Corrosion of Enameled Tin-plated Covers o Glass Containers EDWAKD F. KOEI3IAN Campbell Soup Company, Camden, ,V. J . Canners of vegetables have observed for a number of years the early occurrence of perforations in tin-plate covers when such foods were packed in glass. As manufacturers of strained baby foods shifted from the use of cans to glass containers this type corrosion occurred more often, regardless of the weight of the tin coat on the jar covers. This presaged a heretofore unrecognized corrosion factor. The data presented explain the mechanism of the
corrosion causing the perforations. It involves the ability of the seal to conduct an electric current. Consequently a galvanic couple is set up with metal exposed on the inside of the jar as the anode and metal exposed outside the jar as the cathode. This electric conductivity is due to the conductivity of the gasket or to capillary films of electrolyte on the interfaces of the gasket, or both. Correctibe measures expected to eliniinate electrolyte are suggested.
REQUENTLY, unpredicteti and extensive perforations of enameled tin-plate caps on glass jars containing strained foods appear soon after packing. These perforations occur almost indiscriminately in tin coatings ranging from 0.5 pound electrolytic to 1.5 pounds hot-dipped tin per base box. Products with p H as high as 5 t,o 6 arc involved. There is obvious indication of the operation of a highly effective corroding factor not previously recognized. The data presented here, developed t80 disclose a metins to eliminate t,his fa,ctor, indicate that the perforations are due to a galvanic couple of which (1)the microscopic area of exposed metal on the portion of t.he cap within the jar is the anode; ( 2 ) the relatively large area of the raw-cut, air-exposed peripiicry of the cap is the cathode; and (3) a filni probably maintained by the force of capillarity in the seam, through which hydrogen ions and possibly sodium and pobassium ions can travel, is the electrolyte.
of approximately 4000 jaw, were packed. The storage t,enipc.r:rture was 80" =t 2 " F. A jar representing each variahlc IWS opened for examination after the folloming storage periods:
PERFORMANCE O F CAPS ON .JARS COYTAINING SOUPS
I n the folloit-ing experiments two lots of caps, A and B, both made from the same lots of tin plate but coated with different enamels, were used. The difference in adhesion of the various coatings to the metal was reflected in the rate and number of perforations and the surface darkening of the foods and has been helpful in elucidating the cause of the perforations and the surface darkening of the foods. Experiments included the follon-ing five soups, all of a fairly high pH, varying over a range of 0.8, and each involving scveral ingredients : Date Packed, 194s P1-I Eight vegetables with oatmeal ripril 1 5 13 Liver with seven vegetables and oatmeal hlarch 19 5.44 Beef with six vegetables and egg noodles hfarch 17 5.65 Lamb wit,h five vegetables and barley March 19 5.78 Chicken with five vegetables and semolina March 18 5.93
Storage, Months 6 9
Hot-dipped Hot-dipped Electrolytic Electrolytic
1.oo
0.50
On the dates indicated about 100 jars for each of 40 variables (two lots of caps, four kinds of tin plate, and five products), a total
,Jan. IS, 1949 ,Jan. 19, 1949
9314
10
Table I indicates the iiloliths of storage aft,cr which the cap wvn-; perforated when only onc cap of each variable was examinctl; also the product in the jar and tjhe csoating of tin on the cap. Three of the A raps out of 20 were perforated wit,hin 6 months after paclcing and nineteen out, of 80 wrre perforat.ed whi>ii opened within 10 months after packing. Perforation as usc~l here means any pit on the inside of the cap that has entirely pen('trated the base plate. The outer coat of tin and enamel may still hold the vacuum of the jar for varying periods thereafter, bui, :I mere touch of a probing needle will cause it to break through. During storage there was an even more serious developme,rit, than the perforations. Fine months after packing, the food in jars with A caps showed surface darkening sufficiently serious to make these foods, of a light color, nonmerchantable. This was due t'o excessive iron iniparted hy thc cap as was evident (Table 11) from the iron content, of the t,op one third of the food in the jar. On January 22, 1949, 10 months a.fter packing, every jar was osaniined without' opening. Table 111gives the percentage of the c q s found to be perforated. Had the jars been opened and t.hc caps examined from the inside, the percent,ages would have been higher. This follows from
TABLEI. CORROSIOP~ O F CAPS ox GLASS,JARS COSTAINIAO SOUPS
(Inspection of only one cap repiesenting each variable) Inspectionsa When Caps Were Perforatod -. Vegetable Liver Beef 1.anib Chickou
Caps A and B were made of the following four tin plates: Tin per Base Box, Lb. 1.50 1.25
I!;xiiniincd
Caps, A enamel Tin plate, lb. t i n b a s e box 1.5 1.25
.O
0.5
Caps, B enamel 0.5-lb. tin/baae box a
4th
..... .....
