The Adhesion of Electrodeposits to Plastics

by

Edward B. Saubestre, Lawrence J. Durney, Juan Hajdu & Edwin Bastenbeck

Enthone, Inc.

West Haven, Connecticut, USA

 

The 1966 Carl E. Huessner Gold Medal Award for Best Paper appearing in Plating in 1965

Presented at the AES 52^nd Annual Convention in New York, New York, July 13, 1965

Originally published as E.B. Saubestre, et al., Plating, 52 (10), 983-1000 (1965)

 

Editor's Note: This paper is part of a series on the AES/AESF/NASF Best Paper Awards.  In 1965, Edward B. Saubestre and co-workers received the Carl E. Huessner Gold Medal Award for Best Paper appearing in Plating.  A printable PDF version is available by clicking HERE.

ABSTRACT

The use of electrodeposited plastics is growing rapidly, especially in automotive and appliance fields.  With this growth, there arises a need for valid testing techniques, to measure such properties as corrosion resistance, adhesion of coatings, etc.  Adhesion testing of plated plastics is generally done by means of pull testing, using various adaptations of the Jacquet test.  Although this is a valid test, it is subject to widespread misapplication due to lack of understanding of the fundamentals involved in this test.  Some of the pertinent equations relating to pull testing are developed in the paper, wherein it is shown that a specific parameter (which includes pulling force, width of the test strip, thickness of coating, and modulus of elasticity) measures a combination of adhesion and ductility.  Practical implications for commercial plated plastics are given.

Nature of the metal-to-plastic bond

A. General considerations

Any discussion of the nature of the bond between a plastic and a metal must begin with the seemingly obvious question, Where does the bond break?  The seemingly equally obvious answer, Between the plastic and the metal, turns out to be rarely, if ever, correct.  Bikerman¹ has shown why failure of adhesion can rarely occur exactly at an interface, based on considerations of crack propagation.  In the specific case of the Jacquet Test for peeling of metals from plastics, it will be shown below that dimensional analysis also shows that the break cannot occur at the plastic-metal interface.

The break may actually occur in any one of three places, of which two are of practical importance.  The bond may break within the metal, within the plastic, or along a relatively weak boundary layer between the two which forms at one stage or another within the plastic body.  Primarily because of tensile strength and Young's Modulus considerations, a break within the metal is most unlikely in most practical situations.  Therefore, the bond breaks either in the body of the plastic itself, or along a weak boundary layer within the plastic.  Bikerman² refers to bonds that break within the plastic as "proper bonds" and those that fail along weak boundary layers as "improper bonds."

In practical terms (referring to metal-plated plastics), the former refers to "good" adhesion of metal-to-plastic, the latter refers to "poor" adhesion.  Obviously, then, to eliminate "poor" adhesion of metal-to-plastic, it is necessary to investigate what causes the formation of weak boundary layers, and then to eliminate those causes.  There are two schools of thought on why failure of the metal-to-plastic bond can occur in a weak boundary layer in the plastic, and it is probable that both approaches are of practical significance for plated plastics.

The first which might be called the "molecular theory" school,³ claims that bond failure is caused primarily by a weak chemical-electrical interaction between metal and plastic.  According to this school, strong attraction between the atoms of metal and the polymer molecules of plastic is required for a strong bond.  A valence bond between the different substances found across the interface would be ideal, but interaction between dipoles in the two phases is more likely.  This hypothesis has been objected to by Bikerman⁴ on the ground that substances (such as polyethylene) which contain no dipoles adhere well to metals, with which no chemical reaction is possible; furthermore, he points out that when a brittle polar, a brittle non-polar, and a ductile non-polar material are compared, the two brittle materials behave similarly, while there is no similarity in the behavior of the two non-polar materials.  On the other hand, the "molecular theory" helps to account for a number of readily observed phenomena.  It explains, for example, how oxidation of polyethylene produces a stronger joint with metals,⁵ because of dipole interactions between the metal and polar groups produced by oxidation in the polyethylene.  It could also explain why use of chromic acid-containing oxidizing acids improves adhesion of metals to ABS plastics.⁶  (See below, however, for an alternative explanation).  It also explains why polyethylene, which normally bonds poorly to copper, can be made to produce a good bond to it by chemically oxidizing the copper surface so as to produce a thin layer of black cupric oxide.⁷  A further objection has been made to the "molecular theory"⁴ that bond strengths are not nearly as high as they should be were dipole interactions effective in establishing overall bond strength.  This objection, however, has been met by pointing out that bond strengths are not as high as might be expected in terms of intermolecular forces because, in part, of the large size of the long-chain polymer molecules.⁸  The configurations which these adopt to give maximum interaction with the metal sterically hinder other molecules trying to attach themselves to the surface,⁹ thus limiting the amount of surface available for bonding. In summary, according to the "molecular theory," the way to avoid poor bonding of metals to plastics is to create dipole interactions at the interface between metal and the plastic boundary layer.  In practical terms, this would generally mean the creation of oxidative linkages such as the cited methods for oxidation of polyethylene and ABS before metal bonding.

