RATE OF SEDIMENTATION

Concentrated Flocculated Suspensions of Powders

HAROLD H. STEINOUR
Portland Cement kssociation, Chicago, 111.

Hates of sedimentation are reported for concerlIratecl flocculated suspensions of various finely divided solids, iucluding microscopic glass spheres. Most of these solids were tested at more than one fineness. Each powder embraced a wide range of partirle sizes and was tested at a

series of concentrations. A rate equation previously found applicable to flocculated suspensions of approximately uniform-size emery particles was, in general, supported by the new data. The equation is shown to be compatible with Powers equation for portland cement pastes.

N THE second articlt. of this series ( I S ) 13quatiori 3 for rate of Sedimentation was shown to be applicable to concentrated, flocculated suspensions of emery particles of practically uniform size. I n the present article this rate equation is applied to flocculated suspensions of powders which contained particles of many sizes, and were in this respect, more typiral of suspensions in general. The rate equation is

involves the assumption that the particles can be treated the Same as uniform-size particles having thc same value of u.
DESCRIPTION OF LOWDERS

(Eq. 3, 13) This equation may be re~ardctl as a modification of one applied by Powers (9) to the sedimentation of portland cement pastes. In the present studies i t was developed first for nonflocculated suspensions, in which uniform-size particles were used in order to obtain uniform settlement. I n :ach suspensions of emery particles, empirical constant wiis needed apparently because of liquid which remains with the particles during their fall; during sediment,ation the main mass of liquid moves relative to the partii / ( 1 - wi) per unit volume cles, but a quantity of liquid equal to w of solid appears not to take part in this flow but to remain stagnant a t the surfaces of the angular particles. In the first article of this series (12) wiwas shown to be zero for nonflocculated suspensions of spheres; evidently the rounded surfaces did not keep any of the liquid out of the flow. When concentrated suspensions of the emery particles were flocculated, Equation 3 (13) was still found to apply, but with a higher value of ini. Apparently the conditions that control the rate of settlement remained much the same as before; indeed, experiments indicated that the initial resistance to settlement remained wholly of viscous origin, and the appearance of the suspensions was con.&tent with the assumption that the displaced liquid still flowed past the individual particles. That is, there appeared to be no opportunity, such as occurs a t low concentrations, for the flow- to by-pass groups of particles. To apply Equation 3 (13) to powders containing particlcs of many sizes, V , is formulated as follows:

where u = specific surface value, sq. cm./cc calculated as if each particle were a sphere ha&g same density as particle and same rate of fall when alone in a large volume of viscous li uid. For uniform-size particles it is equal to 3 7 ~ where , T is radius of equal-settling sphere. Numerical value of u is obtained from sedimentation analysis of a very dilute, nonflocculated suspension. With V , calculated as in Equation 1, the application of Equation 3 (13) to flocculated mixtures of particles of various sizes

Tu make the testing of Equation 3 (23) fairly thorough, powders were used which differed in chemical composition, in fineness, and in characteristic particle shape. These powders comprised five finenesses of ground glalis, two of microscopic glass spheres, three of burned shale, one of silica, and two of limestone. The glass and burnet1 shale were ground in a laboratory ball mill. The spheres were made by feeding some of the ground glass into a blast lamp (10, 11). The silica and limestone were purchased already ground. To obtain different finenesses, airseparation and mixing were employed in some cases. Microscopic examination (12) showed that the preparations of spheres contained relatively few particles that were not well shaped. The particles of burned shale werc shown t,o havc a comparatively rough surface texture. The values of u, the specific surfaces of equivalent spheres, were determined from sedimentation analyses, essentially by the Wagner turbidimeter method (14), using the equipment specified for testing cement (A.S.T.M. standard test (3115-42). All turbidimeter tt:sts were made in water, using a little sodium hexxmetaphosphate as dispersant. The size analyses were commonly carried down only to a particle diameter of 5 microns, but the contribution of the finer sizes to the value of u was estimated by a method devised by Dah1 (5). By this method a cubic equation is assumed to fit the size-distribution.curve below 10 microns. Since u enters Equation 3 (13) as a part of V,, its determination by sedimentation analysis is wholly in accord with the theory of the equation. However, since the sedimentation of a concenhated suspension is in some respects similar to flow through a bed of particles, specific surface values determined by the air permeability method of Lea and Nurse (7) were obtained for comparison with the u values. Indeed, since the glass spheres used in this investigation varied somewhat in density because of enclosed gas, the specific surfaces of both preparations were determined solely by this air permeability method; the method does not require that the particles have the same density, and it seems to give the actual specific surface of spheres (6, 12). Table I shows the specific surface values obtained by each of the two methods, sedimentation and air permeability. Values calculated from air permeability tests (by the Lea and Nurse formula) are identified bysymbol So. ,411 values, whether of So or u, are averages from two or more tests, except u for burned shale 1 which is a selected result (4% above the average) believed to represent the best-dispersed sample. Table I gives some of the data from the turbidimeter size analyses and values of wi

