The machining of platinum alloys is considered a difficult and expensive process.  Factorial designed experiments were used to investigate the effect of machining variables for three different cutting materials; ceramic, cubic boron nitride and polycrystalline diamond. Surface texture was examined with both a scanning electron-microscope and contact instrument.

INTRODUCTION
The increasing popularity of platinum jewelry has placed an emphasis on improving the manufacturing processes. One of the important processes is the machining of platinum to shape the metal into its desired final form. An excellent starting point for the study of platinum machining is the publication by Rushforth. Some of his conclusions were as follows:

• Adhesion of platinum to the tool is the most likely source of excessive tool wear during the machining of platinum.
• A contributing factor is the abnormally high hardness values observed in chips produced by the platinum machining operations.
• The use of more positive tool geometry's has resulted in increased tool life while retaining surface finish.
• Improvements in the platinum gloss can be brought about by changing the configuration of the diamond tool and using specific lubricants.
• He suggested that the behavior of platinum at high strains and high strain rates could be a fruitful area for further research.

Recently, Astakhov has written a book that takes a system engineering approach to consider the metal cutting processes.

He arrived at the following definition:

"Metal cutting is a forming process of components that are so arranged that by their means, the applied external energy causes the purposeful fracture of the layer to be removed. This fracture occurs due to the combined stress including the continuously changing bending stress that is the cause of the cyclic nature of the process." The definition overcomes the shortcomings of orthogonal cutting (single shear plane) model which is generally referenced. The orthogonal cutting model requires that the stress gradient in the single shear plane must be infinite and the acceleration of the particles would also be infinite. Hence, the orthogonal theory is based on the chips being formed by simple shear rather than the combination of shear and bending stress as proposed by Astakhov. Initial deformation and outer boundaries can be added when bending stress is included. This permits consideration of chip formation and properties, chip separation and the behavior of the work piece material.

If the model is applied to different work piece materials, it predicts different chip formation. The chip form can vary from rectangular chip fragments to continuous fragmentary (with difficult to distinguish fragments to unstable fragments with variable thickness).

The characterization of the machined surface is an important aspect of the machining studies. The reference we used was the ASME B46.1-1995. We shall use the definitions of surface texture, roughness and waviness from this reference. Surface texture is the composite of certain deviations that are typical of a real surface. It includes roughness and waviness.  Roughness is the finer irregularities of the surface texture that usually result from the inherent action of the production process or material condition. Waviness is the more likely spaced component of surface texture. Waviness may be caused by such factors as machine or work piece deflections, vibration and chatter. Roughness may be considered as superimposed on a wavy surface.

The scanning-electron-microscope was chosen for the initial studies to characterize the surface texture. However, the characterization was not done quantitatively but qualitatively. The surface was ranked from the least rough and wavy to the most rough and wavy.

A Mitutoyo Surf Test 5J-301, Series 178, profiling contact skidded instrument was used to measure two roughness height parameters, Ra and Rq. Ra, the roughness average is defined as the arithmetic average of the absolute values of the profile height deviations recorded within the evaluation length and measured from the mean line. Rq, root mean square (RMS) roughness is defined as the root mean square average of the profile height deviations taken within the evaluation length and measured from the mean line. The result is a quantitative measurement of the surface height.

At this point the original article gives details on the experimental procedures conducted and the concluding results.

DISCUSSION
These observations were made during the dry machining. The dry machining produced the predictable worse results due to the fact that there was quite a bit of heat built up at the cutting edge. This along with lack of lubrication, caused the chips to stay at the cutting point, not only contributing to further heat buildup but marring the newly cut surface. The dull, smeared surface texture is a result of this phenomenon. 

Very thin cuts were not as effective because chips tended to stay at the cutting point of the tool, folding over themselves and building up along the O.D. of the tube before breaking away. This was especially true at low speeds and feeds. Except for one example (PCD effect (AB)) the interactions were negligible. This makes sense since most of the runs were so poor that it would be difficult to separate minor interactions between effects.