The second school of thought, which might be referred to as the "rheological theory,"¹⁰ explains bond failure in weak boundary layers of the plastic as being the result of a type of adhesive (or stress-strain) failure.  Such adhesive failure in a weak boundary layer may occur owing to the formation of oils or low-molecular-weight polymers exuded from the bulk of the plastic into the surface layer, loosely bound oxides, moisture, gases or other substances in which shear or tensile failure may occur more readily than in the main body of the materials being joined.  An outstanding example of this phenomenon is the deleterious effect of moisture on the bonding of metals to ABS plastics.  A bake-out of moisture before conditioning and electroless plating of ABS, or immediately after the electroless plating step, or even after final plating with (for instance), copper, nickel, chromium, can considerably improve adhesion.  In fact, it can be stated almost categorically that the adhesion of metals to ABS plastics improves with aging, owing to elimination of moisture in the weak boundary layer.⁶  Bikerman¹¹ has given further weight to this argument by noting that "syneresis" can be a source of trouble, and that eliminating the result of syneresis will improve bond strength considerably.*  For example, in the specific case of polyethylene investigated by Bikerman, it was found that low-molecular-weight hydrocarbons and oxygenated impurities only loosely bound to the surface constituted the principal phase of the weak boundary layer formed by this syneresis.  These were removed¹¹ and found to comprise less than one per cent of the plastic, and not to affect mechanical properties (Young's Modulus; breaking strength), or interfacial properties (contact angles for water were the same for treated or untreated polyethylene), but to eliminate IR spectroscopic peaks for carbonyl groups.  The result was that poor bonding was converted to good bonding.  This approach also offers another explanation of the effect of the chromic acid treatment of plastics.¹²  For example, it was shown that chromic acid treatment of polyethylene (which admittedly improved bonding to metals) did not result in the formation of an oxidized layer by IR spectroscopy.¹³  Bikerman¹¹ concludes that the effect of the chromic acid treatment is to destroy some of the low-molecular-weight ingredients present in the weak boundary layer, thus eliminating the effect of syneresis in the surface layer.  It has been suggested that another support for this approach is the fact that passing a stream of hot nitrogen gas over polyethylene will have the same effect as the treatments discussed above in improving bonding properties;¹⁴ i.e., evaporating impurities is the key phenomenon, since no oxidation can occur under these conditions.  Some related observations dealing with fiber-elastomer adhesion have also appeared recently.¹⁵  On the other hand, the same author discussed the resistance of adhesive bonds to water, (water reduced adhesion to zero if the specimen were peeled apart when wet; if allowed to dry before adhesion testing, the bonds retained their original strength), and concluded that the reversible behavior noted could most readily be explained by postulating that adhesion was caused by hydrogen bonds.  In the presence of liquids of high energy density, hydrogen bonds were correspondingly weakened.

B. Goal of the current investigation

From this review of the literature on the bonding of metals to plastics, the following conclusions are drawn:

1. An adequate test method for evaluating the bonding of metals to plastics must involve rheological failure, since simple tensile failure at the metal-to-plastic interface can occur only in the case of so-called "improper bonds," which are of no commercial importance except in the very limited field of application for the production of what might be termed "plated-plastic-junk."
2. Experiments should be conducted to determine, on a scientifically acceptable basis, the nature of the bond cleavage between plastic and metal.
3. Experiments should be conducted to determine, on a practical (commercial) plane, which operational factors contribute to the development of good bonding between plastic and metal, and more particularly, in view of major industrial (especially automotive) interest, which factors lead to good bonding of copper to ABS plastics.
4. Comparisons should be established experimentally between rheologically acceptable tests and other industrially acceptable tests for determining the bonding of metals to plastics, particularly ABS plastics.
5. Simple chemical procedures should be evaluated for experimentally verifying the existence of weak boundary layers as previously discussed.
6. A correlation should be made between methods of testing plated plastics and those for testing plated metals.

 

The following sections will detail our attempts to meet these six objectives.

C. The Jacquet test

The Jacquet, or "pull," test is the most commonly used industrial test for adhesion of metals to plastics.  Figure 1 indicates the common geometry of the Jacquet test.  A relatively thin metallic strip (generally, the electrodeposited coating under test) is pulled at a 90° angle from the underlying plastic substrate.  The force required, either to initiate or to sustain at a steady rate, peeling of metal from plastic is recorded as the numerical value for the test (Appendix II contains further details).  It is evident that either the plastic substrate or the metallic film must yield plastically in order to permit the radius of curvature which results from pull testing.  Experimental data presented later in this paper will prove that the plastic yielding occurs entirely in the plastic film.

It will be assumed that both the plastic-yielding substrate and the metallic film obey Hooke's Law, which says that the intensity of the load applied (the stress, commonly expressed in lb/in.²) divided by the resulting elongation (the strain, commonly expressed as in./in.) is a constant (Young's Modulus; the modulus of elasticity).  When the plastic-yielding film is subjected to a loading, Hooke's Law being obeyed, the following relationship must hold:

                                                                     (1)

where E_f is Young's Modulus for the plastic film, S is the stress on the plastic film, y is the deflection of the plastic-yielding film under load, and tf is the thickness of the original (unstressed) plastic-yielding film.

Note that in order for the dimensional analysis to be correct, it is necessary that the yielding under load take place in a region of finite original thickness.  This is proof, from dimensional analysis alone, that the bond must break at a point somewhat removed from the actual interface between metal and plastic.  If the break were to occur exactly at the interface, then, from the above analysis, it follows that the breaking stress would approach zero, i.e., breaking of the bond at the true plastic-metal interface occurs only when the bond strength is virtually zero.  When there is any significant bond strength, the break must occur other than at the interface.