l s t , 4th
.. . . ,
4th
..
itti 4th
3rd
1st :314
,,
9a/4 months; a n d 4th after 10 months.
1578
,
,
1st inspection after 6 months' storage;
,
..
4th
'4th' 4th 2nd,3rd,4th
4th
2nd after 9 inontiis:
l h t , 2nd 2 n d 4th 4th
. ... . 8?d
3rd after
August 1950
INDUSTRIAL AND ENGINEERING CHEMISTRY
CONTENT OF SOUP AFTER STORAGE TABLE 11. IRON Iron, Mg. per 100 Grams B caps Acaps 2.46 1.14 1.96 0.76 1.30 0.70
Lamb soup, single jar Lamb soup, composite of t h e e jars Chicken soup, composite of three jars
TABLE 111. CORROSIOX
O F CAPS ON
GLASSJARSCONT.4INING
. SOUPS (Inspection nithout opening after storage for 10 months) Caps Perforated, Yo Vegetable Liver Beef Lamb Chicken
1 26
10.7
Caps, B enamel
Tin plate, lb. tin/baae box 1.5 1.25 1.0 0.5
3.2 7,7
8.8
16.3
0 0
0 1.1
2.7 1.1 0 1.1
0
0 3.3 0
1.1 1.9
10.7 18.1
the fact, as stated above, that many of the pits that form from the inside are held for some time from breaking all the way through by the outer tin and enamel coats. Many of such points are noticeable from the outside but not all. In summary, of the nearly 2000 A caps, 21.4%, and of an equal number of B caps, 3.8%were perforated 10 months after packing. Two months later over half of the remaining A caps were perforated, The effect of variables such as product, pH, and weight of tin coat was overshadowed by the intensity of the corroding factor. It is a well known fact that coats of enamel vary in their adherence t o metal. The enamel on the A caps adhered excellently; it was broken, viewed macroscopically, only where pits or perforations in the metal had formed. If the development of the pits in the metal could have been followed microscopically from their beginning, it would have been found that the pits form where microscopic breaks occur both through the enamel and possibly through the tin coat. In contrast, the enamel on the B c a p adhered poorly. Over the area of the skirt of the cap it generally peeled, and there was severe detinning of the cap over this area. I n addition blisters formed, often scattered a t random, but more generally where the metal had been bent or drawn in forming the cap. Under each blister there was a pit in the metal, although the pits under a large blister seldom became the point of a perforation. This enamel failure was the reason for fewer perforations in the B caps as well as less surface discoloration of the food due to iron. As will be evident from the data to follow, tin is anodic to iron and enamel failure exposes tin, which in turn inhibits iron corrosion. ELECTROCHEiMlCAL RELATIOXSHIP BETWEEN IRON AND T I N
For the purpose of discovering the cause of these perforations, hundreds of corrosion experiments were made to study the effect of all the cap components. In these experiments sheet iron (steel base plate for tin plate) specimens, 2.5 by 4 em. in dimension, and similar specimens of pure sheet tin were used. These specimens, alone or coupled, treated or untreated with one of the various cap components, were imbedded in the contents of a jar of one or more of the five products involved. The cap ingredients studied included China wood oil, whose unsaturated double bonds theoretically might act as a hydrogen acceptor and thus increase corrosion (3); titanium dioxide (one C.P. and four commercial samples) and zinc oxide, each of which theoretically might supply osygen to enhance corrosion; and the gasket.
