THE VAPOR PRESSURE O F SULFUR DIOXIDE AND' AMMONIA BY J?. W. BERGSTROM
Introduction I n connection with a series of investigations on the physical properties of the constituents of natural gas, Burrell and Robertson' have measured the vapor pressures of ammonia and sulfur dioxide a t temperatures ranging from the normal boiling points down to below -100". The wide discrepancies between the values of the vapor pressures around the normal boiling point of these two liquids as determined by Burrell and Robertson, and the values obtained previously by other investigators, and especially the disagreement, to the extent of a whole degree between the normal boiling points as interpolated from their results and the carefully determined values for these constants made some years ago in this laboratory by H. D. Gibbs2led to a redetermination of the vapor pressures of these two s u b ~ t a n c e s . ~
Description of Apparatus The apparatus used for vapor pressure measurements is shown diagrammatically in Figs. 1, 2 and 3. The manometer and thermoregulator are separately represented in Figs. 2 and 3. The liquid whose vapor pressure was to be measured was confined in the bulb I. A capillary connected this with the left arm, J, of the manometer. By means of the two-way, stopcock K, I could be connected either with the purification train, C, C', C", or with a Topler pump (not shown). E was Burrell and Robertson: Technical Paper 142, U. S. Bureau of Mines. Gibbs: Jour. Am. Chem. SOC.,27,851 (1906). * The measurements described in this paper were completed during the summer of 1918. Publication was delayed because of the appearance of the Bureau of Standards article on the vapor pressure of ammonia. Although the work done on sulfur dioxide is now of greater importance, it was thought worth while also to include the data obtained from ammonia, as confirmatory of the results of the Bureau of Standards. 1
Vapor Pressure of Suuur Dioxide and Anzrnonia
359
a large Dewar tube vessel, 45 X 20 cm internal dimensions, containing petroleum ether for the constant temperature bath. G was a motor-driven stirrer. The resistance thermometer H was connected to three long heavy leads attached to a Wheatstone bridge. Liquid air contained in a five-liter Dewar flask, A, was forced over into the wide test tube M by air pressure exerted on the liquid thru C'. The magnitude of this pressure, and hence the rate of transfer of liquid to M was regulated by changing the depth of immersion of the tube under the water in D . This arrangement is a modification of Hen-
Fig. 1 Apparatus for Vapor Pressure Measurements (Thermoregulator shown separately in Irig. 3)
ning's low temperature thermostat. The petroleum ether bath was cooled with liquid ammonia down to -30" and with liquid air below this temperature. Ammonia from a compressor, after passing thru a pipe and needle valve, entered a wide closed tube, F, thru P, where it evaporated and was pumped back thru a second pipe attached to 0. The manometer, shown in detail in Fig. 2, is a modification of that described by Germann.' The dimensions are Henning: Zeit. Instrumentenkunde, 33, 33 (1913). Germann: Jour. Am. Chem. SOC., 36, 2456 (1914).
F. W . Bergstrom
360
given in the figure. A stopcock, G, separating the two arms of the manometer, was opened only during adjustment of mercury level, and preceding vapor pressure measurements. With G closed, a pressure corresponding to a previously regulated temperature could be read while the bath was being cooled in readiness for the next observation. As the mercury level in 3 was kept near the glass index P during most of the time that liquid was in I, any entrance of air at low pressures could readily
,
Attached to Mera~ry reservoirs Fig. 2 Manometer
Fig. 3 Thermoregulator
be detected. The mercury column prevented direct contact o f t h e enclosed gas with the stopcock. The method of producing a high vacuum above the mercury in J', and the precautions necessary completely to free J' of traces of air are given in the original article by Germann. The apparatus was pumped out a t intervals thruout the work to insure the maintenance of a perfect vacuum.
Vapor Pressure of Sulfur Dioxide and Ammonia
361
A large mercury reservoir for rough adjustments of level was attached below E', and a small one for the more delicate adjustments necessary to bring the mercury level to the tip of the index P was connected to E. As the height of the index was determined a t the beginning of a series of measurements, but one cathetometer reading was necessary for each pressure. The former was checked at frequent intervals. The total height of a mercury column could accurately be measured to within 0.2-0.4 mm. The height of mercury in the manometer was read with a cathetometer a t a distance of eight feet. The instrument used possessed a brass scale 1.2 meters in length, graduated in millimeters, with a vernier reading to twentieths. As it was slightly bent during the earthquake of 1906, it was necessary to level the telescope before each reading.l The mercury used in the manometer was purified according to the method of Hulett.2 A sample of this mercury melted a t -38.9 O (resistance thermometer). The observed manometer height was corrected to 0" for expansion of the mercury and of the brass cathetometer scale. Other corrections were within the limit of experimental error.
