Death to False Metal: Why Aluminum isn’t for Everything

It’s happened again: someone has asked me to weld something out of aluminum that would probably be better made out of something else. Probably stainless.

Aluminum is a great material. It machines well. Its corrosion resistance, even in not-terrifically-corrosion-resistant alloys like 6061, is plenty good enough for many applications. Did I mention it machines well? It’s really a machinist’s dream: soft but not gummy, brittle enough to form chips nicely, and with tons of thermal conductivity to draw heat away from the tool.

The problems usually arise when someone switches from designing aluminum machined parts to designing aluminum welded parts, which is a different animal entirely.

1. The welding itself.

First off, aluminum welding is more challenging than welding many materials, including stainless and mild steel, for a few reasons:

  • You need bigger, more pricey welding equipment. Although its melting temperature (~1100° F) is actually lower than that of steel (~2600° F), aluminum’s thermal conductivity is roughly double that of mild steel, so you really have to pour heat into it to get it to melt, which means bigger, more complicated welders.
    • Aluminum TIG requires AC, not the DC used for almost all other TIG applications. All TIG machines have DC, but not all have AC. In addition, many AC-capable TIG machines have AC-specific adjustments: AC balance, AC frequency, and/or AC waveshape. This adds complexity.
    • Sometimes this equipment is so big that it needs expensive electrical installations. A Miller Syncrowave 250, for example, is a common TIG machine that works well for aluminum. But it requires 93 amps of 230V single-phase power in order to run at its rated output. Have you priced that kind of electrical work lately?
    • If you don’t have a welder with enough juice, you can pre-heat the aluminum, e.g. with an oxy-fuel torch, but this complicates things further, including cleaning (see below).
    • MIG is of course faster than TIG, but aluminum MIG is especially pricey, as aluminum has low stiffness and therefore poor feedability. A spoolgun is the cheapest way to get started (~$800 and up) but they’re finicky little things. A push-pull setup is far better, but the gun alone can easily cost $2,000. In contrast, steel wire will feed well in a sub-$200 MIG gun.
  • The welding technique is harder. Aside from some expert aluminum welders, most folks find aluminum pretty challenging, especially at first. I’ve always taught it after teaching mild steel and stainless.
    • With TIG, the filler metal addition rate is significantly higher than with steels, so filler metal must be fed more quickly. 6061 is prone to cracking unless the foot pedal is tapered off to minimize crater size. Aluminum TIG requires far more pedal manipulation than is required for other metals.
    • With MIG, there are a variety of possible issues, including burnbacks, birdnesting, and porosity (which in my experience is more common with aluminum MIG than with aluminum TIG, and is overall vastly more common with aluminum than with steel or stainless).
  • Cleanliness is everything. Like stainless, aluminum is highly reactive, so much so that what we think of as aluminum is really a block of aluminum with a layer of aluminum oxide covering it. The oxide layer is usually thin, but aluminum oxide is a vastly different material: it’s roughly 3x harder (Source) and its melting point is 3.5x higher (Source and source). Its removal is critical to success in aluminum welding. This is generally best done with a hand-held stainless steel scratch brush. There’s a bit of technique to the brushing, which is best learned in person from someone who knows how, and the brushes used ought to be dedicated to this specific purpose to prevent cross-contamination.
  • Aluminum is its own universe. Many of the work practices, techniques, and consumables that work well for other metals don’t work for aluminum.
    • Aluminum behaves differently than steels as it heats up. There’s no orange or red glow to indicate temperature.
    • It can be terribly difficult to assess the alloy content (and therefore properties) of an unknown aluminum. With steels, the main alloying element is carbon. Carbon content is a huge factor affecting weldability of steels, and conveniently, carbon content can be assessed easily, e.g. by spark test or hardenability test. No such simple tests exist for aluminum alloys, and yet aluminum alloys vary more widely than steels in both alloying contents and properties than steels do, including huge variations in weldability.
    • Selecting filler metals for aluminum can be more complex than with steels. There are dozens of aluminum filler metals available. The two most common ones (ER4043 and ER5356) behave quite differently, so a well-equipped shop needs both, but users must be careful not to mix them up, as they’re essentially identical.
    • In contrast, switching from, say, mild steel to stainless steel, requires only a new filler metal and some small changes in machine settings and/or  torch technique. Someone with 20 hours of mild steel TIG experience can pick up stainless TIG in only a couple hours, but this hasn’t been true of people making a switch to aluminum in my experience.

2. Strength-weight ratio

At this point, you may be thinking, “Sure, it’s a bit of trouble to weld, but aluminum has such a fantastic strength-weight ratio that it MUST be worth all this trouble!” And you’re right…except you’re also wrong.

Let’s consider 6061-T6, as it’s by far the most common flavor of aluminum. Typically 6061 is supplied in the -T6 condition, as 6061-T6. The “6061” part just describes the alloying contents: 97% Al, 1% Mg, 0.6% Si, 0.3% Cu, etc. But 6061 is a heat-treatable alloy, so its properties vary dramatically depending on the heat treatment it’s had. That little “-T6” is a temper designation; it tells us what heat treatment has been done. In the case of -T6, it means the 6061 aluminum has been through a series of heat treatment operations, including heating to ~980° F  for an hour, rapidly cooling in water, then heating again, for anywhere from one hour at 400° F to eight hours at 325° F.

