---

SDN 0013.07 – Heat Treatment of 6061 Aluminum
Revision A

 

1. Introduction

The precise treatment to be used for the aluminum parts in GNIRS needs to be specified. The following discussion represents the results of library investigation coupled with past experience at NOAO. None of this makes me an expert on the subject, and anyone reading the notes below should clearly understand that.

I leaned heavily on Roger Paquin’s report and on the references cited therein. It is worth reading the references, since they put Roger’s recommendations in context. The current revision (A) also takes into account conversations with John Schureman at Alumatherm and Dan Reeves at Magparts.

Heat treatment of aluminum alloys is done for two reasons, which can conflict. The first is to strengthen the material, and the second is to reduce dimensional instability (usually due to stress). Note that stresses are also reduced by mechanical working, which is what distinguishes T651 from T6.

2. Requirements

The parts in the GNIRS cold structure will not be subjected to high stresses. Since the elastic modulus (which might be a concern) does not depend on heat treatment, the primary concern is therefore dimensional stability. Yield strength is a concern for screw holes (thread stripping), and tempered 6061 is also somewhat easier to machine; thus the processing should approximate a T6 condition.

2.1 Optics

A small number of optical components have been made or will be made with aluminum substrates. Even if the final surface is some sort of replica, any distortions in the underlying substrate will also be present in the optical surface.

The tolerances for these surfaces are specified in Ming Liang’s optical tolerance tables, but for the purpose of this discussion we take a value of 0.1 mm rms surface accuracy as representative. For the instrument’s beam size (~100 mm) this corresponds to allowable distortions of a part in 10⁶.

2.2 Structure

The structural requirements are much looser. Material deformations will typically occur over long periods of time or as a result of thermal cycling; as a result the requirements are not those set by short-term flexure (which are very stringent) but rather overall alignment and configuration repeatability.

The alignment requirements typically imply tilts of reflecting elements that are stable to the equivalent of ~10 pixels on the detector, or about 0.1 mrad. This implies dimensional stability of similar size (part in 10⁴), which is substantially less demanding than the requirements imposed on optical elements.

Just as a check, one can examine the alignment tolerances of the lenses in the cameras. If one allocates ~ 10 mm to dimensional problems, and considers a typical spacing of 30 cm (or less), the required stability is about 3 parts in 10⁵, somewhat tighter than the above but requiring an asymmetry in the camera barrel.

A separate concern is warping introduced by the stress-relief process. One solution is to do a final cut on critical surfaces after all heat-treatment is complete. It is worth looking at the extent to which this is worthwhile. For structural elements, the main concern is maintaining reference points or surfaces aligned. These are, to start with, limited by machining accuracy. Mounting points for a single mechanism can probably be machined to tolerances <0.001 inch over distances of 10 inches or so. If dimensional stability during final heat treatment is worse than a part in 10⁴, subsequent machining would be justified. If one is prepared to do a certain amount of shimming, the acceptable distortions would be several times larger. (This assumes that dowel pin holes and similar locating features are produced after the final treatment – these cannot be allowed to move around.)

2.3 Time and Money

Competing with all these concerns is the desire to stay on schedule and within budget. The ideal process might be one in which there are successive steps of treatment and machining, so that the material is only gradually transformed from raw alloy to finished, stable part. But if this results in the part arriving six months late, it is of little use.

3. Standard Processes

3.1 Tempering

Aluminum is strengthened by tempering; these processes are standardized and fairly wll characterized in terms of properties like yield strength. 6061 aluminum is usually treated to a T6 condition, which represents solution heat-treating followed by artificial aging. (Processes like T651 involve subsequent working of the plate; T7 implies over-aging.)

The tempering affects conductivity (electrical and thermal) slightly – about 10% – and has a pronounced effect on hardness, yield strength and the like (factors of 2-3). The modulus of elasticity is not affected. (Details on all this can be found in the ASM Metals Handbook, vol. 2 and 4).

