Calculating Infinity: The Paradox Of Moore's Law

Moore’s law – the observation that the number of transistors in a dense integrated circuit doubles approximately every 18 to 24 months – is a projection and not a physical or natural law. Nonetheless, it has proved accurate for several decades, and has been used in the semiconductor industry to guide long-term planning and to set targets for research and development.

Many advances in digital electronics are strongly linked to Moore’s law: examples are quality-adjusted microprocessor prices, memory capacity, sensors and even the number and size of pixels in digital cameras.

But can this continue indefinitely? Intel stated in 2015 that the pace of advancement has slowed, starting at the 22nm feature width around 2012, and continuing at 14nm. However, in April 2016, Intel CEO Brian Krzanich stated that “In my 34 years in the semiconductor industry, I have witnessed the advertised death of Moore’s Law no less than four times. As we progress from 14nm technology to 10nm and plan for 7nm and 5nm and even beyond, our plans are proof that Moore’s Law is alive and well”.

Physical limits

Silicon-based technologies have nearly reached the physical limits of the number and size of transistors that can be crammed into one chip, while alternative technologies are still far from mass implementation. Down-scaling transistor size is more than an engineering challenge, as there is fundamental physics to consider.

Moore’s law would come to an end because transistors are as small as atoms and cannot be shrunk any further. To address this, fundamentally new concepts for electronics would be needed to produce commercially viable alternatives which meet demands for ever-growing computing power. Moore’s Law may become exhausted eventually unless new technologies come along that will lend it leash.

“The whole semiconductor industry wants to keep Moore’s Law going. We need better performing transistors as we continue down-scaling, and transistors based on silicon won’t give us improvements anymore,” said Heinz Schmid, a researcher with IBM Research GmbH at Zurich Research Laboratory in Switzerland.

Compound semiconductor materials

Schmid’s Zurich-based team with support from colleagues in Yorktown Heights, New York has developed a relatively simple, robust and versatile process for growing crystals made from compound semiconductor materials that will allow them be integrated onto silicon wafers – an important step toward making future computer chips that will allow integrated circuits to continue shrinking in size and cost even as they increase in performance.

The IBM team has fabricated single crystal nanostructures, such as nanowires, nanostructures containing constrictions, and cross junctions, as well as 3-D stacked nanowires, made with so-called III–V materials. Made from alloys of indium, gallium and arsenide, III-V semiconductors are seen as a possible future material for computer chips, but only if they can be successfully integrated onto silicon. So far efforts at integration have not been very successful.

The new crystals were grown using an approach called template-assisted selective epitaxy (TASE) using metal organic chemical vapour deposition, which basically starts from a small area and evolves into a much larger, defect-free crystal. This approach allowed them to lithographically define oxide templates and fill them via epitaxy, in the end making nanowires, cross junctions, nanostructures containing constrictions and 3-D stacked nanowires using the already established scaled processes of Si technology.

“What sets this work apart from other methods is that the compound semiconductor does not contain detrimental defects, and that the process is fully compatible with current chip fabrication technology,” said Schmid. “Importantly the method is also economically viable.”

Spin not charge

One promising approach to developing new technologies is to exploit the electron’s tiny magnetic moment, or ‘spin’. Electrons have two properties – charge and spin – and although current technologies use charge, it is thought that spin-based technologies have the potential to outperform the “charge-based” technology of semiconductors for the storage and process of information.

Scientists from University College London (UCL) have discovered a new method to efficiently generate and control currents based on the magnetic nature of electrons in semi-conducting materials, offering a radical way to develop a new generation of electronic devices.

In order to utilise electron spins for electronics, or ‘spintronics’, the method of electrically generating and detecting spins needs to be efficient so the devices can process the spin information with low-power consumption. One way to achieve this is by the spin-Hall effect, which is being researched by scientists who are keen to understand the mechanisms of the effect, but also which materials optimise its efficiency. If research into this effect is successful, it will open the door to new technologies.

The spin-Hall effect helps generate ‘spin currents’ which enable spin information transfer without the flow of electric charge currents. Unlike other concepts that harness electrons, spin current can transfer information without causing heat from the electric charge, which – as mentioned above – is a serious problem for current semiconductor devices. Effective use of spins generated by the spin-Hall effect can also revolutionise spin-based memory applications.

The scientists have reported a 40-times-larger effect than previously achieved in semiconductor materials, with the largest value measured comparable to a record high value of the spin-Hall effect observed in heavy metals such as platinum. This demonstrates that future spintronics might not need to rely on expensive, rare, heavy metals for efficiency, but relatively cheap materials can be used to process spin information with low-power consumption.

Touch and Go

Overall, perpetuating Moore’s Law in the foreseeable future will require the development of disruptive technologies which take the electronics industry beyond its silicon comfort zone. Whether these can be developed in time for Moore’s Law to be maintained in the immediate future is touch and go, but even if not, there is no reason why it could not resume in the future.