T H E COLLOIDAI BEHAVIOR OF CLAYS AS RELATED TO THEIR CRYSTAL STRUCTURE1 T. F. FORD, A. G. LOOMIS,
A N D J. F. FIDIAM Shell Development Company, Emeryville, California
Received August 91, 1039 INTRODUCTION
Most salts cause precipitation or gelation of clay suspensions. Small amounts of sodium tannate, sodium hexametaphosphate, 'or sodium silicate, on the other hand, remarkably reduce their viscosity. Advantage is taken of this fact in oil well drilling by thc rotary method to produce mobile clay drilling fluids of high density, and such treated suspensions serve various practical purposes (1). Experiments here described develop in detail the view, which has been also suggested by Garrison ( 6 ) , that these chemicals are adsorbed on specific portions of the exterior of clay particles, and that they reduce the viscosity of suspensions by destroying the aggregation of the particles into structures. VISCOSITY REDUCTION PRODUCED BY VARIOUS CHEMICALS
The curves of figure 1 show the viscosity reduction produced by the chemicals which have been mentioned and by sodium pyrophosphate, one of the most effective reducers (4). In figure 2 similar curves are given for a series of sodium polyphosphates, of which sodium pyrophosphate is the first member. The sodium polyphosphates have the general formula NanPn-208,-6 (18) and probably differ only in the length of the PO2 chain, according to the structure NaO
0
\ / \ / NaO-P P / NaO
\ / 0I \
0
0
0
ONa
/ \ / (NaPOd, P \ / \
ONa I n these experiments the viscosities were determined at room temperature with the Stormer viscosimeter. The clay is one found near Frazier Mountain, California, and employed in the Los Angeles area to make Presented at the Sixteenth Colloid Symposium, held at Stanford University, California, July 6-8, 1939. 1
2
T. F. FORD, A. G . LOOMIS, .4ND J. F. FIDIAM
i
R IOOUL. SOLUTION
3 4 5 6 7 MILLILITERS REAGENT SOLUTION PER 120 OM. SUSPENSION
FIG.1. Viscosity reduction produced by various chemicals. All curves corrected for dilution.
GRAMS OF CHEMICAL ADDED AS WATER SOLUTION TO 120 GRAMS OF SUSPENSION
FIG. 2. Viscosity reduction produced by sodium polyphosphates. All curves corrected for dilution.
3
COLLOIDAL BEHAVIOR AND STRUCTURE O F CLAYS
suspensions for well drilling. old.
The suspensions used were a t least 3 weeks
EVIDENCE FOR ADSORPTION OF VISCOSITY-REDUCING CHEMICALS I N THE CLAY-WATER
INTERFACE
The amount of sodium tetraphosphate required to reduce the viscosity of Frazier Mountain clay is greater per gram of clay for fine fractions than for coarse fractions as obtained in the De Lava1 and Sharples centrifuges. Since the analyses of these fractions (table 1) appear to eliminate any explanation based on differences of chemical composition, a t least the initial effect of this chemical must be on the outside surfaces of the particles. Similar results have been obtained with graded Wyoming bentonites. TABLE 1 Chemacal analysis of Frazier Mountain clay fractions
I SiOl. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . FezOl. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Alz08. . . . . . . . . . . . . . . . . . . . . . . . . . . . . CaO, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . MgO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . NazO/KzO as NazO. . . . . . . . . . . . . . . . . . . . . .
