Silicon dominates the electronics industry, but its poor optical properties mean that III-V compound semiconductors are preferred for photonics applications. Photoluminescence at visible wavelengths was observed from porous Si at room temperature in 1990, but the origin of these photons -highly-localized defect states or quantum confinement effects? -has been the subject of intense debate ever since. Since then attention has shifted from porous Si to Si nanocrystals, but the same fundamental question about the origin of the photoluminescence has remained. Here we show, based on measurements in high magnetic fields, that defects are the dominant source of light from Si nanocrystals. Moreover, we show that it is possible to control the origin of the photoluminescence in a single sample: passivation with hydrogen removes the defects, resulting in photoluminescence from quantum-confined states, but subsequent UV illumination reintroduces the defects, making them the origin of the light again.

We have studied the low-temperature photoluminescence of the two-dimensional electron gas in a single GaAs quantum well in magnetic fields up to 50 T over four orders of magnitude of illumination intensity. At the very highest illumination powers, where the recombination is excitonic at zero field, we find that the binding energy of both the singlet and triplet states of the negatively charged exciton (X Ϫ ) increase monotonically with the applied field above 15 T. This contradicts recent calculations for X Ϫ , but is in agreement with adapted calculations for the binding energy of negative-donor centers. At low-laser powers we observe a strong transfer of luminescence intensity from the singlet ͑ground͒ state to the triplet ͑excited͒ state as the temperature is reduced below 1 K. This is attributed to the spin polarization of the two-dimensional electron gas by the applied magnetic field.

We report photoluminescence measurements on stacked self-assembled InP quantum dots in magnetic fields up to 50 T. For triply stacked layers the dots become strongly coupled when the layer separation is 4 nm or less. In contrast, doubly stacked layers show no sign of coupling. We explain this puzzling difference in coupling by proposing a model in which the holes are weakly confined in the Ga x In 1Ϫx P layers separating the layers of dots, and are responsible for the coupling. Since only one such intervening layer exists in the doubly stacked dots coupling is excluded. Our model is strongly supported by the exciton masses and radii derived from our experimental results, and is consistent with available theory.

We have studied the photoluminescence from type-II GaSb/GaAs self-assembled quantum dots in magnetic fields up to 50 T. Our results show that at low laser power, electrons are more weakly bound to the dots than to the wetting layer, but that at high laser power, the situation is reversed. We attribute this effect to an enhanced Coulomb interaction between a single electron and dots that are multiply charged with holes.

We study nanograin size confinement effects, and the effect of the increase of local temperature on the first-order Raman spectrum in silicon nanogranular films obtained by cluster deposition. The local temperature was monitored by measuring the Stokes/antiStokes peak ratio with the laser power up to ϳ20 kW/cm 2. We find large energy shifts, up to 30 cm Ϫ1 , and broadenings, up to 20 cm Ϫ1 , of the Raman-active mode, which we attribute to both laser heating and confinement effects. The phonon softening and phonon linewidth are calculated using a phenomenological model which takes into account disorder effects through the breakdown of the kϭ0 Raman-scattering selection rule, and also anharmonicity, which is incorporated through the threeand four-phonon decay processes. Very good agreement with experimental data is obtained for calculated spectra with nanograin sizes of about 10 nm, and with an increase in the anisotropy constants with respect to those of bulk silicon.

We present the results of photoluminescence experiments on the negatively charged exciton X Ϫ in GaAs/Al x Ga 1Ϫx As quantum wells ͑QW͒ in high magnetic fields (р 50 T). Three different QW widths are used here: 100, 120, and 150 Å. All optically allowed transitions of X Ϫ are observed, enabling us to experimentally verify its energy-level diagram. All samples behave consistently with this diagram. We have determined the binding energy E b of the singlet and triplet state of X Ϫ between 23 and 50 T for the 120 and 150 Å QW, while only the triplet E b is observed for the 100 Å QW. A detailed comparison with recent theoretical calculations shows an agreement for all samples across this entire field range.

A high-purity GaSb/GaAs quantum ring system is introduced that provides both strong hole-confinement in the GaSb ring and electron confinement in its GaAs core. The latter is responsible for a reduced inhomogeous linewidth measured in photoluminescence, in comparison to the previous measurements made on nanostructures with differing morphology in this material system. This allows the resolution of multiple peaks in the photoluminescence due to discrete charging with holes, revealing the mechanism responsible for the excitation-power-induced blueshift.

Dynamic random-access memory (DRAM), which represents 99% of random access memory (RAM), is fast and has excellent endurance, but suffers from disadvantages such as short data retention time (volatility) and loss of data during readout (destructive read). As a consequence, it requires persistent data refreshing, increasing energy consumption, degrading performance and limiting scaling capacity. It is therefore desirable that the next generation of RAM will be non-volatile (NVRAM), low power, high endurance, fast and nondestructively read. Here, we report on a new form of NVRAM: a compound-semiconductor charge-storage memory that exploits quantum phenomena for its operational advantages. Simulations show that the device is extremely low power, with 100 times lower switching energy per unit area than DRAM, but with similar operating speeds. Non-volatility is achieved due to the extraordinary band offsets of InAs and AlSb, providing a large energy barrier (2.1 eV) which prevents the escape of electrons. Based on the simulation results, an NVRAM architecture is proposed for which extremely low disturb-rates are predicted as a result of the quantum-mechanical resonant-tunneling mechanism used to write and erase.