901

902

INDUSTRIAL AND ENGINEERING CHEMISTRY TABLE I. PHYSICAL PROPERTIES OF MATERIALS AND VALUESOF w i OBTAINED IN SEDIMENTATION TESTS

Vol. 36, No. 10

Reading3 were taken with a micrometer microscope sighted on a disk float which rode a t the surParticle Sizea, Cumulative % Densface of the suspension (9, I S ) up to Particle Diam. of: Sp. Surface, w i in Sediitj,, Flocculants were chosen accordGrams/ 5 10 25 60 _____ Sq. Cm./Co. mentation Powder Cc. microns microns microns microns u SO* SO/U Tests ing to the requirements of the Ground glass 1 2.50 6.9 2.3 24.0 80.5 1.25 0.170 2615 3,290 several powders and are given in Ground glass 2 2.50 2.4 8.7 27.1 79.2 2720 3,570 1.31 0.195 Ground glass 3 2.50 5.2 11.8 31.6 83.0 3610 4,040 1.12 0.175 Table 11. Waterwas thesuspenGround glass 4 2.50 6.7 84.9 4220 5,240 1.24 0.190 18.1 40.2 sion medium. The conditions of Ground glaC 5 2.50 12.8 29.2 59.5 87.6 1.16 0.185 6700 7,500 Glass spheres 1 2.35 .,,, .... .... .... ..._ 2,980 (1.00) 0.100 sedimentation ensured laminar Glass spheres 2 2.32 , ,,, .... .... .... .... 3,990 (1.00) 0.095 flow of the displaced water in all Burned shale 1 2.57 7.4 19.7 47.7 79.8 4710 9,130 1.94 0.315 cases as judged by comparison Burned shale 2 2.57 (11.3) (26.6j (60.5) (89.0) (6320) (11,160) (1.76) 0.312 Burnedshale3 2.57 15.2 33.6 73.4 98.2 7940 13,200 1.66 0.315 with the work of Carman (4)on Silica 1 2.65 12.7 22.1 44.9 82.9 5850 8,500 1.45 0.203 (lime) flow through granular beds. 0.165 (dye) The curves of height of susperiLimestone 1 2.78 11.7 19.0 36.2 63.2 6650 6,050 LO7 .. . . . . . , Limestone 2 2.78 18.0 29.4 54.6 82.9 8320 9,980 1.20 ....... sion against time were similar to Parentheses enclose values not obtained by direct experiment. Burned shale 2 was a mixture of I and, 3. those obtained in the previous Glass spheres were not analyzed for size distribution, but microscopic inspections showed that particles varied studies of sized particles in that widely in sine. * Porosity of test bed was sometimes varied in check tests, results of which were averaged. Mean porosities they exhibited an initial straightwere: ground glass,,0.49, except 0.52 for No. 5 ; spheres 1 and 2, 0.47 and 0.45, respectively; burned shale, 0.47; eillca, 0.49; limestones 1 and 2, 0.45 and 0.51, respectively. line portion followed by some rounding off, The coarser the powder, the more dominant was the straight portion of the curve. This is illustrated by Figure 1 which will receive attention later. All powders of a given material or type are numbered in the order of increasing fineness. in which coarse and fine glass powders are compared by plotting so that the initial slopes of the curves are the same. I n most SEDIMENTATION T E S T S cases the curves ended rather abruptly, but a somewhat prolonged slow final settlement was characteristic of the silica powder In the sedimentation tests on the flocculated suspensions the (Figure 2 ) . Limestone 2 showed peculiar behavior; its fineness mixtures were stirred with an electrical mixer for at least 2 minwas about the same as that of burned shale 3, but the settlement utes and were tested in a cylindrical jar 100 mm. in diameter, was much slower a t the same dilutions and continued fora long time using depths of sample between about 30 and 60 mm. Previous at a rate that remained a large fraction of the initial rate (Figure work had indicated that wall effects do not significantly influence 3). The greater the dilution of suspensions of limestone 2, the the settlement a t the center of the vessel under these conditions. more the curves tended to round over and to fail to show a long initial straight-line portion, the opposite of the usual tendency. 100 Rates of sedimentation calculated for the initial constantrate periods are given in Table I1 together with the porosities of the sediments and supplementary data.
90
RATES OF SEDIMENTATION
80