I
MEDIUM F R A C n O N
per cent
52.43 9.34
~
I
F I N E FRACTION
52.11 9.48
so,--. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . c1-. . . . . . . . . . . . . . . . . . . . . . . . . . . . . Loss on ignition. . . . . . . . . . . . . . . . . . . . . Total. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The viscosities of treated suspensions tend to revert to original or even higher values after several hours or days, especially if there is agitation. Among the polyphosphates reversion is the most rapid for the pyrophosphate. Possibly these agents are slowly lost by diffusion into inner particle surfaces where they cannot prevent aggregation but may, on the other hand, promote splitting of crystal sheets with multiplication of particles. The amounts of polyphosphates required to produce the maximum viscosity reduction in 120 g. of suspension containing 14 g. of clay, as calculated from figure 2, are as follows: For NadPzO, For NasPsOlo,prepared by fusing NarPzO7 For NasPaOls,prepared by fusing Na4PzO7
0 00021 mole
+ 1/6 h’a8P0018 0 OOO19 mole + 1/3 NaePsOls 0 00018 mole 0 00020 mole For N a ~ P 7 0 prepared ~~, by fusing NaxPOd + r\Ta6P6O18 For NaloPaOzs, prepared by fusing NarPz07
+ NaaPsO~s
0 00018 mole
approximately
4
T. F. FORD, A. G. LOOMIS, AND J. F. FIDIAM
The experiments suggest that these compounds produce equal effects per molecule because they cover equal areas per molecule. Tannic acid evidently coats the same area as pyrophosphate, since onefifth as many moles are required for maximum effect (figure l), but the cross-sectional area of the molecule, CeH~O~[-COC~H~(0H)20COC~H2(OH)3]6,is about five times the cross-sectional area of the pyrophosphate molecule. EVIDENCE FOR STRUCTURAL BREAKDOWN I N CLAY SUSPENSIONS ON CHEMICAL TREATMENT
Left undisturbed, clay suspensions tend to develop considerable mechanical strength or yield value, but most of this mechanical strength is 7
3
G
I'
i -
-
z 5
I
4
3
k
iP I
0
e
I
nue
3
4
IN HWRS ELAPSED BETWEEN TnoRouw AGITATION
AND YEASURYENTOFML STRENQTH
FIG.3. The relation of gel strength to time of setting, and the effect of sodium tetraphosphate on gel strength.
lost immediately on chemical treatment, as shown by the curves of figure 3. In these experiments the torque just necessary to initiate rotation of a metal cylinder imbedded in the suspension was taken as a measure of the mechanical strength, or gel strength. Similar data are reported by Garrison (6). Structural breakdown of suspensions on treatment is shown not only by loss of gel strength, but by the disappearance of structural viscosity as shown by the curves of figure 4. The relation between viscosity and concentration for treated suspensions is in fact often described fairly well by the Einstein equation 7/70 = 1 2.5C~
+
derived for spherical particles.
5
COLLOIDAL BEHAVIOR AND STRUCTURE OF CLAYS
CRYSTAL STRUCTURE
Clays are hydrated aluminum silicates, most of which may be classed as either kaolins, Altoa.2Si02.2H20, or montmorillonites, A1203.4Sio2.HzO. These formulas represent ideal analyses of the acid or amphoteric clays. Both the kaolins and the montmorillonites (or bentonites) are composed of superimposed layers of aluminum and silicon bound together through oxygen bridges (2). The structures of the kaolin and montmorillonite unit sheets as given by Hofmann, Endell, and Wilm (10) are reproduced in figure 5. Many of these unit sheets may be combined in individual clay particles. Water may penetrate between the unit sheets in montmorillonite and presumably may even cause their complete separation (9, 19).
I200
900
i
2
600
300
0
loo
2 0 0 x K ) 4 o o 5 0 0 6 0 0 7 0 0
800
DRIVING FORCE IN GRAMS
FIG.4. Relations between shearing stress and rate of shear for treated and untreated suspensions of Frarier Mountain clay. Stormer viscosimeter.