Temperature Regulation Liquid air or ammonia was introduced in the manners described, into F and M , Fig. 1, at rates just sufficient slowly to cool the bath. The thermoregulator with its heating element then accomplished the close regulation desired. The thermoregulator, Fig. 3, consisted of a framework of one cm glass tube, F, filled with toluene, and connected by a capillary with two tube reservoirs, A and E, the former for toluene, the latter for toluene and mercury. The heating wires W, strung on the frame, were connected to a 110 volt circuit thru a variable five-lamp resistance and a relay, operated This instrument was recently compared with a new Gaertner cathetometer, and found to be accurate. Phys.Rev., 33,307 (1911).
362
F. W . Bergstrow
by contacts a t C and E. To set for temperatures, a small reservoir attached to E was raised until mercury had risen t o the upper contact. While keeping the reservoir in this position, the bath was cooled to suck mercury up into the capillary B'. Upon closing the stopcock, and turning on sufficient current slowly to warm the bath in spite of the cooling due t o liquid air or ammonia, the desired temperature could be held constant as long as necessary. Mercury should stand at least a centimeter high in B' for the best operation. This type of thermoregulator eliminates the use of a stopcock in contact with the toluene. Temperatures have been regulated by this method t o within 0.02", as shown by the change in resistance of the platinum thermometer element. With liquid ammonia, temperatures could be held constant to within 0.05" without difficulty. Thermometry Temperatures were measured with a three-lead, compensated, platinum resistance thermometer. The resistances were determined with a dial Wheatstone bridge reading directly to 0,0001 ohm, the points of balance being found with a high sensitivity galvanometer. The instruments were manufactured by Leeds and Northrup. The resistance thermometer was calibrated by the Bureau of Standards, with the following results : Resistance a t 0 " = Ro= 9.360 ohms. Fundamental Interval = Rloo-Rc= 3.210 ohms. 6 in Callendar's difference formula= 1.47. The calibration was checked in this laboratory at the freezing points of water and of mercury. Resistances of 9.3590, 9.3593, and 9.3597 ohms were observed a t 0" in three experiments a t different times. The resistance a t the freezing point of mercury (-38.89", Henning;l -38.87", Wilhelm2) was found to be 8.089 ohms. The corresponding temDerature was calculated to be -38.9". Henning: Ann. Phys., (4)43, 294 (1913). U. S. Bureau of Standards., Sci. Paper No. 204, 655-61 (1916).
Vapor Pressure of Sulfur Dioxide avld Ammonia
363
As the simple Callendar formula does not hold below -40 ' the thermometer was calibrated against the vapor pressure of liquid oxygen according to the methods of von Siemens1 and Henninga2 The oxygen was prepared and purified as described by von Siemens, but the apparatus was in most respects identical with that of Henning. A resistance of 2.924 ohms was found a t -190.7" (average of three values). The temperature was calculated from the observed pressure (317 mm), using the vapor pressure data of von SiemensS3 The temperatures below -40' were determined from a formula of Henning's3 involving this calibration. The temperatures as determined by the resistance thermometer are probably accurate to within 0.1 '. Preparation and Purification of the Gases The vapor pressure measurements were checked in each instance by determinations upon a second sample prepared by a different method. Sulfur Dioxide.-The first sample of sulfur dioxide was obtained from a stock cylinder of the commercial liquid, and was purified in the following manner: The gas was condensed in the first of the fractionation tubes C" (Fig. 2), containing phosphorus pentoxide, and allowed to stand for an hour to insure complete removal of moisture. A vacuum of a few millimeters of mercury was produced in the tubes, and some sulfur dioxide distilled into C", with this surrounded by an alcohol bath a t about - 90 '. Most of the gases of high vapor pressure at that temperature were thus removed. The liquid, after reaching room temperature, was caused to boil vigorously by opening stopcock B ' until its volume had diminished one-third (to remove gaseous impurities). Following this, the sulfur dioxide was frozen, and the entire system, including the bulb I, Fig. 1, evacuated as completely as possible. The frozen mass was melted, a portion Von Siemens: Ann Phys., (4)42,871 (1913). Henning: Ihid., (4)43, 294 (1913). 8 LOC.cit.