2 al

-E
C
D c

Figure 4 shows the degree of conformity between Equation 3 of the second article (15) and the rate data from Table 11. If the equation is applicable, the data for a set of tests in whirh

70

60

?
P

50

9 S
L

6
40

v ,

P B
v)

5 30
20

10

0
0
10 20

30

40

50

Time, (unitschoren to give same initial slope)

20

40

60
Time, minutes

100

120

Figure 1. Sedimentation Curves for Ground Glass Powders 1 and 5 in 0.25% Zinc Sulfate Solution at B = 0.60

Figure 2. Sedimentation Curve for Silica Powder in One-Sixth Saturated Lime Water at e = 0.62
(Height o f suspension, 57 Mm.)

October, 1944

INDUSTRIAL AND ENGINEERING CHEMISTRY

903

TARI,E IT. SEDIMENTATION OF FLOCCULATED SUSPENSIONS AT VARIOUSDECREES OF DILUTION

Test Flocculant, % r i f Water Wt. Zinc sulfat,e, 0.26 (Chry nologital) 5 1 2 3
0

No.

PorosFluid ity of ConSeditent, Settling Settlement, ment, % i n Rate Total Cm.8' Iiiitinl % of % Initial Settled Vol. See., Height, VOl. Mm. ( x 102) x 108 Height

of

Flocculant, % of Water Wt. Lime water,

SHtd.

4 7 8 Zinc sulfate, 0.25

8

Ground Glass 1, 26.6' 52.5 43.8 52.5 43.8 55.0 40.3 57.5 49.0 57.5 49.0 52.0 00.0 02.5 50.0 59.0 05.0

C. 8.4 9.1 11.9 14.2 13.7 17.1 20.0 22.7 C. 8.3 8.0 11.1 12.0 14.8 17.1 20.3 22.7 C. 8.4 10.3 12.3 11.4 15.6

48.1 47.8 48.9 50.5 50.8 51.8 52.8 54.7 48.2 48.1 49.4 51.4 53.1 54.8 50.1 67.9 47.9 49.5 50.8 51.4

1 2 3 4 5 6 7

(;round Glass 2, 20.1' 680 43.8 52.5 637 52.5 43.8 843 55.0 40.3 1088 57.5 49.0 1360 52.0 00.0 1810 02.5 56.6 2320 65.0 59.0 2930 07.5 04.0
(;round Glass 37.4 52.3 54.7 39.4 56.9 41.4 50.9 41.4 59.4 44.0

Lime s a t e r . satd.

Zinc sulfate, 0.25

1 2 3 5 4 1 7 9 2 3 8 4 5

3, 25.2' 426 527 689 030 906

Fluid ConTest tent, Settling NO. % in Rate &, Total Cm./ (Chrq- Initial Sec., Vol. nologi- Height, ( 6 X 109 X 108 cal) Mm. Burned Shale 2, 26.0' 57.5 125 1 39.1 60.0 155 2 43.7 02.5 193 3 45.1 4 45.0 65.0 258 07.5 351 5 40.2 445 40.3 70.0 0 579 40.4 72.5 7 75.0 744b 8 46.1 Burned Shale 3, 25.7' 57.5 70 3 39.2 57.5 78 39.8 10 60.0 92 42.0 2 f28 42.2 I 02.5 4 05.0 150 43.7 208 44.4 5 07.5 281 44.1 70.0 6 272 45.5 70.0 7 370 46.2 72.5 8 75.0 507 46.4 9 485 75.0 45.7 11
1 2 3 4 5 0 7

ity of Settle- Sediment, ment, % of % of Initial Settled Height Vol.

Poros-

C.
4.2 6.0 7.8 9.7 11.8 15.5 20.2 25.7 C. 2.4 2.0 3.6 4.8 0.9 11.5 11.1 14.0 19.7 26.9 22.9 55.6 57.4 59.3 61.2 63.1 64.5 65.5 66.4 56.4 56.6 58.5 60.6 62.4 63.2 66.3 65.2 65.8 65.8 67.6 53.0 54.3 54.5 56.0 57.0 57.8 58.4 48.4 50.0 50.0 51.8 52.4 53.2 53.5 48.2 51.0 53.1 54.2 52.3 54.5 56.4 58.6 60.0 62.0 64.1 64.2

01 . B

Lime water, satd.