Clay particles, like the unit sheets of which they are composed, are themselves plates, as can be directly observed in some cases or inferred either from streaming in polarized light or from the nature of the twinkling in the ultramicroscope. GROSS STRUCTURE IN CLAY SUSPENSIONS
The plate-like character of clay particles is emphasized by Lewis, Squires, and Thompson (14), who suggest that overlapping will build up structural agglomerates. Broughton and Squires (3) suggest edge-to-edge contact. The chief objection (9) is that gelation occurs in fine bentonite suspensions so dilute (0.1 per cent) that, for continuous contact, ratios of length and breadth to thickness of square platelets of the order of 1000 to 2000: 1 are
6
T. F. FORD, A. G. LOOMIS, .9ND J. F. FIDIAM
necessary. This objection holds only for suspensions to which salt has been added, however, and in the absence of added electrolytes gelation has not been observed below 1.4per cent (8). The particles used to produce these latter gels might have had ratios of length and breadth to thickness which do permit edge-to-edge contact in semi-rigid structures. To explain gelation Salmang (20) has postulated immobilization in microscopically thick hulls of bound water, but calculations based on the sorption capacity for exchangeable ions (11) and on heats of sorption (17) lead to the conclusion that the thickness of the solvation mantle cannot
4 0+2(OH)
MONTMORILLONITE (BENTONITE)
KAOLINITE 2H,O.A1,0,2Si0,
FIQ.5. Crystal structures of kaolin and montmorillonite (from Hofmann, Endell, and Wilm (IO)). exceed 25 to 40 A; and decreasing setting time with rising temperature is also inconsistent with the bound water theory, as solvation is generally considered to decrease with temperature (9). Although the existence of compressed water in bentonite suspensions is shown by increasing apparent density with concentration (9), at low concentrations apparent densities are found which are equal to or even less than the dry density of bentonite. This suggests that the compressed water is not bound as solvation hulls to single particles but that it must be entrapped between particles. The forces primarily responsible for structure, therefore, do not arise from the bound water.
COLLOIDAL BEHAVIOR AND STRUCTURE OF CLAYS
7
Another view, due to Freundlich and Hamaker ( 5 , 7 , 12), is that the particles in thixotropic clay suspensions are held at equilibrium distances such that forces of repulsion due to interpenetration of the diffuse ionic clouds are just balanced by forces of attraction which are assumed to be the London-van der TTaals forces. Langmuir (13) substitutes for longrange van der Waals forces the Coulomb attraction between the micelles and the oppositely charged ions in solution. The theories of Freundlich, Hamaker, and Langmuir apply to uniformly charged particles, but variations of exchange capacity with Si02-Ri03 ratio (21), orientation in an electric field (15), and the high electrical conductivity of masses deposited on the anode membrane in electrodialysis, as opposed to the low conductivity in non-oriented thixotropic systems (16), all suggest that positive as well as negative areas are to be found on clay particles. According to the crystal structures, the planar and transverse surfaces of particles should differ radically. Since all transverse surfaces must be formed by breaking the lattice, they should expose both aluminum atoms with unsaturated valences or unsaturated aluminum-oxygen groups, which normally will react with water to form hydroxides, and unsaturated oxygen atoms attached to silicon, which normally will react with water to form silicic acid. Over the planar silica surfaces secondary valence forces will induce acid hydrolysis of adsorbed water, and over planar alumina surfaces (as are found in kaolin) the reverse may be expected. Thus, in the same particles, both positive and negative reactive areas, points, or edges are available. Between like surfaces of clay particles, according to the analysis of Freundlich, Hamaker, and Langmuir, high equilibrium distances of separation are possible. However, between different kinds of broken bonds on the transverse surfaces or at the edges or corners of the particles, or between edges and planar surfaces of the opposite charge, or between dissimilar planar surfaces, actual contact which may be regarded as chemical reaction should be expected. Structures in untreated clay suspensions may be attributed, then, to the combined effects of attraction between dissimilar surfaces, edges, or corners, and to repulsion or large equilibrium distances of separation which occur primarily between like planar surfaces. The clay particle may be considered a colloid molecule and symbolized by the following formulas: for montmorillonite (bentonite), in neutral solution,
or for kaolin,
8
T. F. FORD, A. G. LOOMIS, AND J. F. FIDIAM KAOLIN AND BENTONITE
According to the formulas which have been written, in suspensions of hydrogen bentonite elimination of water between particles may occur as indicated by the equation,
Since the particles are never of completely regular shape, less ideal configurations result which permit structures in three dimensions. In the case of hydrogen bentonite, however, the planar ionic clouds are so feeble that there is little or no mutual repulsion; consequently there is no thixotropic structure, but dense aggregates may be formed. Hence welldialyzed bentonites swell only to a limited extent in water, and some clay8 low in alkaline salts, such as are found in wet climates, will not form thixotropic suspensions unless small amounts of electrolytes are added. The hydrogen clays are but little dissociated, in contrast to the ionic clouds which are formed where hydrogen has been substituted by other metallic ions. For kaolin the condensation between the ends of particles should be analogous to that for bentonite, but the dissimilar planar faces should permit the formation of more compact structures, even in the presence of electrolytes. I n strong alkalies, however, it is possible that both faces of kaolin may hydrolyze acidically with multiplication of particles and improved thixotropic behavior. Since the dissimilar natures of the planar faces of the kaolin unit sheets do not permit separation in water to such an extent as is possible with bentonite, there is ordinarily a large difference in particle size between kaolin and bentonite particles. This difference can be observed microscopically or inferred from wide differences in settling rate. THE PLATING ACTION OF VISCOSITY-REDUCING AGENTS
Experiments have been described which indicate that viscosity reduction by small amounts of certain large negative ions is a surface effect. Presumably these ions react with the exposed aluminum atoms on the broken transverse surfaces. Considering only bentonite and representing the viscosity reducing chemical by Nad, the reaction can be symbolized thus:
The contact structure has been destroyed, unipolar particles formed, and the structural viscosity, therefore, reduced.
