A history of electron beam testing – also known as voltage contrast.
1. The early history of voltage contrast
The technique was invent by Graham Plows sometime after 1965, when he was a graduate student in Charles Oately's famous scanning electron microscopy group at Cambridge University in the U.K., and the work he did on the technique got him a Ph.D. (G.S.Plows, "Stroboscopic Scanning Electron Microscopy", Ph.D. Dissertation, Cambridge University; 1969), as well as a pair of patents.
The patents named him and his supervisor, W.C. Nixon as the joint inventors, and were taken out, paid for, and owned by the Cambridge Instruments Company, who in 1965 had started selling a commercial version Charles Oately's original scanning electron microscope. Cambridge Instruments had been a big and long established company, founded in 1878 – as Cambridge Scientific Instruments - by Horace Darwin, Charles Darwin's youngest son, and had produced a very wide range of scientific and industrial instruments, but they had been taken over by the George Kent Group in 1968, and the scanning electron microscopy group had become something of a white elephant, which was eventually floated off as a separate company in 1975, when I happened to be working for the George Kent Group at Kent Instruments in Luton, and was peripherally interested in what was going on. Since then it has been through a few more changes of name and ownership and is now part of Carl Zeiss Electron Microscopy Ltd. I joined the company in November 1982, after it had been bought by Terence Gooding in 1979.
The patents taken out by Cambridge Instruments were
Plows Graham Stuart, Nixon William Charles: Electron beam apparatus. Plows Graham Stuart Sep, 15 1971: GB1246744
Plows Graham Stuart, Nixon William Charles: Electron beam apparatus. Plows Graham Stuart Apr, 15 1970: GB1187901
When I was a graduate student, completing my own Ph.D. in physical chemistry at the University of Melbourne in Australia at roughly the same time as Graham was working on his Ph.D. ( I finally graduated a year later than Graham, in 1970), I'd come across a couple of the “voltage contrast” pictures he'd published in the journal “Electronics Letters”, and had been impressed by the weirdness of the whole idea.
People with a more direct interest in the subject saw the chance of making money out of it.
After he’d completed his Ph.D. Graham Plows worked at Cambridge Instruments for some 18 months, devising a scheme that could be added to a standard Cambridge Instruments scanning electron microscope and sold to semiconductor manufacturers. When the design was far enough advanced to let Cambridge Instruments work out how much it would cost and thus how many they could expect to sell, they determined that the market wasn’t big enough – at that time – to justify the cost of developing the hardware needed to make such a system work.
Graham Plows left Cambridge Instruments at that point and eventually set up his own company, Lintech Instruments, which initially concentrated on rather different products.
Around 1978 Don Ranasinghe at British Telecom’s Martlesham Heath Research Centre found some money to pay Lintech to develop voltage contrast hardware which could be added to a standard scanning electron microscope, and Cambridge Instruments agreed to fit this additional equipment to some of their scanning electron microscopes and take the responsibility for assembling the package for their customers, and maintaining the package after it had been sold. This wasn't anything unusual – at the time Cambridge Instruments was already selling a significant proportion of their electron microscopes pre-configured for X-ray microanalysis with extra hardware supplied by Link Systems.
When I joined Cambridge Instruments in November 1982, I reported to the senior engineer then responsible for all the electron microscope development work – Ralph Knowles – and started off by spending several months with a couple of service engineers getting the first of the Lintech voltage contrast packages working. It proved to be difficult, and one of the things we put together in this period was a twenty point acceptance test to check that all of the various add-on parts that we got from Lintech actually worked.
When Graham Plows had published his thesis work, Don Ranasinghe wasn’t the only person who had appreciated its potential utility.
Siemens Semiconductor – now called Infineon – set up their own voltage contrast group early on, and scored a notable success when Intel licensed their 8051 single chip microprocessor design to Siemens. Feuerbaum and his colleagues examined the bare chip in operation, and noticed that one of the internal clock drivers was more heavily loaded than the others, which prompted them to suggest a small change to the chip layout. This change doubled the maximum speed at which the chip could be clocked, and thus its performance, allowing Siemens to sell a lot more of their improved version of the 8051 chip than they (or Intel) had originally expected.
Eberhard Menzel's 1981 Ph.D. thesis on voltage contrast seems to have been based on work that he did with Feuerbaum's group, and I got a copy of the thesis (via Cambridge Instruments German sales representative) in 1983, and I found it very useful – I could read German well enough to follow what was being said.
