OK, a little to the south, but still pretty close to Silicon Tundra

Pissing away a million dollars

http://www.electronicsweekly.com/blogs/mannerisms/yarns/pissing-away-a-million-dollars-2015-12/

Serendipity and HCL

This contribution from Ian McWalter of CMC

In the 1970s there was a company in Ottawa called Microsystems International, essentially a research wing for Northern Telecom (Nortel for anyone under 30). They had great technical successes but were eventually shut down... this closure and the severance pay that came with it led to a tech boom in Ottawa a decade later and still today.

Semiconductors were plagued by contaminating mobile ions like sodium (Na+) which were especially problematic in the growth of transistor gate oxide. To try to reduce contamination HydroChloric gas.(HCL) was used to flush the quartz tubes in which the gate growth (oxidation) was carried out. One night a technician forgot to turn off the HCl source during a gate oxide growth and the result was spectacularly good transistor performance! Within a short time every foundry used HCl during gate oxidation,

The paper is here http://jes.ecsdl.org/content/119/3/388.abstract

and the patent is here http://www.google.com/patents/US3692571

And more challenges

Short delay while two of my old start-ups get exited and I launch two new ones, sorry

In the early 80`s we had replaced our barrel plasma etchers with more modern Tegal planar etchers. All of our production was converted over and soon after yields dropped drastically (especially on the 8870 DTMF tone decoder), once again everyone went into frantic overtime, weeks and months went by and our entire work-in-process was in peril. After millions of dollars in losses Stuart Boyd was able to measure definite short circuits between polysilicon lines on the wafers. He determined that there were fine 'stringers' left after etch along the edges of any 'steps' in the underlying topography. Polysilicon lines which went over a step were thicker in the vertical direction and since the reactive species in the planar etchers only travelled vertically and not anisotropically they never got completely removed.

A few months later we invited Alan Reinberg, a senior scientist at Perkin Elmer, to come and speak to us. He gave a fascinating talk on plasma etch in general and then came to he subject of the challenges of planar etchers. When he said 'obviously you have to adjust for some anisotropy in the etch to eliminate poly stringers' we all got very interested in our shoes.... if only we had invited him a few months earlier.

Bennett's rock... and Colin Harris (guess which is which)

Colin Harris came to Bromont this week so we had a quick tour around. This is his last week at PMC-Sierra, he is retiring and will be working with a number of start-ups in and around Vancouver.

Colin reminded me of a story from the early days...

One of the HR Directors from Kanata came down for a celebration we were having and, coincidentally, he was thrown in a swimming pool (an astonishing number of bars had adjoining pools at the time). He made the mistake of taking off his wet clothes and piling them on a chair. Needless to say, his clothes disappeared instantly and he never found them. He drove back to Kanata at 3 in the morning, arrived at 6 am and had to run into the house in his wet underwear. Try explaining that to a loving wife.

Just for closure - the clothes were on the floor in the back seat of his car..

This from Peter Gillingham

I too remember the Calma. I joined Mitel in 1983, a little bit late for those helicopter trips from Kanata to Bromont. I remember long nights in the Kanata Calma tents putting together analog sections of the BLIC (later renamed DNIC) and the BPhone. I think there may have been a third chip for the Bullet system but can't remember what it was. We'll have to track down Ken Buttle or Pat Beirne to find out. Also remember fun days in the lab with massive coils of twisted pair tracking down transmission dropouts in the DNIC and SNIC. I think I achieved the record for the tallest stack of equipment actually used in a test. The one piece of equipment I could not use was a weird box with toggle switches and screwdriver adjusted potentiometers that Gregg Aasen purchased to test the DDX. It was only later after I joined MOSAID that I understood what the SRT-1 memory tester did.