Z i n c sulfate,

11.25

10 11 12
%tic sulfate, 0.25
1

Ground Glass 4, 25.7' C. 4.7 36.7 4.1 36.7 4.0 36.7 7.0 38.7 8.5 39.5 8.0 39.5 11.2 43.2 13.8 40.0 16.6 49.0 15.8 29.0 22.4 02.5 23.9 50.6 Ground Gla ss 30.7 50.0 43.2 50.0 38.8 52.5 40.7 55.0 43.3 57.5 39.7 57.5 39.5 57.5 46.0 60.0 49.0 62.5 40.3 02.5 52.4 65.0 50.6 67.5 53.9 67.5 61.2 70.0 5, 24.6' 103 90 113 138 154 202 194 242 300 310 392 533 506 679 C. 3.0 3.8 4.3 6.2 7.0 5.6 6.0 11.2 9.9 9.3 14.0 18.4 16.1 23.2

47.5 47.9 47.9 48.9 50.8 51.0 52.1 53.5 55.0 55.5 54.9 57.3 48.4 48.1 50.4 52.0 54.3 55.0 54.8 55.0 58.3 58.7 59.3 60.1 01.2 01 0 33.2 36.2 36.7 37.9 40.9 41.0 41.2 42.0 44.4 42.3 43.0 40.0 45,9 38.1 38.0 39.7 40.8 40.0 41.3 42.3 42.0 42.3 55.5 56.5 58.6 58.2 60.0 59.9 00.8 61.7 01.7 62.3

49.0 52.0 54.0 58.0 00.5 04.0 68.0 48.5 49.0 54.0 57.0 61.0 64.5 67.0 46.2 45.0 46.0 47.0 30.4 30.2 31.1 31.9 32.7 33.3 34.1 34.2

Silica 1, 24.0' C. 199 6.4 50.0 58.0 254 7.9 318 12.0 00.0 370 13.7 62.0 470 10.3 64.0 576 19.5 60.0 701 23.0 68.0 Silica 1, 24.0' C. 56.0 228 58.0 258 338 00.0 422 62.0 554 04.0 683 06.0 913b 68.0 14.8 16..1 19.0 21.2 24.4 27.4 31.1

Lime wnter, sat,d.

1 2 3 4 5 0 7 2 3 1 4 4 3 1 2 5 6 7 8
1

10 2 3 4 11 13 5
6

Gentian 0.01 violet 0.04 0.01 0.02 Lime water, satd.

Silica 1, 24.0' C. 17.9 57.5 281 296 13.2 57.5 616 25.4 05.0 600 23.0 65.0 Limestone 1, 25.2" C. 5.5 55.0 112 57.5 153 6.6 60.0 190 8.2 62.5 208 9.2 273 12.4 65.0 304 14.5 07.5 467 16.4 70.0 23.0 72.5 910

14 7 8 12 9 Zinc sulfate, 0.25 10 11 12 13 2 3 4 5 1 6 7 8 9 7 8 1 2 9 3 4 5 6' J,iine water, 3atd. 7

Glass Spheres 1, 24.6O C. ( I 6.5 231 5.9 301 9.2 391 11.5 358 6.9 486 11.0 14.9 651 797 18.1 19.1 1025 22.0 1074 1421 24.7 25.1 1865 2550 30.8 Glass Spheres 2, 25.7' C. 3.1 25.6 40.0 3.3 25.0 40.0 4.6 26.0 42.5 20.0 45.0 7.0 7.4 27.8 45.0 10.0 29.2 47.5 13.2 30.6 50.0 32.2 52.5 18.1 34.0 22.2 55.0 Burned Shale 38.9 58.3 60.0 43.4 46.0 62.5 02.5 44.4 44.9 65.3 47.1 65.3 45.2 07.5 45.6 70.0 47 1 70.0 47 1 70 0
1, 25.5' C.

