Historical Articles

July, 1952 issue of Plating

 

Quantitative Measurement of Adhesion of Electrodeposited Metals*

H.C. Schlaupitz, R. Wallace & Sons Manufacturing Co., Wallingford, Conn. and W.D. Robertson, Hammond Metallurgical Laboratory, Yale University

*Abstracted from a thesis presented by H.C. Schlaupitz in partial fulfillment of requirements for the degree of Master in Engineering, Yale University, 1951.

INTRODUCTION
For many practical purposes adhesion of an electrodeposited metal to a metallic base may be evaluated by qualitative bend and twist tests. To obtain an understanding of the factors involved in adhesion, more precise, quantitative data are required, and it is for this reason that numerous tests have been proposed. The extensive literature up to 1945 has been reviewed by Ferguson and Stephan ¹; additional quantitative tests ^2,3,4,5 have since been proposed, but with one exception⁵ they are modifications of tests previously described.

In general, to determine the bond strength of an electrodeposit it is necessary to measure the stress required to produce fracture or separation at the interface. Detailed consideration of the problem indicates, however, that stress measurement is considerably more difficult than it appears and that no test so far proposed is fully adequate for the purpose.

In the case of a weak bond, when little or no deformation of the base metal or electrodeposit takes place during the measurement of the load required to separate a unit area of plate, the values obtained are a measure of the bond strength. When a stronger bond is tested, however, the force required to separate the electrodeposit from the base metal may approach or exceed the force necessary to cause plastic deformation in one or both components of the specimen. In this more general case, the significance of the applied load is unknown, and the stress state existing at the interface is undefined and certainly very complicated. The general problem may, therefore, be stated in terms of the special case of a weak bond.

Actually the problem is analogous to that encountered in measuring the fracture strength of a metal in the absence of deformation, as distinct from the conventional tensile strength which has no fundamental significance. Attempts have been made to measure fracture strength without appreciable deformation by imposing embrittling factors such as- lowering the temperature and notching the test specimen. In the present case, deformation may be minimized by design of the test specimen.

In the following discussion, the problem of measuring adhesion is defined as precisely as possible, certain of the tests that have been proposed are evaluated, and a technique is demonstrated which appears to offer possibilities as a quantitative method for measuring adhesion.

DEFINITION OF AN ADHESION TEST
It has been demonstrated ^6,7,8,9 that under ideal conditions the structure of the base metal, including annealing twins and other detailed structural elements, may be continued in the electrodeposit (Fig. 1). Where this occurs, the measurement of adhesion is, in effect, a determination of the strength of the metallic bond. So far, this bond strength has not been measured unambiguously, because, with the tests employed, ductile metallic materials deform plastically under load, and local stress concentrations reduce the apparent average stress required for fracture. When appreciable plastic deformation occurs before fracture, the state of stress at the ductile fracture is not well known, and the fracture stress is dependent on the amount of prior strain. Furthermore, plastic deformation and fracture obey different laws; the former depends on the critical resolved shear stress, and the latter probably depends on a critical tensile-stress condition. In any event, the conditions for flow and fracture must be separated from each other and considered in the design of an adhesion test.

An adhesion test which includes these considerations will more nearly measure the desired property. Some of the characteristics of such a test method are:

1. The loading force should be applied as simple tensile loading perpendicular to the interface so that the geometrical arrangement of the three principal stresses will be as simple as possible, one acting in the direction of the applied load while the other two are acting in the plane of the interface.

2. The region under test should be a solid cylindrical disk. Past experience indicates that a solid round tensile specimen is best for the evaluation of the mechanical properties of materials, because the symmetry existing in such a specimen permits calculation of stresses and strains. With a round specimen, one principal stress acts along the axis of the cylinder, and the other two are radial and tangential.

3. The adhesion specimen should be so designed that stress concentrations at the interface are minimized. This is especially important owing to the fact that the plastic flow, which might normally relieve local stress concentrations, is also to be minimized. Premature failure would occur in regions of high stress concentration, and a lower average stress value for failure would be obtained.

