1st off, I would caution against assuming GTAW is a low heat input process. It can be (especially with automation as you are using), but due to low travel speeds, the heat input can get very high indeed. However, low heat input does not generally help getting low hardness. As already pointed out, the higher cooling rates associated with low heat input contribute to high hardness. Also, HAZ widths are not really relavent to hardness testing. While the HAZ width is indicitive of the relative heat input (on the same type and thickness of material), there is not necessarily a correlation with hardness.