Seven Elements That Have Changed the World (30 page)

It was not a straightforward problem. The companies owned leases of different pieces of land on top of the field and so what each one owned depended critically on the areal distribution of the oil, and that turned out to be very uneven. A long night turned into an early breakfast when, after a lot of stops and starts, a solution appeared and I went to the office. Overnight working turns out not to be a new phenomenon.

I was doing this in Anchorage’s only ‘computer bureau’ run by Millet Keller, a graduate of Stanford University. It contained only a single computer, an IBM 1130, the state of the art at the time. During the day, Millet would run commercial programs in computer language COBOL, creating financial accounts for local banks. At night, I was able to run my own programs written in FORTRAN, a popular scientific and engineering
computer language. Using the advanced IBM technology, which performed ‘as many as 120,000 additions in a second’, I was able to model BP’s Alaskan oilfields to help in their development.
⁶³
BP has been an innovator in computer technology since its earliest days. In the early twentieth century, it developed programs to calculate the most efficient routes for oil tankers to travel. But the invention of the IBM 1130 provided a huge leap in the processing power available to the industry. As a geophysicist, Millet was interested in the work I was doing and would often stay up to work with me, watching over the temperamental machine or feeding in the next punch card.

The IBM 1130 was the first computer I had encountered outside Cambridge University. It was less powerful than Titan, the university’s monstrous mainframe computer, and far smaller, cheaper and more accessible. Titan had filled an entire room and required a whole laboratory team to work it. IBM sought to take computing technology out of that sort of setting to a wide variety of industries for which computing was becoming a necessity.

Today, exploring and drilling for oil without computers is unimaginable. By the time BP was producing oil from the Thunder Horse field, whose namesake production platform nearly sank as Hurricane Dennis passed close by in 2005, it had been able to use seismic and other data to construct a three-dimensional visualisation of the reservoir, several kilometres below the surface.
⁶⁴
That allowed teams of scientists and engineers to work together and actually see the implications of a decision, such as where to place a well, on the long-term health of the reservoir. Processing the huge quantities of data needed to do this has been made possible by the extraordinary growth of a computer’s capability over the last sixty years. And at the core of all this technology, back in Anchorage in 1971 as in the high-performance computer age of today, is a simple, tiny device made from silicon: the transistor.

In the late 1940s, William Shockley and his team in the solid state physics group at Bell Labs were exploring the unusual electrical properties
of a group of elements called semiconductors. Bell’s telephone networks were still operated using mechanical switches and signals were amplified using vacuum tubes.
⁶⁵
These were slow and unreliable and so the director of research was tasked with finding an electronic alternative. Shockley thought the answer could be found in semiconductors, from which he hoped to create an amplifying and switching device.
⁶⁶
Although its theoretical basis seemed flawless, it did not work. His colleague John Bardeen, a brilliant theoretical physicist, then set his mind to the problem. He realised that electrons were becoming trapped at the surface of the semiconductor, stopping current flowing through the device.
⁶⁷
Working with Walter Brattain, whose skilled hands matched and complemented Bardeen’s brain, Bardeen was able to overcome the surface trapping and, in doing so, turned Shockley’s idea into a practical reality, the world’s first transistor.
⁶⁸

At the end of June 1948, Bell Labs announced the invention of the transistor by Shockley, Bardeen and Brattain; they would later win the Nobel Prize in Physics for this breakthrough. At the press conference, they explained that the transistor had the potential to replace the vacuum tube, the device then used to make radios and rudimentary computers. Like the vacuum tube, the transistor could amplify electrical signals and act as an on-off switch, but do so much faster, in a much smaller volume, using much less power.
⁶⁹
At the time, the media thought all this unimportant and made little fuss. The
New York Times
‘carried the big news on page 46, buried at the end of a column of radio chitchat’.
⁷⁰
The potential for the transistor to change the world had yet to be realised by the wider public. After all, journalists must have wondered, what impact could these devices and their abstract functions have on our everyday lives? Even today, few people make a connection between these minute pieces of silicon and the complex functioning of the computers, with which we create images, manage communications and generate sounds.