4. The state of stress at the interface should be one approaching “hydrostatic tension”, i. e., a triaxial stress condition where all three principal stresses are tensile in nature and equal in value. Such a state of stress meets the theoretical requirements for fracturing. Furthermore, theoretically, with this state of stress, plastic flow cannot occur.

5. The state of stress on either side of the interface should be the same, in order that the evaluation of the state of stress existing at the interface be simplified. This condition can be approached when the base metal is coated on both sides with the electrodeposit. With such an arrangement (Fig. 2), the foregoing requirement will be met in the limiting case when the thickness of the base metal approaches zero.

6. The specimen should be designed so that failure will occur at the interface under investigation rather than in the deposit or base metal. If the bond is weaker than both components, no special problem is involved. If the bond is stronger than one of the components and weaker, than the other, the problem may be solved by making the dimensions of the weaker component approach that of the bonding region. On the other hand, if the bond is stronger than both base metal and deposit, as might be expected for ideal adhesion, then the requirement can be realized only in the limiting case where the thickness of the deposit and the base metal approaches that of the region involved in adhesion.

The actual thickness over which the bonding forces exist has never been established; it may be only a few atomic diameters. But it must also be considered that on an atomic scale the interface is not an absolutely smooth plane even for the most highly polished metals, and consequently the region involved in adhesion testing probably includes the entire distance between the peaks and valleys of the interface.

7. From the aforesaid considerations it is evident that for an ideal adhesion test the deposit and the base metal should be as thin as possible. Some means must be devised for gripping such thin sections in order to apply the load; the gripping means must be able to transmit a state of “hydrostatic tension”, and it must not fail before the composite under test. No method of supplying the gripping means (such as heating) should be employed which might affect the adhesive strength.

The seven items listed above are the principal features of an ideal adhesion test. From a consideration of these characteristics, it is evident that one of the difficulties in adhesion testing is that the dimensions of the volume under test are very small. The application of macro stress analysis on this micro scale may introduce errors, but the errors are probably smaller than those normally encountered in conventional adhesion-test designs.

In order that the data obtained from an adhesion test be evaluated fully the following points should be noted and recorded:

(1) The total elongation occurring during the test or the change in diameter should be noted. This information will provide some measure of plastic flow occurring before fracture. Also, it should be observed whether a maximum load is obtained during loading or whether the load increases steadily to the point of fracture. This feature has been neglected in adhesion data presented in the literature; in most cases it is impossible to determine whether the values presented are for maximum loads or for fracture loads.

(2) The location and type of fracture should be examined, both on a macro and a micro scale. It should be observed whether the fracture started at the surface, in the center, or uniformly throughout the cross section, and whether it occurred at the interface, in the deposit, or in the base metal. These observations indicate the nature of the state of stress that existed at the time of fracture and what the results obtained represent.

EVALUATION OF KNOWN QUANTITATIVE ADHESION TESTS
The quantitative adhesion tests previously proposed can be classified into four types:
(1) The Burgess method ¹⁰
(2) The Ollard method and its modifications ^{2,4,9,11,12,13,14}
(3) The shear-test method ^3,15
(4) The Brenner nodule method ⁵

An evaluation of these tests reveals that they do not meet all the requirements of the “ideal” test. In most, the state of stress existing at the interface is very complicated and cannot be analyzed. This is due mainly to the following:
(a) The geometry of the test specimen and eccentric loading are conducive to stress concentrations.
(b) The location of the interface and the geometry the test specimen is such that plastic yielding can occur if the bond strength approaches the yield strength of the base metal or of the electrodeposit.
(c) Two different metals bound the interface. This unsymmetrical condition produces different stress state on opposite sides of the interface because of differences in moduli and flow characteristics.

It appears that the quantitative value obtained from these tests, while useful as a means of demonstrating trends in plating procedure, are not well defined and that further work on test design is indicated.

