By: tarlinian (tarlinian.delete@this.gmail.com), June 24, 2022 3:38 pm
Room: Moderated Discussions
Mark Roulo (nothanks.delete@this.xxx.com) on June 24, 2022 2:29 pm wrote:
> Doug S (foo.delete@this.bar.bar) on June 24, 2022 1:31 pm wrote:
> > Why are wafers round? Since we make chips that are square or rectangular, making
> > wafers square would minimize or totally avoid wasted silicon on the edges.
> >
> > I mean, I know that the first order is: they are round because all
> > the equipment is designed to handle round wafers not square ones.
> >
> > But was any thought ever given to switching to square wafers when iterating from 100mm to 150mm to
> > 200mm and 300mm wafers? Or is there some obvious reason why they must be round that I'm missing?
>
> Wafers are cut from grown silicon ingots.
>
> The silicon ingots are round.
>
>
In addition to the ingot issue, there are many wafer processes that rely on the rotational symmetry of the substrate. For example, photoresist is spin coated, CMP works by spinning the wafer against a rotating pad. Generally speaking "corners" are bad for many processes. While I'm not expert, I imagine plasma processes behave badly near corners for similar reasons. (In fact on 300 mm wafers, the area near the notch generally is relatively badly behaved, compared to other areas due to types of issues, and there were plans to do away with the notch on 450 mm wafers for similar reasons.)
> Doug S (foo.delete@this.bar.bar) on June 24, 2022 1:31 pm wrote:
> > Why are wafers round? Since we make chips that are square or rectangular, making
> > wafers square would minimize or totally avoid wasted silicon on the edges.
> >
> > I mean, I know that the first order is: they are round because all
> > the equipment is designed to handle round wafers not square ones.
> >
> > But was any thought ever given to switching to square wafers when iterating from 100mm to 150mm to
> > 200mm and 300mm wafers? Or is there some obvious reason why they must be round that I'm missing?
>
> Wafers are cut from grown silicon ingots.
>
> The silicon ingots are round.
>
>
In addition to the ingot issue, there are many wafer processes that rely on the rotational symmetry of the substrate. For example, photoresist is spin coated, CMP works by spinning the wafer against a rotating pad. Generally speaking "corners" are bad for many processes. While I'm not expert, I imagine plasma processes behave badly near corners for similar reasons. (In fact on 300 mm wafers, the area near the notch generally is relatively badly behaved, compared to other areas due to types of issues, and there were plans to do away with the notch on 450 mm wafers for similar reasons.)
