THE RELATION BETWEEN OBSCURING P O K E R AND PARTICLE NCMBER AND SIZE O F SCREESING SMOKES BY P. D. WATSON AND A . L. KIBLER*
Object The object of the work described in this paper was to check the theoretical assumption that the screening power of smokes and clouds is simply related to their particle number and size and to determine the numerical value of that relationship. In addition to this, data were desired on the absorption of moisture by hygroscopic smoke particles, the composition of smoke particles, the efficiency of dispersion, and the stability of smoke clouds. Historical
KO previous experimental work is known which relates the obscuring power of smokes or clouds with the particle number and size. The fact that such a relationship exists was quite accurately forecast by H. W. Walker1 on purely theoretical grounds. A closely allied line of work consists in a correlation of particle size (concentration remaining constant) with intensity of the Tyndall beam in aqueous colloidal suspensions. Relevant data include work by Mecklenburg2 who worked with graduated suspensions of colloidal sulfur particles, and Tolman3 and his collaborators, who worked with graduated suspensions of silica particles. Tolman extended his work in a qualitative way to rosin smokes. Mecklenburg’s work covered the range from 0.0; X IO-^ to 8.4 X I O + cm. diameter. Tolman’s quantitative work covered the range from I O X IO-6 to 206 X I O + cm. diameter. His qualitative work with smokes covered the range from 0.j X IO-^ to I O X IO-^ cm. Although there is some discrepancy in the results of the two investigators named above the following points seem well established : I. For particle diameters below that of the wave-lengths of light, the intensity of the Tyndall beam increases with an increase in the size of the particles, concentration remaining constant. For particle diameters above that of the wave-lengths of light, the 2. intensity of the Tyndall beam decreases with increase in the size of the particles, concentration remaining constant. 3 . For particle diameters of the order of magnitude of the wave lengths of light, the evidence is conflicting but it is evident that this is the transition range. The range of wave-lengths of visible light is from about 4 X IO-^ to about 7.5 X IO& cm. The average wave-length of white light is about 5.7 X IO-^ cm. * Technical Divisions, Chemical Warfare Service, Edgewood Arsenal, Md. ‘Walker: Ind. Eng. Chem., 17, 1061 (1925). Mecklenburg, Kolloid-Z., 16, 97 (1925). Tolman et al.: J. Am. Chem. SOC.,41, 300, 575 (1919).
OBSCURING POWER O F SCREENING SYOKEb
IOi5
Theoretical It is known that the absorption of light by true solutions depends upon the thickness of the layer traversed and on the molecular concentration in that layer.’ Svedberg has shown that the mechanism of light absorption may be the same for colloidal solutions composed of observable particles and for molecular solutions.2 Ostwald has also shown that, in many cases, absorption by a colloidal solution of a very high grade of dispersion approximates absorption by the corresponding molecular solutions.3 I t is by no means certain, however, that the mechanism of screening by smokes is as simple as the absorption of light by the smoke particles. Among other factors which might conceivably affect obscurance are reflection and refraction of light, composition of the smoke particles, relative position of observer, smoke and target, and intensity and location of the light source. Suture of the Problem.-The relation between the obscuring power of smokes and the number and size of the smoke particles may be expressed specifically as the ratio between the sum of the projections upon the surface of the target for all the smoke particles between the eye and the target, and the area of the target when it is just obscured in the smoke cloud. This ratio is referred to in this paper as the “screening ratio” and designated by the letter R. In order to establish a relation of this kind it is necessary either to use smokes in which the particles are all of the same size, or to know the proportionate number of particles of each size present. We may assume, however, from a study of the Brownian Movement, and from other considerations, that the greater number of particles in any stable smoke are essentially of the same size. Most of the efficient smokes are hygroscopic in character, absorb large quantities of water, and produce spherical liquid particles. Therefore, the area of a great circle through the sphere may be regarded as the projection of the particle upon the vertical target. From the above it is obvious that the essential data necessary to establish the desired relationship are the number of smoke particles per unit volume, the average size of these particles and the screening power of the smoke. The number of particles may be found with sufficient accuracy by means of the ultramicroscope. Knowing the number of particles per cubic centimeter, the average diameter of the particles may be calculated from the weight and density of the smoke. This involves sampling and analyzing the smoke from a known volume of air. The obscuring power of the smoke may be determined by noting the distance a t which a target just disappears from view while under constant illumination. Experimental Experimental work was carried out under accurately controlled conditions in a smoke chamber of approximately 63,000 cubic feet capacity and 1,000 square feet cross-sectional area. It was fairly air-tight and provided with ~~
Taylor: “Treatise on Physical Chemistry,” 2, rzoi. Svedberg: “Colloid Chemistry,” 194 (1928). Ostwald: Kolloidchem. Beihefte, 2, 409 (1910-11).
