Tuesday, May 15, 2007

WISDOM of PAOLO GARGINI (process technology)

I thought this series of quotes from Paolo, was very apt in guiding successful device technology development, or in fact any R&D effort which is complicated and substantive in technical challenge.

Oftentimes, people who ignore these words of wisdom, or sadder yet don't understand the nuggets within, find poor success in solving technology development challenges, and will often not comprehend the organizational root causes of project or team failure.

The best teams in development are often fluid and agile, taking this wise advice to heart.


1] The Right Things may Happen for the Wrong Reasons

2] Predictors of Engineering Limits have Always been Proven Wrong by the Right Improvements

3] It Would be Wrong to Believe that the Right Fundamental Limits Don’t Exist

4] It is Wise to Look For The Right Solutions before Things Start to go Wrong

5] It Would be Wrong to Delay Taking Action and not Doing the Right Thing at the Right Time

quote from PAOLO GARGINI
(re: process technology)
Director of Intel Technology (process) Development Strategy
Intel Fellow, and Chairman of ITRS - Intl Tech Roadmap Semiconductors

Mine own edits to this are -> Try your best, hang in there and think about assumptions that might be able to be circumvented, where presently assumed to be impenetrable barriers. This is the essence of 2], and while one cannot bank on unusual solutions magically appearing that destroy assumptions of limits, one can look passively, actively, like a crouching tiger, or a ravenous reader, or experiment prolificly and efficiently, to find the fix to one's ills, that ails one's quest.

Unspoken here is that true breakthroughs, development thereof, and recognizing them when they might have been staring you in the face for some time, is a bit of a soft art. Overmanage and tightly control dissent, and you tightly often squeeze your endeavor into a rather tight corner...

An example of this was when an older researcher was almost grudgingly accepted to find a corner to do some work in a lab of a rather prestigious research scientist, and he proved key to the concepts to enable laser based microcantilever deflection sensing so critical to practical AFM - atomic force microscopes. All from attempts to test laser rangefinding to the moon on the lab ceiling, we now see molecules and atoms like never before.

Apparently complementary Heterostructure FETS (or something similar) were stumbled upon because of an error in the intended MBE EPI epitaxy layer growth sequence.

The major single yield improvement for the worlds first DRAM, The Intel 1101 256bit pMOS DRAM, occurred when a bright, humble, yet confident engineer, Tom Rowe, in the early 1970s, in the face of a plant wide edict by Gordon Moore to prohibit ANY and ALL process experiments on the 1101 Dram process (then running a miserable ~5-10% yield ), was ignored in secret by Tom Rowe.

Tom then found a means to >5x increase yield of the worlds first DRAM using, as Les Vadasz called it, a SOOPER DIP, which was Tom Rowe's secret wet etch chemical recipe to clear the metal contact openings without destroying the CVD deposited Field Oxide... And yes that was a Sooper Dip apparently and it both saved the day and led to the commercial success of Intel's efforts in fielding the first MOS Dram memories. Tom is a hell of a guy, fun to talk with and a kind people person. And a great raconteur...

Overconstrain technical innovation and problem solving, overcontrol and we all lose. Work your tail off and get a little loose, and we all win.

Work hard and think like the devil.

Paolo is not a personal friend, but I worked at Intel Livermore '84-86 when he headed Intel Logic TD......quoted from Paolo's 2004 Semicon West presentation

Originally posted 10/16/05
bumped up to refresh on May 15th 2007, as apt as ever.

Sunday, May 06, 2007

Tim May's 1991 whitepaper on Parallel Nanoprobe Lithography

In the early 90's, in the course of participating in the Palo Alto Nanotechnology Working Group, Tim May wrote an interesting whitepaper hypothesizing what scanned nanoprobes might offer in advanced potentially inexpensive high rate lithography. It was a very interesting series of ideas that predated much of the more formal IP generation in parallel nanoprobes, esp. from universities.

Tim is a brilliant scientist, and in the early days of DRAMS at Intel, had discovered the alpha particle effect causing RAM memory bit loss was from trace radioactive materials in some ceramic IC packages. He was an Intel R&D scientist and often had avant garde innovative ideas. There is some story of some lead bricks under a famous Intel executive's house that were used in Tim's experiment to verify that cosmic rays were not at root cause of the memory errors.

For some interesting perspective on the future foretold by Tim, as yet unrealized, I am posting his paper - scanned, to the web, with his full permission and credit to him.

I thought and continue to think that Tim May is far more creative than often given credit for. Brilliant, sometimes cuttingly sarcastic like some fraction of great scientists are(esp if in contact with similar Hungarians??!!), and a very witty guy.

I even remember his efforts in describing how AI techniques might be used to benefit wafer fab process diagnosis and control, by rule based analysis of electrical test data and in-process wafer fab metrology results, far predating others efforts much later...

After I received this hardcopy, I later joined Digital Instruments, and a few years into my stay at DI, when Minne began talking to Virgil Elings of the Quate group's later research into parallel nanoprobes, I gave a copy of this to Virgil, who in his usual effervescent manner, was slightly bemused.

Here are scanned images (with watermarks) of Tim May's brilliant whitepaper on Parallel Nanoprobe Lithography from around 1991 or earlier. (undoubtedly Tim will correct me later with a more accurate date and other details).

This posting is a series of 14 clickable high resolution images, reduced in size on the web page for faster page loading, but each will render in high resolution if you click on the smaller images. Or right click on each, to directly download in windows the high res images.

This does beg several unspoken questions re the IP filings on parallel nanprobes later by several other groups who may have read of, or listened to Tim's pathfinding ideas in the very early 90's, at the Palo Alto Nanotechnology Working Group, or same through the net.