Limestone 2, 25.8' C. 35 .. .. 38.2 05.0 .. 33 .. 31.7 05.0 3 .. 54 .. 32.5 67.5 4 33.0 70.0 50 .. 5 .. 90 .. 34.0 72.5 6 .. 75.0 169 .. a 46.1 0 Sedimentation curve for this highly concentrated suspension was exce tional in t h a t the constant rate established initially changed abruptly to her one. Channeling was not observed and was im robable because of a the %igh concentration. The curve was somewhat like ttose previously ohtained with nonflocculated emery a t high concentrations (1.3). b Channeling observed. Lime water, satd.

only

has been varied should fall on a straight line for which

4 5 8 1 9 10 2 3

205 243 354 333 442 447 593 720 782 787

6.3 7.9 9.9 10.3 13.3 13.5 17.0 21.7 21.0 20.4

where the value of w,gives the intercept of the line on the -axis. The solid lines drawn through the data points in Figure 4 have these theoretical slopes. The graphs for ground glass and glass spheres (Figure 4)show good agreement between the data points and the theoretical lines, except for glass spheres 1 a t high values of e. The lack of agreement in that case is similar to what was observed previously (13) a t high dilutions of emery powder. The exceptionally high points can be attributed to flow of liquid through fine channels

904

INDUSTRIAL AND ENGINEERING CHEMISTRY

Vol. 36, No. 10

developed in the floc structure because of the relatively low concentrations of solid. Fair agreement with the rate equation is shown by the graph fo; burned shale. At the lowest values of E the points are a little high, but otherwise agreement is good.
6.4

5.6

stances have given unusual results (9). Some suspensions seem t.o be too strongly flocculated for Equation 8 ( I S ) t o apply; in such cases structural resistance may sometimes be involved. Certain other suspensions are apparently too weakly flocculateti -for example, the less concentrated suspensions of silica in one sixth saturated lime water. However, the fact that so many of the powders that have been investigated have settled in approximate conformity to Equation 3 ( I $ indicates that the equation has a considerable range of application (although only at concentrations great enough to avoid channeling and t o permit of B cont,inuous floc structure).
EFFECT OF SPECIFIC SURFACE

48
e -

9
E

A.o

d
L

3 . 2

a VI *

2.4
C

Since u was varied widely, conformity of the experimental data to Equation 3 of the second article (13) (with V 8formulated as in Equation 1 of this paper) is support for the assumption that u has the effect indicated by the equation. To show more directly the degree of this support, straight lines representing the data were first drawn on plots of [&(1 e ) ] 3 us. E without recourse to the theory. If these lines had conformed strictly to the t)heory, their slopes would have been given by Equation 2 which, b-ith V , expanded, is:

x B 2

1.6

(3)

Solved for u this becomes

0.8

a
0

20

40

60
Time,

80
minutes

100

120

140

if __2y(p ) is designated by F. To show the degrcv ~ (1 ~i)~(slope)3 to which the data conform to Equation 4,the experimental values of u are plotted in Figure 5 against F 1 / z calculated from thc: slopes and w i values found empirically from the plots of

Figure 3. Sedinientation Curve (Incomplete) for Limestone 2 in Tap Water at E = 0.125

(Height of auspension, 34 mm.)

[&(1

- 4]1/3

us.

The tests on silica (Figure 4) agree in part with the theory. However, the high point at B = 0.68 for a test in saturated lime water and most of the point>s for tests in one-sixth saturated lime water are not represented by the theoretical line. These exceptions, like those for glass spheres 1, can he attributed to breaks in the floc structure. If this is the correct explanation, the breaks evidently occur a t much lower values of B in one-sixth saturated than in saturated lime water, an indication that the particles cf silica have less cohesion in the weaker solution. The breaks in floc structure assumed here were not, in general, actually seen. Only in the tests in one-sixth saturated lime water a t E = 0.68 (and in a test on burned shale 2 a t E = 0.75) were distinct channels detected. I n contrast, Powers (9) nearly always found channels and boils in tests,on cement pastes when the rates of sedimentation were too high to agree with his equation. This has also been the writers experience in testing cement pastes. I n suspensions in which channels are suspected but not seen, the cohesion between particles may be weaker and may result in many breaks but smaller, shifting ones. At low fluid contents the rates of sedimentation of the silica were nearly equal in the two different lime solutions, even though the differences in total settlement were pronounced (Table 11). Results of this kind had previously been obtained by Powers (9). The floc structure is evidently weaker in the lower concentrations of flocculant, but this weakness does not appear to affect the rate significantly unless actual breaks develop in the structure. It is not clear from Figure 4 whether the suspensions of limestone 2 are exceptions to the theory, but obviously those forlimestone 1 fail to conform. Apparently the nature of the floc structure is not the same in all suspensions, a fact that has been indicated also by tests with organic liquids which in some in-

