As electronic devices shrink, the interaction between electrons and the silicon crystal lattice, described in terms of 'quasiparticles', is a central issue. Ultrashort laser pulses can track the birth of such a quasiparticle.
With few exceptions, electronic devices are based on silicon: using this semiconductor material, an extremely high density of basic switching elements (transistors) can be incorporated in complex circuits. Soon the size of transistors on commercial silicon chips might drop below 10 nanometres1. At this level of integration, the physical dimensions of each element are similar to the distance between atoms in the material, and the conventional description of charge flow in electronic devices breaks down. On page 51 of this issue, Hase et al.2 present an experiment that gives a flavour of the colourful dynamics that might be expected under such conditions.
In a material such as silicon, the motion of electrons is influenced by the surrounding ion cores of the crystal lattice. These processes are complicated; but, on time and length scales that are much larger than molecular ones, they may be described in a straightforward way by assuming that the electron mass is altered and that its specific state of motion has a finite lifetime. In the language of condensed-matter physics, the electrons are 'renormalized' by the surrounding matrix, becoming 'dressed' as 'quasiparticles'. Until now, microelectronics engineers could quietly assume that interactions between these quasiparticles and their matrix occur instantaneously — that when an electron bumps against the ion lattice, there is a sudden change in its kinetic energy and momentum. But at nanometre scales, where this semiclassical picture is no longer applicable and quantum effects become relevant, it is not clear at present how to model the electron behaviour.
There is another complication too. In principle, it should take a finite amount of time to establish the quasiparticle in a non-equilibrium situation or to complete a scattering with the lattice. The typical timescale for this is given by the oscillation cycle of the lattice vibrations. These are quantized, and are given the name 'phonons'. As transistors shrink to nanoscale dimensions, they might respond so quickly to a pulse of electric current that their internal switching times approach this limiting timescale.
The build-up of quasiparticles has been observed3 in compound semiconductors such as GaAs. But quantum kinetic processes have not previously been accessible in silicon. To measure them requires extremely high temporal resolution, which can currently only be achieved using ultrashort laser pulses, of a few femtoseconds duration (1 fs = 10−15 s). However, the electrons moving inside silicon are typically of low energy and their coupling to the photons of the laser pulse is relatively complicated, so direct insight has been hampered.
Hase et al.2 have worked around this problem by using femtosecond ultraviolet pulses first to excite the electrons to much higher energies, and then to probe the subsequent dynamics (Fig. 1). The absorption of high-energy photons by the electrons is a relatively simple process, and interpretation of these data is straightforward. The authors then trace what happens to the electrons using the electro-optic effect, which describes how the excited electrons and lattice vibrations change the optical properties of the material. These measurements are truly a tour de force: the electro-optic effect is absent in bulk silicon (because it is a centrosymmetric system); but Hase et al. have sampled reflected light from the silicon surface and show that it is modulated. Their apparatus needed to be a hundred times more sensitive than in previous experiments4 (on polar materials), because only a few atomic layers at the surface contribute to the signal.
The time traces recorded by Hase et al. consist of two parts (Fig. 1). One is an exponentially damped oscillation, which originates from the interactions of lattice phonons and the photo-generated electrons. The other is an aperiodic, transient signal from the electrons themselves that appears only for 50 fs. Other researchers might have been content to focus on the phonon oscillations alone. But with an experiment capable of 10-fs time resolution, Hase et al. realized that the electronic response does not simply follow the excitation pulse. Instead, it contains more information on the build-up, or dressing, of the electronic quasiparticles. Cleverly, they transformed their time-dependent data into time–frequency space, creating a sequence of spectra defined by varying the delay between the application of the pump and the probe laser pulse. And there, they have found new physics.
Extremely close in time to the initial excitation, Hase et al. have uncovered signatures of the force exerted by the electronic charge on the lattice, and by the lattice on the developing quasiparticles. The most remarkable feature is a dip in their spectra where the electronic and lattice responses overlap in the time–frequency plane. The authors attribute this to a so-called Fano interaction5, originating from a coherent superposition of phonons and the broad continuum of electronic excitations. Solid-state physicists talk about 'quantum correlations' in this context; researchers from the field of quantum optics might use the term 'entanglement'. In any case, Hase et al. have watched, step by step, the ultrafast birth of quasiparticles in silicon, as electrons are dressed by the lattice.
This paper2 stimulates many fascinating questions, some general in nature, others specific to future research. For example, specialists might wonder how theory will succeed in explaining quantitatively the complicated phenomena that are seen in this experiment. Also, how closely does the behaviour of the very energetic electrons observed by Hase et al. mimic that of their lower-energy counterparts travelling through silicon chips? The day will come when quantum physics directly influences the functionality of computers and other electronic equipment that we use in everyday life — the question is, when?
Doyle, B. et al. Intel Tech. J. 6, 42–54 (2002); http://www.intel.com/research/silicon
Hase, M., Kitajima, M., Constantinescu, A. M. & Petek, H. Nature 426, 51–54 (2003).
Huber, R. et al. Nature 414, 286–289 (2001).
Cho, G. C., Kütt, W., Kurz, H. & Ibach, H. Phys. Rev. Lett. 65, 764–766 (1990).
Fano, U. Phys. Rev. 124, 1866–1878 (1961).
About this article
Physical Review B (2019)