The invention of the STM

The scanning tunneling microscope (STM) was invented in 1981 and resulted in a Nobel Prize for Gerd Binnig and Heinrich Rohrer of IBM in 1986, That must be some sort of record, usually Nobel prize winners have to survive many decades to be recognized. It uses vacuum tunneling of electrons from a sharp metal tip to a surface and can obtain precision to the level of single atoms When they first submitted their paper it was refused as being preposterous, they had glued together pieces of thick cardboard which they had cut out after tracing single scans on them and then taken a Polaroid picture of the resulting 3D surface, 

I was at the Ecole Polytechnique de Montréal in 1988 and they had a guest speaker from one of the IBM research labs involved in STM development. He showed astounding video of real-time scans of cleaved graphite surfaces, identified lattice defects, and even demonstrated perturbations caused by single gas atoms when the vacuum was broken. At the end of his talk he said 'I have brought a few microscopes along to show you' and picked up a shoe box that had left on the counter! He pulled out little blocks of teflon with stacked piezoelectric devices glued to them and a metal probe tip with wires hanging off. I had been envisioning room-sized ultra-hightech equipment... How many more simple but powerful inventions are out there to discover?

This just in from Jacques Laneuville

Hello Gord,

Got the link from LinkedIn Pulse.

I remember some of those stories. I remember the Calma (and it’s temperamental pen). I remember the entire plant being encircled by trailers (including Bennett’s rock) that were offices, conference rooms, library, etc... I remember a party at Melodie Lee’s where everyone wound up in the pool (helped by you in some cases).

Got some information from Luc Carrier about the missing names in your blog picture:

“A cote de Francois Tanguay c’est Paul Aubin, a la droite de Gord c’est Tino Varelas je crois et a la droite de Gord il ressemble a Pierre Codere mais je ne me rappele pas qu’il ait travaille a Bromont. »

For the purchase of the Bromont fab by Mitel, it was 1976.

Have a good day,

Jacques Laneuville

Ed. Note - These edits are now made. Bennett's rock is still there on the lawn in front of the old part of the building (I will take a picture if nobody has one), Ralph Bennett was responsible for the semiconductor division in its early days and when he heard how much it would cost to blast it and remove it he said 'leave it there' - an early example of lean management.

The trailers formed a maze both in the parking lot and in front of the building. I remember a group of engineers getting together in the far end of one series of trailers where they had set up a monitor for a micro-probing station with a TV tuner so we could watch one of the early shuttle takeoffs.

Concerning Melodie Lee's pool party - no comment... 

A challenge now and again....

We had a few tough periods... at one point we were having field failures from a variety of components and could not figure it out. They didn't fail if they were mounted in ceramic packages (expensive) but seemed to stop working after about two years in the plastic packages we wanted to use (cheap). We mounted a campaign called 'ISO in plastic' and studied everything we could think of: conformal die coatings, new package materials, new packaging techniques, sources of contamination in our fabrication line (even to the extent of firing off the halon fire extinguishers to see if they were leaking contamination). 

It was an incredible time in the industry, we were all just geeks working on cool problems and we instantly shared information if it would help someone else. The head of Process R&D, Jack Morris, asked a friend about this problem and he found out that we had been increasing the amount of phosporous in our passivation glass (PSG) to make it more resistant to cracking. At about 7% phosphorous we had reached the point where moisture coming through the package could leach it out and make corrosive phosphoric acid which was attacking the aluminum interconnect on the chips. The answer was to replace half the phosphorous with boron and make borophosphosilicate glass (BPSG). It had cost us millions to sort this out...

I remember the GPT

Test equipment for semiconductors was EXPENSIVE ( we were shocked at the idea of a MILLION dollar tester from LTX, Teradyne or Megatest) so people tried all sorts of workarounds; software on a PC controlling a rack full of benchtest equipment using HPIB or RS-232, multiple dedicated test stations (one for analog, one for digital, one for or RF) or custom standalone test boards using an on-board processor, etc.

The approach at Mitel was to take the chassis from an SX-200 PBX and to retarget it as a test system by developing custom cards for the power supplies, switch matrix, system controller etc. The hardware was the responsability of Russ Fields and the software was done by Brad Snoulten (sp?). It was a very cost effective system, programmed in a language called SEMTEL, but it had a personality.

Every few months when the new coop students came in Claude Auclair, head of test department operations would plunk down a box of 1 Ohm, 1 Watt resistors. These were the limit resistors for the 1 Ampere range on the power supplies and new programmers routinely blew them up. You could judge the quality of the new programmers by the smell in the test area.