Thus, the Jacquet test meets the rheological requirements of Bikerman's models^1,4,10 in that it tests the breaking of bonds in the plated plastic and not at the interface.  For these reasons, we believe that the Jacquet test is a valid and useful one for general industrial use in determining the adhesion of plated plastics.  It must be remembered, however, that the Jacquet test measures, not pure adhesion, but a complex of factors.  Appendix I contains a mathematical development of the Jacquet test, from which it is concluded that the resulting numerical reading for "adhesion" is actually a measure of all of the following:

1. Thickness of the electrodeposited metal
2. Thickness of the plastic-yielding film
3. Young's Modulus of the deposited metal
4. Young's Modulus of the plastic-yielding film
5. Tensile strength of the plastic-yielding film.

 

Note particularly the first item: thickness of the electrodeposited metal.  It follows from the equations developed in Appendix I that the thicker the electrodeposited coating, the higher the Jacquet test value obtained, all other things being equal.  For example, for a given chemical conditioning process, on a particular plastic substrate, the film properties will likely be essentially constant, so that, as a first approximation, the Jacquet test adhesion value is proportional to the % power of the thickness of the electrodeposited coating.  Naturally, it does not follow then that mere obtaining of a higher Jacquet test adhesion value with the use of thicker coatings is indicative of better adhesion.  Indeed, as will be shown below, the converse is frequently the case.

For copper-plated ABS plastics, substitution of the proper physical constants (Appendix I) leads to the following relationship:

                                                                                            (2)

where F is the reading obtained by use of the Jacquet test on a 1 in. wide strip (lb.), and t_m is the thickness of the electrodeposited copper (in.).  Thus, for example, for a copper coating of 2 mil thickness, F is 9.5 lb. for a 1 in. wide strip.  In the data to be presented below, the results obtained experimentally with the Jacquet test will be compared with the above figure, and it will be found that correlation is quite good in general.  The constant (shown as 1000 above) will be referred to as "K."  Its physical significance is discussed in Appendix I.

Appendix II contains detailed information on how the authors conducted the Jacquet Test, using two types of apparatus.

In the light of these observations, efforts were next made to isolate and examine directly the detached plastic film, which could generally be observed visually as a slight "whitening" of the underside of the peeled-back copper strip.

E. Nature of the plastic film

First, an attempt was made to photograph cross-sections of the peeled-back copper strip (Fig. 4).   As will be noted, this method did not permit the direct photographing of the detached plastic film because of its fragility.  All that appears in Fig. 4 is a trace of evidence that one side of the copper strip has a little plastic on it.

G. Effect of conditioning on plastic surface layers

Studies of plastic surface layers up to this point have concentrated on the nature of the detached plastic film.  As is evident from Fig. 1, however, during Jacquet Testing tearing takes place somewhere between the metal-plastic interface and the boundary of the layer of plastic which yields plastically during pull testing.  In the theoretical development of Appendix I it was shown that t_f, the total thickness of the plastic-yielding film, is the key parameter and not merely the detached portion of the plastic film.  Accordingly, in the next part of this study, attempts were made to determine whether the plastic-yielding film could be identified before Jacquet testing, so that a better estimate could be made of the value of t_f.

Cross-sections of copper-plated specimens were prepared by cold mounting, followed by polishing with dry 120 grit emery paper, wet 400 grit paper, immersion in 25% ammonium persulfate solution to remove the copper smear; they were then rinsed in methyl ethyl ketone, very lightly polished, dipped a few seconds in persulfate, solvent rinsed and dried.  Two characteristic cross-sections are shown in Figs. 8 and 9.  Immediately evident is the existence of a dark layer in the plastic immediately below the metal-plastic interface.  More faintly detectable is a second, deeper layer, which has a slightly altered appearance relative to the bulk of the plastic.  In these photographs, the copper electrodeposit is 2.8 mil thick; the first degraded plastic layer is 0.42 mil deep; the total depth of the two types of altered plastic is 1.04 mil.

For Figs. 10 and 11, the same mounting procedure was followed as before, except that an additional step was added at the end to help bring out the chemical differences in the various plastic layers.  This final step consisted of immersing the mount in 0.5% fuchsin dissolved in glacial acetic acid. The dye is then eluted with ethyl alcohol for three minutes.  As can be seen in the photographs, this procedure brings out more clearly the existence of two different altered plastic layers beneath the metal coating.  In Figs. 10 and 11, the copper electrodeposit was 0.835 mil thick.  The first layer of degraded plastic immediately beneath the metal was about 0.31 mil deep; the combined depth of the two types of plastic layers was between 1.15 and 1.25 mil.

since B should be a constant.  Table 3 indicates some typical results obtained, in which:

                                                                                                   (5)

a quantity that should equal unity if experimental results are to follow theoretical considerations.

One interesting result of Table 3 is that k ranges from 0.81 to 1.28.  This remarkable agreement with theory indicates that the Jacquet test for "good" adhesion conditions is in agreement with theoretical models previously discussed, in which failure occurs not at the interface, but at some distance from it.

Another interesting observation in Table 3 is the fact that B appears to be much better for electroless copper than for electroless nickel, even though Jacquet test values are comparable for a given thickness of electrodeposited copper.  In like fashion, the value of F_c is higher for electroless nickel than for electroless copper, i.e., there is a higher level of Jacquet test values for electroless nickel in which improper bonding is obtained.