364
F. W . Bergstrorn
boiled off under reduced pressure, and the process repeated. This treatment served to remove the last traces of dissolved gas and to free the liquefaction bulb and connecting tubes from air. Two-thirds of the purified sample were distilled into I, and half of this was pumped off to make sure that no air had been trapped in the transfer. With stopcock K closed, mercury was raised to the pointer, sealing the opening. The second sample of sulfur dioxide, prepared by the action of concentrated sulfuric acid on copper, was washed successively with water and concentrated sulfuric acid, ther passed over ignited pumice saturated with water (to removc HzS04),and thru drying tubes to remove the greater part of the moisture before condensation in C”. The subsequent treatment was the same as described above. Ammonia-Ammonia was obtained (1) from a small stock cylinder of the commercial liquid, (2) by distillation of C. P. ammonia water. The gas was condensed in a tube, C” (not the one used for sulfur dioxide), containing sodium and allowed to stand for an hour to remove moisture. Onethird of the sample was rapidly boiled off by opening stopcock B’. The ammonia was fractionated from C” to C’, and thence to C, traces of sodium which ordinarily would have been carried over having been removed by an ignited asbestos filter sealed between C” and C’. The sample was frozen in C with liquid air, the entire system evacuated, and dissolved gases removed by a process of alternate freezing, melting, and boiling, as described for sulfur dioxide. The introduction into the vapor pressure bulb I was carried out with the same precautions as before. The second sample was partially dried with soda-lime, condensed, and treated as described.
Results of Measurements The results of the vapor pressure measurements made are presented in Table 1, while the values interpolated for every five degrees of the temperature scale, together with similar data of other investigators, are given in Table 2 (for sulfur dioxide).
Vapor Pressure of Suljur Dioxide and Ammonia
365
TABLE I Sulfur Dioxide Observed resistance ohms
Corresponding :emperatwe
9.179 9.133 9.068 9.039 9.001 8.949 8.845 8.716 8.640 8.498 8.374 8.365 8.221 8.054 7.846
O
C
Observed pressure corr. to 0" mm
-5.5 924.8 -7.0 867.2 -9.0 799.6 -9.9 765.4 -11.0 727.5 -12.6 676.7 -15.8 583.6 -19.7 483.2 -22.1 430.7 -26.4 345.5 -30.2 282.5 -30.5 278.2 -34.9 219.1 -40.0 162.9 -46.3 110.7 Ammonia Observed pressure 'corr. to 0" mm
8.383 8.261 s.109 7.956 7,.795 7.631 7.440 7.228 1 Triple
-30.0 -33.7 -38.3 -43.0 -47.8 -52.8 -58.6 -65.0 point.
903.3 751.3 593.6 460 .'O 350.8 261.4 182.2 117.4
Observed resistance ohms
7.591 7.471 7.259' 7.091 6.975 6.927 6.850 6.805 6.572 6.391
Corresponding temzerature C
0bser v ed pressure corr. to 0" mm
-54.0 66.2 -57.6 52.0 -64.0 32.4 -69.1 21.4 -72.6' 16.5 -74.1 14.1 -76.3 11.7 -77.7 11.o -84.7 4.4 -90.1 2.5 Sec:ond Sam] .e 9.169 1 -5.9 I 911.6 9.047 774.4 8.911 -13.8 641.2. 8.765 519.2 Second Sample
1 -;:::1
CorreObserved sponding Resistance temperature ohms O C
8.374 7.716 7.152 6.951 6.843 6.732 6.579
-
-30.2 -50.2 -67.3 -73.3 -76.5 -79.9 -84.4 -77.91
Observed pressure corr. to 0" mm
891 .O 304.4 101.2 64.8 50.6 37.2 24.2 45.51
F. W . Bergstrovvt
366
TABLE 2 Sulfur Dioxide Vapor pressure measurements interpolated for every five degrees of the temperature scale compared with data of other investigators. Temp. " C Bergstroml mm
0 -5 -7.2 - 10 -11.5 - 15 -17.8 -19.5 -20 -25 -30 -35 -40 -45 -50 -55 60 -65 -70 Solid -75 -80 -85 -90
-
-
Burrel12 mm
Pictet mm
Bliimcke mm
-
947 860 762 712 607 530 488 478 373 286 217 162 120 87 62 43 30 19.9
-
792
-
639
-
-
507 396 306 234 174 115 89 62 43 30 20.4
-3
-
12.8 8.0 4.1 2.5
12.8 8.0 4.1 2.5
In all of the observations above -30" ( i. e., when the bath was cooled with liquid ammonia) the temperature was held constant for from twenty t o thirty minutes before making readings; otherwise, from five to ten minutes (with liquid air cooling). 8
Other Measurements Several years ago, Murray E. Tucker and George Ziser independently made measurements of the vapor pressure of ammonia and sulfur dioxide in this laboratory, Observations 1
3
Calculated from formula. Interpolated from large-sized plot of Burrell's data. Observed.
