The nineteenth century was the century of the atom. In that century, the early ideas of Stratos and Lucretius became the systematic chemical rules of Dalton (combining elements in strict proportions) and Mendeleev (the periodic table). The nature of crystals (or at least, of facetting and cleavage) was understood. Kinetic theory and Brownian motion drew attention to the dynamics of atoms. The electrical properties of atoms made their mark in electrochemistry, through Faraday’s work. The century culminated in electromagnetism and in those phenomena which shaped and verified it: the discovery of radio waves, X-rays, and the electron. Yet there were clouds on the horizon, as Kelvin observed. One cloud (from experiments on the velocity of light) led to relativity theory. The other cloud (specific heat and black body radiation) led to quantum physics. 

The twentieth century brought radio, X-rays and the electron to some degree of fruition: it was the century of the electron. In the 1950s, it was said an electronics person did not need to know of the existence of the electron. At the end of the twentieth century, we are pushing the manipulation of the electron to the extremes of the materials we can create. The pressures have been ultimate miniaturisation, ultimate speed of operation, low-power operation, and a series of practical demands of compatibility and manufacturability. We beginning to accept a quantum theory which we are not sure we understand intuitively. Will the twenty first century be that of the quantum, in which quantum statistics, quantum tunnelling, and quantum coherence will be the realm of engineers, not just scientists? 

Learning From the Past

What can we learn from these events? First, that change can be very rapid. It took about twelve months for the electrical telegraph to replace the Pony Express over the Rockies. Secondly, that engineers may be able to make things work before we, as scientist, understand them. Even before the electron was discovered, let alone understood, the electric telegraph had ousted semaphore signals and the Pony Express and their like. Technological pressure from semiconductor applications made engineering solutions essential. In the 1960s, dislocations in Si were harmful. They were eliminated. So were swirl defects in the 1970s. And the DX and EL2 centres have been avoided by judicious choices of composition more than by atomistic science. GaN is being tested in traffic lights before we, as scientists, have mapped and understood the defects. So we should be learning that the applied sciences have ways of their own, and that they provide a powerful source of basic scientific ideas. Thirdly, we should recognise that external social pressures affect us all, and in ways which make our predictions unrealistic. If one looks at 1970’s predictions for the next 25 years, then what do we find? Simplifying a little, the 25-year forecast for the information technologies was achieved in about 5 years. The 5 year forecast for environmental changes seems to have taken 25 years. The immediacy of the information technologies has enhanced their power (the cynic would say that computer games will always be more popular than taking environmental improvement from posturing to reality). 

Given the outstandingly poor record everyone has with predictions, I shall be in good company if I fail. But let me begin with two quotations. Ken Olsen (founder of GEC) said "There is no reason why anyone should want to have a computer in their home." And in 1968, an IBM representative asked "But what is [the microchip] good for?" These quotations are not so silly as they seem. Do we really want a computer in our home? Or do we want computer power, in the form of capability for our equipment to do intelligent things? In the 1920s, General Electric was proposing that each home needed a large electric motor to supply mechanical power. This was superseded by many small motors in individual appliances. And what is the microchip for? Is there a limit in size and power for most of our daily purposes? And is this limit imposed by hardware or software (including both algorithm design and interfaces to the human, rather than to that New Person of Silicon Valley, the nerd)? Could it be that Moore’s famous law (of exponentially increasing density of components on a chip) will end, not because of physics, but because of changing social demands? 

The Software Challenge

We are all aware of software problems: how many of us has never felt the need to apply destructive force to our computers? So we must remember the human component, with all its failings. I would hope for more adaptability to suit the user, avoiding complicated or counter-intuitive procedures (there is a Haiku: "To close down, switch off / Should be quite intuitive / But it’s click on Start"). Software raises broader issues. Validation of software is becoming less and less assured as it moves from checks by independent academic groups to competing commercial software houses. The opportunity for mischief is growing with access to the Web and to computer power, whether mischief in the form of viruses, or as a means of propaganda or persuasion. Yet the opportunities for improving the quality of life are enormous. Education could reach the currently-deprived. Culture and traditions could be conserved. Lives could be enriched, not merely diverted. There might be believable and sympathetic (yet electronic) companions to those who age alone. There is room for providing resource for the intellectual and for the spiritual, especially now that there are more people with the time, education and inclination to seek leisure activities without triviality. 

Hardware Challenges: does Semiconductor Physics have a Future?