Any computational problem can be broken down into a set of simple logical steps, such as the decision to combine two numbers or choose one or the other. These steps are controlled by ‘logic gates’, which are the basic building blocks of digital circuits. Logic gates are made from transistors
and other simple components, and they use transistors as switches to send signals. Most logic gates have two on-off switches which together act as inputs. Each switch can either be off or on, known as ‘0’or ‘1’, and the logic gate’s output is determined by these two inputs as well as the type of logic gate. For example, an ‘AND’ gate will give an output of 1 only if both the first ‘and’ second inputs are 1. All other input combinations (0 and 1; 1 and 0; 0 and 0) will result in an output of 0. A computer, at a fundamental level, is simply a number of these transistor-based logic gates linked together to produce a complex output. The capability and complexity of the computer rises as more and more gates are connected.

Transistors allow this to happen because they are very small, very cheap and use only a little power. Those features allow enormous numbers of them to be put together in one computer. It is, however, their speed that makes computers really useful. A transistor’s on-off switching function is controlled by a small electric current. Its small size and the speed of the electrons enable it to be turned on and off well over 100 billion times each second. If you used your finger, it would take around 2,000 years to turn a light switch on and off as many times. Silicon’s semiconducting properties make it ideal for making these switches, although other semiconductors, such as germanium, were originally used and today transistors can be made of many different alloys. None of them, however, rival silicon’s combination of high performance and low cost.
⁷¹

The first commercial applications of the transistor were not, however, computers, but technologies that used its other function as an amplifier. The first of these was a hearing aid produced by Sonotone in 1952. The same principle was applied in radios, amplifying the electromagnetic waves received from transmitting stations. The small size of the transistor dramatically reduced the size and cost of radios, making them portable and opening up their ownership to a vast new market. The transistor radio, the ‘trannie’, heralded a new era of popular music heard anywhere by everyone. As these new products took off, the importance of the transistor began to be widely recognised. In March 1953,
Fortune
published an article entitled ‘The Year of the Transistor’. ‘In the transistor and the new solid-state electronics,’
Fortune
wrote, ‘man may hope to find a brain to
match atomic energy’s muscle.’
⁷²
Silicon had joined uranium and titanium in a class of post-war ‘wonder elements’.

Soon after Shockley, Bardeen and Brattain had invented the transistor, their relationship began to break down. Shockley, paranoid and competitive in the extreme, felt he was not being given sufficient credit for the invention.
⁷³
He became unhappy at Bell Labs where his apparent lack of management abilities, combined with his foul temperament, led him to be overlooked for promotion. In 1956, he left to set up his own company, Shockley Semiconductor, in California, having been encouraged to move there by Frederick Terman, the Dean of Stanford’s School of Engineering. Terman had the vision to see the potential of the semiconductor industry and wanted his graduate students to become a part of it. Together, Shockley and Terman shifted the centre of the industry from the East to the West Coast of the US, laying the foundations of Silicon Valley.

At Shockley Semiconductor, a number of brilliant individuals began investigating the potential of silicon. ‘Neither the processing nor the physics of [silicon] was well understood,’ wrote Gordon Moore, an employee at the company. ‘We were just exploring the technology and figuring out what could be done and we had a lot of things to make work before we could try and build something.’
⁷⁴
However, working at Shockley Semiconductor was a trying experience. Shockley’s bad temper and poor management practices wore down his workforce. He was known to stage public firings and demand lie-detector tests over trivial matters. Shockley and his employees disagreed not only about the direction of the company but also about which new inventions should be commercialised. After about a year, a group of eight of Shockley’s most talented and ambitious employees decided to leave the company. The ‘traitorous eight’ were put in touch with Bud Coyle and Arthur Rock, the latter the father of venture capital. Coyle and Rock persuaded them that, rather than being employed by another company, they should set up on their own. With US $1.4 million of funding from Sherman Fairchild, an inventor and businessman with a
large stockholding in IBM, the group founded Fairchild Semiconductor, in nearby Palo Alto, California.