Therefore, it is concluded that, for a given electroplating system (UBAC copper), electroless copper plating offers evident advantages over electroless nickel plating, insofar as these tests are concerned.

Table 3 - Crack propagation in Jacquet testing.

Effect of processing factors on adhesion of metals to plastics

A. Introduction

As previously noted, the Jacquet test relates to the % power of the thickness of the electrodeposited coating.  The experimental data were therefore analyzed with this assumption in mind, that is to say, that "adhesion" test data were related to "K" values, where:

                                                                                                     (7)

As noted in Appendix I, the theoretical value of K for copper is 1000, and for nickel 1100.  Likewise, this same relationship will be used to express the theoretical value of F as a function of t_m in the graphs presented below.

The other commercially important variables which affect the Jacquet Test include:

1. The type of plastic substrate (this paper is limited to ABS plastics).
2. The electroplating system chosen (three copper solutions are considered).
3. Chemical pre-treatment of the substrate prior to electroless copper plating (some consideration will be given to this factor below).
4. Aging of the electrodeposited plastic (periods up to one week will be considered).

 

B. Effect of electroplating thickness

In these studies, the simplified strain-gage method of reporting data was used (see Appendix II).  The panels used were 4 × 6 in. plaques of ABS plastic (Cycolac EP-3510, Marbon Chemical).  The following generic treatment cycle was used:

1. Clean in a mildly alkaline solution designed for ABS.
2. CWR (cold water rinse).
3. Neutralize in an acidic solution.
4. CWR
5. Condition in an acidic, dehydrating, oxidizing solution containing hexavalent chromium (Patent applied for).
6. CWR
7. Secondary conditioning in oxidizing acid.
8. CWR
9. CWR
10. Sensitize in an adsorbing reductant, as in tin (II) or titanium (III) solutions.
11. CWR
12. CWR
13. Activate in noble metal salt, as of Pd, Pt, Au, Ag.
14. CWR
15. CWR
16. Electroless copper plate in an alkaline formaldehyde type of solution.
17. CWR
18. Bright acid copper sulfate electroplate.
19. CWR
20. Cure in hot water.
21. Dry.

 

The specific cycle used was a proprietary one, which corresponds to the above steps:

   1. Enplate PC-452, 60 g/L (8 oz/gal), 65°C (150°F), 3-5 minutes.

   3. Enplate AD-480, 7.5 g/L (1 oz/gal), Room temperature (RT), 15 seconds.

   5. Enplate Conditioner 470, full strength, 60°C (140°F), 1 minute.

   7. Enplate Conditioner 471, RT, 2 minutes.

   10. Enplate Sensitizer 432, 1:15, RT, 1 minute.

   13. Enplate Activator 440, 1:15, RT, 20 seconds.

   16. Enplate CU-400, 2A, 5B, 9 water (vol), RT, 10 minutes.

   18. UBAC copper, air-agitated, to desired thickness.

   20. Age 20 minutes at 70°C (160°F).

The results obtained are shown in Table 4.  In the table, each panel was plated to two different nominal thicknesses, and two actual thicknesses were measured in each section of the panel using the Kocour tester.  These are referred to as the thick and thin sections respectively of the panel.  In the “A” test on each panel, the pulling was done from the thick to the thin section of the panel.  In the “B” test, the pulling was in the reverse direction.

Table 4 - Effect of plating thickness on the Jacquet test.

In all cases, the panels contained two nominal electroplating thicknesses, in order to permit evaluation of the effect of electroplating thickness independently of the effects of the chemistry and molding conditions of the substrate.  The pulled strips were 0.5 in. wide, but the data were transcribed to the more conventional lb/in. value.  Data were taken in both directions, i.e., from thick to thin section of the plated panel and vice versa, in order to determine if the direction of running the pull test is of significance.  Data were taken immediately after the completion of Step #21 in the above cycle and again after two and seven days had elapsed.

The data show that the direction of pulling is of no significance.  Therefore, data for both directions of pulling were averaged in subsequent studies.

The effect of plating thickness is shown in Figs. 12 and 13.  It was found that in every case, the thicker portion of an individual panel has a higher F value and a lower K value.  That the thicker portion of the plated plaque should have a higher value of F is to be expected, based on previous discussions.  What is rather more surprising is that K obviously decreases with increasing plating thickness.  This is unexpected.  The explanation is not yet clear; perhaps it is due to the influence of internal stresses in the electrodeposit.  This effect has not been considered in the theoretical discussion.  It is also possible that the decrease of K is due to Newtonian behavior of the plastic.  This point will be considered below.  In any event, two general conclusions may be drawn from these data:

1. F increases with increasing plating thickness.
2. K ("adhesion") decreases with increasing plating thickness.

 

These conclusions are important commercially in that they amplify and serve to condition the oft-repeated statements to the effect that "greater plating thicknesses may be required to meet the minimum pull test values."  It follows from our results that such statements may be quite misleading; indeed, increasing the thickness of electrodeposit, while increasing the value obtained in pull testing, may actually result in decreased "adhesion," as determined by "K."

We were gratified to note that the experimental data correspond in order of magnitude to the theoretical predictions.  It is concluded therefore that the theoretical analysis serves to set a limit on the ultimate "adhesion" values which are attainable for electroplated plastics.  It is clearly evident from Figs. 12 and 13 that for copper electrodeposits less than about 1.2 mil thick, the Jacquet test values obtained experimentally are slightly greater than the theoretical value, while for thicker deposits, experimental values are usually less than theoretical values.