-1point for limestone 2 is not shown because its location was too uncertain. The single point for silica represents the tests in saturated lime water; the data for emery were obtained in p r p vious work ( I S ) . The solid line has the slope 0.35 required t)y Equation 4. As was to be expected, the point for limestone is distant from the rest. Otherwise, the points are represented rather well by the theoretical line. Thus, the use made of u in determinirip tho rate of sedimentation seems justified. The specific surfaces determined by the Lea and Kurse air permeability method (designated XO as in Table I) are alm plotted in Figure 5. Since tho points are widely scattered ant1 cannot well be represented by any straight line through the origin, it is evident that So is, in general, unsuitable for use ii! place of u in Equation 3 ( I S ) . Whether this situation may bv altered by recently suggested modifications (3, 6) of the Lea arid Nurse method is not apparent from this study.
FACTORS AFFECTING w i

I n the formula for rate of sedimentation, as given by Equatioii 3 (IS), the wi term differs from the others in not being determinable a t present except by actual sediment,ation tests on concentrated suspensions. The problem of prediction is complicated 1 ) ~ the increase in w i caused by flocculation. When a powder is rvactive chemically, like portland cement, there are further complications. However, some qualitative inferences from presriit. data are possible. From Table I, w i for inert powders seems to show no pronounced change as the fineness is changed. -4lthough W i increased with the fineness of flocculated emery powders of practically uniform particle size ( I S ) , inert powders that are not closely sizcd may ordinarily show no such effect. Aqueous pastes of portland cement show a consistent increase in the value of Vi with inrrease in fineness (Q), hut it nwms J)rObahl(~ that this is rausctl

October, 1944
0.10

INDUSTRIAL RND ENGINEERING CHEMISTRY

0.07

905

0 . 0 s

0.06

0.ot

0.05

0 . 0 7

0.04

0.oc
-,m

0.03

7 z

5 0.0:
0 . 0 1

0.02

0.01

0.0:

?oo
E

0.0:

0.01

0.K

E
'/a

Figure 4. [ Q ( l - e ) ]

wvs. e for Various Powders

The solid lines conform to Equation 3 of the second article (13).

in a flocculated suspension is evident from the fact that, for the silica powder, the value was less when the flocculant was the dye than when it was saturated lime water.
COMPARISON WITH POWERS' EQUATION

F.l

0.2

0.3

0.4

0.5

0 . 6

0 . 7

1)) u layer of hydrate formed over the grains when they are first nuxed with water. An added layer should have the same elTect on w,as stagnant liquid; the greater the quantity per unit volume of solid, the greater would be the value of w,. Hence, since the quantity would probably be approximately proportional to the q)ecific surface, w,would increase with iacrease in fineness. I t is noteworthy that fair-sized values of wtwere found neces+iry in the rate equations for the flocculated glass spheres. Siiice the value of w,for nonflocculated spheres has been shown t o be zero ( l a ) ,the data support the previous evidence from tests o i l emery (IS) that flocculation increases w,. That the nature of the floccrilant ran affect the value of ID$

This series of studies, from which Equation 3 (IS) was developed, was undertaken to help explain the sedimentation of portland cement pastes, especially the significance of the w, term which Powers (9) found to be necessary in his rate equation for i waR the sedimentation of such pastes. The significance of w given detailed consideration in the second article of the series (IS), and views advanced there have now received further support. However, Equation 3 ( I S ) which embodies those ideas differs somewhat from the Powers equation and, indeed, differs with respect to the use that is made of wi. I f w i actually has the same significance for the cement pastes that it has been concluded to have for the other systems, then Equation 3 (IS) may be expected to be practically equivalent to the Powers equation within the range of conditions that prevail in the pastes That there is approximate equivalence can be shown as followt:: Powers' equation is. = 0.2 d P s - P / ) ( e wd3 (5) nu2 ? 1 -

906

INDUSTRIAL AND ENGINEERING CHEMISTRY

20

Vol, 36. No. 10

12

18

1 6 10
14

8
*
c

12
0
X

y o

10

6
8

2
2

0 0

10
!