1579
To join two corrosion specimens, usually one of iron with one of tin, so they might constitute the two poles of a galvanic couple, a strip less than a millimeter wide was cut half the length of the width of the specimen and the tips of the strips were then fused together. An iron specimen was fused t o a second iron specimen with a tiny drop of molten tin a t the tip of the narrow strips. This avoided signifirant exposure of tin. The iron specimen was weighed before it was fused to the tin specimen and the couple was weighed after the fusion. After the corrosion test the couple was first weighed, then the tin contact was melted and all tin on the iron was wiped off, while molten, with a towel. This made it possible to determine what loss was due to iron corrosion and what loss was due to tin corrosion while the two were in the couple. -4s in previous work of this kind ( I ) , all specimens were pickled t o remove any oxides of the metals. Every test was made in triplicate. Since only one of the commercial samples of titanium dioxide and the gasket significantly affected the corrosion, the averages of the triplicate tests with these alone are given. Table IV shows the loss in milligrams of the 2.5 X 4 em. specimen treated with titanium dioxide and gasket, exposed in liver soup for 26 and 28 days, respectively. T o expose a specimen to titanium dioxide, about one half of one side was smeared with a thick paste of the titanium dioxide made with distilled water. This paste was then partially dried. There appeared to be free contact of the food liquid with the metal through the titanium dioxide layer. To expose a specimen to the gasket, two gaskets were each wrapped twice around the specimen. The iron specimen of experiment 1 shows that both the titanium dioxide and the gaskets increased the corrosion of the iron. Experiment 2 merely indicates that when two iron specimens are coupled each corrodes about the same an uncoupled iron specimen. From experiment 3, it appears that the increased corrosion of the treated specimen of iron decreased the corrosion of the untreated specimen of iron coupled with it. I n other words, these materials tend to increase iron corrosion directly in contact with them, and this decreases corrosion on adjacent iron. Since the perforations represent iron corrosion adjacent to, not in direct contact with the gasket or titanium dioxide, one can scarcely incriminate these materials. This behavior of the gasket is somewhat comparable to the sulfur-bearing compounds suggested by U. S. Patent 2,168,107 ( 3 ) to be incorporated into the gasket to inhibit corrosion.
TABLE IV. CORROSION TESTS (Specimens exposed in liver soup: titanium dioxide samples, 26 days; gasket samples, 28 days) Loss in Weight, Mg. TiOn Gasket UnUnExpt. Specimens treated Treated treated Treated NO. 2.5 X 4 Cm. controls samples controls samples 1 Iron 3.27 5.00 3.67 5.17 2 Iron couoled with iron 1st specimen 3.20 3.73 ... 2nd specimen 3.27 ... 3.27 Couple iron with iron 3 2.96 4.73 1.37 9:i3 Tin 4 0.53 0.66 0.73 0.61 5 Iron coupled with tin Iron specimen 0.9 1.23 ... Tin specimen 10,96 11.43 ... 6 Iron specimen ... 0.9 1.13 Tin specimen 11.7 ,.. 16:4 , . . 7 Iron specimen 0.86 1. 0 Tin specimen ... io:ij ,.. ii,'43
...
... ...