1076
P. D. WATSON AND A. L. KIBLER
ventilating ducts and a large circulatory fan for keeping the smoke uniform in concentration throughout. The chamber was also provided with facilities for humidifying and drying the air to any degree of humidity desired. Obscuring Power. Observations of the obscuring power of the smoke were made through a small window in one end of the chamber upon a circular, black, Io-inch target mounted 4 feet above the floor. The target was mounted upon an iron rod set in a supporting frame directlyin line with the observation window. The target was moved back and forth by means of the iron rod which was graduated in feet. The target was illuminated by two rows of overhead lamps, each row containing fourteen 60-watt lamps, set over the target and about 16 feet above it. Visibility readings were taken one or two minutes after the start of each run and a t intervals thereafter. These values were plotted in a visibility-time curve and the values used in the work were read from these curves. Particle Number. Particle number was determined by means of an ultramicroscope arranged upon a shelf on the outer wall of the smoke chamber near the observation window. The smoke was drawn into the cell through a 1/4-inch glass tube about I 2 inches long which projected into the smoke chamber. The source of light for the ultramicroscope was a direct current carbon arc. An optical system projected a fine pencil of light into the observation cell. An absorption cell of cupric chloride solution was used to eliminate the short rays and prevent convection currents in the cell. I n order to reduce the personal error of the observer to a minimum the illuminating beam was cut off for brief regular intervals by means of a rotating sectored disk placed in its path. The depth of focus of the microscope was 0.0444mm. A diaphragm with two small circular apertures was inserted in the eyepiece. The apparent diameters of the apertures were 0.150and 0.120 mm., determined by focusing on a millimeter scale. The average of fifty successive counts was taken. From the particle count, the depth of focus and the diameter of the aperture used, the number of smoke particles per cubic centimeter was calculated. Smoke Samples. Samples of the smokes were taken by drawing 20 liters of air-smoke mixture through specially prepared filter tubes with filter beds of ground, fritted glass and thin asbestos mats. The filter tubes were weighed immediately before and immediately after the collection of samples. Great care was exercised to keep the tubes at equilibrium with the moisture of the room. The composition of the smoke was determined by analysis for the chief constituent. From the composition of the smoke particles their density was determined either by reference to standard tables or by special experimental work. Materials. The smokes used were those which, upon dispersion in air of ordinary humidity, will become liquid droplets. For the liquid smoke materials, dispersion was accomplished by means of an efficient spray. White phosphorus was burned in an open pan and zinc chloride was dispersed by means of a smoke candle containing the so-called H.C. smoke mixture. Ammonium chloride was dispersed by atomizing liquid ammonia and 32%
OBSCURING POWER OF SCREENING SMOKES
I077
aqueous hydrochloric acid. The several smoke materials used are shown below: Titanium tetrachloride White phosphorus Oleum Zinc chloride Chlorsulfonic acid Ammonium chloride In the case of most of the smokes generated, upon dispersion in a humid atmosphere, hydration, hydrolysis or deliquescence occurs. The resulting smoke particles are aqueous solutions of the resulting products. I n the case of chlorsulfonic acid, analysis of the smoke revealed only traces of hydrochloric acid and it was assumed that this product of hydrolysis escaped from the liquid droplets. The case of titanium tetrachloride presented a dificult problem on account of the complicated chemical reactions which occur between TiCl, and water. As a consequence of the uncertainty as to the composition of this smoke some special work was done on the densities of titanium compounds. Mixtures were made up containing varying percentages of titanium tetrachloride and water and these mixtures were analyzed for titanium and their densities determined. It was found that when the densities of the titanic acid solutions were plotted against their titanium content, a straight line was obtained. The values for the density of the particles of titanium tetrachloride smoke were taken from this graph. The several smoke materials used are tabulated below along with the bases to which the analyses of the smoke samples were calculated. Smoke Material
Analyzed for
Zinc chloride White phosphorus Titanium tetrachloride Ammonium chloride Oleum Chlorsulfonic acid
ZnC12
P206 Ti02 NH4Cl
so3
SO3 Method of Calculation. The radius of a particle is determined from the equation:
where
M = total mass of the particles in
I cc. (g). d = density of the particles (g./cc.) Tu' = number of particles (no./cc.) r = radius of particle (cm.) The screening ratio, R, is determined from the equation: R = x D'KV __
4
where D = diameter of a smoke particle (cm.) V = distance between the eye and the target when the target is just obscured (cm.) R = the screening ratio.
1078
P. D. WATSOS AND A. L. KIBLER
TABLE I Zinc Chloride R.H.
90
90
60
60
Diam. of part. Dx104 em.
Density of part. d.
2.17
0.547 0.691 0.826
1.53 1.42 1.36
I53 198 268
3.94 2.80
0.657 0.711
1.39 1.35
0.776
1.22
I43 192 259
1.63 1.80 1.80
Time Number Et. ZnCll after particles sample by dispersion N X 10-8 from 201. anal. min. mg mg.