C. Effect of aging of the electrodeposited plastic

Figures 12 and 13 also illustrate the effect of aging of the electrodeposited plastic on the Jacquet test.  After two days of aging, F increased about 0.5 lb/in. on an average, and about 0.75 lb/in. after one week.  It should be borne in mind that a preliminary aging step was incorporated in the plating cycle.  If the plaques had not been so cured, the effect of aging would have been even more pronounced.  It is evident from the data that the 20-minute cure in hot water suffices to develop substantially the ultimate bond to be attained for the given system, although further aging does improve results a bit further.

D. Effect of type of ABS plastic

Three types of ABS plastics were examined in these tests.  Plastics A and B are designed for electroplating use, but "A" is produced to exacting quality control standards, while "B" is not.  Plastic C is not designed for electroplating use.  The cycle used was that noted previously, except that Steps 6 and 7 were eliminated.  In addition, times and temperatures differed slightly; differences are noted:

Step 1:             2 minutes

Step 3:             10-15 seconds

Step 5:             65°C (150°F), 15 minutes

Step 13:           1 minute

Step 16:           15 minutes

All other conditions remained as before.  The test results were obtained on a tensometer (Instron Tester), operated at a rate of pull of 1 in./min.  The maximum and minimum readings on the recording chart were noted, the arithmetic average was used as the average F value.  The results are shown in Figs. 14 and 15.  It will be readily noted that Plastic C, which was not designed for electroplating, gives markedly poorer results than Plastics A and B.  Thus, the type of ABS used is of considerable importance in electroplating.  Within the limits of the experiments conducted, no difference could be detected between the quality-controlled grade of ABS, and the one which is not controlled, with respect either to the test values obtained, or to the reproducibility of the data.

In comparing the results of these experiments with those of Figs. 12 and 13, it should be remembered that a different Jacquet test method was used. The F_MAX and K_MAX values of these data correspond essentially to the F and K values of the previous studies. With this in mind, it is seen that the same effect of thickness is observed on F and K.  However, the results obtained in these experiments on Plastics A and B were generally slightly lower than those obtained in the earlier experiments.  This was thought to be due to the different conditioning cycles used, and this point was examined briefly as reported below.

Figure 15 - K vs. thickness of bright acid copper on various types of ABS plastics.

E. Effect of conditioning

These studies did not include any comprehensive investigation of the effect of conditioning, in that only commercially useful conditioning methods were employed.  As noted above, however, some quantitative differences were noted between the data of Fig. 12 and of Fig. 14, which were thought to be due to the different conditioning cycles used.  Accordingly, a brief study was made of three conditioning cycles:

No. 1. The same as in Section D above.

No. 2. As in No. 1, except that the chromic acid content of the conditioner was doubled.

No. 3. The same as in Section B above.

The results are shown in Fig. 16.  Plastic A was again used in these studies.  Although no startling differences were found, to the extent that any differences exist, it appears that cycles No. 2 and No. 3 are slightly superior to cycle No. 1.  This probably accounts for the slightly superior results shown in Fig. 12 relative to those in Fig. 14.

G. Effect of composite coatings

Most commercial electroplated plastics are copper-nickel-chromium plated.  This means that the theoretical analysis of the Jacquet test results obtained will be more complex than that which has been considered in this paper.  Although we made no study of composite coatings as such, the submitted data just referred to above contained some information on the effect of depositing two types of copper coating, whose K values are quite different.  This is illustrated in Table 5.  It is noteworthy that even when the coating with the higher K value constitutes over 20% of the total coating thickness, the value of K closely follows that of the coating of the lower K value.  We suspect that this important relationship would apply also in a similar manner to copper-nickel-chromium systems.

H. Effect of plastic molding conditions

It is generally recognized commercially that it is quite difficult to injection-mold an ABS piece and maintain absolutely uniform molding conditions on all surface areas.  Although many factors enter in, gating is a prime example of how improper molding conditions can affect the subsequent adhesion of electrodeposits.  Improper design of the gate and incorrect ram speeds can lead to the presence on the surface of the molded piece of stressed areas in the vicinity of the gate which are unsuitable for electroplating.  The presence of such stressed areas is most easily detected, before chemical treatment of the plastic surface, by immersing the plastic piece in glacial acetic acid for one to two minutes at room temperature.  The development of white areas on the surface is indicative of undesirable surfaces for plating.  In checking the test plaques used in the experiments reported in Figs. 14 and 17, it was noted that a pronounced white area developed during the glacial acetic acid test in the immediate vicinity of the gate area.  Reviewing earlier experiments, it was noted that there was excellent correlation between F_MAX and F_MIN and the location of the gate on the test piece.  For illustrative purposes, some of the data shown in Fig. 14 were examined in detail.  The results are tabulated in Table 6, and two tracings of the original chart recordings are shown in Fig. 19.  Strikingly shown in Fig. 19 is the gate effect, along with its effect on FAVG as used in earlier tables.  It is because of this gate effect on F_AVG that we prefer to concentrate attention on the value of F_MAX and K_MAX.

Table 6 - Effect of Gate on Jacquet test - Plastic B.

                                                                                                    (8)

where m is less than unity.  Using the same values that were given in Appendix I for the various constants involved, we can calculate that for m = 0.5, K is 647; for m = 0.55, K is 767.  Thus, we might be able to account for the lower K values noted at the greater thicknesses of copper.

