Imprimir Republish


Revolution on the horizon

A possible substitute for silicon in a new generation of computer chips has come forward

ibmIBM’s new computer memories produced with hafnium oxideibm

At the beginning of this year the international giants IBM and Intel announced, almost simultaneously, that they have already mastered the technology for producing chips made from a new raw material, hafnium oxide (HfO2), which could substitute silicon dioxide (SiO2). These two substances, one or the other, are used to create an insulating layer around the transistors existing within a chip, with the objective of avoiding that electrical current leakage occurs and, in this manner, making them more efficient. The announcement was received as a great advance, because the basic materials used over the last four decades at the level of transistors had not been altered. “To substitute silicon dioxide with an alternative material is a revolution in the materials area to which the industry resisted considerably, but which today appears inevitable”, says the physicist Israel Baumvol, a professor at the Physics Institute of the Federal University of  Rio Grande do Sul (UFRGS), who, since 1999, has been working in partnership with IBM.

The announcements by the two companies are important because hafnium oxide presents a series of advantages in relation to silicon dioxide, such as a reduction in energy loss, an increase in chip processing speed and the reduction of the physical size of the transistors. “With hafnium oxide, it’s possible to reduce the current leakage of the more critical transistors in dozens or hundreds of thousands of times and, through this, to reduce the dissipation of the chip’s heat”, explains Baumvol, who has already produced and published, together with his group, more than 25 articles in scientific magazines in partnership with IBM. As to the size of the transistors, they could be reduced by half in four years and by eight times in ten years. “The practical result of this is an increase of processing speed of up to 50% over the next two to four years and of 100% in eight to ten years”, says the UFRGS professor.

Although the major world manufacturers of chips and computers already master the technology and manufacture processors with transistors based on hafnium oxide, they might only put them on the market starting from next year. “In order to be able to commercialize them, one needs to attain the requisite of trustworthiness of ten years, demanded by commercial microprocessors”, says Baumvol. “It’s possible to carry out accelerated tests and verify this ten year trustworthiness in much shorter times. This is what the companies are doing. I believe that the new processors will be put up for sale within one to two years”, says the researcher. Intel has already announced that its first processors with the new technology will arrive on the market in 2008.

Thus like silicon dioxide, hafnium oxide is a material of elevated dialectic (or high-k). The main property of these materials is their high capacity to store electric charge. The problem with the current existing technologies is that the fine layer of silicon dioxide that surrounds the transistors does not impede that large leakages of electricity occur in the integrated circuits, bringing about overheating and an increase of energy spent on the equipment. This limitation also places in doubt the Moore Law, the maxim concerning the information technology industry prophesized by Gordon Moore, the ex-president of Intel, according to which the number of transistors of a chip doubles every two years. Without the discovery of a new material, of a technical and commercially viable form, the Moore Law is under serious risk of coming to a standstill.

With hafnium, a highly expensive chemical element, as rare and as expensive as gold, the manufacturers hope to solve this problem and to build chips even smaller and quicker. The practical effect of this is that batteries of cell phones, notebooks and other electronic devices could last for longer, doing away with the need to recharge them so often. As well as this, the use of hafnium oxide in the production of transistors will allow for a new generation of chips having only 45 nanometers of channel length, around two thousand times finer than a hair. “We don’t know how far we can decrease the transistor size using alternative dialectics, but perhaps it’ll be possible to reach dimensions of even 16 nanometers”, affirmed professor Baumvol. “With this, the consumer will continue to have electronic devices smaller and smaller, cheaper, using less energy, and everything indicates, with greater processing capacity.”

The use of hafnium (chemical element number 72 in the Periodic Table) in the manufacture of semiconductors for the electronics industry would make the mining industry to increase the metal’s production, which is not found in abundance in nature. Only 50 tons are produced per year in the world. The greater part is destined to the manufacture of nuclear reactor components and metals of high specification for the aeronautic industry. As the quantity of hafnium used in the production of transistors is very small – only 50 molecules are sufficient to make a layer of 3 to 5 nanometer in thickness -, the specialists discard a shortage of supply. In the opinion of Bernard Myerson, vice-president of technology at IBM, in a communication issued by the company, one cubic centimeter of the metal, a size equal to a small sugar cube, would be sufficient to cover ten football fields of silicon plates, containing hafnium transistors, used in the manufacture of chips.