-Corrosion Mayne, J. E. O., J. Sor. Chem. Ind. (London),65,196 (1946). Menard, P. L., and Dunn, T. H., Petroleum Develop. and Technol., 165,26(1946). Olsen, E., and Szybalski, W Corrosion, 6,N o . 12,405-14(1950). Parker, I. M., Oil Gas. J.,45,No. 28,255-9 (1946). Rhodes, F. H.,and Kuhn, W. E., IND.ENG.CHEM.,21, 1066 (1929). Shock. D. A,. and Sudburv. - . J. D.. Petroleum Enar.. - . 23. N o . 8. 86B’(1951): Speller, F. N., “Corrosion, Causes and Prevention,” 2nd ed., pp. 186.360.New York. McGraw-Hill Book Co.. Inc.. 1935. laid., ’3rd ed., pp. 212-13, New York, McGraw-Hill Book Co., Inc., 1951.
.
I
by Water
(49)Ibid., pp. 396,409. ( 5 0 ) Speller, F,N., Proc. Natl. Dist. Heating Assoc., 24,203 (1933). (51) Saeller. F. N.. and Chsppell, E. L., Chem. & Met. Eng., 34, 421 -(1927). (52) Uhlig, H. H., “The Corrosion Handbook,” pp. 480-1, New York, John Wiley & Sons, 1948. (53)Ibid., pp. 909-13. (54) U. S. Bur. Mines, Bull. 433, 68,71 (1941). (55) Vernon, W. H. J., J. SOC.Chem. Znd. (London),66,138 (1947). (56) Warner, J. C., Trans. Electrochem. SOC.,83,328 (1943). (67) Whalley, H. C. S. de, J. Soc. Chem. Ind. (London), 56, 569-70 (1937). ’
RECEIVED for review February 21, 1952.
ACCEPTED June 20, 1952.
Inhibito rs for Eliminati ng Gorrosion 0
in Steam and Condensate Lines R.
c.
ULMER AND J. W. WOOD
Power Chemicals Division, E. F. Drew 8, Co., h e . , New York, N. Y.
steam condensate, the followONTROL of corrosion Control of corrosion in steam condensate i s of vital iming are cited: W h i t m a n , in steam condensate portance to the power industry. In general, two types of Russel, and Altieri (11)noted systems is of vital imtreatments have been advocated to control corrosion: the corrosiveness of carbon portance to the power indusdioxide in solution and the introduction of alkaline substances to neutralize the acidity try. Not only does such corincreased aggressiveness of rosion damage the equipment and introduction of film-forming materials to coat the metal. systems containing both carwhich is attacked, but the Corrosion test data were obtained in 600-pound-perbon dioxide and oxygen. corrosion products are resquare-inch boiler condensate with representative maSkaperdas and Uhlig ( 16) and turned to the steam-generterials of both types. In general the type that neutralizes Uhlig (10)discuss most asa t i n g e q u i p m e n t . There pects of the subject, includacidity was found to give the best results. Raising the pH they promote formation of ing the anodic and cathodic deposits, e.g., magnetic iron to around 8.0 decreased the corrosion to an inappreciable reactions, the effect of pH oxide, and the development amount in the case of iron. Use of film-forming materials and oxygen content, and the of electrolytic corrosion cells. also decreased the corrosion rate but not to such a low effect of temperature. The Straub (17, 18) and Corey value as in the case of the neutralizing type. With both extensive studies of Collins (8),among others, citenumerand Henderson ( 7 ) present a types i t was found that the rate of corrosion increased at ous examples of boiler tube wealth of information confailure which can be traced to higher concentrations of dissolved oxygen. This also was cerning the quantitative redeposits f o r m e d i n t h i s true in the case of the controls. In the case of copper and lationships among pH, carbon fashion. brass both types gave good results. In selecting a treatdioxide and oxygen content, Considerable work hasbeen ment for a given case, the metals in the system and the flow rate, and rate of cordone on the general subject dissolved oxygen content of the steam should be considered. rosion. Of particular imporof steam and condensate cortance are their findings conrosion. Numerous corrective cerning- the effect of volume means have been proposed. flow rate, as distinguished from velocity and carbondioxide conThere is, however, no clear-cut understanding in industry regardtent, on the corrosion rate. They suggest (6) that corrosion rate of ing the applicability of such methods. I n fact, there is some steel is a function of the rate of delivery of carbon dioxide t o the concern about some of the modern recommendations as they zone of corrosion, according to the equation: may involve the use of ammonia or ammonia-base materials and in some cases a material is added to the steam that will actually R = 5.7W0.” cause a “deposit” in the system. With the need for supplying where R = corrosion rate, milligrams per square decimeter per answers to some of these questions an investigative program was day, and W = rate of delivery of carbon dioxide to specimen in started some years ago, the results of which are presented in this pounds per hour X 100,000 = carbon dioxide concentration in paper. parts per million x flow rate in pounds per hour X 0.1. The principal source of oxygen and carbon dioxide in condenBackground of Study sate has been shown to be the boiler feed water. The gases are expelled from solution in the boiler and pass into the steam lines, The generally accepted