Significant figures in Table IV are for the untreated, noncoupled iron specimen of experiment 1 and the untreated noncoupled tin specimen of experiment 4, compared t o the coupled iron and tin specimen of experiment 5. This comparison 8hows that by virtue of being in the galvanic couple, the corrosion of the iron specimen is reduced by about two thirds whereas the corrwion of the tin specimen coupled with it is increased 15- to
1580
INDUSTRIAL AND ENGINEERING CHEMISTRY
20-fold. Comparing the control in experiment 5 with experiments 6 and 7 in which the iron and t,in were t,reated, the anodic relation of the tin to the iron overshadows any effect, that the titanium dioxide or gasket might have had. This anodic relation of tin to iron explains jvhy the pit under a blist'er on the B caps does not generally perforate. As the blister forms, a greater and greater area of tin surrounding the pit is exposed to inhibit iron corrosion in the pit. The areas of tin exposed by peeling of the enamel where blisters are not formed tend t o inhibit, iron pit'ting to form perforat,ions wherever iron may be exposed. Enamel failure is, however, not a satisfactory measure against perforations. The anodic reaction of tin in contact with iron has been shown previously ( 1 ) within the range of the acidity of the f r u h tested. That it is also decidedly anodic, in the range of p H in the products herein discussed is highly important. CORROSION OF CAPS OK GLASS JARS CONTAINING FRUITS AND VEGETABLES
Corrosion in cans of fruits and veget'ables generally involves hydrogen which is commonly evolved as gas. Enough hydrogen thus evolved will cause one or both ends of the can to bulge, resulting in so-called springers and swells; these, rat'her bhan perforation of the metal, will make the cans unmerchantable. Because of the anodic effect of the tin, not enough iron is imparted t,o the food in cans to darken it even when hydrogen has caused springers or swells. No hydrogen formation in jars of strained baby foods is ever in evidence. hIany jars, with normal vacuum, when examined closely have been found to be perforat,ed up t o the outer layer of t,in and enamel and to have numerous other pits. An analysis of the gas in a considerable number of jars closed with caps A and B and packed more t'han a year has revealed not a trace of hydrogen. The blisters that form on the inner surface of the B caps are filled with a clear liquid. Spot test's of this liquid indicate it to have a p H comparable tjo that of the food in the jar. Analysis disclosed a chloride ion content equivalent to 3.5% of sodium chloride, but a sodium content equivalent to only 0.11 yosodium chloride in one sample and 0.053% in another sample. The jar contents contained approximately 0.6% sodium chloride. The clear liquid in the blisters contains ferrous iron equivalent to the chloride present. Exposed to the air this liquid develops a brown flocculent precipitate which is judged t'o be ferric hydroxide. This seems to be conclusive evidence that the pit. under a blister represents only the anode of a galvanic couple. The fact that no hydrogen appears in the jar with the type products tested in this work together with the other facts herein revealed is evidence that the cathode of this couple lies outside the jar. The electrolyte of the galvanic cell must permit hydrogen ions and possibly sodium and potassium ions t'o travel t'o the cathode. This can be made possible by a film of liquid formed by the force of capillarity through the seam or by a film of food entrapped in the seam to make contact with the periphery of the cap, or hydrogen ions may travel through the gasket. This would result in the relatively large uncoated metal area of the periphery of the cap, acting st.rongly cathodic (because of its contact, with the oxygen of the air) t80the relatively small areas of metal where the perforations occur. A somewhat analogous phenomenon has been observed in cans. At one time paper gaskets were used on the bottom end of the can which could be seamed dry. In such cans there formed just outside the seam a white, crusty material referred to a t the t,ime as "white rust." Analysis revealed it to be composed of sodium, potassium, tin, and carbonate. No negative ion from the can contents accompanied the sodium and potassium, which were no doubt accompanied by hydrogen ions, since they travel faster than sodium or potassium ions. The thinner gasket of rubber now used, and the greater pressure with which the double seam of the can is compressed, effectively eliminates contact of the electrolyte with the oxygen of the air.
Vol. 42, No. 8
The penetrating force of capillarity is exemplified by t'hc double seam of a can in anot'her connecbion. It is not uncommon t,hat spoilage results if the water in the cooling t'ank becomes contaminated with bacteria in considerable numbers. Thir occurs kl-ith normal seams that are adequate to hold the vacuum of' the can. Chlorinat'ion of cooling rvat8erhas been developed as a corrective measure. This phenomenon can be explained on the basis of a film of bacterial thickness through the seam. In February 1949 an additional major experimental pack was made consisting again of 100 jars for each of 29 variables. The same five soups as were used in t'he previous experiment were packed in jars with six lots of caps, among them those recoinmended a t the time for conimcrcial use, as well as some experimental lot8s, varying as t'o tin and enamel coating. Besides generally substantiating the results of the previous experiment, this pack demonstrat.ed a serious drop in vacuum between 10 and 13 months after packing. At, the time of packing each ,jar \vas "tapped" as is customary iit bhe industry and every Ion. vacuum jar was discarded. As a further check the vacuum of one jar of each variable was periodically taken with the following results: ~
29 jars each test
Average
Vacuum, Inches Min.