The initial solution heat treatment is fairly critical, and for this reason everyone does it the same way (or almost). The standard specifications for this process can be found in Vol. 4 of the Metals Handbook, MIL Spec H-6088, or SAE AMS 2770.

The material is then quenched. The quenchant can be water (at different temperatures) or a water/polymer mix (e.g., polyalkylene glycol and water). (Air-cooling can be used as well; it is even slower, obviously.) The faster the quenching process, the stronger the material, but the greater the internal stresses. Roger Paquin recommends the PAG/water mix, although it appears from the ASM Handbook that hot water (80+ C) might work about as well. Since nothing in GNIRS is going to be highly stressed, this appears to be a reasonable recommendation. One is constrained here somewhat by vendor’s standard solutions – they typically are set up to work at certain concentrations, and a different percentage of PAG will require a special set-up at extra cost (and time delay). Roger Paquin recommends a PAG concentration of 28% (for thicknesses characteristic of our bench parts) whereas vendors appear to work around 20%. It does not appear that this is critical.

After quenching the material is artificially aged by maintaining it at a relatively elevated temperature (160-175 C) for a period of time (hours). The aging at higher temperature is faster and is also supposed to reduce residual stress. Over-aging reduces strength somewhat but is alleged to also reduce stress.

Another means of reducing residual stress is to leave extra material (1/8 inch or so) on the piece for the tempering, which can be machined off later. This is because the surface layers develop stress that is a strong function of depth as a result of the quenching. This also allows distortions in the material to be corrected.

3.2 Further Stress Reduction

Further stress reduction processes are described by the experts themselves as “black magic” and “cookbookish”. The basic principle is one of cycling the material between a moderately high temperature (but not one that will alter temper significantly) and as low a temperature as is practical (basically liquid nitrogen). The recipes vary as to whether the high temperature should be 100C, 150C, or 195C and whether the cycling should be fast or slow.

There is also disagreement as to whether one cycle is enough, or whether additional cycles are required.

Also, if one is in control of the tempering process, and not just working with existing T6 or T651 stock, it is argued that the stabilization process is best done prior to the aging, immediately following the quench (within 2 hours). This is because the aluminum has not yet hardened significantly, and is thus likely to be most responsive to the stress-relief process.

If the cycling is done after the tempering process is complete, NOAO has traditionally done 5 cycles. We have no direct experience with cycling prior to aging; Dan Reeves of Magparts recommends 4 cycles, though he says that the changes between cycle 3 and 4 tend to be marginal.

Dan Reeves also suggests aging after finish machining rather than before (the material is then more or less in a T4 condition for the machining); he concedes that the advantages of this sequences are probably modest if one has adequately thermal-cycled the part beforehand.

4. NOAO Experience

NOAO has made and tested some metal cryogenic optics. These are the Phoenix grating and the GNIRS Offner primary. In both cases, stock 6061 T651 was used without further tempering; the optic was manufactured except for a final finish cut, uphill quenched (at least 5 cycles between boiling water and liquid nitrogen) and then completed. A final optical surface (replica grating or diamond-turned Alumiplate) was then applied. Both of these have shown good performance (optical quality, stability) with thermal cycling; the Phoenix grating is the larger of the two, and has been cycled many times by now.

5. Conclusions and Issues

The process used by NOAO to produces cryogenic metal optics (5 cycles uphill quench) is simple and straightforward. This process should be used for all metal GNIRS optics. Note that the relatively small size of these pieces means that it is not necessary to go through a tempering step. Although Roger Paquin’s process may well be better, it is also more difficult (expensive) to implement; our experience suggests that it is not required to achieve the desired performance.

What is at issue, then, is whether the entirety of this process – plus tempering to T6 – is required for the large structural pieces in GNIRS, given that dimensional stability requirements are many times less critical. Also, in this case, we can do the thermal cycling prior to aging; is this advantageous or not? One should note that there is no mechanical stress relief – what is produced approximates T6, not T651.