Initial experimental work was undertaken to investigate the reproducibility and significance of data obtained with Knapps modification⁴ of the Ollard adhesion test. Because the ultimate purpose of the tests was to determine the effect of various factors on the adhesion of silver deposits, it was decided to evaluate the testing procedure by means of tests on silver sheet, without an electrodeposit. Specimens of the shape specified by Knapp (Fig. 3) were machined from anode-silver sheet approximately 0.125, inch (3.2 mm) in thickness. Some of the specimens were tested in the original Knapp fixture, and others were tested in the slightly modified fixture shown in Fig. 4. The modifications consisted of inserting dowel pins, threading both the plunger and guide, and reducing the clearance between the plunger and the holder. They were made in an attempt to improve the loading conditions. The load was applied to the plunger by a Southwark hydraulic tensile machine with load ranges of 0-500 lb (0-227 kg), 0-2000 lb (0-909 kg), and 0-12,000 lb (0-5454 kg). Representative results obtained with silver sheet are presented in Table I. Also presented are the tensile strengths obtained on the same material using the standard A. S. T. M. sheet tensile specimen.

Table 1. THE TENSILE STRENGTH OF SILVER--A COMPARISON OF VALUES OBTAINED WITH KNAPPS
ADHESION SPECIMENS AND THE STANDARD A.S.T.M. SHEET TENSILE SPECIMENS

 

The tensile strengths obtained on adhesion specimens of similar dimensions are reproducible within ±5 percent of the average. No significant difference is noted between specimens tested with the original fixture and those tested with the modified fixture. For the annealed material, the strength value obtained with the adhesion specimen is, approximately 2000 psi (1.40 kg/ mm²) higher than that obtained with the standard A. S. T. M. sheet specimens taken parallel to the previous rolling direction. For the cold-rolled specimens the average is approximately 1400 psi (0.98 kg/mm²) below that obtained on the standard A. S. T. M. tensile specimens tested in the direction parallel to the direction of rolling and approximately 5450 psi (3.83 kg/mm²) below that on the standard specimen in the direction normal to the direction of rolling. For adhesion specimens of nickel plated aluminum alloys, Knapp also obtained slightly lower values (700-3900 psi; 0.49-2.67 kg/mm²) than those for the standard sheet tensile specimen. He did not, however, indicate the direction of testing with reference to the rolling direction of the standard specimen.

Specimens 23 and 24 in the table show the effect of the dimensions of the adhesion-test specimen on the strength values obtained. The high value exhibited by specimen 23 can probably be attributed to the low annular height of the test section. Knapp compares this effect with that of a notch in a tensile specimen, which causes the tensile strength to increase because of the constraint of the surrounding material. The low strength exhibited by specimen 24 is probably associated with the extensive plastic deformation that occurred around the plunger, which in turn resulted from the low value of the thickness of the bottom section.

In order that the plastic flow in the Knapp adhesion test be investigated, sections of fractured specimens were examined under the microscope. The photomicrograph of specimen 14, in Fig. 5, shows a typical structure. The material surrounding the cold-rolled silver specimen is a heavy silver electrodeposit which was used to join the fractured sections of the specimen for polishing purposes. It is evident that considerable plastic flow occurred both in the vicinity of the fracture and in the bottom section of the specimen in contact with the plunger. The appearance of the test section in the photomicrograph corroborates the fact that simple tension does not exist throughout the test and that a very complex stress state prevails instead. It is also evident that failure takes place by the process of necking, or reduction in area of the annular test section almost to a point, which is characteristic of ductile failure as distinguished from fracture.

In conclusion it can be said that these tests with the Knapp type of adhesion specimen tend to confirm the previous evaluation of this and other Ollard methods. Although the tensile strength (maximum load divided by original area) of specimens of silver sheet of similar dimensions is reproducible within 5 per cent, the state of stress prevailing throughout the test is difficult to define; certainly it is not the same as that in a standard A. S. T. M. sheet tensile specimen, and there is no reason to expect that comparable tensile strengths should be obtained, though such comparisons have been attempted.
(To Be Continued in August)