At the time, one of the biggest technical problems was making the use of a transistor very reliable, unlike the cumbersome and unreliable vacuum tube. Each transistor had to be connected in a circuit by wires installed by hand. As the number of circuits in a computer grew the chances of one of these connections failing rose significantly. That was very risky. Other components in the circuit, such as resistors, were not made of silicon but of carbon and other materials. This made circuit production an expensive and inefficient process.
⁷⁵
In 1958, Jack Kilby, a scientist at Texas Instruments, started making changes to these circuits that would eventually lead to the development of the ‘integrated circuit’ by making all the components out of silicon. But Kilby’s circuits were still connected with fine wires.

At Fairchild Semiconductor, a method had recently been developed to package and to protect these components by using the silicon dioxide layer that naturally forms on the surface of silicon.
⁷⁶
Robert Noyce, a co-founder of Fairchild, described the process as being like ‘building a transistor inside a cocoon of silicon dioxide so that it never gets contaminated. It’s like setting up your jungle operating room. You put the patient inside a plastic bag and operate inside of that, and you don’t have all the flies of the jungle sitting on the wound.’
⁷⁷
Noyce began to think about what else could be done with the new process. He realised that the oxide layer could be used to simplify the production, and so could reduce the cost, of entire electronic circuits. The insulating properties of the layer enabled all the components of a circuit to be produced simultaneously on one piece of silicon. Instead of using wires, the components of the circuits could be connected by using a thin sheet of metal spread on top of the dioxide layer. Wherever a hole was punched through it an electrical connection would be made with the component underneath. Electrical connections could now be ‘printed’ on to a circuit, rather than made with fragile wires. He called this invention the ‘integrated circuit’, in which transistors, capacitors and resistors were printed and connected simultaneously on a piece of silicon.
⁷⁸
This dramatically increased reliability, so much so that NASA used integrated circuits on the early Apollo space missions. Production costs were also
dramatically reduced. Bardeen believed that recognising the natural tendency of silicon to form a protective dioxide layer had led to an invention which was as ‘important as the wheel’.
⁷⁹

Fairchild Semiconductor became the leader in the development and production of integrated circuits as a result of Noyce’s invention. The company grew rapidly; revenue was US $500,000 in 1958 but was over forty times that by 1960.
⁸⁰
Around it many more computer technology and software companies were created and their location came to be known as Silicon Valley.

In 1965, Gordon Moore, one of the Fairchild ‘traitorous eight’, noticed a consistent trend in the way in which the price and size of silicon transistors fell; that trend has underpinned the extraordinary growth of Silicon Valley ever since. Moore’s eponymous law states that the numbers of electronic devices (such as transistors, resistors and capacitors) which can fit on to a computer chip will double every year.
⁸¹
In 1965, Moore expected that this rate of increase would continue for at least ten years so that by 1975 the number of components that could fit on to a computer chip would have grown from 60 to 60,000. To everyone’s surprise he was right. But the realisation of Moore’s law did not stop in 1975. The exponential rate of increase in computing power and the consequential reduction in the cost of that power has been going on ever since.
⁸²

Today the most advanced microprocessors contain over two and a half billion transistors, the size of which has now shrunk to an incredibly small 22 nanometres, just nine times wider than a DNA chain. When combined with the overall growth of the computer industry, this leads to an extraordinary result: more than 10
¹⁸
(a one with eighteen zeros following it) transistors were produced in 2011. This is more than the number of grains of rice grown across the world each year and more than the world’s yearly output of printed characters. It costs even less to produce a transistor than it does to print each letter in a book, newspaper or magazine. The process of miniaturisation, described by Moore’s law, produces faster and cheaper chips. And when chips became smaller and cheaper, they were used in
more and more devices and were embedded into our daily lives. As Moore wrote in the article in which he first outlined his law: ‘the future of integrated circuits is the future of electronics itself.
⁸³