2366
INDUSTRIAL AND ENGINEERING CHEMISTRY
work, and to J. F. Foster, supervisor of fuels research, and Fred Benington, assistant supervisor of fuels research, at Battelle, for their helpful interest and consideration. LITERATURE CITED
(1) Brinsley, F., and Stephens, S., Nature, 157,622 (1948). (2) Coward, H. F., and Greenwald, H. P., Bur. Mines, Tech. Paper 427 (1928). (3) Crone, H. G., private communication to Gaydon mentioned in (6),pp. 91-2. (4) Fowler, A., and Vaidya, W.I f . , PTOC. Roll. SOC.( L o n d o n ) , 132A,
310 (1931). (5) Gaydon, A. G., “Spectroscopy and Combustion Theory,” pp. 88. 90-2, London, Chapman and Hall, 1948. (6) Kurz, P. F., “Stability Studies with Mixed Fuels. I. Hydro-
Vol. 45, No. 10
carbons, Hydrogen, Hydiogen Sulfide,” Battelle Tech. Rept. 15036-3 to Wright Air Development Center, Contract AF 33 (038)-12656 (April 30, 1952). ( 7 ) Xurz, P. F., Battelle Memorial Institute, Columbus, Ohio, unpublished results. (8) Payman, W., J . Chem. SOC.,115, 1438-82 (1919). (9) Smith, F. A., and Pickering, S. F., Xatl. Bur. Standards J . Research, 17, 7-43 (July 1936). (10) Steacie, E. TV. R., personal discussion with P. F. Kurz at Columbus, Ohio, 1950. RECEIVED for review M a y 15, 1953. ACCEPTED July 3 , 1953. Presented before t h e Division of Gas and F u e l Chemistry, AMERICAN CHEMICALSOCIETY,Pittsburgh, P a . , .4pril 1953. Work done under t h e sponsorship of the Flight Research Laboratory, Wright Air Development Center, Wright-Patterson Air Force Base, Ohio.
Rate of Evaporation of Glycerol in High Vacuum D. J. TREVOY Research Laboratories, Eastman Kodak Co., Rochester, N . Y .
T
HE rate at which molecules pass through an imaginary window in a uniformgas has been derived from the kinetic theory (8) and is given by
where r is the mass per unit area per unit time, p is the vapor pressure a t temperature 2‘, degrees absolute, ilf is the molecular weight, and R is the gas constant. For a liquid in contact with its equilibrium vapor, the maximum rate of two-way interchange of molecules a t the interface is also expressed by this relation. Langmuir ( 7 ) ,and later Knudsen (6),considered the evaporation process to proceed independently of the condensation process at the interface. It follows that if the vapor is removed irreversibly as rapidly as it is produced, the maximum rate of evaporation into free space is also given by this equation. By experiment Knudsen found that the rate for liquid mercury was sometimes less than that predicted, and he introduced an “evaporation coefficient” to absorb the discrepancy. Evaporation coefficients for a number of liquids were subsequently measured by Rideal ( I O ) , Alty arid Nicoll ( 2 ) , Alty ( I ) , Baranaev (3),Priiger (Q),and Wyllie ( I S ) . The results of these experiments suggest that nonpolar substances, such as carbon tetrachloride aud benzene, have evaporation coefficients near unity, while polar substances, such as water or ethyl alcohol, have much smaller coefficients (0.01 to 0.04 for water). It is implied by these workers that each substance has an evaporation coefficient which is a characteristic physical property, and that the value of the coefficient for different substances may vary widely. Experiments in these laboratories, already described (5j, suggest that this may not be the case, but rather that fresh surfaces of all liquids evaporate a t the maximum theoretical rate. The fact that lower than optimum rates are often observed is an indication that the evaporating surface is not truly representative of the bulk material. Then an evaporation coefficient less than unity is not a basic characteristic of the substance but is determined by specific conditions at the surface. The emission characteristics of an old surface are consequently variable, and may depend on many factors, such as adsorption to the interface of soluble impurities from the liquid, orientation in the interface of polar molecules with or without subsequent adsorption of impurities from the liquid or vapor, and adsorption of impurities from the vapor. Alty (1) measured the evaporation of freshly formed drops of