6 16 26
3.50 1.94
6 16 26
3.49
14.4
2.24
11.4
I.j6
9.3
1
6 16 26
347
6.5 4.2 3.6
2.52
0.474
2.2j
2.04
0.472 0.jjo
6 16 26
4.16 2.24 1.43
1.07
2.12
1.15
9.3 9.5 8.6
3.25 2.70
'
54
.o
3.76
0.j
5.1
I .jj
0.529
4.2
1.36
0.572
I2
jo
1.49 1.47 I. jo
Distance Screening of ratio target R. cm. 1.25
1.44 1 . 54
1.69 1.71
1.91
195
I . 23
293
390
1.09 I .06
198
2.10
2 90
1.43
403
I.48
Ave. I .49
TABLE I1 White Phosphorus R.H.
90
90
60
60
60
Time Xumber Wt. after particles sample dispersion N X 10-6 from 20 1. min , mg.
6 16 26
2.68 2 5 . 7 1.84 19.7 I. IO
15.2
Pl05
Diam. of Density Distance Screening
by anal. mg.
part. D x 104 cm.
of pari.
3.69 2.48 1.93
0.938 0.976 I ,063
1.11
198
1.10
247
I. IO
326
3.66 3'40 3.18
1.08 I .08 1.08
204
j , 2j
d.
of target em.
ratio
R.
6 16 26
3.65 35.1 1.89 2 5 . 5 0.94 19.2
3.71 2.03
0.947 1.061 1.218
6 16 26
2.78 1.35
4.9 3.3 3.4
3.23 1.96 1.91
0.467 0.j26 0.573
1.6; 1.60
311
1.19 0.91
1.57
412
1.17
6 16 26
2.78
5.9
3.46
0.502
I .60
27j
I .j I
1.2j
4.0
2.77
0 . j70
3.8
1.73
0.610
384 494
1.22
1.10
1.65 1.45
6 16 26
1.86 1.89
8.j 4.2 4.2
3.46 1.91 1.85
0.678
1.40
0.527
1.45 1.44
262 357
1.47
j19
1.87
1.10
I.2j
2.62
0.606
4.18 317 3.47 Av. 3.86 250
2jO
Av.
1.59 1.76
1.41
OBSCURING POWER O F SCREENING SMOKES
I079
Experimental Data. The experimental data and the screening ratios calculated therefrom are shown in Tables I t o VI. For each smoke tested, runs were made a t 60 and go per cents relative humidity in order to test the effect of humidity upon particle number and size. The actual humidities were within about one per cent of the humidities desired. Temperatures were recorded but as they had no discernible effect upon the results of the tests they are omitted from the tables. Samples of the smoke were taken and observations were made a t 6, 16, and 26 minutesafterdispersion of the smokes.
TABLE I11 Titanium Tetrachloride R.H.
90
90
90
90
60
60
60
60
Time Yumher Wt. after particles sample dispersion S X IO-^ from 20 1. min mg.
by anal. mg.
Diarn. of Density of part. Dx104 part. d. em.
Distance Screening of ratio target R. cm.
0.614 0,599 0,796
.44
IO1
1 .14
I .82
171
0.90
I .46
2 j0
I
I.jI
0.513
I .24
I34
I .7r
I .21
0.421
I
247
I .02
0.95
0.391
.61 I .89
320
0.71
3 22 3.18
0. j89 0.664
I .48
146
1.28 1.15
6.4
2.54
0. j27
I .82
207
I .04
3.62 I .38
9.6 4.1
2.02
0 .j66
1.40
I .68
0,555
I
I .02
3.4
I
.30
0 .j63
I .78
j.8j 2.37
6.5
4.08
0.3j8
2.32
i .3
3.22
.4j
3.6
2.32
0.j36 0.461
.91 2.42
8.96
10,3
2
162
I
5.4 3.5
5 .oo 3 .54
0.379
3 .60
0.391
2.39
302
I .31
2.64
0.472
2.61
15I
0.96
9.9 6.1 4.9
3.58 2.78 I .8j
0.46j 559 0.672
1.74 1.95
162
1.49
0
28;
I .20
1.77
448
I
7.3 7.9 3.5
3.46
0.384
.io
0.5'0
1.99 I .69
162
2
290
1.16 1.99
I
.95
0.165
2.17
464
I .20
6 16 26
4.82
6 16 26
6.17 1 0 . 8 2.96 3.8 I .86 2.2
6 16 26
5 73 2.86 2 .30
6 16 26 6 16 26 6 I6 26
TiOI
I .86 I .lo
1
1.22
6 16 26
0.8;
6 16 26
3,37 1.53
j
.42
I .;I
6.21
16.8 7.6 10.8
16.7 13 . o
3.86 3 .04 2
.54
I
I
.36
.78
I
01
82
104 204
320
186 326 197
.i4
0.95 0.68 0.81 I .09 I
.74
I.20
.64
.38
I080
P. D. WATSON A N D A. L. KIBLER
TABLE IV Ammonium Chloride R.H. Time Number Wt. KH&I after dis- articles sample by persion x Io* from 2 0 I. anal. min. mg. mg.
