V O L U M E 21, NO. 3, M A R C H 1 9 4 9 Table X.
401
(Glass bulbs 9- t o 10-mni. O.D., 0.1 grain of glass per bulb)
KO,
Cs20
359 367 368 370
Lit0 25 25 25 25
364 379 365 441 456
24 25 26 27 28
2 2 2 2
429 426 411 36,;
26 26 26 26
- _ _ _____ ~
0 0 0 0
~
sodium error has been made of 28 lit'hium oxide-3 cesium oxide-4 lanthanum oxide-65 silica. The p H characteristics of a very large number of glasses have been briefly discussed from the Resistance viewpoint of molecular composition. The pHIncrease Factor responsive glasses have been considered to be 2.0 structures comprising a silicon-oxygen network 1.8 1.4 in which the alkali metal ions and other metal 1.1 ions occupy the interstitial positions within the 1.0 silicon-oxygen network. The charact,erand dimen1.0 1.0 sions of the ions govern the base exchange rc1.5 actions and hence the sodium error and stability 3.1 of the glass. The relative degree of hydration of 1.5 1.3 the lithium ion and its low coordination number 1.2 with oxygen ions are important considerations 1.2 for ~- an ideal p H glass. The composition of the most practical pHresponsive glass represents a compromise in ~ h i c h greatest stability with minimum sodium error and relatively lon. direct current electrical resistance are obtained.
Composition-Electrical Resistance of pH Glasses at 25" C. Composition, _ _ Mole _Per Cent _ ~ B e 0 hIgO CaO BaO LazOa Si02 7 0 0 3 65 0 0 I) 3 65 0 7 0 3 65 0 0 0 7 3 65
'6
ResistanceAfter Original 100 140 500 900
1 year
198 250 712 960
0 0
4
4
0 0
2 4 3 3
j
0
2
0 0 0 0 0
4 2 2
3 3 1 3 2
83 63 63 63 61
5200 1300 1300 265 110
5200 1500 1600 390 344
1 2 2 2
0 0 0 0
0 0 0 0
2
3 2
1
4 2 4
2
P
67 64 64
1
63
200 500 450 1300
300 640 550 1600
4
4
~ _ ~ .~
-..--
.
- .~~ __
cesium oxide type of glass favors low sodium errors. A strontium oxide or barium oxide constituent also favors low sodium errors. h relatively low silica content, in combination with the correct modifying oxide is favorable for lox sodium errors and the highest stability of pH glasses. The direct current electrical resistance of a pH-responsive glass is related to the radius of the ionic constituents of a glass. A 28 mole yolithium oxide content, an alkaline earth metal of small ionic radius, and a 70 mole 70silica content favor a relatively low
The use of 60 to 63 mole yo of silica in combination with from I to 3 mole % ' of lanthanum oxide, or rare earth oxide, and lithium oxide and cesium oxide, results in very favorable I-"-rewonsive glasses with relatively low sodium errors. It is not essential that an alkaline earth metal be present in a pH-responsive glass. .4 pII-responsive glass with relatively low
ACKNOWLEDGMENT
The author desires to acknodedge the help of Mrs. John S. Penny for considerable experimental work in connection with these studies. LITERATURE CITED
(1) Bates, Hatner. Manov. and .Icree. J . Research S a t l . Bur. Stand-
115 (1945). 21i 391 (lg4'). G. .INAL. (7) Perley, G . A . , U. S.Patent 2,444,845(1948). Jahrb .\fineraZ. Geol. Beitage Bd., 9, 602 (8) Thugutt, s. J., -veltes (1895). (9) Vesterberg, K. h.,Z . U ~ O T Q Chem . , 110,48 (1920). (')
RECEIVED
, . ' A
September 7, 1048.
DIFFERENTIAL THERMOMETER H. E. KITSON
AND
JOHX MITCHELL,
JR.
A m m o n i a D e p u r t m e n t , E. I . du Pont d e Nemours & C o m p a n y , Znc.. W i l m i n g t o n , Del.
Differential thermometers filled with liquids other than w-ater have been constructed and tested. Although no radical increase in thermometer sensitivity can be achieved by their use, it is possible to secure maximum sensitivity for a wide range of ebulliometric solvents by the proper choice of thermometer liquid. These thermometers have been used in semiroutine analyses for some time with satisfactory results.