water, the surface temperature being deduced from the surface tension which was measured simultaneously. The evaporation coefficient obtained was only 0.04. It is thought that the low value may perhaps be due to uncertainties in the surface temperature, or, as the drops were formed very slowly, there may have been sufficient time for the formation of inhibiting surface layers. In earlier experiments (5),di-Zethyl hexyl phthalate and di-2ethyl hexyl sebacate were evaporated from the constantly r e newed surface of a stream of liquid falling through an evacuated space. l i p to a vapor pressure of about 5 microns, the observed rate was exactly that of the theoretical. The rate then fell off slightly as the vapoi pressure increased and Tas about 7 5 % of optimum at 100 microns. The 2 5 % discrepancy in rate is believed due, not to an evaporation coefficient less than unity, but to a reduction in temperature a t the surface resulting from a steady-state thermal gradient in the fluid stream. Although these data support the thesis that the evaporation coefficient is unity for new, clean surfaces of all liquids, it seemed desirable to measure the coefficient for a liquid with very different propert’ies than those of the phthalate and sebacate esters. A search of the literature revcaled that in esperiments x-ith glycerol an abnormally low rate was obtained by Wyllie ( I S ) , an evaporation coefficient of 0.052 being reported a t 18” C., with a vapor pressure less than 0.1 micron. This fluid appeared suitable for falling-stream experiments, and vias particularly interesting because it is different in structure from the esters and is highly polar. The falling-stream technique was used in a tensimeter specially adapted to handling glycerol. METHOD
In principle, the apparatus was the same as that described earlier (6),but it differed in many details and is shown diagrammatically in Figure 1. Tubing a t least 18 mm. in outside diameter was used around the entire liquid path to minimize pressure drop. Two centrifugal impellers in series were needed to circulate the glycerol, and the temperature was controlled by a bimetallic thermoregulator which fitted snugly into a large thermowell in theleft arm. The glycerol wa8 distributed radially by flowing i t through a cylindrical wire screen, 11/3 turns of 30 mesh, just before it entered the jet tube, having an inside diameter of 1.27 om. and a length of 11.0 cm. The large jet tube was necessary to obtain even a small stream a t
October 1,953
I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
2367
When vacuum was broken, the test tube with condensate was weighed immediately to avoid absorption of water vapor from the air. For runs a t 30" C., the coolant was pumped through a rubber tube wrapped around the left arm of the tensimeter, t o remove heat generated by the impellers, and the drainage time was increased to 90 minutes. To eliminate uncertainty in the starting time and also the error owing to incomplete drainage, shorter runs a t each temperature served as blanks. The weight of condensate and the time from the short run were then subtracted from the weight and time for the long run. Every detail of manipulation wfts precisely duplicated in short and long runs to eliminate error. In some cases, a small correction was applied to the blank where the temperatures of short and long runs were not quite identical. The apparatus and procedure described were suitable for operation in the temperature range 30" t o 70" C., but to measure the rate at 18' C., the temperature a t which Wyllie obtained data, several modifications were necessary, The same condensing jacket was used, but the diameter of all tubulation except the jet tube and the pump was increased to 27 mm. to reduce pressure drop in the viscous liquid. A screw lift (4) was employed t o circulate the fluid. This was made by inch in diameter, on a piece of steel winding copper tubing, drill rod, 1/8 inch in diameter and 14 inches long, forming a spiral with turns about '/z inch apart. The copper tube was then fastened to the drill rod and sealed a t each end with silver solder.
V
Figure 1. Falling-Stream Tensimeter Used for Glycerol i n Range 30" to 70" C.