Vapor Pressure of,Sulfur Dioxide and Ammonia
367
were made with practically the same apparatus described in this paper, but the methods of use were not the same. Calibrated toluene thermometers were used thruout. The measurements, while insufficient to determine the entire vapor pressure curve with accuracy, are especially numerous near the normal boiling and triple points. These observations are plotted in Figs. 4 and 5.
Fig. 4 Vapor Pressure Determinations near the Triple Point of Ammonia
The Triple Point of Ammonia No definite information regarding the triple point of sulfur dioxide was obtained. A well-defined triple point was found for ammonia (see Fig. 4). The constant temperature bath was allowed to warm up slowly, after being cooled to -85', and the pressure observed just as the solid began to melt. This pressure a t the triple point was found to be 45.5 mm. The corresponding temperature calculated from the formula for the vapor pressure curve is -77.9'. The measurements made by Ziser and Tucker in the region of the triple point are plotted in Fig. 4. Tucker's observations were made on different occasions, and the curves, while not in agreement, are both markedly discontinuous from liquid to solid. He was able to obtain a t one time two points upon the supercooled ammonia curve. The observations of Ziser and Tucker, with the exception of some scattered results, show the triple point to be slightly below - 78 '.
F. W . Bergstrom
368
Brill1 found a discontinuity between the vapor pressure curves for liquid and solid phases, altho no mention is made of the fact. Burrell found no such break. The Bureau of Standards2 has recently accurately determined the pressure and temperature a t the triple point of ammonia. The observed pressure was 44.9 mm. The observed temperature was -77.70', while that calculated from their vapor pressure formula was -77.84'.
50 0
780 E = !P
LlJ
760
5
CD [o u.l
740 720
Fig. 5 Measurements near the Normal Boiling Point
Formula for Vapor Pressure Cdrve The experimental results were-expressed by the partially theoretical Nernst formula :
Log,, P = -
-+ 1.75;logTA
4.571T --
Ann. Phys., (4) 21, 170 (1906). *Jour. Am. Chem. SOC., 42, 222 (1920).
- + C. €1
4.571
Vapor Pressure of Sulfur Dioxide and Ammonia Sulfur Dioxide ( P in mm of mercury) 1577.3 Log10 P = - 1.75 log T - 0.006411T
T
+
369
+ 6.3286
The average deviation from the experimentally determined values (disregarding sign) is about one mm. Discussion of Results and Comparison with Previous Work. Sulfur Dioxide The only vapor pressure determination covering the range from the normal boiling point to solid was made by Burre1l.l Regnault's2and Pictet's3results extend only to -30 ". Faraday4 and Bliimcke5 have made a few observations below 0". Burrell's results vary considerably from previously determined values. The normal boiling point, - 11.0"' interpolated from his data, is nearly a whole degree lower than determinations made by Regnault2 (- 10.08"), Pictet3 (- 10.0"), Gibbs6 (- 10.09O) and the author (- 10.05", manometric and direct). His sample was prepared by the action of concentrated sulfuric acid on copper, and purified by fractionation a t low temperatures. The sulfur dioxide finally used for measurements distilled within 0.2 O, altho the method of determining this range is not indicated. Burrell's work appears accurate from his descriptions, but the vapor pressures of ammonia and sulfur dioxide have been carefully determined in this laboratory, and results have been obtained agreeing in each case much more closely with previous observations than do his. The close agreement with the first determinations of the check measurements made upon the second samples affords a good indication of the purity of our material. Regnault's results agree very closely with those of the author (see Table 2 and Fig. 9). The maximum variation is Loc. cit. a
Regnault: M6m. de l'Acad., 1863, 535. Pictet: Arch. de Genhve, 13, 212. Faraday: Phil. Trans., 135, 1, 155 (1845). Blumcke: Wied. Ann., 34, 10 (1888). Gibbs: Jour. Am. Chem. SOC., 27, 851 (1905).