Ideally, one would investigate this as a science project. This would provide definitive answers (one would hope) in a few months’ time. But faced with a schedule, it is necessary to be less scientific. Given the risks, it is also necessary to be conservative.

Faced with this, I propose the following:

5.1 Recipe

5.1.1 Parts from 12-inch Plate

This sequence applies only to parts manufactured from the large (12 inch) 6061 plate. (Note: it is a plate, not a forged billet.)

·        Part is to be rough machined until there is roughly 1/8 inch of material remaining.

·        Part is solution heat-treated according to the ASM Handbook or similar spec (530C for 65 min for first half-inch plus 30 minutes per additional half inch).

·        Part is quenched in PAG (ideally 28% concentration, otherwise highest concentration vendor routinely supports).

·        Part is thermally cycled 4 times, starting within 2 hours or less of preceding (quench) step. Cycle consists of the following:

§         Quench in liquid nitrogen for 30-60 minutes

§         “Uphill quench” in boiling water for 30-60 minutes

·        Part is artificially aged at 175C (not 160C) for 8 hours or so.

·        Part is finish machined

·        Part is cleaned and painted (Aeroglaze Z306)

5.1.2 Other Parts

This process applies to all other parts except small non-structural parts (e.g., clips, cable clamps). Such small parts should be finish machined and anodized (if required) without stress-relief. Note that the starting material will almost invariably be 6061-T651.

·        Part is machined to final dimensions +0.025 inch per surface.

·        Use NOAO standard uphill quench process (5 cycles). This is not the same as Roger Paquin’s process.

·        Finish machine part, except for critical interfaces that should not be anodized. In GNIRS, these are mechanical/thermal interfaces, where good thermal contact is needed. Locations of resistors, cooling straps and pre-cool attachment points are critical thermal interfaces in this context, and should not be anodized. If they can be masked off, it is not necessary to leave material to be machined away, since they are not precision mechanical interfaces. Parts that will not be anodized at all, either because they are painted or left as bare metal, can be completely finish machined.

·        Anodize (mechanism parts, not bench)

·        Finish machine critical interfaces to remove anodize

5.2 Issues

The tempering process is directed more at producing a very strong material and only secondarily at reducing stress. But unless one skips it altogether, it doesn’t seem worthwhile trying to generate a special recipe that will be less familiar to vendors. Some improvement in yield strength over the untempered alloy is needed.

An area where one can economize is the number of times the part is rough machined. The process described in 5.1.1 provides for no intermediate rough machining; the complete sequence of heat treatment is carried out and then finish machining takes place. This is clearly economical, and our expectation (hope) is that it will be effective.

Second, the whole issue of number of thermal cycles, maximum temperature, rates, times, etc. is very much a matter of taste. So far, I haven’t seen a procedure involving “eye of newt” but maybe I didn’t look hard enough. In the absence of a consensus, I am inclined to rely on our direct experience (for smaller parts and optics) and to work by analogy (and consider cost) in dealing with the bench parts.

Appendix A: Heat Treatment Specification.

This is a restatement of the process described in section 5.1

All parts shall be heat treated and stress relieved as followed:

1.        Part is to be solution heat-treated at 530°C (985°F) according to MIL Spec H-6088. The maximum thickness for each part will be supplied in order to permit the vendor to determine the correct soak time.

2.        Part is to be quenched in a room-temperature mixture of polyalkylene glycol (PAG) and water. The preferred concentration of PAG is 28±2%, but concentrations as low as 20±2% are acceptable. Vendor must notify NOAO of concentration used. Maximum quench delay will be according to MIL Spec H-6088.

3.        Part is to be stress-relieved by means of 4 cycles of cold stabilization followed by boiling-water quench. The stress-relief process must begin within 2 hours of the quench carried out in step 2. The stress-relief cycle consists of the following steps:

3a. Part is cold stabilized in a liquid nitrogen bath for 30 to 60 minutes.

3b. Part is uphill quenched in boiling water for 30 to 60 minutes.

NOAO Intranet Services

NOAO Copyright  Statement