I think you misunderstood a diagram; grams of Al per square centimeter is a standardized unit of shielding in the literature, not an engineering choice.
That doesn't matter though. Since it's not representative of the real shielding, it's an inapplicable standard to use for this problem.
You should realize that there's no simple way to convert from aluminum to polyethylene: it's not as simple as "PE is X times better." Due to secondaries aluminum stops improving after 30 g/cm^2,[1] hence why their chart (bottom line) levels out. PE doesn't have that problem, and lower dose rates are achievable versus what the author states.
TL;DR regardless of if you call it a "standard" or not, the author is still using bad math to overstate the actual risk.
I didn't choose AL shielding as the reference standard to use in the literature; I'm sorry it bugs you.
You can read the author's other work, that goes into great detail about different types of shielding, if you want to gain confidence in his math. The upshot is you need many meters of polyethylene to effectively shield the heavy ion component of GCR, which is what the fuss is about.
But the point of that particular diagram is not shielding, but to illustrate the 2-3x uncertainty in estimates of tumor risk based on our poor understanding of high-Z ion exposure.
The same thing could easily have been communicated in a less misleading way. Conspicuously the error doesn't sandbag the author's primary thesis, but does the opposite. Choosing to include the graph without caveat mostly serves to obfuscate (and at worst deceive), not illustrate.