9
COLLOID.4L BEHAVIOR AND STRUCTURE OF CLAYS THE GELLING ACTION OF SALT6
Figures 6 and 7 show that the amounts of specific viscosity-reducing chemicals required for maximum effect with Frazier Mountain clay are almost in proportion to the amounts of clay, but the amounts of salts required to produce gelation seem independent of the amount of clay in a given amount of water. This should be true if viscosity-reducing chemicals plate certain parts of the exteriors of clay particles, and salts affect the ionic atmospheres.
600
i. 500 '
(L:
d
8
400
5
W K
5
300
H$
200
s
9 B 0
100 60
r
v)
0
55
0
0
I
2
3
4
5
6
7
8
9
IO
MILLILITERS REAGENT SOLUTION ADDED TO 120 GRAMS SUSPENSION
FIQ. 6. Amounts of viscosity-reducing chemicals required at different concentrations of clay.
With increasing concentration ordinary salts first raise and then lower the yield point and the apparent viscosity of a suspension. According to the ideas of Freundlich and Vinograd (12), with increasing ionic concentration the thickness of the double layer is progressively reduced while the forces of attraction are constant, and therefore the force required to move the particles out of their oriented positions is greatest a t an intermediate salt concentration. These effects of ordinary salts are superimposed upon and apparently are independent of the presence of specific viscosity reducers. The latter are needed only in small amounts and so do not appreciably affect the ionic
10
T. F . FORD, h. G. LOOMIS, AND J. F. FIDIAM
atmospheres but perform the single function of preventing adhesion between particles with consequent destruction of that portion of the gel strength and viscosity attributable to structural contact. The gel strength may be reduced by adding one of the polyphosphates and then progressively increased by adding salt, or the salt may be added first, so that many possibilities exist for the control of this or other properties.
MILLILITERS REAGENT SOLUTION PER 120 GRAMS WATER IN SUSPENSION
FIG.7. Amounts of salt required for gelation at different concentrations of clay
The effectiveness of salts in both gelation and coagulation of clay suspensions increases steeply with the valence of the positive ion, just as in the coagulation of other colloids. AMPHOTERIC REACTIONS AND THE EFFECTS O F ALKALIES
Addition of water to a dry montmorillonite particle may be indicated by c $ s i /
+ HP + H O / S ~ O H
COLLOIDAL BEHAVIOR .4ND STRUCTURE OF CLAYS
11
and for this hydrated particle the possible reactions of hydrolysis are:
HO/Ss---Xj7OH
In accordance with these formulas, two minima in viscosity or yield point curves are observed with increasing pH (1, 14). On adding sodium hydroxide to a neutral suspension the first effect is to induce the acidic hydrolysis of alumina and then to affect the ionic clouds in the same way as salts. This dual action is illustrated by the curves of figure 6. The viscosity is reduced a t low concentrations of sodium hydroxide because particles with uniform sign of charge are formed, and at higher concentrations the viscosity is increased because the ionic clouds are affected. The viscosity reduction obtained is less than with pyrophosphate because the acidic hydrolysis is induced by a mass action effect of the alkali, and this electrolyte concentration required for viscosity reduction also simultaneously produces some gelation. SUMMARY