T
HE theory of the differential thermometer was first outlined
by Menzies (6) in 1920. Together with ) r i g h t ( 7 ) he demonstrated its value in ebulliometric molecular weight determinations. Although several workers have reported using the thermometer ( 3 , 4 , 8 ) ,none has published any attempt to improve the thermometer's range or sensitivity. The ebulliometric method of Menzies and Wright ( 7 ) has been used, with slight modifications, for molecular weight determinations in this laboratory for a number of years. The method was ieverely circumscribed, however, because of the limitations of the water-filled thermometer. The sensitivity of the thermometer, in ' C. per mm., drops off rapidly below the boiling point of water, thereby greatly limiting the number of organic solvents with which it could be used. Furthermore, according to the procedures .of the above noted investigators, the thermometer was not used
*
much above the boiling point of water. Menzies ( 6 ) pointed out these difficulties and an obvious solution, which was to fill the thermometer with a liquid other than water. The present research was undertaken to prepare thermometers filled with liquids other than water and to devise, if possible, a more sensitive differential thermometer. Since its completion, Barr and Anhorn ( 2 ) have described another method of filling thermometers with nonaqueous liquids. 4PPARATb-S AND CHEMICALS
A standard high vacuum bench, capable of reaching vacuums better than 10-5mm. of mercury, was used in much of the filling work. Liquid nitrogen, dry ice-methanol, and water icemethanol were used as refrigerants for handling liquids used in filling the thermometers. A small tube furnace, 2.5 em. (1inch) in diameter X 30 em. (12 inches) long, was used for outgassing the
402
I
t h e r m o m e t e r bodies. The current to the heater was adjusted to give a temperature of 530" C. Thermometer bodies were constructed of No. 774 Pyrex brand glass tubing of suitable size. KO special effort was made to secure tubing of uniform bore. Pure liquids for filling the thermometers were secured by a variety of methods. All except the p-xylene and sulfur dioxide were distilled a t least once from the best obtainable grades. The former was Standard Sample 215-5 from the N a t i o n a l B u r e a u of Standards and was used A C without further purification. The latter was taken directly from a cvlinder of refrigeration grade materia1.Figure 1. Possible Differential of the compounds were Thermometer Designs passed as vapor over Ascarite and phosphorus A . Menzies original design Dentoxide, and then B , C. Proposed designs triply distilled in v a c u u m immediately before being placed in the thermometer. This latter procedure was intended to remove any traces of carbon dioxide, water, or inert gases. EXPERIMENTAL WORK
Thermometer Design. Several different designs of the thermometer body are possible. Three of these, including hlenzies' original design ( 6 ) ,are shown in Figure 1. The two proposed designs ( B and C) have the advantage of a narrower cross-sectional area and, so far as the authors know, suffer no disadvantages over the original hlenzies type ( A , Figure 1). The length of the main capillary depends either on the ebulliometer to be used with the thermometer or the temperature differential to be covered. The length of the bulbs in turn is dependent on the length of the capillary, as they need be only large enough to contain liquid to fill the capillary. In most of this work the capillary was 18 cm. long, and the bulbs had a capacity of about 1.5 ml. each. There as no difference in results a i t h the three designs. Filling Thermometers. Menzies (6) says littl,e about his method of filling the water thermometers except that "permanent gas is removed by the process of boiling out. . . I ' This method cannot be used to fill nonaqueous thermometers, however, as there must be no decomposable vapor present a t the sealing point (G, Figure 2) during the sealing operation. The following procedure was developed and used to fill thermometers with a variety of liquids. I t is necessarily detailed, because manipulative difficulties arose \There some variations TTere tried. The series of traps and drying tubes, shown in Figure 2, was attached to the vacuum bench. The degree of vacuum necessary is not known accurately, but almost certainly it should be better than lp. The system should be "tight" enough to hold this vacuum without the pumps for 2 or 3 hours. Having achieved this condition, a blank thermometer body was sealed on the bench a t point A . Tube E was filled with suitable chemicals for the removal of carbon dioxide and water (usually Ascarite and phosphorus pentoxide), and fastened to the line, using a resin to seal .I1.Stopcock SJwas closed and S pand SIwere opened. The small tube furnace was placed around the thermometer body, heated to 530" C., and maintained at this temperature for approximately 2 hours. The furnace was then shut off and, after 15 or 20 minutes, removed from around the thermometer. The desired amount of filling liquid was placed in tube F, frozen, and attached to the bench a t J z with a resin sealing com-
ANALYTICAL CHEMISTRY pound. With the material in F still frozen, 8 3 was opened cautiously and most of the air w&s pumped off (to approximately 1 0 ~ ) . S3 was closed and the material in F allowed to liquefy. After refreezing, S3 was opened and any released gas was pumped off. This offgassing was repeated at least twice. Trap D was cooled with liquid nitrogen and the majority of the material in F was allowed to evaporate through E into tra D. When sufficient material had distilled over, SZwas closed ancftrap C cooled. Trap D was allowed to warm up. When all the material had distilled from D to C, B was cooled and the material distilled into this trap. With all the material in B, SI was closed and the contents of B were allowed to warm up. The thermometer body (which should have been a t room temperature) was cooled with a refrigerant which did not freeze the filling material. (If suoh a refrigerant is not readily obtained, the filling is