Why does this matter? Because it means that even though 6061-T6 is super strong stuff, as soon as you weld it, all that fancy heat treatment goes out the window, along with almost half the material’s tensile strength:

  • Typical 6061-T6 ultimate tensile strength (UTS): 45 ksi
  • Typical 6061 ultimate tensile strength as-welded: 27 ksi
  • Minimum expected 6061 ultimate tensile strength as-welded: 24 ksi (Source)

Oops, you know that super strong 6061-T6 you just welded on? It just got 40-47% weaker. To be fair, this weakening happens on a gradient, with the weakest part being nearest the weld (see Figure 2 here for more). In fairness, if we weld together two sheets of 6061-T6, there’s typically at least some part of each sheet still at the pre-welded strength of 45 ksi. But weakness near welds is really unfortunate, because welds and stress concentrations tend to be co-located.

If the goal was to have a material that’s strong and light (i.e. high strength-weight ratio), then this post-weld weakening business is bad news, because your aluminum weldment hasn’t gotten lighter, but it has gotten weaker:

  • UTS/density, 6061-T6: 462
  • UTS/density, 6061-T6, typical as-welded: 277
  • UTS/density, 6061-T6, minimum expected as-welded: 246
  • UTS/density, A36 mild steel: 204
  • UTS/density, 304 stainless: 253
  • UTS/density, 4130 “CrMo” steel: 342

(These are unitless ratios; higher numbers indicate a material that’s stronger per unit of density. Note that neither A36, 304, nor 4130 lose significant strength when welded properly.)

The main takeways:

  • 4130 has a strength-weight ratio 23% higher than as-welded 6061-T6 typically does. The major downside of 4130 is corrosion resistance, and while paint is a great solution to that problem, it adds weight and doesn’t stick around forever.
  • Even 304 stainless (not known for its super strength) may have a slightly higher strength-weight ratio than as-welded 6061-T6. And 304 is both easier to weld and more corrosion-resistant than 6061-T6. Its main downside is its relatively poor machinability.

3. Added uncertainty

Worse yet, the strength loss in as-welded 6061-T6 is variable depending upon a host of factors. As this aluminum welding guide puts it:

The as-welded mechanical properties of the 6xxx base metals are very sensitive to welding variables such as heat input and joint design

I’ll go a little further. As-welded, weldments rendered in 6061-T6 can vary in strength (and therefore strength-weight ratio) by up to 88%, depending on factors including:

  • location relative to the weld (farther from the weld being stronger)
  • heat input
  • joint design
  • filler metal selection
  • welding process

Welding 6061-T6 introduces undesirable uncertainty about the strength of the base material, which is stacked on top of other uncertainties, e.g. about weld quality.

4. Some caveats

A couple notes for my more technical readers (all both of you).

It’s possible, though typically not at a prototype scale, to post-weld heat treat (PWHT) a 6061-T6 weldment and thereby restore it to roughly the -T6 condition, i.e. with roughly 45 ksi UTS. I doubt this is within the realm of possibility for my readers. It requires an oven that goes up to about 1000 F, can maintain a consistent temperature for hours, and is large enough to fit your weldment inside it. Also, you’ll need to be able to move the nearly-1000-degree weldment out of this oven to dunk it in water. Also, your weldment will distort as a result of all this heating and cooling.

There are aluminum alloys besides 6061, and some of them lose significantly less strength after welding. The 5000 series in particular is attractive for welding applications, for a variety of reasons cogently elucidated in the Maxal aluminum welding guide I’ve referenced repeatedly. Sourcing 5000 series alloys can be a little tricky, but they are generally better for welding applications than 6061. Many of the 5000s bend nicely, too, whereas 6061-T6 often just cracks when bent (try it, it’s fun). Be sure to choose the right filler metal; many 5000-series alloys are best welded with ER5000-series filler metals, as opposed to the ER4043 (and, recently, ER4943) often used when welding 6061.

Similarly, there are aluminum alloys for which the PWHT process is vastly simpler. Alloy 7005, for example, can be post-weld aged at temperatures as low as 200° F, restoring much of its pre-weld strength. But you probably won’t find some kicking around the shop–outside of industries like bicycle frames, 7005 is rare bird.

5. Aluminum isn’t all bad

I’m not saying you shouldn’t use the stuff. Just don’t assume it’s the perfect material for all applications because you’ve used it for machined parts. Know that when you switch from machining it to welding it, you’re adding some significant risk and complexity.

And material selection is always a matter of compromise. 304 stainless, for example, welds easily, looks cool, and has great corrosion resistance…but it’s a pain to machine, it’s a little heavy for its strength, and it’s expensive. 4130 would be considered downright magical were it not for a couple limitations: it needs paint in most applications, and it’s not nearly as machinable as 6061-T6.

Until a perfect material is developed, we’ll have to settle with an assortment of less-then-perfect options…including aluminum.