Regarding hardware, there are many constraints imminent. So long as applications need cheap, long-lasting (10 years is a standard requirement), general-purpose devices, there are problems. Silicon-based devices are reaching a limit determined by economics (fabrication plant costs) and perhaps by other issues, like reliability of the gate oxides. So we shall have to ask questions like: How much miniaturisation is useful? How fast are operations needed for routine devices? 
 

The major effort of semiconductor scientists, world-wide, is (surprisingly) not on silicon, but on III-Vs. Niches are recognised for diamond, SiC, Si/Ge, the II-VIs, and organics. Other niches for semiconductors are recognised for photonic materials, for conductors on chips, for dielectrics, for information storage, and for still further roles. Organics, especially, suggest wide-ranging opportunities because of the impressive ra nge of control possible through synthesis (and even the simpler living things provide an impressive demonstration of what can be synthesised). 

Yet Si still remains the supreme semiconductor. Why, therefore, are there so many outstanding problems in Si, largely unaddressed? Self-diffusion and the accurate prediction of shallow level central-cell corrections are either unresolved or just beginning to be resolved. Processes involving excited states offer novel approaches to processing with low thermal budgets, yet the understanding of excited states has made dismal progress since the 1970s. If there is to be a growth area, that of excited state phenomena seems to me to be the most promising for semiconductors, with natural links to quantum methods and novel materials creation. I worry that semiconductor science might render itself irrelevant by concentrating on technology problems which verge on the obsolete. If semiconductor technology is changing dramatically, and its possible future sketched in the Semiconductor Roadmap, why are the major physics funding agencies not following such leads?

It is common to suggest that biology will be the leading science of the next century. The reality is that the biomedical sciences are being rewritten through physics. The descriptions of neuro-transmitter/receptor interactions, or of protein folding, are being framed in the language of the physical sciences. Major aspects of the activity of the brain are being modelled as neural nets. How much of this rewriting will involve the science of semiconductors is less clear. Yet again, silicon has a special role through the skills developed to shape it into micromechanical devices, as well as through its electronic properties.

Given that the development and characterisation of new materials could become more and more routine, one should ask whether innovation in semiconductors could lead to broader innovation. For example, there is strong commercial pressure in microelectronics to move to very low power operation. Can low-power electronic devices enable large energy savings through control? Many processes which, historically, have involved large energies might be managed with modest power sources if control (e.g., of heat entering or leaving a building) can replace producing and wasting energy. The worries about global warming arise, in part, because of the way European and North American societies evolved in the last two hundred years. Yet, even if Europe and North America were to cut CO2 production entirely, global warming would continue as a threat if other parts of the world were to follow the past Western approach. Is there another option, one which enables the quality of life to be extended without the same massive energy supply? It is not obvious that there is a solution. But semiconductor science should look more widely than to issues of information.

Connecting to Our Lives

Data transmission from one device to another is limited by bandwidth. Data transmission to a human mind is far, far slower, and the effectiveness of data transmission addresses different problems. We are well aware of these problems as physicists, where we have the enormous benefit of electronic journals. Yet probably we all recognise we have not achieved the convenience of the printed book, since we are tied to a computer and to a display with modest contrast and resolution. Communication is not the same as talking, and the active mode of electronic mail differs from the passive WWW. Data display is limited by the cathode ray tube and the mobile phone. New displays are sure to emerge, perhaps even flexible displays as part of one’s clothing. Mobile telephones will become worse (too complex to use, too poor as displays, too irritating to bystanders) before they become better. The subjective response to new promises cannot be ignored.

Yet the most important limitations associated with data are more subtle. Information is not the same as knowledge. Even if there were some way to link the human brain to a major information store (like an encyclopaedia), one can doubt that this would enhance greatly the way of life for the individual. After all, history has plenty of examples of people with incredible memories, and very few have had significant impact on the world around them. It is not clear how quickly information without judgement can become a long-lived business by itself, in that most WWW businesses are surviving on future prospects, not current achievement. And there are nagging practical worries. Will you be able to read your current information in ten years (can you read your old magnetic tapes?). 

Knowledge is not the same as wisdom. Wisdom implies responsibility. The development of global information technologies with virtually unlimited access raises questions of vulnerability, whether from viruses, hackers, or conflict. It is hard to doubt that major technological developments for humanity will be international, and that this global trend will raise many difficulties in law and in resolving conflicts of interest. It is even more certain that the public will want its say in deciding ethical issues which were once left to the professional. Whether the issue is education, new routes in healthcare, threats or worries about industries, or worries about privacy and intrusion, the discussion must be far wider than in the past. We, as responsible scientists, should welcome public interest and questioning. We should be prepared to be challenged, and to use our own knowledge constructively to help the public (of which we are members) to make wise choices.