explanation for corrosion of piping and redissolving when the steam is condensed. Collins ( 4 ) and Mcrtuxiliary equipment in condensate return systems is acid attack Kinney, McGovern, Young, and Collins (13) have shown that on the metal by the carbonic acid which is formed when carbon when steam consumption of pressure equipment is high, very dioxide dissolves in the condensate. Dissolved oxygen, when great concentrations of noncondensable gases may accumulate in present, aggravates such corrosion appreciably, owing t o depolarthe vapor space; this leads to solution of excessive quantities in ization of cathodic areas on the metal surface. the condensate. Deaeration, alone or followed by chemical treatOf the many investigators who have contributed t o presment of the feedwater, is of great importance in reducing corent-day understanding of the mechanism of corrosion by
C
August 1952
INDUSTRIAL A N D ENGINEERING CHEMISTRY
1761
AGE TMK
TEMPERATURE --+ REOWATOR &NO RECORDER
Figure 1.
-
of treatment may include amotig othdra
1STEAM
polyphosphates, and dispersions of high molecular weight, ineoluble amines. Hanlon (11) has noted the impermanence of fdms formed by the phosphate compounds. It is necessary to maintain the dosage of these materials a t the original value to prevent dipsolution of the films. Maguire (14) has found that the films formed by the highmolecular weight amines (octadecylamine) are morepermanent in character. After the initial protective film has been formed, it is claimed that it is possible t o reduce the dosage considerably. To the class of compounds, which for convenience may be called filming amines, may be added several high molecular weight quaternary ammonium bases, which the writers have found to exhibit similar properties Studies of the performance of one member of this class ofcompoundsareincluded in the rxperiments discussed in this papei.
-
AHI OKRATED R E O W A T I M VALVE
Layout of Corrosion Test Equipment
roaion in condensate systems, Dwwation is but partially successful in removing carbon dioxide which is mesent as bicarbonate 'or carbonate; these decompose in the boiler and so introdrice carbon dioxide t o the steam. Means whereby corrosion in condensate s y s t e m can be prevented or mntrolled have been the subject of much study and experimentation by a large number of workers. One method (with which this paper is concerned) consists in chemical treatm a t of the condensat[+. There are tffo principal types of treat.. nlent. With one type alkaline substances in solution are intraduced t o the system to raise the pH and neutralize the acidic constituents of the water. Where appreciable oxygen is present, reducing substances such as sodium sulfite may be used in addition. A second type of treatment consists of adding to the water substances which form protective films on the metal surfaces, insulating the metal from contact STith the corrosive media. In the frst type of treatment, volatile alkalies, e.g., ammonia, ammonia base, and volatile amines, are widely used. An advantage of such substances is that they can be introduced into the boiler or into the steam lines and will then pass through the system with the steam, going into solution when the steam condenses. Their presence in the condensate at all times is thus ensured. Straub (17,18),Straub and Ongman (19), and Leick ( l a ) ,among others, have demonstrated g~ the effectiveness of ammonia. The use of ammonia-base materials has been discussed by Ulmer (21) and the studies made by Berk ( 2 ) Micorla and Berk and Nigon (3) on the relative effectiveness of several volatile aminefi are well known. Methods of introducing the alkaline materials to the steam-condensate system are not standardized. Sonle among them Straub and ulmer, recommend adding them t o the boiler feed water or injecting them into the boiler itself. One possible objection to this method, as point,ed out by Hanlon (11),is that the procedure makes it necessary to treat the entire system to obtain adequate protection in a desired localized section. In the latter case an alternative method is to inject the treatmentdirectly into the or condensate lines by means of a chemical feed pump, in such quantities as are necessary to bring the condensate to the desired pH. Film-forming substances which may be used in the second type
Experimental
I
1262
h1experinlental was desiglled for colllyarative evaluation of several materials of the type discussed. The test procedure is best explained b y reference to the diagram in ~i~~~~ 1. Steam condensate from the boiler of E. F. I ) co., ~ ~ I ~ ~ at .Boonton, , N. J., was withdrawn from the condensate age tank and passed through a single-pass heat exchanger. A temperature regulator, which controlled the steam supply to the exchanger, maintained an average temperature of the heated condensate of 185' F. The condensate then entered a manifold from which extended six branch lines of l-bch pipe, each terminating in a tee. A National District Heating Association cor-
Suppoiling Base
Figure 2.