90
90
90
90
90
90
90
75
75
60
60
60
60
6 16 26
5.26 2.65 1.53
6 16 26
4.02 3.62
6 16 26
5.54 3.04
6 16 26
4.94 1.91 1.35
6 16 26
4.66 2.24 1.28
2.07
4.4
0.14
4.2
1.22
-_
-
-
-
0.521
0.648 0,452 -
1.04 -
I43 165
1.01
207
1.08 -
2 IO
I95
1.60 I .04 1.26 -
241
-
I .03
I43
0.592
1.03
I 56
1.89 1.30
0.545
1.03
I95
1.25
0. 512
I .os
1.09 0.95
0.481 0.612
1.12
0.52
1.08
I43 I49 189
0.478 0.582 0.547
1.05
I80
I .06 1.05
201
1.51 1.20
2.3
0.95 0.95 0.41
i5
0.82
1.9 2.6 IO.4
0.68 0.68 0.41
0.377
1.10
2.j
0.82
1.30
3.8 1.9
4.34 1.66
2.4 -
2.75
3.07
6 16 26
2.67 1.61
6 16 26
I .09
Density Distance Screening of of ratio part. d. target R. cm.
0.551
6 16 26
6 16 26
a. I -
Diam. of part. DXro4 em.
2.19 1.20
6.8 4.8
0.95 0.68 0.41
7.3
1.22
10.0
2.5
3.5 5.6 4.9
0.71
-
2.91
-
I .IO
1.2
1.17
1.9
0.472
1.08
207 217
0.936
I .OI
247
1.09
1.22
434 0.593
1.08
0.68
0.502
1.10
I .36 0.82
0.356 -
1.17
146 156 192 650 695
0. b8
1.22
592 708 839
1.16
I ,08
1.21
0.54
0.417 0.400 -
1.19 -
0.449 -
1.09
576 601
1.63
0.340 0.342 0,334 0.420 0.3ii
1.23
461
1.18 1.18
552
1.81 1.03
677
0.71
1.78 0.96
0.653
1.14
351 439 j76
3.19 1.94
2.0
1.17
0.7
1.36 0.68 0.41
6
4.34
16
2.02
2.2 1.0
1.22
26
1.30
0.6
0.68
6 16 26
3.66 I .96 0.89
3.4
2.18 2.04 0.95
1.9
781
0.58 0.69 0.49 2 .so -
0.524
6 16 26
1.3
-
-
-
0.71 0.83 2.04
1.26
3.2; I .63 1.15
-
0
0.75
0.436
95 0.54 2.04 0.82 0
6 16 26
3.o 1.3 -
2
1.46
-
1.20 I . 18
SO3 604 750
-
Av.
2.11
2.24 I . 24 I . 84
-
1.28 I. 30
OBSCURING POWER O F SCREENING SMOKES
I081
TABLE V Oleum R.H.
90
90
90
90
60
60
60
Time Number Wt. after dis- articles sample persion k x r o b from zol min mg.
6 16
3.60 2.70
5.9 0.8
26
1.17
0.9
6 16 26
8.20
21.4
3.19
so
3
by anal. mg. 2
.o
Diam. of part. DX104 cm. 0
I
5
0
0
Density of part. d.
Distance Screening ratio
R.
I .25
2.26
.52 .19
0.49
.I8 19
2.08 I .6j
0.5
0.265
I
0.25
0.395
I
5.4 4.2
0 . j96
I
15.7
I .
2.07
12.7
3.4
0'734 0 .j 9 0
10.8
3.6
0,582
1.25
1.16
.6
I .20
0.77 0.93
I ,
19
0.51
I
.67
6
4.18
16 26
2 .07
1.5
0.599
.07
7 .O
2 . 2
0.798
1.23
3.9 2.9
0 . 542
1.33
0.687 0.686
I .22 I
.23
0.98 0.98 I .04
I
j
6
j .02
11.1
16 26
2.27
.68
9.4 7 .O
6 16 26
6.29
12.9
4.4
.25
I
3.65 1.45
2.8
0.539 0.452
I
5.1
I
I .02
4.6
2.4
0.599
I
.44 .41
6
3.27
-
-
-
16 26
I .TI
7 .3 6.5
3.7 2 .6
o ,663
I
.40
I .I2
0.699
I
.30
1.33
6 16 26
3.90 I .86
12.5
6 .o
6.7 5.6
1.5
0.607 0.667
2.6
0.735
1.37 1.16 I .36
I .34 0.99 0.88
I
I
.40
0.99
2.2
.88
0.97
-
-
AV. I
.20
1082
P. D. WATSON AND A. L. KIBLER T.4BLE
\?