1. Certain chemicals, such as sodium tannate and various phosphates, which in low concentration reduce the apparent viscosity of concentrated clay suspensions appear to be adsorbed on specific parts of the crystal faces of the clay particles, and to destroy the aggregation of the particles into structures. 2. The gelling and coagulating effects of most salts are attributed to changes in concentration of the inter-particle ionic atmospheres. 3. On the basis of the detailed crystal structures of clays, chemical formulas are written for colloidal clay particles and are used in explaining the various phenomena encountered with clay suspensions. REFERENCES (1) AMBROSE AND LOOMIS: Physics 1, 129 (1931). (2) BRAQG: Proc. Roy. Inst. Gt. Brit. 30, 39 (1938). (3) BROUQHTON AND SQUIRES: J. Phys. Chem. 40,1041 (1936). (4) BYCK,H.T. : Unpublished results, Shell Development Laboratory, Emeryville,
California. (5) FREUNDLICH: Thizotrop~. Hermann et Cie., Paris (1935). (6) GARRISON: Mining Technology 3, No. 2, and Petroleum Technology 2, No. 1,
Technical Publication 1027 (1939).
(7) HAMAJKER: Rec. trav. chim. SS, 1016 (1936).
12
ARTHUR A. VBIRNON AND RARRISON A. NELBON
(8) HAUSEBAND LEBEAU:J. Phya. Chem. 42,1031 (1938). AND REED: J. Phys. Chem. 41,911 (1937). (9) HAUEER AND WILM: Z. Krist. 66,340 (1933). (10) HOFUNN, ENDELL, (11) HOUWINK:B h t i c i t y , Plasticity, and Structure of Matter, p. 336. University Press, Cambridge (1937). (12) HOUWINK:Eketicity, P h t i c i t y , and Structure of Matter, p. 338. University Press, Cambridge (1937). (13) LANGMUIB: J. Chem. Phys. 6, 873 (1938). AND THOMPEON: Trans. Am. Inst. Mining Met. Engrs. 114, (14) LEWIS,SQUIBES, 39 (1936). (15) MABBHALL: Trans. Faraday SOC.26, 173 (1930). (16) MABEHALL: J. Phys. Chem. 41, 935 (1937). (17) MATTSON:Soil Sci. SS, 301 (1931). (18) MELLOR:Treatise on Inorganic Chemistry, Vol. VIII, p. 990. Longmans, Green and Company, London (1928). (19) REED:Petroleum Eng. 9, No. 4, 48 (1938). (20) SALMANG: Kolloid-Z. 46, 377 (1929). Soil Sci. 41, 26 (1936). (21) WINTEBKOBN:
CHEMICAL PREPARATION OF COLLOIDAL SUSPENSIONS I N NON-AQUEOUS SOLVENTS. I METHYLALCOHOL AND BENZENE' ARTHUR A. VERNON'
AND
HARRISON A. NELSON8
Department of Chemistry, Rhode Island State College, Kingston, Rho& Island deceived February bS, 19S9
Tomaschewsky (9) prepared fairly stable alkali-metal organosols by the simultaneous condensation of the vapors of the metal and the dispersion medium. Fodiman and Kargin (1) used a method by which they condensed metallic vapors in an organic liquid. The change of solvent method was used by Horiba, Otagari, and Kiyota (5) and by Von Hoessle (lo), while Svedberg (7) used an electrical dispersion system. A silver suspension in alcohol was prepared by Formstecher (2) by reduction of silver nitrate with formaldehyde. Hydrogen sulfide was used by Lottermoser (6) to suspend cupric sulfide and mercuric sulfide in alcohol. In view of the scant information on the subject it seemed of 1 This article is condensed from a thesis submitted by Harrison A. Nelson t o the Faculty of Rhode Island State College in partial fulfillment of the requirements for the degree of Master of Science in Chemistry, June, 1938. * Present address: Department of Chemistry, Northeastern University, Boston, Massachusetts. a Present address: Department of Chemistry, The Rice Institute, Houston, Texas.
