Incremental development of existing product, process or technology is very cost effective up to certain level. When it reaches its limit, each and every incremental improvement demands exponential growth in efforts and cost. What we need now is “break-through technology”, “manufacturing process” and “distribution”.
Why can't we undertake a research on the following:
1. Why not reduce the size of wafers from the current 450 mm to 1 mm?
2. Why not eliminate the mask and use portable laser writer for lithography?
3. Why not mass produce Semiconductor manufacturing equipments for small scale production of semiconductors?
4. Why not educate the semiconductor experts to start “Cottage Industry” of semiconductor production?
5. Why not $1million fabs than $ 3 billion fabs?
Photomask producers and their suppliers rarely, if ever, return the cost of capital to their investors, unless they have a monopoly (e.g KLAC in inspection tools, Hoya in blanks). The pattern generator market is very small and very demanding. Who will make the investment in R&D and manufacturing to produce a multibeam tool?
There are all sorts of constraints once you get past a single beam -- otherwise it would have been done by now. Perhaps the 10 beam number comes from an energy delivery limitation.
It already takes two passes of the beam to deliver enough energy to expose the resist. The passes are separated in time to prevent excess localized heating that distorts the substrate.
Assuming you don't want to have even more passes (which would reduce throughput) you have to use the same energy beams. More than 10 beams might apply so much energy to the mask to cause serious heating issues.
It's not clear me why "10 beams or less" should be the sweet spot for an ebeam mask writer. Certainly at some point, as with the multibeam wafer writers, you just cannot push patterning data to the tool fast enough. But do you reach that point at only 10 beams for a mask writer? Even with 10 beams you would still likely have write times of several hours for advanced masks.
This is a good step to increase the production and reduce the cost of the product. The multi beam system idea is good and the alignment of the beams ,correlation between the beams and controlling them will will make the R&D expensive
If the semiconductor manufacturers had taken a long view to pay their reticle suppliers a profit in line with their own profits, the reticle suppliers might have had enough money for timely investments in new technology and engineering.
Starving critical vendors has been a short lived purchasing strategy. Think of it like driving a truck carrying a full load of gas drums racing down a road. To increase mpg, the accountant casts off the heavy gas drums to lighten the load so they ACHIEVE more miles per gallon. They certainly will until they run out of gas much sooner than expected and with a look of surprise. Common sense was thrown off the truck with the drums. When customers find themselves outrunning their vendors, there is a root cause.
Blog Doing Math in FPGAs Tom Burke 24 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...