Here are the page images of Tim May's brilliant whitepaper on Parallel Nanoprobe Lithography.

And yes, I used to kibutz with him over his name for the "pizza crawler" as he wittily called it....We had a few yuk yuks over it..

Thursday, May 03, 2007

SmartTip 's NanoProbe Fountain Pen - a Dip Pen Nanolithography Tool

SmartTip is an unusually innovative spinoff of the Netherlands' University of Twente's division of Transducer Science and Technology (formerly called MicMec), which has done prodigious amounts of R&D in micromachining of all kinds of microsensors and actuators, including the more conventional mems structures like pressure sensors and accelerometers and other more novel devices.

I enjoyed reading dissertations from Twente's institute, as invariably the research was fascinating, and often novel, and always well characterized. I learned of Twente's efforts when I was at IC Sensors, and Hal Jerman and Steve Terry had a couple of the interesting dissertations around the offices, which made for fascinating reading. Plus on staff there was an engineer Dedrick DeBruin, who came from Twente if I am not mistaken...

So recently it seems after SmartTip's few earlier efforts at mildly novel nanoprobe developments for SPM / AFM (notably the Canticlever for magnetics MFM and related devices), SmartTip is fielding an inking type scanned nanoprobe, derived from a standard atomic force microscope cantilever. The firm is calling the devices Fountain Pens and is going into a beta type customer sampling to learn of the application subtleties before attempting to field a product on a larger scale.

While there is little reference to Dip Pen Nanolithography at SmartTip's page, close examination of the mechanism used, seems like this is what it may very well be doing (ie using DPN(tm) 's described ink transfer methods), although not with dipping to preload the tip for wicked type inking from a solid tip nib. It is a curious connection...

The Smart Tip [aka DPN(tm)] Fountain Pen devices have taken a more common low stress silicon nitride AFM cantilever and incorporated an inking well in the body of the otherwise solid pyrex glass chip, drilled 1-2 holes in the pyrex and incorporated some recesses in the glass using the otherwise normal anodic bond side glass etch, and then added a fluid channel in the cantilever arm(s), plus an inlet coupling the cantilever fluidic channel to the pyrex glass region's ink well, and a tiny etched outlet from the far end of the fluidic channel just near the back of the pyramidal molded silicon nitride AFM tip.

[ed - postscript MAY 24th 2007 - this picture linked above, hints at the potential for IP in useful systems integration of continuous feed microfluidics, beyond the mere static reservoir shown in the hole drilled in the glass existing chip's design.

A systems level integration for near continuous feed, would embody an external larger reservoir, linked to the chip level reservoir which might include tubing, a demountable compression seal at the top of the drilled glass substrate chip and various combinations of external / internal valving, sensors for flow and level / volume remaining, and even integrated electrochemical sensors.

This basic device platform also might permit /enable microvalve fluidic chemistry switching either remotely with longer cycle time / dead volume or more closely coupled fluidic switching closer to the cantilever or a multi microvalve manifold microfabricated on or near the cantilever, even best near the tip to reduce the dead volume of chemistry switching from multiple microvalves - Mark Wendman]

Interesting structure, but possibly not novel enough? Granted this seems to be the 1st commercially fielded AFM cantilever that incorporates and integral microfluidic channel(s) in the cantilever arm, but it still seems to be a device using Dip Pen Nanolithography methods, as DPN(tm) if I am not mistaken, is described more generally as what transpires at a solid Tip apex to wick a meniscus of fluid to the scanned surface just as an ink nib does conventionally.

Also DPN(tm) is not restricted to applications that require dipping per se to provide the ink supply, but I might be mistaken, even if the name DPN(tm) implies dipping, the actual technology might not stipulate this with restrictions of DPN(tm) to dipped supply of "inks".

In short, DPN(tm) likely includes cases in which the fluidic supply to the tip is integrated into the cantilever, and specifically with cases of cantilever incorporates a microfluidic channel.

There seems to be a novel configuration enabled by the integrated inkwell and fluidics channel - where on this web page at Twente is described that the MFP probe enables AFM-based electrochemical conductor deposition by connecting the metal electrode on top of the probe chip (anode) and the sample substrate where conductor deposition is intended (cathode) to a bias power supply. The electrical circuit is formed using the electrolyte solution in the cantilever microchannels. The deposition is voltage regulated since well performing currents for this method are very tiny.

Regrettably they as yet have not acheived electrodeposit linewidths much smaller than 250nm, which I suspect might be due to the substrate spreading resistance, or the gap between the hole at the end of the microfluidics channel being further away from the tip apex. It could also be possible that the electrical biases and writing speeds need further optimization?

I like the structures, even if the ideas might not be as novel with respect to technology as Twente's MicMec / SmartTip might otherwise hope for.

In the Beta Field testing, I'd guess that 3 critical aspects of the device's features are implicitly being evaluated - 1] the fluid channel of the cantilever(lever stiffness and fluidics flow for various sample fluid viscosities), 2] the diameter of the fluid channel opening, in proximity to the tip (just behind the tip ) and 3] the desired sharpness of the tip...

Additionally one might ask if the reflected spot quality from the lever backside AFM laser reflection is in any way affected by the fluidics channel. This largely will be determined by how the fluidics channel fabrication was integrated into the cantilever process ( ie how flat the resulting reflecting surface is so that laser spot quality is not excessively degraded).

Fascinating stuff ...

note - DPN(tm) is a trademark of NanoInk inc.
Dip Pen NanoLithography is also a trademark of NanoInk, but has been used in academic contexts before trademarking

Enter your email address:

Delivered by FeedBurner

Subscribe to Wendmans Views on Nanotech by Email