Chlorsulfonic Acid R.H.
90
90
90
60
60
60
60
Time Number Wt. after dis- particles sample persion S X IO-^ from 20 1. min.
6 16 26
6 16 26 6 16 26 6 16 26 6 16 26 6 16 26
6 16 26
Diam of. part DXIO'
Density of part. d.
-
I .63
0
2.12
o 516
I .02
j .22
o 460 o 486
I .13
2
.io
2
.83
482
I
1.08
130 496
5.10
0
2.17
0
I .84
0 512
IO
-
I
.08
Distance Screening ratio of target R. rm. I22
-
146 198
0.88
131 16j
0.83
0.43
1.13
-
232
131
0.9;
I .08
1j6
I .08
201
0.6; 0 . j6
j
.26
0
lil
1.16
232
2.1j
2
.i5
0
329
I.27
275
0.64
I .81
0 jII
I . I j
348
I .29
6.81
0
3i2
I
214
.17
0
I
I .58 0.53
1.73
0
3.59 508 348
2
.33 .62
I I8
241 296 16; 192 262
I .04
j .02
0
2.24
0 311
I .27
1.25
o 380
I ,2 5
139 533 -
I.27
28;
I
I.27
3'4 390
I .25
3.82 1.79 2.22
0
0
I
.22
-
0.79
0.33 0.37
.66 -
h v . 0.96 Discussion of Data Ratio. ?jot including phosphorus smoke at 9 0 7 ~relative humidity the average of the screening ratios is 1 . 2 7 . If results be omitted which are manifestly out of line, due mainly to very small smoke samples, the average ratio would be somewhat higher than this, in the neighborhood of 1.35. For spheres with uniform packing, the theoretical ratio for complete blocking out of the background is 1.57. Again excepting phosphorus smoke at 90Tc relative humidity it will be noted that the diameters of the smoke particles lie substantially between the limits 0 . 4 0 X 10-4 cm. and 0.75 X IO-?cm. which define the range of the wave lengths of visible light. The average diameter of the white phosphorus
OBSCURING P O W E R OF S C R E E N I N G SMOKES
1083
smoke particles at 90% relative humidity is 1.03 X IO-^ cm. while the average screening ratio is 3.86. The obscuring value of white phosphorus at high humidities is little better than it is at low humidities in spite of the large amount of water absorbed by the particles and their consequently large size, These data point to the conclusion that for smoke particles whose diameters are within the range of the wave lengths of visible light screening is accomplished by a simple blocking of the light rays. For particles of this size the composition is not a factor. For particles whose diameters are larger than the wave lengths of visible light, screening is accomplished by a different mechanism and the refractive index or other physical properties of the smoke particles may be a factor. These and other conclusions in this report apply only to white smokes having liquid particles. Some work was also done on a black carbon smoke having solid particles but on account of the irregularity in size and outline the projections of the particles could not be calculated. It should also be mentioned that the screening ratio named above has been proved to apply only to the experimental conditions which obtain in the smoke chamber used. There is some evidence that smokes are more efficient in screening out of doors than in the smoke chamber. If this is the case, it is probably due to such factors as the relative position of observer, smoke, and target, to the intensity and reflection of light, variation of particle size, and other such causes. Efect of H i m i d i t y . The obscuring power of all the smokes was greater at 90Tc than at 6 0 7 ~relative humidity, and, in some instances, this was very marked. This was due to the larger size of the particles at the higher humidity caused by the adsorption of moisture. If the absorption of moisture increases the diameter of the particle beyond the wave lengths of light, as in the case of white phosphorus smoke, the increase in screening power is not proportional to the increase in particle size. In the case of ammonium chloride smoke the increase in screening power at 90% relative humidity was very large. By a consideration of the data of Owens' with the aid of the vapor pressure of saturated ammonium chloride solutions2 we were able to show that a t 6 0 7 ~ relative humidity the ammonium chloride particles were composed of fine crystals, at 90e; they were liquid while for the two runs at ; 5 % the particles were just on the border line between liquid and solid. These data check w r y well with the observed screening power. The data in the tables give us a means of estimating roughly the amount of water absorbed by the smokes studied. Such an estimate, based on a careful analysis of the data, is shown below. Equilibrium with the moisture in the air is, in all cases but one, reached before 6 minutes and the values shown in the table represent an average of all observations. J. S. Owens: Proc. Roy. SOC.,llOA, 738 (1926). Edgar and Swan: J. Am. Chem. Soc., 44, 570 (1922).
1084
P. D. WATSON AND A. L. KIBLER
TABLE VI1 Parts of Water absorbed by Smokes At 90% R.H.
Smoke material
At 6 0 7 ~ R.H.