Siemens' success prompted Intel to put quite a lot of money and effort into setting up their own voltage contrast system, based on an S.100 electron microscope that they'd bought from Cambridge Instruments. The S.100 was cheap and rather spartan, and probably made the job a bit more difficult than it might have been. Cambridge Instruments didn’t have any contact with Intel about this system (beyond sending in service engineers to maintain the microscope from time to time) but we had a rough idea of what they were doing.
Cambridge Instruments was rather better informed about the Siemens project and eventually signed an agreement with Siemens to commercialise the Siemens voltage contrast work.
My second task at Cambridge Instruments turned out to be putting together a close copy of the Siemens voltage contrast system to be sold to Thompson-EFCIS – another European semiconductor company – for their development laboratory in Grenoble, in France. Shortly after I'd joined Cambridge Instruments, the then technical director, Ian Cruttwell, had dragged me along to a meeting with Thompson-EFCIS at Grenoble, in my capacity as the man who was going to put the system together and get it working. I'm a quick study, though not in Ian Cruttwell's class, and what little I needed to say at the meeting didn't reveal quite how new I was to the project.
During the actual process of putting the machine together I was again supervised by Ralph Knowles.
The machine was supposed to be as close as possible to a direct copy of the Siemens machine – Thompson-EFCIS weren't prepared to wait for Cambridge Instruments' commercialised version of the Siemens voltage contrast system - but we were putting the voltage contrast add-ons onto a Cambridge Instruments S.250 electron microscope, which wasn't the electron microscope that had been used at Siemens, so we were obliged to improvise. In the process, I ended up inventing a slightly improved electron beam-blanking system (which Cambridge Instruments patented) and we tidied up a number of the Siemens physicists’ improvisations to produce a slightly more manufacturable system.
US patent number 4,614,872 “Charged particle deflection” May 1, 1984
We probably didn't actually need the fancier beam-blanking system, but Ralph Knowles wanted our machine to be able to offer EBIC – electron beam induced current probing - as well as voltage contrast, so he wanted a beam blanking system that worked at beam voltages up to 15kV (where the electrons had enough energy to get deep enough into the integrated circuit to generated charge carriers inside individual transistors) as well as at beam voltages of around 1kV which were just high enough to kick secondary electrons out of the metal patterns on the surface of the integrated circuit. EBIC was useful in some applications, but some circuits died if you kept doing it for too long, and it was never a widely popular technique.
I installed the machine at the Grenoble laboratory late in 1983 and it worked. The customer was never wildly happy with the machine – it could be made to do what they needed it to do, but it was never easy to use, and the Princeton Applied Research (PAR) Box-Car Integrator which Siemens had chosen to handle the stroboscopic signal processing was relatively slow.
The PAR unit was sold as being able to sample at up to 5MHz (which wasn't quick), but in fact couldn’t go faster than 2.2MHz. When we were putting the system together at Cambridge Instruments I'd noticed this, and had got into the machine and replaced a critical TTL integrated circuit with a faster part, which got the speed up to 2.7MHz, but nobody at Cambridge Instruments or PAR could come up with a practicable scheme to make it go any faster.
As soon as we’d installed the system in Grenoble, Graham Plows sued Cambridge Instruments for violating the patents on voltage contrast imaging that Cambridge Scientific Instruments had taken out in his (and Bill Nixon’s) names back in 1968 and 1970. When Cambridge Instruments had run out of money in 1979, they’d stopped paying the annual fees on both patents, abandoning them, and Graham Plows – as a named inventor – had taken advantage of the opportunity to take possession of the patents by taking over paying the annual fees.
Cambridge Instruments had been aware of the risk that Graham would sue, and had paid quite a lot of money to an expensive patent lawyer for a “counsel’s opinion” which had assured them that Graham would not have much chance of winning his case if he did chose to sue.
Once Graham Plows had got the legal machinery in motion, the expensive patent lawyer started being less confident about winning the case, and Cambridge Instruments eventually chose to buy off Graham – though not before spending more money on lawyers than had been spent on putting together the machine that I’d installed at the Thompson-EFCIS laboratory at Grenoble.
The money Cambridge Instruments used to buy off Graham Plows was largely a European Economic Union development grant intended to pay for the development of a specialized electron microscope for electron beam testing, and it was transferred into Graham Plows pocket – in his capacity as managing director of Lintech – by paying Lintech to develop a substitute for the PAR Box Car Integrator which we could add onto the Cambridge Instruments S.200 electron microscope (essentially a more powerful version of the S.100 microscope electronics, up-graded to the point where they could drive the S.250 electron-optical column) which had gone into production late in 1983 – I’d spent six months as part of the team sorting out the S.200's numerous teething troubles while the negotiations with Lintech had been going on.