NDHA Corrosion Tester
rosion tester was inserted in each branch of the manifold through the tee fitting. The condensate, after flowing over the corrosion tester, passed through the side arm of the tee, then through a cooling coil to the discharge, Samples of this effluent were used for analysis and control. Flow rate was regulated by means of globe valves in the discharge lines. The water level in the storage tank was approximately 30 feet above the discharge point. Since the water level never varied more than 1 foot from the midpoint of a sight glass on the tank, the test system operated under a near-constant head m,itha maximum head variation of 1 3 . 3 % . Reasonably constant flow rate, therefore, existed throughout the test. The chemicals used for treatment were fed to four of the six
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 44, No. 8
~
Corrosion by Water manifold branches by constantvolume chemical feed pumps. The other two were used for control. The treatment was introduced 5 feet ahead of the corrosion testers. A set of small baffles was placed in each test pipe, downstream from the entry point of the chemicals, t o ensure thorough mixing of treatment and condensate. This arrangement kept the metal surface a t a minimum, which is of importance in the case of the filming-type materials. The heat exchanger, manifold, and fittings were of Type 304 stainless steel, so that contamination of the condensate through corrosion of the test equipment would be minimized. The corrosion tester shown in Figure 2, developed by the National District Heating Association, was used in this work because it has been demonstrated to give corrosion rate measurements in a comparatively short time which agree well with the rate of corrosion by the same media in plant equipment. It has been adopted by the American Society for Testing Materials ( I ) for determination of corrosion rates in aqueous media; details of its applicability, accuracy, and limitations have been described by Collins (6). The corrosion test specimens are short coils of wire in helical form, supported in a metal frame by short pieces of plastic material in such a way t h a t the specimens do not touch each other or the metal of the frame. The frame is attached a t one end to a standard pipe plug so that the tester can be introduced into a piping system through a standard fitting. Before use, the coils are cleaned with a solvent to remove grease, and are dried and weighed. At the end of the test period, the specimens are brushed lightly and cleaned in dilute inhibited acetic acid t o remove corrosion products ( I ) , and are dried and weighed again. Corrosion rate is a function of the change in weight according to the equation
(
p = - x - 365 12 n
42)x
1000
where P = penetration rate in inches per year X 1000,D = initial diameter of wire, n = duration of test in days, w1 = initial weight of specimen, and w 2 = k a l weight of specimen. The condensate used for these tests was the combined condensate returns t o the company's 600-pound-per-square-inch, 100,000-pound-pephour boiler plant. The steam in this plant is mainly used for process heating and operation of pumps. After the condensate had passed through the heat exchanger, its charaeteristics were as shown in the following table: Temperature,
a
F.
E a l i n i t y , p.p.m. as CaCOa Cot. p.p.m. 01. D.D.m. Fe .
cu
180-190 8.10-6.25 2-4 7.0-8.0 0.009-0.016 Trace None
The flow rate past each specimen averaged 300 pounds per hour; hence the average velocity in the 1-inch pipes was 0.22 foot per second. Two materials commonly used for each of the two types of
Table I. Effect of pH Control by Means of Volatile Alkalies on Penetration Rate of Common Metals Avera e Penetration, In.,%r. x 1000 Yellow Treatment pH steel Copper brass None .... 6.10- 8.25 19.9 1.7 18 Ammonia-organic" 0.0228 8.00- 8 20 5.2 5.8 4.6 0.0684 9.00- 9.05 4.1 7.0 5.6 2.907 9.93-10.05 4.8 5.9 5.9 125.4 10.97-11.06 5.5 5.1 8.0 Amine typeb 0.0735 7.95- 8.20 5.6 4.5 5.6 0.2310 8.93- 9.08 4.1 8.6 6.0 18.48 10.00-10.10 4.8 6.5 4.8 535.0 11.00-11.06 4.8 8.0 7.7 0 Sold commercially as SLCC by E. F. Drew & Co Inc New York N. Y. b Sold commercially a8 Drewamine,by E. F. Drew'& Co:, Ino., New York, Dosa e Me./%.'