370
F. W . Bergstrow
about 3 millimeters. He made two series of observations which were not in good agreement. In one of the series he admits that he did not completely remove air from the apparatus. This may also have been true of the second series. Pictet's results do not fall on a smooth curve. Bliimcke determined the vapor pressure of sulfur dioxide at two temperatures below -10". His measurements were not intended for an accurate and complete determination, but merely as a check upon an apparatus used for finding the vapor pressure of solutions. His normal boiling point, - lo", agrees with that of the author. Faraday likewise made two determinations below 0 ". They do not agree well with the other measurements mentioned.
Ammonia Regnault,l Pictet,2 Brill,3 Perman and D a ~ i e s B , ~~ r r e l l , ~ Keyes and Brownlee,6 and the Bureau of Standards' have previously made measurements covering the whole, or a part of the range from -30" to -85". The determinations of Brill, Burrell and the Bureau of Standards are the only ones extending below the freezing point. With the exception of Perman and Davies, the others made observations only as low as the normal boiling point, -33.4". Brill made one series of measurements around -79" by passing hydrogen thru several layers of solid ammonia, and determining the partial pressure of the latter in the mixture of the exit gases. The second series, made by the static vapor pressure method, covered the range from -33" to -80". Brill's vapor pressure-temperature curve is lower than the author's in the region of the normal boiling point, and below -61 ", but higher a t intermediate points. The differences may be due to thermometry. Regnault: MCm. de 1' Acad., 26, 535 (1862). Pictet: Arch. Gensve, 13, 212. a Brill: Ann. Phys., (4)21, 170 (1906). 4 Perman and Davies: Proc. Roy. SOC.,78, 28, 42. Burrell: loc. cit. 6 Keyes and Brownlee: Jour. Am. Chem. SOC.,40,39 (1918). Cragoe, Meyers, and Taylor: Ibid., 42, 206 (1920).
Vapor Pressure of Sulfur Dioxide and Ammonia .
371
Burrell's observations were made by the static method upon a sample prepared by boiling C. P. ammonia water. It was purified by repeated distillations until the entire final sample distilled over within 0.2O (or less?). As the molecular boiling point elevation constant of ammonia is only 3.4", the boiling point of a normal solution would be -33.06". It can readily be seen that the constancy of the boiling point of ammonia, unless very accurately measured, is no indication of its purity. Burrell's curve agrees with that of the author below - 54 ", but diverges a t higher temperatures, the difference a t the normal boiling point amounting to 1.2". The value found for this constant, -34.6", is over a degree lower than any determination previously made by manometric means. Perman and Davies made three measurements below -30 O in order to determine the normal boiling point, which they found t o be 33.5'. The results are in agreement with the writers' only near -34" and a t -46". As only three measurements below -30 O were made, an error in one or more of the values would seriously affect the entire vapor pressure curve. The data of Brill, Burrell, Perman and Davies, and the author are compared graphically in Fig. 9. Regnault and Pictet, respectively, found the normal boiling point to be -32.8" and -33.0" by manometric means. Regnault removed water by passage thru a U tube surrounded with liquid ammonia and thru a calcium oxide drying tube. Air was washed from the apparatus by pumping ammonia from the condensed liquid. The latter method, especially, is inadequate. Keyes and Brownlee by manometric methods found the normal boiling point to be -33.25", and by direct determinaTwenty-eight vapor pressure measurements tions -33.21 were made in the region of the normal boiling point. The pressures, reduced to 33" for comparison, vary from 747.6 mm to 786.9 mm and are too erratic for conclusions drawn from the mean. The Bureau of Standards has carefully determined the Their vapor pressure between the limits of +70" and -80".