II
L
the lower operating temperatures. Condensate was collected in a single zone, 6.38 cm. high, by circulating acetone a t -70" C. through a cylindrical jacket coaxial with the condensing surface. The acetone was cooled externally with dry ice, and was circulated continuously by a total-immersion centrifugal pump. Samples could be withdrawn and weighed or allowed to return to the main bulk of fluid by use of a two-way stopcock connected to an auxiliary mechanical vacuum pump. The shape of the alembic condenser was modified to avoid drippings from the area above the main condensing zone during drainage. Before use, the glycerol was dewatered by distillation under high vacuum in a pot still. The water collected rapidly in a dry ice-acetone trap and was removed from the system before distillation was continued. The first and last fractions of glycerol were discarded, and no difference in refractive index of the eight middle fractions could be detected with the precision refractometer. The flow tensimeter was charged with about 550 ml. of dewatered glycerol, and circulation continued a t 70" C. under vacuum for about 8 hours to degas the fluid. Prior to each run, the correct temperature and head were maintained for about an hour, without coolant in the jacket, to establish a reproducible holdup on the walls of the condensing zone. Cold acetone was then admitted t o the jacket, and the timer was started when the coolant reached an arbitrary level in the jacket. During a run, the temperature was recorded a t regular intervals, the head being held approximately constant by occasional manual adjustment of the motor speed. At a temperature of -70" C., the condensed glycerol was frozen in a layer on the inside wall of the condenser. T o end a run, the diffusion pump was turned off, the stopcock was turned to the sampling position, and glycerol circulation was stopped as the timer passed the end point. The cold acetone was immediately drained from the jacket, and air a t room temperature was blown through the jacket for 60 minutes while the melted condensate drained into a weighed test tube.
4l
2
01 15
1
20
'
25
1
30
I
35
I
'
1
40 45 50 Temperature,'C.
'
55
1
60
'
65
1
5
70
Figure 2. Area of Steam i n Condensing Zone A glass tube in which the screw could turn freely was mounted in the apparatus, and, in operation, the screw was rotated by an external direct current motor through a shaft seal (16). A symmetrical stream was readily obtained without using a wire-screen distributor. The fluid was cooled b y passing water a t 10" C. through a jacket concentric with "the screw lift. The flow of water was turned off and on by a solenoid valve controlled through a thermoregulator. The latter was sensitive to changes in capacity and operated from a clip on a thermometer immersed in the glycerol above the jet tube. I n this way it was possible to limit the temperature variation during a control cycle to 0.3 C., the average temperature during the cycle being constant within 0.1" C. for the duration of a run. As overnight operation was necessary, acetone for the condenser was cooled by circulation through a coil in a 14-inch bell jar containing dry ice and a little acetone. All electrical circuits were operated through a set of relays which turned off motors and cold water if, for any reason, the head dropped '/z inch. It was necessary to charge the bell jar with 40 pounds of dry ice once every 24 hours, and the apparatus O
2368
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 45, No. 10
IV-
\ \ 7-\\ 5b, b,,
\
Temp
c.
io 9: -5
?
g
-:z Z
\
010 07 00 7 -0050040030 02 -
0 004 0 003 0 002
\ \
\ \O
\
, , , ,
75 70 65 6 0 55
I
50 4 5
l
I
40
35
30
25
, ,, 20
15
Temperature,oC
Figure 3.
Rate of Evaporation of Glycerol
0. Experimental, falling stream
0 . Experimental, open cup (Wyllie)
- - -.