20

30

10
F?

20

30

FT

Figure 5. Respective 3Ierits of Specific: Surface Values,

and So, as Factorb i n the Rate Equation

To conform to the theory, the points should fall on one straight line through the origin (Equation 4). * Ground glass A Silica 0 Burned shale Glass spheres A Limestone 0 Emery

where uu. = specific surface value, sq. cnl./cc., calculated as if the particles were spheres of the same density as the particles and the same rates of fall at infinite dilution except that particles smaller in dianieter than 7.5 microns are assumed to have t,he average diameter of 3.75 microns (A.S.T.31. method C115-42). Ttic, :issumption regarding the particles smaller in diameter than 7.5 microns distinguishes u,,, from the u defined previously; u is assumed to represent the equivalent-sphere surface rathcir closely. In contrast with Powrs equation, F:qnation 3 ( I S ) with V, expanded is:

age is close t o 0.27; evidently Equation 6 is approximately vahi under average conditions. In general, however, the slopes of the data lines in plots of [&(I < ) ] 1 3 us. E should be in&i if Equation 5 is correct, but should pendent of the value of w vary about 4% from the mean if Equation 6 is correct. Sinw Equation 5 has been found applicable to cements, this might be thought to rule out Eyuation 6, but the data for cement p are not sufficiently precise to justify such a conclusion. I t is believed, therefore, that the ideas regardirig the nature of 713; H I N I the effect of specific surface embodied in Equation 6, and itr equivalent, Equation 3 of the serond article ( I S ) , are appliczat~ltto cmicnt pastes.

COh c 1.u SIOYS

Equating the two formulations of Q and solving for u lu,, gives

u
~

E----

1.109

u w

1 - wz

The magnitude of the ratio u/u, can be e\tablished a5 follow? Lea and Nurse (7) found that So/u averaged 1.22 for cement\ They also found that So/u, ranged between 1.6 and 1.8. However, in other investigations the ranges in So/u, have been 1.771.92, 1.70-1.99, and 1.70-2.00 (1, 2, 8). From these data anti much additional unpublished work, a value of 1.85 appear3 to be reasonably representative of Sg,uwfor American cements. Comhining the ratio of SO to uu with the ratio of SO to u, gives.
u/uU

The initial rates of sedimentation of highly concentrateil YIIS1)ensionsof the materials investixat,cd, except, the limestonc:, :iprt:o fairly well with Equation 6 . Comparison of specific surface values determined b y thtz :tir I)ermeability method of Lea arid Nurse ( 7 ) with u values. n . ) i i t , l l :&re the specific surfaces of equivalent spheres as determined IJ). iedimentat,ion analysis, indicates that the former arc di9tinc.t I!. l(w suitable for use in the rate equation. Comparison of Equation G with the equation developccl 11). Iowers (9) from studies on portland cement, indicates thitt t Iiv t,wo should give approximately the same rrsults for r x n ~ e i i t
p:istes.
ACKNOW LEDGMEKT

= 1.52

(8)

Equation 7 can then be solved for u,,and the result is 0.27 For cements, ut, generally lie5 between 0.24 and 0.32, and the aver-

Tlie writer was assisted at different timeb by Lynn . I 13mii?r, . Richard G. Brusch, and Herbert IT. Schultz in the expvrimcmt:il work reported in this article.

October, 1944

INDUSTRIAL A N D E N G I N E E R I N G C H E M I S T R Y
YOMEhCLATURE
vqD =

907

F =
g

cni.-2, in which w i and the slope are q(l wi)Z(slope)3' determined experimentally from a plot of IQ(1 )]*'a
I_____

zg(ps

- ")

vs.

specific surface value, sq. cm./cc., calculated the same as u, except that particles smaller in diameter than 7.5 microns are assumed t o have the average diameter o f 3.75 microns
LITEHA'L'I'RE CI'rEI)

Q =
r =

= acceleration of gravity, cm./sec.Q

S a =

v, =
wi =
9
E

=
=

PI Pa

= =
=

initial rate of settlement of top surface of suspension, cm./ sec. radius of a sphere, em. specific surface of a powder as determined by air permeability method of Lea and Nurse (T),sq. cm./cc. 2g(p,"), cm./sec. (velocity given by Stokes' law for a 162 single particlr) dimensionless constaiit, experimentally determined coefficient of viscosity of a fluid, grams/(cm. X sec.), or po1ses proportion of total volume of a sus ension occupied by liquid, analogous to porosity in bels of particles density of a fluid, grams/cc. density of a solid, rams/cc. specific surface vaue, sq. cm./cc., calculated as if each particle were a sphere having same density as particle and same rate of fall when alone in a large volume of viscous liquid; for uniform-size particles it, is equal to 3 / r

(1) Bates, P. H., Proc. Am. Soc. T'estCag ,Mate~iuls,41, 224-34 (1941). (2) Blaine, R. L., A.S.T.M. Bull. 108, 17-20 (Jan., 1941). (3) Blainc,R. L., in Rept. of Working Comm. on Fineness of A.S.T.M. Comm. C-1 on Cement: ASTM Bull., 118, 31-6 (Oct., 1942). (4) Carman, P. C., J. SOC. Chem. Id., 57, 225-34T (1938). (5) Dahl, L. A., Portland Cement Bssoc., unpublished work.