Max.
After 7 months After 9 months -4fter 9.5 months
These figures, except as reflected by the minimum, indicate no significant loss in vacuum. An occasional vacuum of the order of these minima may escape the t,apper. The next examination of the pack was made 13 months after packing when external examination revealed 49 perforated caps and tapping revealed many jars wit'h low vacuum. A few jars, up to a dozen for each variable, were selected for vacuum determination. Most of those select,ed had indicated a low vacuum when tapped, but some with high vacuum were purposely included. Of 189 jars selected, 43 had a vacuum of 0 to 10 inches. Of these, 32 were perforated. The remaining 146 jars with vacuum ranging from 11 to25 inches had an averageof 16.1 inches. At the next examination (14.5 months after packing) the 405 jars containing the product that has always manifested the least surface discoloration and the least perforated caps were tapped, and 109 jars were selected for low vacuum tests. The average vacuum of these was 15.3 inches, minimum 10 inches, and maximum, 19.5 inches. The only possibilities for this loss in vacuum are : A. Generation of hydrogen due to corrosion: The gas in t h e head space of many jars was collect,edfor analysis but no evidence of hydrogen was detected in the Moorehead gas buret. B. Generation of carbon dioxide due to spoilage: The food in these jars was not spoiled nor did the gas in the head space contain more than normal amounts of carbon dioxide such as one encounters immediately after packing. C. Air seepage: The presence of both ferrous iron and exposed metal would prevent any- accumulation of oxygen at the rate of air seepage. This oxygen in turn would enhance the rate of corrosion. Having learned the n~echanismand cause of the perforations, corrective measures are suggested. Basically, an electrolyte through the seam must be prevented, whatever may be the reason for its being there. Briefly, the following suggestions are offered: 1. Omit the enamel coat on that port'ion of the cap covered by the gasket. Enamels generally are not impervious and the minute particles of the titanium dioxide increase perviousness, possibly by behavior similar to soils in which capillary forces 18,000 times the force of gravity have been demonst'rated. Corroded areas, with severe detinning and prominent pitting of t,he base plate, have been observed under almost unbroken pigmented enamel. A cap coated with an unpigmented enamel now under t'est for 10 months gives evidence of the soundness of this principle. 2. Provide a gasket of water-repellent material or one coated
August 1950
INDUSTRIAL AND ENGINEERING CHEMISTRY
with such a material to counteract capillarity, which might conceivably cause a film of electrolyte between the gasket and the cap to reach the outer edge of the cap, a gasket that will not itself conduct an ionic current. Probably one of the polyethylenes or a film of paraffin would be most effective. Merely treating the surfaces with one of the many detergents may be effective, by virtue of its effect on surface tension. 3. Avoid entrapment of a film of food in the seam. ACKNOWLEDGMENT
Acknowledgment is made to A. D. Bowers, chief chemist, control laboratory, Campbell Soup Company, for the microdeter-
1581
mination of chloride and t o Jackson B. Hester, soil technologist, Campbell soup Company, for the determination of sodium by the flame photometer method* LITERATURE CITED
(1) Kohman, E. F., and Sanborn, N. H., IND. ENG.CHEIM., 20, 76
(1928).
(2) Stevenson, A. E., and Flugge, S. L., U. S. Patent 2,168,107 (Aug. 1, 1939). IND* ENG* CHEM** 754 (1909)* Ws (3) ' 1
R~~~~~~~jUly 19, 1949.