Comparison of the Jacquet test with other tests

A. Tensile tests

Although it is generally more practical to test plated ABS plastics by means of peeling tests of one sort or another, it is often desirable to compare the readings obtained in such tests with those obtained in tensile tests, since the latter are so commonly specified in testing plated metals.  From the analysis shown in Appendix I, it follows that, for the values of physical constants used:

            Tensile strength (psi) = 5.5 K

Thus, in Fig. 21, where it was noted that the steady state value of K was about 750, the tensile strength of the bond would be about 4000 psi.  In the same manner, the data in Figs. 12-21 may be converted to the equivalent tensile strength reading.

B. Relation of Jacquet test and thermal cycling tests

In addition to the Jacquet Test, thermal cycling tests are frequently used in the industry to determine the quality of the adhesion of electrodeposits to plastics, the theory presumably being that the difference in thermal expansion coefficients between plastic and metal will set up sufficient strains at the neighborhood of the interface as to cause rupture of the bond if the adhesion is relatively poor.  There are a number of variations of the thermal cycling test, but the version which we used is believed to be representative:

1. Immerse the plated part in water at 82°C (180°F) for one hour.
2. Allow the part to return to room temperature by removing from the water for 15 minutes.
3. Immerse the part in a cold box at -18°C (0°F) for one hour.
4. Allow the part to return to room temperature by removing from the cold box for 15 minutes.
5. This constitutes one complete thermal cycle. Steps 1-4 are repeated as often as desired.  Our criterion was 15 cycles.

 

Table 7 - Relation of the Jacquet test to thermal cycle tests.

The results of a comparison between the Jacquet test and the thermal cycling test are shown in Table 7.  The parts used in these tests were syrup server tops which were selected because they were known to be good quality moldings specifically designed for subsequent electroplating.  All the parts tested passed 15 cycles of thermal testing without any sign of blistering or peeling.  Thus, in the range of 1-3 mil of copper coating, parts whose adhesion to the plastic is at least 50% of the theoretical amount and whose Jacquet test value is about 5.5 lb/in. or more will withstand the thermal cycle test.  In order to explore further the correlation between the Jacquet test and the thermal cycling test, we next selected an ABS plastic said by its manufacturer to be unsuitable for electroplating use because of the pigments employed.  The thought was that if lower pull test results were obtained, failure of the thermal cycling test might occur.  Accordingly, test plaques of such materials (Cycolac H-9731; T-9723, Marbon Chemical) were processed using the double conditioner method and electroplating with UBAC copper.  Results obtained are shown in Table 8.  It will be noted that, indeed, lower Jacquet test results were obtained, and there was corresponding failure of the thermal cycling test.  Sometimes, the thermal shock test is used as a quality control test, rather than the thermal cycling test.  These two tests are quite similar, except that in the thermal shock test, no relaxation period is provided at room temperature.  As a typical example, we carried out the thermal shock test by immersing the part in 82°C (180°F) water for one hour, and quenching in brine at -18°C (0°F), and holding for one hour.  This constituted the first cycle.  After examination of the part, the second cycle was carried out by holding the part for one hour in brine at -18°C, then immediately immersing in 82°C water and holding for one hour.  Then after another examination of the part, the test repeats as above.

To compare the thermal cycling and thermal shock tests, we repeated the work reported in Table 8, using identical processing conditions and substrates.  Based on the meager results reported, no great differences can be noted between the two tests.

Table 8 - Relation of the Jacquet test to thermal cycle tests.

However, in comparing these two thermal tests with the Jacquet test, a problem arises in that failure in the thermal tests does not occur over broad areas of the plaque, but only in highly localized areas, which are believed to be areas of concentrated stresses in the plastic.  This observation accords with general commercial experience.

Accordingly, some further studies were made to determine whether or not thermal tests measure the same interfacial parameters as the Jacquet test.  In these studies, ABS plastic plaques (Cycolac EP-3510) were cleaned and acid dipped as previously and, while being handled with great care to avoid surface contamination, had small circular patches of plater's vinyl tape applied to four areas of the surface, as shown in Fig. 22.  The plaques were then conditioned using the double conditioning cycle.  After the second conditioning step, the patches were removed, and the plating cycle was continued in the normal way through copper and/or nickel electroplating.  Thermal cycling tests were conducted as above.  In these studies, the acid copper electroplating bath was operated to produce deposits of 4100 psi tensile stress, and the nickel bath was operated to produce deposits of 7500 psi compressive stress.

The results obtained are shown in Table 9.  In all cases of failure in the thermal cycling test the blister corresponded almost exactly to the size of the patch, and furthermore, the bond broke at the interface of metal and plastic, there being no evidence for the formation of any detached plastic film.  The latter statement agrees with commonly reported observations on industrial parts which fail thermal tests for one reason or another.  This would also be consistent with the theoretical development of this paper in that separation of the bond at other than the interface implies a degree of true adhesion.  Thus, we come to the following important conclusion: by the very nature of the failure obtained, the Jacquet test and thermal cycling (or shock) tests measure different things.  Therefore, one test cannot substitute for the other.  The only obvious exception to this statement is the case in which the adhesion of the electrodeposit to the plastic is so poor that even Jacquet test failure occurs at the interface.  It is self-evident that such cases are of no commercial interest whatsoever.

