool (3 hours heating in vacuo at a temperature of zoo "C.) had, for the first time, proved inadequate. Not only did the points obtained fail to yield a linear graph, but the sorption values differed so widely from those of the first experiment that the conclusion that the sorbent still contained a residue of gas was unavoidable. Continued heating a t the same temperature, though effecting the removal of some small further quantities of gas (circa I 50 cmm), led to no material improvement in the results. We were reluctant a t this stage of the work t o increase the out-gassing temperature for fear of altering the condition of the sorbent. Proc. Roy. Soe. 105A,487 (1924).
549
SORPTION BY GLASS
yT
earlier experiments. The conclusion that the presence of an impurity in the sorbent may depress the rate of sorption in the
0
1.80 v
j552
igg 5’ w
q
=
later, appears to be well established’.
TABLEVI Exp. I. 2.
3.
Solvent range of time covered d log t/d log by experiment
Sorbent
animal charcoal (1
L(
LL
(6
toluene benzene benzene
5 mins to 3 days to I
Trans. Am. Electrochem. SOC.36, 153 (1919). Trans. Faraday SOC.14,202 (1919).
years years day to 1 2 0 days 11
11
17.5 13.4 12.0
=
5 50
D. H. BANGHAM AND F. P . BURT
It is noteworthy that the value of the coefficient (which appears to vary with the solvent) is of the same order as yielded by our various experiments with glass. Mention may also be made of F. Bergter’s experiments on the sorption of air by cocoa-nut charcoal1, as some of these accord very closely with the linear formula. Thus, an experiment carried out a t 735.6 mm. yields the graph depicted in Fig. 9.. With the exception of the first observation (t = 0.75 mins.), all the points lie closely about a straight line having a reciprocal slope of 18.5.
FIG.g
Summary The sorption of nitrous oxide by glass reproduces the essential characteristics observed with carbon dioxide. Considering any series of constant-pressure experiments, log (sorption) appears to be a single-valued function of log (p6%),the relationship approximating more closely to the linear than in the case of carbon dioxide. Sulphur dioxide also follows the linear log s, log t law, but as we did not succeed in out-gassing the sorbent we were unable to examine the pressure exponent. We wish to acknowledge here the receipt of grants from the Royal Society Committee and from the Brunner Mond Research Fund, which partly defrayed the expenses of this investigation. Chemistry Department, University of Manchester. Ann. Physik, (4) 37, 498 (1912).
’