Graham's purpose built electron beam tester was initially very successful. Motorola used a Lintech electron beam tester to debug prototype MC68000 processor chips and claimed that it saved them some six months of development time.
Some of Lintech's electronic hardware was developed by a Neil Richardson, who had a Ph.D. in electrical engineering, but was used by Graham in many different roles, one of them being on-site customer training.
When Graham sold his first electron beam tester to Fairchild in California, Neil spent a couple of weeks at the Fairchild laboratories at Palo Alto in California installing the machine and training the engineers who were going to use it. He seems to have done a good job, because Fairchild eventually hired him away from Lintech to work at Palo Alto as a research engineer.
The commercial success of Graham's electron bean tester attracted the attention of companies that sold test equipment to semiconductor manufacturers.
Feuerbaum’s group at Siemens metamorphosed into ICT GmbH and developed a – not all that successful - electron beam tester in conjunction with the Japanese Advantest Corp. Don Ranasinghe bought one for British Telecom and used it happily for eight years, until British Telecom stopped making its own integrated circuits, when the machine was sold on to the UK Ministry of Defence Research Centre in Gloucester, where it is reputed to be still in use. The sample chamber on Don Ranasinghe's original voltage contrast electron microscope had proved inconveniently small.
At Fairchild Neil Richardson transferred to Schlumberger Test Systems in San Jose, California - Fairchild Semiconductor had been taken over by Schlumberger back in 1979. At San Jose he seems to have been one of the senior members of a team that developed a better version of the Lintech electron beam tester.
Neil presumably knew exactly what was wrong with the Lintech system, and seems to have influenced the Schlumberger team to make sure that the Schlumberger system did pretty much exactly the same job more easily, slightly more quickly and significantly more reliably.
One of his then collaborators – Mike Engelhardt who now works for Linear Technology Inc. - had a high opinion of Neil's contribution. In January 2004 he wrote “"Neil's vision of an e-beam probe did extremely well" in a thread on the newsgroup sci.electronics.design.
Amongst other things, the Schlumberger system up-graded the user interface from Lintech's menu-based system running on a IBM personal computer to a graphical user interface (GUI) running on a work-station.
By then, the Apple Lisa had entered the personal computer market with a graphical user interface (GUI), but it had barely enough processing power to make it work.
The Apple Lisa did make it widely obvious that a GUI made a system easier to use, and the “professional” software packages commercially available at the time were almost always written to be run on workstations powerful enough to support a GUI. The Metheus computer-aided electronic design package we bought at Cambridge Instruments ran on a specially designed work-station which incorporated no less than three Motorola 68000 processors – one to look after the display, another to do the number-crunching and a third to look after the hard disk.
Graham Plows’ approach to development was always biased by his role in selling the machine – he wanted to be able to offer every possible bell and whistle, and he had preferred to direct Lintech’s development effort towards adding yet one more gizmo to the machine rather than getting the bugs out of the existing gizmos. Neil had spent enough time operating the machine and training customers in using the machine to appreciate the disadvantages of this approach, and – according to Mike Engelhardt – the Schlumberger machine ended with a 98% market share.
From the day the Schlumberger electron beam tester hit the market in 1986, Lintech didn’t sell another machine. Lintech had a two year order book at that point, and stayed in business until they had shipped the last of the old orders.
By then Schlumberger had built up a similar back-log of orders, which meant that one semi-conductor, manufacturer - Samsung – had become so desperate to get hold of any electron beam tester that they were prepared to order a Lintech system, but their order didn’t reach Cambridge until an hour after Lintech had stopped trading, early in 1988.
Cambridge Instruments bought up Lintech's assets for about £30,000 shortly after they'd stopped trading, and accepted Samsung's order for a £200,000 Lintech electron beam tester. Graham Plows became the new technical director at Cambridge Instruments (his predecessor - Ian Crutwell - had recently been promoted to a similar position in the Leica group – which did include Cambridge Instruments) and Cambridge Instruments hired most – but not all – of Lintech's employees.
What happened after that is covered by my weekly reports from the 8th July 1988 to the 8th November 1991 accessible here after you've ploughed through a bit of my personal history .The reports for 1988 can be read here, 1989 here, 1990 here, and 1991 here, but they will make more sense if you start your reading here.