ZnC12 3 2 P2Or 7 5 I Ti02 1 5' I KH,Cl 5 O S SO3 (from oleum) 2 5 I SO8 (from chlorsulfonic acid) 5 2 I t is interesting to compare the concentrations represented by the proportions of substance and water shown in the above table with the concentrations of solutions in bulk a t equilibrium with air under the conditions of humidity and temperature which obtained in our experiments. In addition to ammonium chloride, which is discussed above, the smokes from phosphorus, oleum, and chlorsulfonic acid lend themselves to such a comparison because for these smokes the composition of the smoke particles is known with a fair degree of accuracy. From the analytical data it is assumed that the smoke from chlorsulfonic acid is composed essentially of sulfuric acid. For sulfuric acid sufficient data exist in the literature for complete vapor pressure curves at any concentration and temperature. For phosphoric acid the data are not so complete. By the aid of the Clausius-Clapeyron equation, however, vapor pressure curves were drawn from the data available for phosphoric acid. A comparison of the theoretical concentrations a t equilibrium with the air a t the temperatures and relative humidities obtaining in the experiments, with the concentrations represented by the analytical data in tables, is shown below. The comparison is expressed in terms of the ratio of water to SO3 and P20s,respectively.
TABLE VI11 Comparison between Theoretical and Observed Parts of Water absorbed by Smokes Relative Ratio H20/SOa or PIOs Smoke humidity
Oleum Oleum Chlorsulfonic acid Chlorsulfonic acid White phosphorus White phosphorus
Theoretical
7
Experimental
90 60
2
I
90
8
5
60 90 60
2
2
6
7.5
1.4
1 . 0
2.5
The comparison in the above table is as good as could be expected considering the roughly approximate nature of the ratios shown in Table VII. There In this case the absorbed water at 6 minutes is distinctly more than a t 16 and 26 minutes, and amounts to about 3 parts.
108;
OBSCURING POWER OF SCREENIXG SMOKES
is a tendency for the experimental values to be lower than the theoretical ones. This is probably due primarily to the well known fact that the vapor pressure of small particles is appreciably higher than for the same liquid in bulk, tending to concentrate fine droplets of solutions. Eficiency of Dispersion and Stability of the Smoke Cloud. The efficiency of the method of dispersion employed may be judged by the amount of smoke material in the air at the end of 6 minutes while the stability of the cloud may be judged by the relative amounts still remaining in the air after 16 and 26 minutes. From the analyses of the samples of smoke collected from 2 0 liters of air the averages of the amounts of the several original materials found present a t the end of 6, 16, and 26 minutes have been calculated and expressed in terms of percentages of the weight of substance originally dispersed. Both humidities have been averaged in together as the recovery was about the same for both. These figures are shown in the following table together with the percentage decrease from 6 to 26 minutes. These last figures represent the stability of the cloud.
TABLE IX Efficiency of Dispersion and Stability Amount found1 in % of Substance
original substance after 16 min. 26 min.
6 min.
H. C. candle
W.P. F.M. Oleum Chlorsulfonic acid IL”4C1
53.0 68.3 74.0 53.6 I5 . 2 20.4
37.3 45.4 59.0
28.0 12.4 16. I
28.3 36.8 44.8
26.0 4.6 9.0
Percentage decrease 6-26 min.
48 46 40 51
63 56
When the percent of smoke still in the air from Table IX are plotted on the logarithmic scale against the time in minutes, on semi-log paper the points define straight lines which are practically parallel. The intersection of these lines on the ordinate representing zero time represents the actual efficiency of dispersion expressed in terms of per cent of the original substance taken. The efficiencies determined in this manner are shown in Table X.