possible with a refrigerant that freezes the liquid, but is much more difficult.) When the thermometer body was filled to the desired extent, the material was frozen in place in the lower bulb and a t the same time the unused material was frozen in trap B , C, or D. (This is a relatively difficult operation, and some experience is necessary to do it satisfactorily.) Siwas opened and the system allowed to remain on the pumps 10 to 15 minutes. With a small flame, 1 to 2 cm. of the capillary were collapsed slowli just above the upper bulb and the sealed area was carefully annealed. The material in the thermometer was kept frozen until the seal cooled to room temperature. As soon as the sealing was complete, SI was closed, Sp and S3 were opened, and the cooled trap was allowed to warm up. F was cooled with liquid nitrogen and kept cold until all the unused material had passed through E into it. Then Sa was closed, F disconnected, and the remaining material discarded. SI was opened and the system pumped clean in the usual manner. With care, very little material ever passed through SI into the main bench. Thc amount of liquid distilled into the thermometer was controlled carefully, so that enough liquid was present to fill the entire capillary a t the lowest temperature a t which the thermometer was to be used, and the amount of liquid present was somewhat less than the volume of either bulb. Otherwise, in the first case, the full length of the thermometer could not be used, while in the sccond case, the vapor space in the upper or lower bulb disappcared under certain conditions and the thermometer failed t o operate. (In judging the amount of cooled liquid in the thermometer, its expansion upon heating must be considered.) Experience with differential thermometers is the best guide to their correct filling. Sensitivity of DifFerential Thermometers. As Menzies (6) has pointed out, the theoretical basis for the differential thermometer's sensitivity is the small difference in the temperature of the two bulbs; this causes a difference in vapor pressure bet\$een the two arms of liquid which is registered as a substantially linear height function. The sensitivity of a thermometer depends entirely, therefore, on three factors: the vapor pressuretemperature differential of the filling liquid, the density of the filling liquid, and the average temperature of the two bulbs. Menzies has calculated the sensitivity of the water thermometer over a range of temperature from 30" to 102' C. Unfortunately, data such as Menzies used for water do not exist for most organic liquids, but reasonably accurate calculations can be made from existing data. ii recently published set of tables from
Figure 2. Vacuum Bench for Filling Differential Thermometers
403
V O L U M E 21, NO. 3, M A R C H 1949
T, C. 30 36.1 40 50 60 68.7 7n ."
80 80.1 90 98.4 100 110
110.6 120 130 138.4 140 150 160
n-Pentane 2.07 1.7Za 1.55 1,19 0.93
...
0.74 ,..
...
(Sensitivity, C. per mm. X 10') n-Hexane Benzene n-Heptane Water
...
... ...
...
...
3.20 2.40 1.88O 1.82 1.46
...
. .
...
L59
...
... ...
...
...
...
...
... ... ... ...
3.34 2.56 2.48a 2.00
1.13
... ... ...
...
6.03 4.44
...
... ... ... ... ... ...
... ...
24.7
15.8
. .
10.4
7. .. ,1 5.0
...
3.28
...
,..
2.50 2.02a 1.95 1.54
3.56
1.24
1.46 1.13
i:60. 1.92
...
, . .
... ... ...
...
0.88 0.70 0.56
...
...
Sensitivity a t boiling point of filling liquid.
the National Bureau of Standards (1) gives constants of many organic liquids for the Antoine equation (A).
B log10 p = A - -
C+t
where p is the vapor pressure in millimeters of mercury, and t is the temperature in C. A, B, and C a r e constants for any given liquid. Differentiating this equation gives: dt - = - -(C
dp
pressure tables for water.] However, attempts to achieve these theoretical values experimentally with water bep-Xylene tween 130" and 180" C. have failed so ... ... far. Good checks were attained a t ... ... ... ... 118" C., using glacial acetic acid as ... ... the ebulliometric liquid. Above about ... ... ... .. .. .. 130' C., corresponding to an* internal ... ... ... pressure of 3 atmospheres, the zero read... ... 4.46 ... ing of the thermometer was a t the top ... , . of the capillary and the thermometer 3.43 . . 2.68 ... became useless. Even a t lower tempera2 . 63a ... tures, it had been noticed that the zero 2.25 4.58 1 .. 7. 1. 3.60 2.954 reading was considerably higher than ... 2.86 predicted by capillarity, although this 2.30 .. .. .. 1.87 was not a source of trouble up to 118" C., a t which temperature several centimeters of clear stem still remained. Similar phenomena were observed with thermometers filled with n-pentane, benzene, and toluene. The point a t which individual thermometers became useless varied slightly but was a t a temperature corresponding to 2 or 3 atmospheres internal pressure. No explanation of the phenomenon is readily apparent. In only one case were the authors successful in using a differential thermometer a t internal pressures over about 3 atmospheres. Two thermometers filled with sulfur dioxide operated reasonably well a t 35" C., and one of them was taken t o 80" C. without undue increase in the zero point. At the lower temperature, the theoretical sensitivity of 8 X lo-' O C. per mm. was approached. Even a t the lower temperature, however, the thermometer was slow in reaching equilibrium, presumably owing to the internal pressure of 5.3 atmospheres, and was not usable as an analytical tool under these conditions. On the strength of these experiments, further work along this line was abandoned. N o practical method is apparent a t the present time for increasing greatly the sensitivity of a differential thermometer. In order to cover the range of ebullioinetric solvents usually encountered with maximum sensitivity, it was decided to use a series of thermometers. The thermometers used to cover the range of 30" to 160" C. are listed in Table 1. These thermometers have been used for some time in the routine determination of molecular weights. The ebulliometer, thermometers, and technique, with typical results, are discussed in a second paper ( 6 ) .