For the ceramic insert, no lubrication, the fact that all the effects for a, b, and c were negative, means that to get the best results we would want to keep all levels at the low end. This would seem to make sense since increasing speed feed and depth would rapidly increase heat buildup and the ceramic insert doesn't provide any heat conduction. In fact, samples cut during this run were all very hot, especially at the high levels.

The PCD insert showed that we would want the feed and the speed toward the high level as indicated by the positive response but the depth of cut kept at the low level, once again, probably relating to heat buildup. Since the PCD insert could wick away some heat we were okay as long as we kept the cuts very light. The PCBN showed kind of the opposite response with a negative effect on the feed effect and positive on the spindle effect and depth of cut. It is not evident why this would be so different from the PCD.

Using the ceramic with the cold air gun to keep the tool and work piece cool, switched the effect of the spindle speed from a large negative effect to a small positive effect. Interestingly, it slightly increased the negative effects of a and c by the same amount. The cooling allowed us to increase spindle speed as long as we didn't overdo it by increasing feed and depth of cut.

Once the use of lubrication was begun, the effect from different types of lubrication was so negligible that it was zero. Interaction effects were also driven to zero. This shows that really what counts is feed and spindle speed. With the PCD insert the influences show that we would want to keep the spindle speed down and feed speed high at a cut depth of .010". The ceramic insert showed the exact same effects.

We then looked at cut depth again as a c factor. With the ceramic insert, we once again would want to keep the spindle speed on the low side of the range (800) and the feed speed near the high-end (45).  Cut depth does show a positive influence meaning we would want to stay near the .012" depth.

The next discussion will center on the final experiments. The results show some interesting trends. In general, the cermets produced rougher surfaces than the PCD inserts, whether the material was standard Pt-4.8%Ru or the machineable platinum. The machineable platinum, however, performed better than the standard Pt-4.8%Ru. Whether PCD or Cermet was used. The best results were using PCD with the machineable Platinum alloy. The worst results were using Cermet tooling with the standard Pt-4.8% Ru alloy.

There are some interesting trends to make note of. On Set #1, which was the Pt-4.8%Ru with PCD, we see a small positive influence from feed speed and a slightly larger negative influence from spindle speed. If one were to rerun this set, raising the feed speed slightly and lowering the spindle speed might yield better results.  For Set #2, which was the Pt-4.8% Ru with Cermet, we see a very large influence from feed speed, and a small positive influence from spindle speed. If one were to rerun this set, one would raise the feed considerably while leaving the spindle speed unchanged.

For Set #3, which was the machineable Pt with PCD tooling, there was negligible positive influence from feed speed and a slight negative influence from spindle speed. This set was to be rerun, the feed speed should be left alone which the spindle speed decreased slightly.  For Set #4, which was the heat treatable platinum with Cermet tooling, there was a decent positive influence from the feed speed and a very large positive influence from the spindle speed.
It is interesting to note that the strength of influences are reversed between the Pt-4.8%Ru with Cermet and the machineable Pt with Cermet.

Also interesting is that the influences between the Pt-4.8%Ru with PCD and the machineable Pt with PCD are almost the same, except lower with the machineable Pt. This could show us that we are at optimum feed speeds and spindle speeds with the PCD tooling and that the levels for those factors would be very different if we were to use Cermet tooling.

This shows very nicely that changing the type of tooling would almost certainly mean a change in process parameters.

SUMMARY
Since cutting or machining is important in jewelry operations, a study was undertaken to demonstrate the methodology to use in determining proper conditions. The experiments demonstrated that cutting materials other than diamond can give acceptable surface textures. Further studies are required to compare the life of the cutting materials and to evaluate the economics.

V7N6

Machining of Platinum Alloys for Jewelry
Costantino Volpe
Tiffany & Co.
Richard D. Lanam
Engelhard-CLAL LP

This is an abbreviated version of the original work. For full technical details, please consult the original paper.