Because of the evidently limited scope of the work reported in Table 9, it is not feasible to draw too many detailed conclusions.  However, we offer the following general observations:

1. Thermal cycling test results improve with increasing total plate thickness.  This seems to offer experimental support for the time-honored "dodge" of offsetting poor intrinsic adhesion between plastic and electrodeposited metal by "encapsulating" the plastic with relatively thick electrodeposits.  By poor intrinsic adhesion, we refer to relatively low values of K in the Jacquet test (say, arbitrarily, less than 50% of theoretical).
2. Within the limits of the particular experiments conducted, it would appear that the use of a substantial copper undercoat may be helpful under nickel.  For example, 0.27 mil of direct nickel led to blistering after plating on the patch areas, whereas this was not the case for 0.24 mil of copper.  Likewise, 0.41 mil of total plate led to blisters over the patches after plating when the deposit was roughly 50-50 copper and nickel, whereas much thinner straight copper coatings did not lead to failure after plating.
3. In general, results obtained with the 3-mm patch were worse than the results obtained with the 6-mm patch.  This is contrary to what we would have expected intuitively.  However, as noted in the case of Panel 4, the results were perfectly reproducible.  We offer no explanation at present; however, if these data could be further verified in additional experiments, it would seem to imply that thermal tests may be particularly sensitive to highly localized adhesion failures.

 

C. Glacial acetic add test

This is frequently used as a quality control test to determine whether or not a given ABS plastic molding is suitable for plating.  It consists of immersing the molding in glacial acetic acid at room temperature for about 2 minutes.  Two sorts of observations may be made:

1. White-colored areas may appear on the plastic; these areas may be particularly pronounced at the gate area.  They are indicative of stressed plastic surfaces, and a part which contains such areas alternating with areas which have not become discolored will be difficult to plate satisfactorily.  In general, the white areas require a minimum conditioning time, while the unaffected areas require more severe conditioning conditions.
2. Cracks may appear on the surface of the plastic, with or without formation of the discolored areas noted above.  Such cracks indicate that the molding has not been properly annealed.

 

Regarding the correlation of this test with others, it was noted repeatedly that the whitish areas on the surface after the acetic acid test almost always correlated well with the areas of failure in the thermal cycle test.  Thus, these two tests probably measure the same thing.  From this, it naturally follows that the whitish areas in the acetic acid test do not measure the same thing as the Jacquet Test, and would not be expected to correlate with it.

On the other hand, the crack patterns which develop in the glacial acetic acid test do not correlate with thermal test results.  In general, however, parts which crack in the acetic acid test yield low readings in the Jacquet Test.  It does not, however, necessarily follow from this correlation that the crack pattern and the Jacquet Test measure the same phenomena.

D. Correlation of Jacquet tests of electrodeposits on metals and on plastics

It has been noted that the bond failure between a plastic and an electrodeposit is most unlikely to occur at the plastic-to-metal bond when examined by the Jacquet Test, except in cases of extremely poor adhesion.  The same should hold true for bond failure of metal-to-electrodeposited metal when tested in this manner.  The bond break could be either in the electrodeposited coating itself, or in the metallic substrate.

In the AES-sponsored project on the adhesion of electrodeposited metals headed by Ferguson,¹⁷ a literature survey turned up no references on the use of the Jacquet test.¹⁸  Brenner, however, had done some work in this area and Ferguson learned of it privately and was able to get full details.¹⁹  The work of Linford and his co-workers²⁰ is not germane to the present discussion, since they were dealing with failures in "molecular" bonding and not in “rheological” bonding failures.

Brenner, in his communication to Ferguson, assumed that the bond failure would occur, not at the interface, but somewhere in the upper layer of the substrate; this was confirmed by his ability to determine the locus of the break by weight loss of the substrate after testing.

Table 10 - The Jacquet test on electroplated metals.

Table 10 covers the principal results obtained by Brenner for copper plating on steel and nickel plating on copper and extruded brass.  In evaluating K, the following values were used:

E_m (Ni) = 24,000,000 psi²¹

E_m (Cu) = 16,000,000 psi²²

The following data are from the ASM Metals Handbook:

E_f (Cu)             = 17,000,000 psi

E_f (brass)         = 15,000,000 psi

E_f (steel)          = 28,000,000 psi

σ_t  (Cu)                        = 55,000 psi

σ_f  (brass)         = 48,000 psi (extruded)

σ_f  (steel)          = 100,000 psi

Specific gravity (Cu)   = 8.9

Specific gravity (brass) = 8.4

From Table 10 it will be noted that the theoretical analysis offered in this paper is in agreement with experimental data, at least in order of magnitude.  This is quite interesting, in view of the fact that we have considered only metal-to-plastic bonds, where F is of the order of 5-10 lb/in., and in Brenner's work, we are considering metal-to-metal bonds where F is 50-750 lb/in.  Thus, over very widely divergent systems, it is seen that the equations offered in this paper appear to be valid.

Another important and interesting observation is that Brenner also found that the bond does not break at the interface, but in the substrate, exactly as found by us for electroplated plastics.  Brenner's weight loss data show that for copper and brass, the value of t_f is 0.75-2.5 mil.  Most interestingly, this too is in the same order of magnitude as the value of t_f for plastics.  Brenner did not determine the value of t_f for copper-plated steel.  However, by using the earlier theoretical analysis, it is possible to calculate its value.  As noted in Table 10, for steel, t_f is 0.001 to 0.02 mil.  Even if we assume, based on Brenner's data for copper and brass, that the actual value of K is three times the calculated value, it is evident that the value of t_f for steel is much smaller than that for brass, copper or ABS plastics.