TABLE X Actual Efficiency of Dispersion of Smokes a t Zero Time Substance Efficiency as yGof original substance H.C. 65 W.P. 80 F.M. 87 Oleum 66 Chlorsulfonic acid 20 xH4C1 32 1
Calculated back to the original material dispersed
1086
P. D. WATSON AND A . L. KIBLER
The low efficiencies of dispersion for chlorsulfonic acid and ammonium chloride smokes are probably due to insufficient heat to bring about decomposition in the case of the former and incomplete reaction between K H 3 and HC1 in the case of the latter. The stability is about the same for the six smokes studied, chlorsulfonic acid being the most unstable. This again is probably due to inefficiency of the method of dispersion. The average decrease between 6 and 26 minutes is 50%. On the whole, Table X indicates a much lower efficiency of dispersal than is ordinarily assumed. It should be emphasized that these efficiencies are calculated in a manner which eliminates the effect of precipitation and leakage in the smoke chamber. There is, therefore, no reason to believe that dispersal out of doors, by the same methods, is more efficient than the figures shown in this table. The stability of a highly non-volatile cloud out of doors is certainly greater than in the smoke chamber. In the main two opposing factors are operative. One of these is the precipitation of smoke particles in the chamber, due to forced ventilation. This occurs particularly around the fan blades and at the bends in the ventilating ducts. (There is also a small but unknown amount of leakage.) The other factor is the increased tendency to evaporation out of doors where the cloud is being continually diluted with fresh air. Composition o j the Smoke Particles. We are justified in the assumption that smokes from the H.C. smoke candle, white phosphorus, oleum, and ammonium chloride consist, respectively, of droplets of ZnCl?, P205, SOs and NH4C1with water since there is no likelihood of any further reaction of these constituents. I n the case of chlorsulfonic acid it is known that heat is necessary for an efficient dispersal. It is stated in Thorpe’s “Dictionary of Applied Chemistry,” 5 , 301, that chlorsulfonic acid is decomposed by heat according to the following equation : c1
The reaction is complete when the vapors are heated to zoo°C. Chlorsulfonic acid also reacts with water according to the following equation : C1
/ + H20 \
SO2
=
H2S04
+ HCl
OH
If the first reaction takes place before the chlorsulfonic acid is exposed to the moisture of the air, and if the temperature is sufficiently high to bring
1087
OBSCURING POWER O F S C R E E N I N G SMOKES
+
about complete decomposition, the part of the products represented by SOz CI?, being gaseous, is lost for the formation of smoke while the part represented by H20 SOs is completely utilized. This latter part represents only 42y0 of the original molecule. If chlorsulfonic acid is atomized in the cold we might expect a more complete utilization of the original molecule, but the fact that this is not true indicates that the cold chlorsulfonic acid may fall out of suspension before it has a chance to react with the moisture of the air. In the work described in this report chlorsulfonic acid was atomized by means of a rapid current of COS. Both the container holding the chlorsulfonic acid and the CO,line were heated by steam, the temperature varying from 11j to 13j°C. The reaction with heat was, therefore, incomplete. The evidence, therefore, indicates that at least a portion of the original chlorsulfonic acid, not decomposed by heat, fell out of suspension soon after atomization. This theory is strengthened by the fact that analyses of the samples collected from this smoke showed but traces of HC1 and also by the very low recovery of this smoke after 6 minutes. This theory indicates that the most efficient method to disperse chlorsulfonic acid is to heat the vapors to a temperature somewhat over zoo°C before discharging them into the air, but then only 4270 of the original material could be utilized. A somewhat higher efficiency than this might be obtained by a very efficient atomizer, using steam as the atomizing agent. Disregarding the HCl an efficiency of about 70yo is theoretically possible by this method. Some work has been done on the use of steam and it has been found that superheated steam gives a good smoke. The composition of the smoke by either method would be droplets of sulfuric acid, the HC1 being driven out by the heat of the reaction. The case of the composition of the smoke from titanium tetrachloride is more obscure and little seems to be known about the reaction of this substance with water. Consequently, some study was made of the data for this smoke and some additional experimental work was done in an effort to determine its composition. Unfortunately, the samples of smoke taken during the counting experiments were not sufficiently large to analyze for both Ti02 and HCl. Therefore, additional samples were collected and analyzed for HCI. The data from these analyses admit of only a rather rough estimate of the amount of HC1 in these smokes. The percentage of HC1 is somewhat more a t 90% than at 6070 relative humidity. The percentages of HC1 present in the smoke a t different time intervals and humidities is approximately as follows :
+
At 6 min. At 16 min. At 2 6 min.
6oC; R H.
9oC&R.H.
I3 '3
'9
8
IO
'7
Calculations of the composition of the smoke represented by the data in Table I11 were made on the basis of the above average contents of HC1. Though the individual variations in composition are quite large, it is prob-
1088
P. D. WATSON AND A. L. KIBLER
able that average figures are fairly representative of these smokes. The results expressed in percentages of TiO,, HzO, and HC1 are shown below.
TABLE XI Composition of Titanium Tetrachloride Smoke in Percentages Time
6 minutes
16 minutes 26 minutes Average
607~ R.H.
90%
R.H. Hz0
TiOz
HzO
HCl
TiOt
49 47 59
38
I9
62
I9
40
13 I3
28
I7
33 37
8
36
55 54
I1
52
HC
IO
The composition of the smoke a t 60% relative humidity appears to be fairly constant from 6 to 26 minutes, and is represented by the following: 2
TiOz
+ 6 H20 + HCl
or the composition may be written Ti(OH)4
+ HzO +
1/2
HCl
Under ordinary conditions of moisture, it appears that the compound originally formed, TiC14.gHz0, begins a t once to hydrolyze, splitting off HCl which escapes from the particles quite readily since most of it is gone in 6 minutes. At 90% relative humidity the relative proportions of HzO and HC1 change rather rapidly after the first 6 minutes. The compositions a t 6, 16, and 26 minutes are represented approximately by the following: At 6 minutes.. . . . . . . . . , . . , , , . .Ti(OH)a At 16 minutes.. , . . . . . . . . . . . . .Ti(OH)a At 2 6 minutes., , . . , . . . . . . . . . . .Ti(OH)a
.
