Skip to content

Taiwan Semiconductor Manufacturing Corporation (TSMC) stole my patented invention to make TSMC "Professor" Chenming Hu of UC Berkeley look good !

See, Chenming Hu of UC Berkeley was ashamed that I solved a problem which he and his entire BSIM3 group could not solve even with all the money that TSMC gave him, so TSMC just copied my invention !! It is a dirty thing that an honourable company would not do, but TSMC is not an honourable company.

The figure above was in a paper to the ACM which I sent to Eva Tardos of Cornell University in 2006, and it was exactly the same as my patent #7,623,995 read my patent on google, read my patent at the USPTO. Eva Tardos replied that the ACM had no interest. So, then I put the ACM paper submission on my website and it was there for years. And then TSMC submitted the exact same thing to the USPTO pretending they “invented” my correction equation method ! read #8,370,774 on google, read on the USPTO. Compare my figure above to the TSMC patent Fig. 2 where they simply renamed my “correction equation engine” as a “REP mapping”. See, Eva Tardos said my paper was not “truly outstanding” but TSMC thought it was outstanding enough to copy !

I made a working solver that made a better fit to the BSIM3 compact model and invented my new “Levenberg-Marquardt-Yechuri Algorithm” with my own two hands. But I did not make a profit, I made a loss. Taiwan Semiconductor Manufacturing Corporation (TSMC) / Chenming Hu of University of California at Berkeley profited from my research ! The Chinese company won and I lost ! This webpage is to tell all people around the world how the American Dream died for me, so you can learn from me. Don’t let this happen to you !

Here's what happened

In 1995 I graduated with a PhD in Electrical Engineering from the University of Texas at Arlington.  My speciality was numerical analysis.  I invented an algorithm to obtain flat-band AlGaAs heterojunctions across a Bragg reflector in a Light Emitting Diode/VCSEL manufactured by Honeywell. With Honeywell, I published a paper in the 1996 IEEE Transactions on Electron Devices. I have a special genius for solving even the most complex problems using numerical analysis.  The flat-band was never done before until I invented it !  The closest anyone came was an MBE sheet dipole by Bell Labs which I referenced in the paper.

In 1996-1998 I worked for Meta-Software (later Avanti) which made HSpice. I was a device modeling engineer who made BSIM3 models. But we had to use a software which ran on top of HSpice and the fit was really bad. The market leader for fitting BSIM3 models was BSIMPro sold by Berkeley Technology Associates which was affiliated to Chenming Hu’s UC Berkeley BSIM3 group. I told the Chinese Avanti management that I am an expert on numerical analysis and I can research/develop a better BSIM3 solver than Chenming Hu and that I would like to lead the Device Modeling group and do the same. But the Chinese Avanti management did not want to compete against Chenming Hu who is Chinese and supported by the Chinese Taiwan Semiconductor Manufacturing Corporation.

In 1998 I quit Avanti and incorporated “Die Soft Corporation”. First I got the (publicly available) source code for BSIM3 model from UC Berkeley and then from July 1998 to April 1999 I solved the BSIM3 model and developed my solver. But I made the mistake of choosing the Sun Sparc/Solaris platform to do my research and this was too expensive, this was a critical mistake, I should have developed on the Windows platform. I developed a solver but it was not perfect and before I could actually sell anything, I ran out of money !!

From mid-1999 to early 2000 I worked for IBM in Burlington, VT as Device Modeling Advisory Engineer. When I got the job at IBM, I was so deeply in debt that I almost could not get to IBM in Vermont, but my mother lent me a few hundred dollars and so I got to IBM. At IBM, I actually used BSIMPro by Berkeley Technology Associates and I realized that the sub-threshold fit was really bad. Until this time I had just intended to compete with Berkeley Technology Associates. But after seeing the really bad sub-threshold fit, I realized that my software was significantly better in the sub-threshold. But IBM wanted me to give them all the secrets I had learned in my research between July 1998 to April 1999. I offered to sell my technology to IBM for a price, but IBM was not interested. See, in return for a job IBM wanted all my research for free, which is not fair, and they really put pressure on me.

By April 2000 I had left IBM and I went back to Arlington, TX to re-start my research This was the time that everything I touched worked. I took the source code for the Levenberg-Marquardt Algorithm from the book “Numerical Recipes in C” and modified it to create for the first time my Levenberg-Marquardt-Yechuri Algorithm. The Levenberg-Marquardt Algorithm is implemented using matrix computations which is standard. Every line of research I had done from July 1998 to April 1999 fit perfectly with my modified Levenberg-Marquardt Algorithm. I now had a working BSIM3 solver which worked magically better than BSIMPro !!  Then I tried to sell my solver. But by this time no company would even return my calls and without anyone trying my software, I could not sell anything.