Based on limited evidence available, we conclude that out studies on the adhesion of electrodeposits to plastics apply also to the adhesion of electrodeposits to metals.  It would be interesting to see more work done on the latter class of systems to determine if this conclusion is indeed correct.

Summary and conclusions

1. It is concluded that the “rheological theory” as expounded by Bikerman^1,10 best explains the nature of the metal-to-plastic bond.  An important consequence of this theory is that the bond failure, for a proper bond, occurs not at the interface but in a boundary layer of the plastic surface.
2. It is concluded that the Jacquet test is a convenient and theoretically sound test for determining the merit of plastic-to-metal bonds, and two methods for carrying out this test are detailed.
3. The equations which underlie the Jacquet test have been developed, and show that this test does not measure adhesion, pure and simple, but a complex of factors, of which an important one is the thickness of the electrodeposited metal on the plastic.
4. On the basis of various assumptions, and estimated values of certain physical properties, it is concluded that the following relationship holds for copper-plated ABS plastic:

                                                                               (9)

where F is the Jacquet test reading in pounds for a 1 in. wide strip, K may be regarded as figure of merit for adhesion as evaluated by the Jacquet test, and t_m is the thickness of the electrodeposit in inches.  Thus, in a given study, a value of K of 1000 would indicate that a "perfect" bond had been obtained.

5. That the bond between plastic and metal breaks in a boundary layer in the plastic was proven experimentally, both by microscopic examination of the plastic surface after peeling away a copper strip (Fig. 3) and by actually isolating the detached plastic film thereby produced (Figs. 5 and 6).
6. The thickness of the detached plastic film is about 0.1 mil, and is quite uniform in structure.  The structure is affected by the chemical conditioning method used, and its thickness increases with increasing conditioning time.  It is not affected by thickness of copper electrodeposit, use of direct nickel plating, or the values of F or K.
7. The thickness of the total layer of plastic which is affected by chemical conditioning, and which apparently takes part in the plastic yielding of the Jacquet test is about 1 mil (Figs. 10 and 11).
8. A study of various processing factors on adhesion of metals to plastics showed general agreement with theoretical considerations, with the value of F increasing with increasing thickness of electrodeposit.  In general, K was highest for copper coatings of 1.25 mil or less, and lower for thicker electrodeposits.  It is therefore concluded that statements to the effect that thicker electrodeposits on ABS plastics are desirable in that they yield higher values of F in Jacquet tests are ambiguous and must be carefully interpreted.
9. Exposure of electroplated ABS to hot water for 20 minutes (after plating) essentially "pre-ages" the part, although some modest further aging occurs on standing.
10. The type of ABS used is of importance, some grades being more amenable to plating than others.
11. Conditioning treatments affect F values, even within the scope of commercially useful conditioning cycles.
12. The effect of the type of copper electroplating solution on F values has not been established.  When stresses were controlled so as to be equal to one another, both conventional acid sulfate and proprietary bright acid copper solutions gave comparable results.  On the other hand, other data were adduced, in which differences were noted between different copper plating solutions.
13. When composite copper electrodeposits are used, which have widely different K values individually, the value of K for the composite coating follows the value of K for the coating with the lower K value, even when over 20% of the composite coating consists of copper with the higher individual K value.
14. Plastic molding conditions have a major effect on the results of the Jacquet test.  Specifically, stresses induced in the surface of the plastic in the vicinity of the gate result in markedly lower F and K values.
15. It was concluded that the relationship between the tensile strength of the bond and the results obtained in the Jacquet test is: Tensile strength (psi) = 5.5 K
16. The Jacquet test was compared with various thermal tests, and it is concluded that the two types of tests measure different things, so that one test cannot substitute for the other.  In a general way, however, parts which have low Jacquet test readings are more likely to fail thermal tests, although there is no simple relationship between the tests.  Thermal tests seem to measure highly localized substrate surface phenomena.
17. When plastics are electroplated with 1-3 mil of copper, they pass thermal tests as long as the adhesion is 50% or more of the K value predicted and the peeling force in the Jacquet test is 5.5 lb/in. or more.
18. Thermal test results improve with increasing total plate thickness, even though Jacquet test results may be comparable.  This justifies the "encapsulation" approach when intrinsic adhesion is relatively poor.
19. Use of a substantial copper undercoat may be helpful before nickel plating in passing thermal tests.
20. The glacial acetic acid test was found useful in observing two sources of surface weakness: presence of locally stressed areas (especially in the vicinity of the gate) and the presence of an improperly annealed surface (as shown by cracking of the surface).  The former observations correlate well with thermal tests, the latter observations correlate with Jacquet test results, at least on a go-no go basis.  Thus, it is concluded that the acetic acid test can be used as a simple chemical test for observing weak boundary layers which will subsequently cause the formation of improper bonds between ABS plastics and metal.
21. The work in this paper was compared with work done many years ago by Brenner on the use of the Jacquet test on plated metals.  It is concluded that the studies presented in this paper apply also to the adhesion of electrodeposits to metals.

 

Footnotes:

*The term "syneresis" is used by Bikerman to refer to the tendency of solidifying thermoplastic materials to "throw out" onto the surface a potentially weak boundary layer consisting of relatively more liquid substances.

Theory of the Jacquet test