F. W , Bergstrorn
372
normal boiling point, -33.35", is 0.1" higher than was found by the author. The average deviation of all the experimental results for ammonia cited in this paper from their curve is three-quarters of a percent. The average deviation of the values ,above -50" is only half a percent. The scale of1Fig. 9 is too small to show these differences clearly. M. Tucker and G. Ziser, whose measurements upon ammonia with the apparatus described in this paper have previously been mentioned, made a number of observations around the normal boiling point. These are shown in Fig. 5 , in comparison with a portion of Burrell's curve. The normal boiling point from Tucker's results is -33.5", and from Ziser's dataj-33.4". Both of these support our values, and constitute further evidence that Burrell's results are in error. The maximum variation of Ziser's measurements for any temperature is 33 mm a t -34.3". The variation at the normal boiling point is 10 mm.
.7800 ,7760 .7720
.7680 ,7640 .76 00
.7560 I
I
270'
770
300' Fig. 6
760'
Vapor Pressure of Sulfur Dioxide awd Anznzonia
373
Data Plotted by Method of Ramsay and Young The vapor pressure measurements described in this paper, together with data of other investigators, have been tested by the Ramsay-Young relation: R'=R c(T'-T) c is a constant for every pair of substances.l R' is the ratio between the absolute temperatures of any liquid and of a standard substance (suchas water or methyl alcohol) a t a given vapor pressure. R is the ratio at another vapor pressure. T' -T = the difference between the absolute temperatures of one of the liquids at the two vapor pressures. The graph of the equation is a straight line, as c is a constant. The measurements of Brill, Burrell and the author .6560 7060 on ammonia, and of Burrell 7020 and the author on sulfur di- 6520 oxide, are plotted in Fig. 6, 648c 6980 with methyl alcohol as standard. The results of the au6940
+
thor are separately plotted in .6400 ,6900 Fig. 7, with water as standard. As the vapor pressures 6360 6860 of methyl alcohol were taken 6720 .E1820 from a fairly large-sized plot 300' 370' 560' '590' of Ramsay and Young's data,2 Fig. 7 absolute accuracy could not be expected in the interpolation of the temperatures, especially a t low pressures. The general trend of the curves would not be affected. The vapor pressures of water were taken from Marx and Davis' steam tables, and represent the most accurate work yet published. The remainder of Burrell's measurements are plotted in Fig. 8 to a smaller scale, with methyl alcohol as standard. The author's measurements give straight lines whether 2
Zeit. phys. Chem., 38, 666 (1901). Phil. Trans., 188, 320.
F. W. Bergstrom
374
plotted with water or methyl alcohol as standard. The observations of Brill and Burrell plotted in Fig. 6 are not straight lines, and accordingly do not satisfy the RamsayYoung relation.
I
270"
700'
I
330'
TCH&I
Fig. 8 (The second curve from the top should be labelled C4Hlo-isoand not CdH,o-l)
Most of Burrell's measurements plotted in Fig. 8 show some curvature, and so do not satisfy this relation. The points for solid nitrous oxide make a fair straight line. With
Vapor Pressure of Sulfur Dioxide and Ammonia
375
the exception of two points on each, the graphs for ethylene and ethane are nearly straight lines.
PRESSURE MM. Of M€RCURY I
I
I I
I I I I
I I
I I.
_c_
I
i I
Summary 1. Vapor pressures of ammonia and sulfur dioxide were determined from temperatures slightly above the normal boiling points to solid. Two samples were used in each case, and the results of the measurements are in excellent agreement.
I
376
F. W . Bergstrm
2. An improved form of apparatus for measurements of vapor pressures below 900 mm and a t low temperatures has been described. 3. The measurements of Brill and Burrell do not satisfy the Ramsay-Young relation, while those of the author do. It has been experimentally shown in this laboratory that Burrell’s vapor pressure measurements on ammonia and sulfur dioxide are in all probability erroneous. 4. The triple point of ammonia has been determined. In conclusion, the writer wishes to acknowledge his great indebtedness to Professor E. C. Franklin, a t whose suggestion and under whose direction this work was undertaken, and to Prof. H. P. Cady for many valuable suggestions.