One night, test engineer Gerry Ebata and I were sitting over a GPT working on a production problem. Gerry got really frustrated and, in his patented fashion (he destroyed many a keyboard 'Return' key), bashed down on the keys to spell out F*** YOU on the tester command line... the machine came back with 'Same to you fella'. We were completeley paralyzed with laughter for minutes, it turned out that Brad Snoulten had programmed SEMTEL to respond to the word F***. We spent quite a long time testing other phrases to see if we could find more 'Easter eggs'.

Mitel Semiconductor Product Engineering in 1981

Back L-R Colin Harris, Charles Morin, Andrée Girard, Paul Aubin Francois Tanguay, Andre Chartrand, Marcel DeGrandpre, XXX, Gord Harling, XXXX. Front Row L-R Steve Kruse, Andre Houde, Frank Lee, Pêter Perry

This in from Jacques Laneuville and Luc Carrier

Got some information from Luc Carrier about the missing names in your blog picture:

“A cote de Francois Tanguay c’est Paul Aubin, a la droite de Gord c’est Tino Varelas je crois et a la droite de Gord il ressemble a Pierre Codere mais je ne me rappele pas qu’il ait travaille a Bromont. »

Quel droit d'abord? and could Edouardo Gotti be one of those? (suggested by Jean Marcoux)

Little help?

Aaaah, the Calma

Keeping a semiconductor fabrication line full was a problem then, as it is now. Mitel took all sorts of contracts including, at one point, the manufacture and test of a car racing game for Magnavox. The two chip set was mounted (die-bonded) on a hybrid substrate but the design was so quirky (and process control so variable) that batch-to-batch the yield would vary from 0 to about 60%, if I recall. The solution was to bond up three or four and then test them,.. testing involved getting people (usually young women from the operations staff) to play the game several times and note down any anomalies. I remember seeing hybrids with notes scribbled on them like 'No sound', 'Left car doesn't work'...

The best tester was Monique Poulin, she was considered to have such a light touch with electronic devices that they made her the Calma operator. The Calma was a station for doing layout but at the time there were no colour screens so all layout was done in various shades of green differentiated by different dashed lines or cross-hatching. I did some manual modifications to the 8804 on it, a chip with fewer than 1 000 transistors it it took days of painstaking effort, not the least of which was the checking. The Calma computer was serial number 18 and it had been created using wire-wrapped components, It took up several large cabinets (each a bit smaller than a refridgerator), it used a digitizing pen and x-y tablet and came with a huge 6 foot by 6 foot by 4 foot high Xynetics 1100 plotter. When Mitel wanted to get rid of it I bought it at auction and kept it in my basement for years, I still regret that I sold it off for metal scrap when I moved to Calgary but I still have a few of the peripherals and a complete set of printed manuals.

Siltek to Mitel

In the 1970s there was a company called Siltek International with a factory in Bromont, Quebec. They were a very highly regarded supplier of 4000 series CMOS but didn't seem to be able to make a go of it financially. (Does anyone know why?)

Around 1976 Mitel was in a bind, the key component they used in their electronic Private Branch Exchanges (PBX), the 8804, was being obsoleted by Motorola. The 8804 replaced reed relays used on competing systems and so with no moving parts, very low power, and much higher reliability it was a key differentiator for the company. The component was an 8 by 4 analog switch array built in 9 micron metal-gate CMOS. They wouldn't be able to ship product without it! Fortunately around that time the Siltek foundry was up for sale at the ridiculous bargain price of $800k, It included the design tools (enormous Calma GDS-II machines), the complete semiconductor fabrication facility, test equipment, and a packaging facility. They bought it and redesigned the 8804 to solve their supply problem.

Some legendary semiconductor people were around at that time including Alan Aitken (the inventor of ISO-CMOS), Ralph Bennett (of whom many stories are told), Tam Nguyen, Peter Dakin, Peter Kung.

Welcome, and some background

As a result of a friendly dinner between semiconductor industry veterans/survivors we decided to record some of our stories of the semiconductor and microelectronics industry in Canada. If you have a good story, please pitch in (heck, give us the bad ones too), if you were involved in one of these stories, please comment, Pictures and links are welcome.