Sensitivity of Differential Thermomieters
Table I.
+ t ) * - logioC(C + t)'
____ B X p
log10C
*
B X p
(2)
density of Hg (0' C.) gives densitv of liauid It" C.) the sensitivity of an organic liquid-filled \hermometer. Table I shows the sensitivities of thermometers filled with different liquids between 30' and 160" C. and indicates that, for any given operating temperature, greater sensitivity can be obtained by using thermometers filled with lower boiling materials. However, in molecular weight determinations a knowledge of the thermometer constant is not necessary. The authors have found experimentally that a scheme of comparative measurements, similar to that proposed by Swietoslawski ( 9 ) , gives more accurate results by ruling out a number of small variables. The molecular weight in this case is based on the folloLving equation: Division of the d t l d p value by
I
\
ACKKOWLEDGiMEh-T
wumaAta ?nu = -
waAtu
(3)
where m u and m, are the molecular weights, respectively, of the unknown and an added known, u,, and w Sare the weights of unknown and standard, and Atu and At, are the observed temperature increases. -4s long as both are expressed similarly, At,, and At8 can be as readily expressed in millimeters as in O C. This obviates a knowledge of the thermometer sensitivity as such. Attempts to Increase Sensitivity of DiiTerential Thermometer. One of the prime objectives of this work was to seek a method of increasing the differential thermometer sensitivity. Inspection of Table I shows that, a t the boiling point of the filling liquid, the thermometric sensitivity is of the same order of magnitude for all the liquids listed, and they would be even closer if differences in liquid density were removed. However, another possible approach was the use of the thermometer a t temperatures above the boiling point of the filling liquid. Superficially there are only two limitations to such a scheme: the pressure which the glass body would withstand and the critical point of the liquid. Table I shows the sensitivity of a water thermometer from 40 O to 180 O C. [The data a t 100' C. and below are from Menzies (6'1, the data above 100" C. are calculated from the usual vapor
The authors wish to thank D. M. Smith and A. Pi. Oemler for their interest in and criticism of this work. LITERATURE CITED
(1) Am. Petroleum Inst., Research Project 44, National Bureau of Standards, "Selected Values of Properties of Hydrocarbons,
(2) (3) (4) (5)
(6) (7) (8)
(9)
Vapor Pressures and Boiling Points, a t 10 to 1500 Mm. Hg," Table 1 k, June 30, 1944; Table 2 k (Part l), March 31, 1944; Table 2 k (Part 2), May 31, 1944; and Table 5 k (Part I), June 30, 1944. Barr, W. E., and Anhorn, V. J., with Hanson, TV. E., Instruments, 20, 342 (1947). Colson, -4.F., Analyst 57, 757 (1932). Hanson, W. E., and Bowman, J. R., IND. ENG.CHEM.,ANAL.ED., 11,440 (1939). Kitson, R. E., Oemler, A. N., and Mitchell, J., Jr., AXAL.CHEM., 21, (1949). Menzies, A. W. C., J . Am. Chem. Soc., 43,2309 (1921). Menzies, A. W. C., and Wright, S. L., Ibid., 43, 2314 (1921). Smith, J. H. C., and Milner, H. W., Mikrochemie, 9, 117 (1931). Swietoslawski. W., "Ebulliometric Measurements," p. 172, Reinhold Publishing Corp., Kew York, 1945.
RECEIVEDM a y 18, 1948. Presented before the Division of Analytical and Micro Chemistry, a t the 113th Meeting of t h e ArERIcaN C H E i f I c A L SOCIETY, Chicago, Ill.