N. Y.
August 1952
Mild
treatment were selected for the experiments. These were an ammonia-organic treatment (SLCC); a volatile amine (Drewamine); a filming amine (active ingredient, octadecylmine); and a film-forming quaternary ammonium salt (SLCC-Q). The ammonia-organia treatment and the volatile amine were tested at several different dosages for comparison of results at. different p H values.
t
\
2o
1: \ ---$\
I
I
I
I
I
I
Figure 3. Corrosion of Metals by Condensate Showing Effect of Increasing p H with Ammonia-Organic or AmineType Inhibitors Water from condenaate returns of Boonton boiler plant Mild steel ----....I Coppar 0 0 0 0 0 0 Brass Flow rate, 300 povndr per hour Velocity, 0.22 foot per s-nd Temperature, 180' F.
It was desired t o observe the effect of each type of treatment on several of the metals commonly used in condensate systems. Therefore, each tester was fitted with coils of mild steel (0.1% carbon), copper (high purity), and yellow brass (60 parts copper, 40 parts zinc). Tests with the ammoniatbase and amine-type inhibitors were carried out at p H values of 8, 9, 10, and 11. Required dosagas and test results are given in Table I. The corrosion rates are shown graphically in Figure 3. Corrosion rates of copper and brass increase with increased dosage of both materials, indicating some attack on these metals by these chemicals. However, in all cases, the penetration values are well within the safe range, penetration numbers greater than 10 being regarded as excessive (6). Other investigators (19)have stated that ammonia in the concentrations employed in corrosion control does not attack copper or brass. These observations were based on failure t o detect copper in the condensate after i t has passed through the system. The writers agree that such corrosion may generally be regarded as negligible. However, i t is suggested that in the case of such very low corrosion rates, the amount of copper pickup may be less than can be detected by the usual analytical methods. Another possibility is that the copper corrosion products may remain on the metal as solids. The presence of oxygen in the water probably increased the attack by these agents. Splittgerber (18) has remarked that in the presence of oxygen, 1 p.p.m. of ammonia will attack brass. The writers' results with condensate containing larger quantities of oxygen substantiate this hypothesis. However, the attack is negligible, as was pointed out, when the p H is 8.0 or below. At low oxygen contents pH's as high as 9.0 or 10.0 are satisfactory. I n a series of tests with film-forming compounds, specimens were exposed for periods of 14,21, and 28 days. In these tests,
I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
1763
!- !
-20:
-15
-5
-10
P2
L n
20
IS
x
No treatment
0
Octadecylamine t y p e Quaternary ammonium salt t y p e
A
25
the octadecylamine type (Permacol) was fed at a dosage of 30 p.p.m.; in the case of the quaternary ammonium salt type (SLCC-Q), 20 p.p.m. were used. Table I1 and the curves of Figure 4 show that the penetration rate was found t o decrease as the time of exposure increased and that at the end of 28 days the penetration figures were in the vicinity of the critical value of 10. This may be interpreted as indicating that the protective films build up slowly, even on such small surfaces as those of wire coil specimens, and that corrosion continues until the films are complete. Evans (IO)has remarked that a discontinuous film may cause accelerated local corrosion, owing to increased anodic action at the exposed points. Results of these tests were inconclusive as to the effectiveness of the materials tested; reports from a few field installations give more favorable but still inconclusive evidence of their value. Further tests were undertaken to ascertain the effectiveness of these treatments in condensate having higher concentrations of oxygen. To accomplish this, a small quantity of aerated condensate was pumped to the test system, just ahead of the heat exchanger. It was not possible to obtain close control of the oxygen content of the test water by this means. I n one series, the average the oxygen - - content varied from 0,031 to 0.068 _p.p.m.; _ value of 0.052 p.p.m. was taken for comparison purposes and tabulation. I n the second series, the oxygen varied from 0.132 to 0.078 p.p.m., with an average value of 0.11 p.p.m. Another result of this addition was an increase in carbon dioxide content of the test water, with a corresponding drop in p H to 5.6 to 5.5, Data from this test are compared with those from the earlier tests in Table 111. pH adjustment gave appreciably better protection of the steel specimens than did the filming materials, especially at the higher oxygen level. Conversely, attack of copper and brass was aggravated by addition of the neutralizing chemicals, approaching critical values a t the higher dosages in the presence of appreciable oxygen. The film treatment reduced corrosion of these metals in most cases. These points may be better illustrated by the bar graph, Figure 5 , in which most of the significant data are shown for ready comparison. The lightly shaded bar in Figure 5 is a value for corrosion of iron a t p H 7 , interpolated from Figure 3. This figure is substantially the aame as those obtained with film treatment a t the Figure 5. lower oxygen content. 1764
30
Some further reference to the two types of corrosion control would appear warranted. It should be borne in mind that pH control has the marked advantage of eliminating some oral1 of the corrosiveness. pH control may have the disadvantages of in some cases only partially eliminating the corrosiveness, e.g., with iron which requires a pH of 8.2 min. ( 9 ) , of being costly where very corrosive liquids are concerned, and of being likely to attack nonferrous metals, especially if appreciable dissolved oxygen is present. Theoretically the so-called film-form-
may not be formed. Anodic action leading to serious localized corrosion may well take place a t the exposed portions, although general attack is reduced. Furthermore, film-forming materials have the disadvantage of possibly contaminating the system or possibly building up so thick as to be objectionable. In considering the best corrosion prevention for a given case, consideration must be given to the metals present. If the system
Table 11.