Cold Compression Set of ~
Elastomer Vulcanizates ROSS E. MORRIS, JOSEPH W. HOLLISTER, AND ARTHUR E. BARRETT Rubber Laboratory, Mare Island Naval Shipyard, Vallejo, Calif. T h e significance of the cold compression set of elastomer vulcanizates as regards internal viscosity, second order transition, and tendency to crystallize is discussed. It is shown experimentally that the ability of a gasket to maintain a seal when compressed for extended periods at low temperatures can be foretold from compression set tests.
S"""
'RAL years ago the Mare Island Rubber Laboratory proposed a cold compression set test for elastomer vulcanizates ( 1 2 ) in the course of work on t,he low temperature properties of rubbers for the Bureau of Ships. Subsequently, other governmental and industrial laboratories have adopted this test for the evaluation of rubbers a t low temperatures (8), and the test has appeared in a military specification for gaskets (14). The purpose of the present paper is to review the significance of c o l i compression set and to demonstrate the relation between cold set and the sealing ability of gaskets a t low temperatures. PROCEDURE FOR COLD SET TEST
The cold compression set test follows the procedure of the A.S.T.M. hot compression set test, method B ( I ) , except for time intervals, temperatures of conditioning and recovery, and load on the presser foot of the dial micrometer used for the thickness measurements. In the cold compression set test, the specimen is compressed at room temperature between chromium-finished plates and held a t constant deflection by means of spacer bars. The clamped specimen is immersed in a cold methanol bath or placed in a cold chamber within 5 minutes after compression. The specimen is released from the clamps a t the end of the conditioning period without removal from the cold conditioning medium. After allowing the specimen to recover in the cold medium for a definite period, its thickness is rapidly measured with a dial micrometer gage. The time intervals and temperatures of conditioning and recovery have not been definitely established for the cold compression set test, except that the temperature of recovery is always the same as the temperature of conditioning, and both are generally below 40' F. A 30-minute recovery period was used for the work reported here. It was found advisable to debrease the load on the hemispherical presser foot of the dial micrometer used for the thickness measurements from 3 ounces to l/z ounce in order to lessen the indentation of the rubber specimens by the foot. For example, a l/n-inch thick specimen of 35 Shore A hardness was indented 0.010 inch more when the 3-ounce load was used than when the '/$-ounce load was used. This greater indentation, of course, reduced the accuracy of the thickness measurement by a corresponding amount. It follows that the present A.S.T.M. hot compression set test would be more accurate if the thickness measurements were made with less weight on the presser foot.
THEORY
Cold compression set is not caused by the same phenomena in the rubber as hot compression set. Hot compression set is caused by plastic flow and sometimes by further vulcanization of the rubber while compressed ( 2 ) . Cold compression set is caused by slow rate of recovery due to high internal viscosity, and also may be caused by crystallization or by second order transition if conditions are favorable for either of these phenomena. In the case of rubbers which do not crystallize, cold compression set is due only to slow rate of recovery, which will be referred t o as the viscosity effect, or to second order transition. A test specimen of such a rubber, rpleased after compression, strives to recover its original unstressed shape during the recovery period at, low temperature. The effort put forth increases with the temperature of the rubber. I n other words, the recovery stress is a manifestation of the entropy of the rubber. The recovery of the specimen is retarded by its own high viscosity, but eventually the specimen recovers its original shape unless it is below i t s srcond order transition point. An understanding of this behavior can be gained by referring to Figure l , which shows a highly simplified mechanical model of rubber known as the Maxwell unit. The viscosity of the fluid in the dashpot of the Maxwell unit retards the recovery of the compressed spring, and the viscosity of this fluid, of course, rises as the temperature falls. When the fluid in the dashpot is frozen, the compressed spring cannot recover and a condition similar to rubber below its second order transition temperature results. In the case of rubbers which crystallize, the combined effects of high viscosity and crystallization are responsible for cold c o m p r e s s i o n set. Crystallization blocks full recovery of the deformed elastic network a s long - as the rubber is kept at a sufficiently low temFigure 1. Maxwell Unit perature. The ex-
t
-7-