5 5
54 54 54 64
4 4 6 5
50 4 50 4 50 4
43 46 44
42 42 42 42
3 5 4 1 4 4 4 4
2 3 4 16 39 40 41 42
38 38 38 38
2 3 1 1
30 30 30 30 30 30 18 18
2 2 1 0 1 1 3 2
17 19 20 21 23 25 24 26 61 52 55
66 62 62 62 62 62 62 62 62
'a\\ b
l
58 58 58 58
66 66
\
Maximum theoretical rate
required only minor adjustments ill motor speed ovcr thc 4-day run. A condenser temperature of -67" C. was obtained. To establish nearly identical starting conditions, the operating history for 2 days before the blank run began was carefully duplicated before the run itself was commenced. The drainage timr waq increased to 3 hours for runs a t 18" C. RESULTS
The area of the stream in the condensing zone was measured photographically, as previously described ( 5 ) ,and the data appear in Figure 2, a some%-hatlarger jet tube being used in the 18-degree model. Rate of evaporation mas measured in the range of vapor preisure 0.07 to 17 microns, with a residual gas pressure about 0.1 micron (Table I). The quantity A&!is the difference in duration of long and short runs, and Atu is the difference in weight of condensate collected. Mean valueq of the rate are summarized in Tahle 11, where observed and calculated rates are compared. The calculated rates are obtained by the equation cited abovr, using values of the vapor pressure reported by Stedman (11). The comparison between observed and calculated values of the rate is shown graphicallv in Figure 3, and, within the limit? of experimental error, the coefficient of evaporation is unity over the entire range. The rate obtained by Wyllie at 18' C. is indicated in Figure 3 by a solid dot, and differa from present data by a farto1 of 11.3. The criticism might be raised that droplets were detached from the stream or were produced when the stream struck the wall of the lower tube before rejoining the main body of fluid, and that these droplets eventually reached the condensing zone. A simple arid convincing proof that this was not the case was provided as a rewlt of the formation of colloidal aluminum in the fluid in the first apparatus. The finely divided metal was produced by rapid iotation of the aluminum impellers in their glass housings. A solution was made up containing 0.73% of distilland containing colloidal aluminum and 99.27% of glycerol containing no aluminum. This was compared with the distillate from a 4-hour run at 30" C. by observing the 90-degree light-scattering using a mercury arc source and a Corning Glass color filter, No. 3486. Scattering was much greater in the case of the solution containing 0.73% of distilland, indicating that gross mechanical transport from stream to condenser accounted for less than 0.73% of the distillate, and possibly none. I t was also demonstrated, within the limits of ex-
No. 27 28 29 30 31 32 33 34 13 14 15 18 47 48 49 50 35 36 37 38 9 10 11 12
70 70 70 66
\o \
0 02 2 --
1 1
Run
1 1
2 2 2 2 6 7 6 5 4 4 5 5 5 5
50 4
46 46 46 46
Temp., C. 70.1 66.2 62.5 58.5 54.5 50.4 46.4 42.4 38.2 30.1 18.2
45
56
Time, h .
eA
5.0 10.0 5.0 10.0 7.0 15.0 7.0 15.0 11.0 25.0 12.0 20.0 10.0 20.0 11.0 20.0
42.0 72.0 31.0 80 0 59.0 110.0 99.0 58.0 71.0 195.0 90.0 181.0 102 238 150 253 349 693 346 646 675 365
15.0 35.0 15.0 31.0
56.0
26.0 45.0 20.0
2970
5728
vapor Pressure, Microns 17.0 12.0 8.5 5.8 3.85 2.58 1.72 1.13 0.72 0.29 0,069
Weight of Condensate,
w ,G.
2.417 4.917 2,242 4.953 2.354 5.235 2.243 5.073 2.765 6,644 3.444 5,174 2,440 4.939 3.155 5,058 2,426 5 837 2 417 5.122 6,043 2.754 4.996 2.131 2.876 5.095 1.994 5.449 2,329 4.650 4.149 2.173 1.522 4.757 2.005 4.324 1.106 3.183 1.995 3.355 0.414 1.910 0.473 1,395 1,870 0.638 2.182 4.483
G.
Area, Sq. Cm.
Rate of Evap.. G./Sec. SQ Meter
5.0
2.501
18.0
4.63
5.0
2.711
18.0
5.02
8.0
2 881
17.5
3.43
8.0
2.830
17.5
3.37
14 0
3,854
16.8
2.73
8.0
1.762
16.8
2.18
10.0
2.499
16.8
2.48
9.0
1.903
16.8
2.10
20.0
3.411
1.5.8
1.80
16.0
2 , 70.5
15.8
1.78
30.0
3.289
14.7
1.24
25.0
2.844
14.7
1.29
40, Min.