(7) Lea, F. M., and Nurse, R . W., . I . Soc. Chem. l,nd.+ 58, 277-83T (1939). (8) Meyers, S. L., Rock: Products, 44 (la), 56-9 (1941). (9) Powers, T. C., Research Lab., Portland Cement Assoc., Bull. 2 (1939). (10) Sklarew, Samuel, IND.ENG. CHEM., AN~L ED., . 6, 152-3 (1934). (11) Sollner, K., Ibid., 11, 48-9 (1939). (12) Steinour, H. H., IND.ENG. CHEM., 36, 618-24 (1944). (13) Ibid., 36,840-7 (1944). (14) Wagner, L. A., Proc. . 4 w SOC. Testing Materials, 33, Part 2, 553-70 (1933).

(6) Keyes, W. F., to be published.

Chemical Nature of Redwood Tannin and Phlobaphene

M.A. BUCHANAN, H.F. LEWIS, AND E . F. KURTH
The Znstitute o f Paper Chemistry, A p p l e t o n , W i s .

HE wood of the redwood tree (Sequoia eernperuirens) coiitains a relatively high proportion of extractives, consisting chiefly of tannin and a water- and ether-insoluble material called phlobaphene. These extractives represent a potential source of valuable by-products and as such should be removed from the wood before it is converted into pulp or is used as a starting material for alcohol by the Scholler process. Certain other uses of redwood (Le., the manufacture of a plastic pulp) depend upon the presence of these same materials. Therefore, a knowledge of the nature of these extractives is important in a program involving the utilization of redwood. Redwood tannin is one of the many natural products which has the property of converting animal hides into leather. These tannins from various sources have different compositions. The n:ttural tannins may be classified as hydrolyzable tannins and phlohatannins (2, 4, 6). Tannic acid is a well-known example of t h e first class; on hydrolysis with enzymes or mineral acids,
_ _
~

it yields glucose and gallic acid. The phlobatannins, in comparison, when heated with dilute mineral acids, form insoluble condensation products which are called phlobaphenes. The phlobatannins are built on the catechin model and, on alkaline fusion, yield a phenol and phenolic acid, or two different phenols. Thus, phlobatannins from several sources yield phloroglucinol and protocatechuic acid on alkaline fusion. Although the natural tannins differ in composition, they all appear to contain several phenolic hydroxyl groups. These groups are responsible for the solubility of the tannin in water and for the leather-making properties. Water-insoluble materials called phlobaphenes normally occur in the wood Yith the phlobatannins. Little is known concerning the true nature of these materials, but they are generally considered to be condensation products derived from the corresponding tannins by the elimination of water. The naturally orcurring phlobaphenes differ somewhat from those

~
~

Tannin and phlobaphene have been isolated from redwood (Sequoia sempervirens), and their chemical nature han been investigated. Both tannin and phlobaphene can be acetylated and methylated. The purified tannin contains 2.8% methoxyl, 63.6% carbon, 5.6% hydrogen, 15.4% phenolic groups, and 20.3% total hydroxyl groups; it still contains 24% of material which is not adsorbed by hide powder but which contains approximately the same methoxyl and total hydroxyl groups as the total fraction. Classification reactions show that the product is a phlobatannin. The phlobaphene contains 6.9% methoxyl. 66.8% carbon, 5.9% hydrogen, 10.1% phenolic groups,

and 13.7% total hydroxyl groups. The high methoxyl content indicates that this fraction may contain some acetone-soluble lignin. Alkali fusion of both the tannin and phlobaphene yields only protocatechuic acid and catechol; the corresponding methylated products yield only veratric acid. Destructive distillation of phlobaphene yields Catechol and small amounts of phenol. Ultraviolet absorption spectra for both tannin and phlobaphene are similar to those for mimosa tannin and alcohol-soluble spruce lignin. Methylation with diazomethane does not cause any significant change in the ultraviolet absorption spectra.