Effect of Film-Forming Compounds on Corrosion of Common Metals
Treatment None Octadeoylamine type" (30 p.p.m.) Quaternary ammonium salt type (20 p.p.m.)b
Time of Exposure, Days 14 21 28 14 21 28 14 21 28
Bvera e Penetration, In.,%r. x 1000 Steel Copper Braaa 20.0 1.7 1.6 19.5 2.2 1.8 21.2 2.0 1.9 20.0 1.6 1.5 17.3 1.8 1.7 11.2 1.6 1.4 18.0 1.7 1.5 16.6 1.7 2.0 12.0 1.9 1.3
a Representative of products being used industrially. b Sold commercially by E. F. Drew & Co., Inc., under trade name SLCC-Q.
1 TESTS 1 AT o.oie PPM oe TESTS AT 0.01PPM Oz TESTS AT 0.11 PPM O2
r{
AYIIIIIILL-OIWNIOTYPE AT p~ 7, INTERPOLATED FROM FISURE a.
Corrosion of Mild Steel b y Condensate Showing Effect of Various Inhibitors
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 44, No. 8
Corrosion by Water ~~~
~
Table 111. Effect of Oxygen Concentration on Action of Various Inhibitors in Controlling Penetration of Metals Penetration Rate, Oxygen In./Yr. X 1000 Content, Treatment P.P.M. pH Steel Copper Brass None
0,012 0.052 0.11
6.2 5.6 5.5
21.0 40.0 46.3
2.0 2.8 8.1
1.6 3.1 8.8
Ammonia-organic
0.012 0.052 0.11 0.012 0.052 0.11 0,012 0.052 0.11
8.0 8.0 8.0 8.0 8.0 8.1 9.0 9.0 9.0
5.2 12.0 11.3 5.6 8.0 12.4 4.1 7.0 10.2
5.3 7.6 13.3 4.5 8.7 11.8 7.0 5.4 14.6
4.6 8.1 12.2 5.6 10.0 15.1 11.0 15.0 11.7
Amine type
0.012 0.052 0.11
9.0 9.0 9.0
4.1 8.0 9.8
Octadeoylaminetypea
0.012 0.052 0.11 0.012 0.052 0.11
6.2 5.6 5.5 6.2 5.6 5.5
11.0 17.0 30.0 12.0 22.0 38.5
6.6 12.0 15.2 1.7 2.4 7.2 1.7 2.2 8.9
6.0 13.0 12.6 2.4 2.8 8.4 2.2 2.0 12.3
Amine
type
Smmonia-organic
(30 p.p.m.)
Quaternaryammoniumb type (20 p.p.m.)
Q
b
Representative of products being used industrially. Sold commercially under name Drew SLCC-Q.
consists entirely of iron, those methods involving adjustment of p H are t o be preferred since they give the best protection.
If copper and brass are present good corrosion reduction of iron can still be obtained if the p H is maintained in the range 7.0 to 7.5. I n this range corrosion of copper and brass is low and in most cases can be considered negligible. Where corrosion of copper and brass owing to high p H from treatment may be a problem or where the corrosiveness is so great that cost of treatment t o protect steel would be excessive, a filmforming treatment may well be considered. Such a material will give marked reduction in corrosion, although the reduction is not so great as for the methods based on p H adjustment.