+ + +
12
HzO
7 HzO 4 H20
+ z HCI + 1.5 HC1 + 0.5 HC1
The change in composition is due to the loss of both H20 and HC1. Hydrolysis of the TiCh.5HrO, which was probably first formed, is complete before 6 minutes when the composiion of the smoke is a solution of Ti(OH), with some HC1 dissolved in it. Limits of Accuracy. The effect of several sources of error in these experiments are briefly discussed below. The depth of focus of the microscope was utilized in determining the volume of the field containing the particles to be counted. The normal eye is capable of accommodation within limits; consequently, the depth of focus utilized by the observer in viewing the particles may differ from the depth of focus as determined experimentally. It is also not necessarily true that only the particles falling within the depth of focus appear as points of light and not as disks. Particles too small to be detected by the ultramicroscope may have been present, but it is thought that the effect of this error would be
OBSCURING POWER OF SCREENING SMOKES
1089
negligible. These above mentioned errors would affect all the results to the same degree. J17hytlaw-Gray,1in a recent article discusses the errors due to the scattering of light within the cell by the smoke particles. I t was thought possible that the filters might not collect the smallest particles present. X check was made by placing two filters in series, but the second filter showed no increase in weight. If any errors were introduced in tbis manner they were quite small. The density of the smokes read from tables and determined experimentally probably does not exactly represent the density of the particles in the cloud, and the errors may be considerable in some cases. M o s t of the smokes examined are hygroscopic and hence very sensitive to conditions of humidity and temperature. It was, therefore, important to weigh the samples of smoke immediately after collecting them to avoid errors due to a change in weight due to loss or gain of moisture. The temperature of the observation room was kept as near as possible to that of the smoke chamber but the humidity could not be controlled. TVhen the humidity of the observation room was markedly different from that of the smoke chamber and the smoke being tested was a very hygroscopic one, like phosphorus, the crror due to this cause was likely to be considerable. I t is thought that errors in the visibility readings of the target were probably of less importance than many of the other errors. The illumination and the point of observation from the cloud were constant for all smokes, and any divergence in this respect from the true obscuring power would be constant. The chief error would lie in the personal factor of the observer’s eyesight. Conclusions I. For smokes consisting of spherical particles of diameters within the range of the wave lengths of visible light the average of the screening ratio (ratio of the total cross-sectional areas of the smoke particles to the area of the target) is 1 . 2 7 . I t is indicated that the correct ratio is somewhat higher than this, probably about 1.35. This ratio holds true for the conditions under which the smoke measurements were made in the smoke chamber. 2. For particles of diameters within the range of t,he wave lengths of visible light (0.40 to 0 . j 5 X IO+ cm.) the screening power is independent of t h c composition of the smoke particles and dependent solely upon the number and size of the smoke particles between the eye and the target. 3 . Phosphorus smoke at 90% relative humidity consists of particles cond e r a b l y larger than the wave-lengths of visible light. The average value of thc screening ratio for this smoke is 3.86. It is indicated that particles of this qizr screen by a different law from that applying to particles of diameters cqual t o the wave-lengths of light. Some light may go through the largc phosphoric acid solution particles. It is possible that for particles larger than the wave lengths of light the physical properties of the substance of which the particles are composed, in particular the refractive index, may be a factor. I
Proc. Roy. Soc., 116A, j40 (1927).
I090
P. D. WATSON AXD A . L. KIBLER
4. For smokes having particle diameters within the range of the wavelengths of visible light, there is no evidence of any regular change in the value of the screening ratio with the change in size over the comparatively small range within which the particles for these smokes vary. j. The results suggest that under the conditions obtaining in the smoke chamber, smokes consisting of particles whose diameters are within the range of the wave-lengths of visible light obscure a non-luminous target when the number of particles between the eye and the target are sufficient, when projected, to form a uniform layer of approximately one particle in thickness. The actual average ratio as stated above was 1 . 2 7 , but it is indicated that the correct ratio is somewhat higher than this. For spheres with uniform packing the theoretical ratio for complete blocking out of the background is 1.57. 6. The obscuring power of efficient smoke-producing materials is to a large extent due to atmospheric moisture which is absorbed from the air by hygroscopic smoke particles. This is the explanation of the superior obscuring power of smokes a t high humidities. It is indicated by the case of phosphorus smoke, however, that there is an upper limit beyond which the absorption of water from the air fails to be reflected proportionally in increased obscuring power of the smoke. This limit is reached when the smoke particles exceed in diameter about 0.75 X IO-^ cm. 7 . For all the smokes studied the precipitation in the smoke chamber followed a logarithmic law, according to which the logarithm of the amount remaining in the air a t any given time was proportional to the time.