Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!cs.utexas.edu!usc!apple!fox!portal!cup.portal.com!mmm
From: mmm@cup.portal.com (Mark Robert Thorson)
Newsgroups: comp.arch
Subject: Re: GaAs considered dead , indium phosphide, rod logic?
Message-ID: <24952@cup.portal.com>
Date: 13 Dec 89 03:49:26 GMT
References: <7269@hubcap.clemson.edu> <1989Dec4.175854.456@mentor.com>
  <1989Dec6.014149.26600@ux1.cso.uiuc.edu> <733@ra.cs.Virginia.EDU>
Organization: The Portal System (TM)
Lines: 25

mac@ra.cs.Virginia.EDU (M. Alex Colvin) says:

> nanomachines (or anything else) built with scanning microscopes can't be
> very complex, or they'll take to long to build. the nice thing about
> photolithography is that it's relatively independent of the nummber of
> features.

On the contrary, they are the same.  An STM-fabricated device is built
piece-by-piece from atoms and small molecules.  An IC mask is made by
a machine called a "pattern generator", which has a precision slit
controlled by motors.  Each line and rectangle is made individually.

When it comes to production, photolithography is at an enormous disadvantage.
Every single IC must be exposed to a copy of the mask (actually several
masks, a dozen or more).  An STM-fabricated self-reproducing nanomachine
would only need to be manufactured once, after which each unit could be
used to make others, much in the manner of a bacteria or algae.  The problem
is keeping it under control.

***

With regard to GaAs, I heard that the oxide is conductive.  Think about
that for a moment, and you'll understand why it could be so much harder
to make than silicon (in which oxygen and water vapor can be used to
grow excellent insulating material).