AlO,
30.0
2.219
13.5
0 913
49.0
3.455
13.2
0.870
51.0
2.274
12.2
0.609
41.0
1.910
12 2
0,636
124.0
3 , 235
11 . o
0.395
91.0
2.319
11.0
0.386
13B
2.06Z
9.7
0.261
103
1.360
9.7
0,227
344
1,496
7.4
0.098
3 00
0.927
7.4
0.070
310
1.232
7.4
0,090
2758
2.333
5.4
0,0261
R a t e of Evaporation, G./Seo. Sq. Meter Calcd. Obsd. 5 14 4.83 3 65 3.40 2 GO 2.37 1 78 1.79 1 19 1.27 0 805 0.892 0 539 0.623 0 356 0.391 0 228 0.244 0 093 0.086 0 0226 0.0261
Difference Obsd. andbetween Calcd. Values of Rate, % -6.0 -6.8 -8.8 f0.G +6.7 f10.8 f15.5 +9.8 f7.0 -7.5 +15,5
perimeiital error, that the presence of colloidal aluminum in the distilland did not altw the measured rate of evaporation. The possibility of splashing at the freezing point (18" C . ) is remote, in view of the sirupy consistency of the fluid a t this temperature. There seems little doubt that a fresh surface of glycerol has an evaporation coefficient of unity over the range of vapor pressure used in the present experiments. The possibility exists that in the rase of stagnant surfaces polarity in the molecule results in an oriented surface layer which is instrumental in reducing the emission of vapor. Much further work will be necessary to make clear the mechanism whereby the emission of vapor from a pool of glycerol may be reduced by a factor of 11.3. It is evident, however, that the surface of the falling stream is being renewed too rapidly to permit establishment of conditions at the surface which inhibit evaporation. and t h r present data for glycerol substantiate
October 1953
INDUSTRIAL AND ENGINEERING CHEMISTRY
the hypothesis (6) that fresh surfaces of all liquids evaporate at the maximum theoretical rate. LITERATURE CITED
(1) Alty, T., Phil. Mag., 15, 82 (1933). (2) Alty, T., and Nicoll, F. H., Can. J . Research, 4, 547 (1931). (3) Baranaev, M., J . Phys. Chern. (U.S.S.R.), 13, 1635 (1939). (4) Hickman, K. C. D., IND.ENG.CHEM.,to be submitted. (5) Hickman, K. C. D., and Trevoy, D. J., Ibid., 44, 1882 (1952). (6) Knudsen, M., Ann. Physik, 47, 697 (1916).
2369
(7) Langmuir, I., Phys. Rea., 2, 329 (1913). (8) Loeb, L. B., “Kinetic Theory of Gases,” p. 88, New York, McGraw-Hill Book Co., 1927. (9) Priiger, W., 2. Physik, 115, 202 (1940). (10) Rideal, E. K., J. Phys. Chem., 29, 1585 (1925). (11) Stedman, D. F., Trans. Faraday Soc., 24, 296 (1928). (12) Trevoy, D. J., Anal. Chem., 24, 1382 (1952). (13) Wyllie, G., Proc. Roy. Soc. (London), 197A,383 (1949). RECEIVED for review December 17, 1952. ACCEPTED June 24, 1953. Communication 1531 from t h e Kodak Research Laboratoriefi.