Then I turned to the US government, I applied for patent protection to the USPTO in 2005. I hope God will bless the USPTO Examiner who examined my patent #7,623,995 read my patent on google, read my patent at the USPTO “Method of extracting a semiconductor device compact model”. This was the first professional person I dealt with in years who did not play politics and actually examined my application on it’s merits. He issued me a patent in 2009.

Then, I submitted a journal paper to the IEEE in 2005, and to the ACM in 2006, and I put all my sourcecode on the internet on my webpage. I am a genius and a member of Mensa and I can solve even the most complex engineering problems using numerical methods, but my business acumen is poor. I knew nothing about business and nothing about protecting my algorithm. I thought since I had a patent submission, I was safe, so I tried to license my technology to large semiconductor companies. I put my source code on my webpage , thinking this would convince them that my solver was worth licensing. Both the IEEE paper submission and the ACM paper submission were also on my website for years. It never occurred to me that any honourable company would steal my invention. And the IEEE refused to publish my paper in order to help their buddy Chenming Hu of UC Berkeley. Eva Tardos of Cornell University also replied saying ACM had no interest.

So according to the ACM and the IEEE my algorithm was not worth publishing, but TSMC thought it was worth stealing. Such theft is common, see this article and this article.
Then, Taiwan Semiconductor Manufacturing Corporation applied for duplicate of my patented algorithm in 2010. Taiwan Semiconductor Manufacturing Corporation applied for patent #8,370,774 read on google, read on the USPTO which is 100% EXACTLY what I had patented in my patent #7,623,995 read my patent on google, read my patent at the USPTO. Taiwan Semiconductor Manufacturing Corporation COPIED my patent. Taiwan Semiconductor Manufacturing Corporation is a dishonourable Chinese company which is willing to steal. And Taiwan Semiconductor Manufacturing Corporation had an accomplice in the USPTO, a corrupt USPTO Primary Examiner, who is willing to issue a patent falsely to help Taiwan Semiconductor Manufacturing Corporation. This corrupt Primary Examiner is not alone. The corruption goes all the way up through the Supervisory Primary Examiner and the TQAS and the Director, so this Primary Examiner was empowered to steal from me to give to Taiwan Semiconductor Manufacturing Corporation. It is the policy of the USPTO to help companies like Taiwan Semiconductor Manufacturing Corporation and everybody knows it. The so called mapping that Taiwan Semiconductor Manufacturing Corporation describes is already in my patent Application. Temporary values are stored and then used in computations. Parameter extraction uses a series of extraction steps and earlier parameters and temporary values obtained are used in later steps, this is STANDARD and so is the incremental addition of measurement I-V curves data and the correction of the model parameter deck based on the new information. So is the use of the simulation model, that is the BSIM3 model source code that is freely available for 20 years and which is what is used to simulate the model. The model parameters are updated in this model. My code which was on my website has extensive use of matrix computation and this is standard in numerical analysis and the Levenberg-Marquardt Algorithm is implemented using matrix computations which is standard. Any thief company or “tough” lawyer representing them who claims that matrix multiplication is novel is telling a lie. There is NOTHING new in the Taiwan Semiconductor Manufacturing Corporation patent application, there is nothing novel and this Primary Examiner simply helped Taiwan Semiconductor Manufacturing Corporation steal my invention as his Supervisory Primary Examiner and the TQAS and the Director told him to do. This is the most dirty, dishonourable thing that only a company like Taiwan Semiconductor Manufacturing Corporation would ask of the USPTO . No other company in the world, not IBM or Infineon or Toshiba would steal from a pro se Applicant in this way. Taiwan Semiconductor Manufacturing Corporation / Chenming Hu are not the inventors of this solution method, they COPIED me and they stole millions of dollars from me in this way.

My invention, my Levenberg-Marquardt-Yechuri Algorithm which was stolen from me was almost CERTAINLY used to extract the BSIM3 models used to design EVERY chip in the WORLD for the last 20 years. So when large semiconductor companies tell you that they don’t steal, believe me they stole Millions of dollars from me by using my invention through theft.  But the joke is on now them. Large semiconductor companies legitimized theft from me for the last 20 years, but now the semiconductor companies in the USA are being wiped out of existence because China is becoming the sole producer of semiconductors for the whole world. You see these “white” companies though being honourable to an immigrant from India does not matter, and now China is getting ready to wipe the large American semiconductor companies off the map. RACISM hurts everybody, because what goes around, comes around ! Even the $250 Billion bill that the U.S. Senate passed recently will do more to help China and TSMC than it will to help any large American semiconductor company, because what these companies lack is morality or honesty and now they have met people even MORE immoral and dishonest than them  !!