When corrosion is known t o be restricted to certain sections of the system, either type of treatment may be added directly to the affected zone. If corrosion is not thus localized, i t appears to be most practical t o feed volatile alkalies to the boiler; filming agents should, under these conditions, be introduced to the system as close to the boiler as can be managed.
literature Cited (1) American Society for Testing Materials, Designation D 935-49. (2) Berk, A. A,, Natl. Dist. Heating Assoc., 38, 273-86 (1947). (3) Berk. A. A..and Nirron. J.. U. S. Deat. of Interior. Tech. Paver 714 (1948). Collins, L. F., Heating, Piping, A i r Conditioning, 17, 36-41 (1 R4R). Ibid., 20,77-80, 83-7 (1948). Collins, L. F., Power Plant Eng., 48, 88-9 (1944). Collins, L.F..and Henderson, E. L., Heating, Piping, A i r Conditioning, 11 (September to December 1939); 12 (January to May 1940). Corey, R. C., paper presented before Am. Soc. for Testing Materials, Detroit, Mich., June 1948. Corey, R. C., and Finnegan, T. J., Proc. Am. Soc. Testing M a terials, 1939. Evans, V. R., “Metallic Corrosion, Passivity and Protection,” p. 546,New York, Longmans, Green & Co.,1946. Hanlon, R. T., Proc., Midwest Power Conference, 138-9. Chicago (1948). Leick. J.. Arch. Metallkunde. 3. 100-5 (1949). McKhney, D. S., McGovern, J. J., Young, 6. W., and Collins, L. F.,Power Engr., 17, 97-104 (1922). Maguire, J. J., Power & Works Engr., 45,No. 6 , 61-3,69 (1950). Skaperdas, G. T., and Uhlig, H., IND.ENG.CHEM.,34, No. 6, 748 (1942). Splittgerber, A., Vom Wasser, 16, 177-82 (1949). Straub, F. G.,Proo., Midwest Power Conference, 295-6, Chicago (1948). Ibid., pp. 439-42. Straub, F. G.,and Ongman, H. D., paper presented before Natl. Assoc. Corrosion Engrs., New York, March 14,1951. Uhlig, H.H., “Corrosion Handbook,” pp. 125-30,538-42,New York, John Wiley & Sons, 1948. Ulmer, R. C.,Power Plant Eng. (October, November 1947). Whitman, W., Russel, R., and Altieri, V., IND.ENQ.CHEJI., 16, 665 (1924). l__--,_
RECEIVED for review August 22,
ACCEPTED June 11 1952.
1951.
Action of Sodium Silicate as a Corrosion Inhibitor in Water Piping LEO LEHRMAN AND HENRY L. SHULDENER The City College of
New
York,
New
York 3 1 ,
N. Y., and W a f e r Service Laboratories, Inc., New York 27, N. Y.
HE continuous addition of a small amount of sodium silicate (8 to 12 p.p.m. of silica) to water has proved to be one of the most effective ways for controlling the corrosion of domestic water piping (17, pp. 359, 366-9; 23). A self-heding thin film forms over the surface of the pipe, practically stopping corrosion, Even though this method of corrosion inhibition has been in use for more than 25 years, no experimental work had been done to explain the mechanism of the silica film formation. The present work was undertaken with this purpose in mind. Two hypotheses had been offered to explain the film formation-chemical reaction between metal ions, or corrosion products of metals, and negatively charged silicate ions or silica micelles (19); and neutralization of oppositely charged particles, resulting in coagulation and adsorption (17). Thomas ( I S ) , Texter @I), Stericker (ZO), and Evans ( 4 ) were of the opinion it was the former, while Vail (17, 24), Hazel ( 8 ) , Schwartz (16),Merrill (fs),and Leaf ( 1 1 ) believed it was the latter. However, only meager experimental work had been done to confirm either hypothesis.
T
August 1952
The problem involved here concerns the nature of sodium silicate solutions having extremely low concentrations. However, the literature pertaining t o the composition of sodium silicate solutions is almost completely confined to concentrated solutions, as reported by Kohlrausch (IO),Bogue ( I ) , Stericker (18), Cann ( I ) , Harman ( 7 ) , Main ( l a ) , Ganguly (6),Debye (3), and Nauman (14). Generally, they indicated that sodium silicate solutions contained both ionic and colloidal forms in equilibrium. Thus it could be postulated that the following equilibria existed in sodium silicate solutions: Solution
-1 t
Solid
*\L
% Ionic
@ Colloidal
It is seen that the necessary components for the two hypotheses advanced to explain the film formation are present.
INDUSTRIAL AND ENGINEERING CHEMISTRY
1765