CORRESPONDEN,CE P
Effectiveness and Safety of Fluoridation of Public Water Supplies SIR: Before undertaking to advise the public in regard to the desirability of fluoridating a local water supply, a chemist should become familiar with the researches that demonstrated that the dental health of a community is, t o a large degree, dependent upon the concentration of the fluoride ion in its water supply. These are to be found in two symposia arranged by the American Association for the Advancement of Science and published in two small volumes (1). RELATION O F FLUORIDE CONCENTRATlON IN A WATER SUPPLY TO DENTAL HEALTH
*
-1
During the first decades of this century a few dentists noted the high incidence of stained and mottled teeth in certain communities in the Rocky Mountain and Texas areas and their absence in other nearby towns. In the affected communities, the abnormality was encountered among only those persons who had lived there during the period in early life (up to about the %ge of 12) in which the dentine and enamel of their permanent teeth were being formed. This distribution suggested that the condition was due to some water-borne cause. Proof of this came in 1931 when it was found that the water of St. David, Ariz., a town where mottling was prevalent, when concentrated to one tenth its volume and fed to rats, induced a somewhat similar defect in their teeth. Fluoride was found present in the water from the wells of this town in concentrations ranging from 3.1 to 7.1 parts per million. The fluoride content of public water supplies in the United States varies from traces to 7 or 8 p.p.m., and in one town it is 15 p.p.m. The incidence and severity of mottled enamel have been shown by extensive dental surveys to increase with the fluoride content of the water. The threshold concentration that induces mottling of a degree detectable only on careful examination by a dentist trained in its recognition is, in the North Central States, about 1 to 1.5 p.p.m. Beginning at about 2 p.p.m., an increasing proportion of the children have mottling of a grade that is easily apparent, and, a t still higher concentrations, the affected areas of the teeth tend to assume a brown stain. I n hot areas in the South where the water intake is greater, the threshold for barely detectable mottling may be so low as 0.5 to 0.7 p.p.m. Over a million persons in 500 communities use water that contains naturally more than 1.5 p.p.m. Fluoride occur8 normally in the teeth and bones of all persons. Most of it comes from the water used, the food furnishing only about 0.2 to 0.3 mg. daily, an amount insufficient to affect the dental health materially. As early as 1916, Frederick S. McKay, a Colorado dentist, noted that mottled teeth seemed less subject to decay than normal ones, an observation that soon found confirmation by others in both this and other countries. Beginning in 1938, H. Trendley
Dean, United States Public Health Service, undertook a systematic epidemiologic study of the relation between the inciden. e of dental caries and the fluoride content of water supplies, which showed clearly that quantities of fluoride too small to cause cosmetic damage are capable of conferring resistance to caries. I n some 21 cities in Indiana, Illinois, Colorado, and Ohio the caries rates calculated from the numbers of decayed, filled, and missing teeth, when plotted against the fluoride content fell on a smooth curve, the rates decreasing from 1037 in Michigan City with 0.1 p.p.m. to 236 in Galesburg, Ill., with 1.9 p.p.m. By plotting a similar curve for the incidence of mottling on the same graph, it became apparent that the concentration of 1.0 to 1.2 p.p.m. is optimal for dental health, the teeth being cosmetically attractive and freer from decay than those in areas with lesser fluoride concentrations. I n recognition of their leadership in the painstaking research which established this relationship, McKay and Dean were jointly given a Lasker Award of the Bmerican Public Health Association. TESTING O F FLUORIDATION
Their work made it seem possible that by adjusting the fluoride content of water supplies to that optimal for dental health, progress could be made in reducing the prevalence of caries or mottling throughout the country. It was soon found possible to reduce the incidence of mottling in such towns as Bauxite, Ark., where the water had the grossly excessive content of 14 p.p.m., by changing the source of the water to a more nearly fluoride-free well. Although it seemed possible that the larger number of communities whose water was deficient in fluoride might be benefited by the addition of fluoride to their water supplies, i t was thought unwise to recommend this until it had been shown in a few selected communities that it is technologically feasible when fluoride is added to maintain the desired concentration within narrovi limits and that the incidence of caries could in practice be reduced to the extent predicted from the epidemiological studies. Accordingly, in 1945, carefully supervised fluoridation was begun in Grand Rapids, Mich., and Newburgh, N. Y. Muskegon, Mich., and Kingston, N. Y., were selected as control cities and agreed to continue to use their fluoride-free water sources. Each year, the teeth of thousands of school children in all four cities were examined by teams of skilled dentists. Since the children born after the start of fluoridation did not enter school until they reached the age of 5 or 6 years, examinations made during the earlier years of the test necessarily included children who had not had the benefits of fluoridation from birth. It was believed, therefore, that the test would have to be continued for a t least 10 years in order to learn the maximum benefit obtainable by fluoridation. These tests have now been in progress for more than 8 years and
