The regeneratively amplified Ti:Al2O3 laser system used in these experiments can be configured to produce pulses with widths of 40 fs or 2 ps, providing an opportunity to study the effects of pulse width on QD height. QD height distributions extracted from AFM data of ablation using the two different pulse widths is shown below, in Figure 10.
Fig 10. Pulse width dependence of size distributions of GaAs QDs produced with (a) 40 fs, 0.15 J/cm2 pulses and (b) 2 ps, 0.30 J/cm2 pulses.
It is difficult to directly compare data from the two pulse widths at identical fluences, as the threshold fluence for measurable QD formation is significantly higher for 2 ps pulses. We thus compare data at a fluence of 0.15 J/cm2 for the 40 fs pulse to data at 0.30 J/cm2 for the 2 ps pulse. No QDs were observed with 0.15 J/cm2, 2 ps pulses, indicating that the ablation mechanism changes significantly in this pulse width range. The size distribution of the QDs shown above is significantly narrower for the 2 ps pulses, a finding that was repeated at other laser fluences.
The laser fluence has a significant impact on QD size, with increasing laser fluences resulting in larger QDs. Topographical AFM images of GaAs QDs deposited on mica substrates are shown below, in Figure 11.
Fig 11. Topographical AFM images of GaAs quantum dots produced at fluences of (a) 0.15 J/cm2 and (b) 1.0 J/cm2. Note the difference in the color scale, which is representative of the QD height. Laser pulse width = 40 fs. Measurements were performed directly across from the ablation spot.
In Figure 11(a), the maximum height (denoted by lighter colors) is 3.36 nm, while in Figure 11(b) the maximum height is 70.73 nm. The size of the QDs can thus be varied from less than 3 nm to more than 70 nm by tuning the laser fluence. The small QD size achievable through PLD techniques is significant for two reasons. First, these QDs, which have been shown through previous transmission electron microscope measurements to be largely spherical, have sizes significantly below the exciton Bohr radius of most semiconductors. They thus exhibit strong three dimensional quantum confinement regime in which significant changes to optoelectronic properties occur. This is in contrast to QDs produced through conventional strain induced growth methods, which exhibit strong confinement in only the growth direction (the lateral dimensions of strain induced growth QDs are often greater than 20 nm). The sharp increase in total QD volume tracks well with previous reports of exponential dependence of material removal from semiconductor targets near the ablation threshold. QD size increases may result from the increased plume densities arising from higher laser fluences. It is important to note that this AFM data was taken at the point directly opposite the ablation spot. This corresponds to the central region of the plume, which is expected to produce the largest QDs.
The distribution of QD sizes can critically affect device performance. We have written software that extracts QD heights from raw AFM data. This software has been extensively checked with measurements performed by hand using the line section feature of the AFM operating software, and was found to be accurate for well separated QDs. QD height distributions for two laser fluences are shown below, in Figure 12.
Fig. 12 Height distributions for QDs produced with laser fluences of (a) 0.50 J/cm2 and (b) 0.15 J/cm2. Measurements were performed directly across from the ablation spot.
The distributions are fairly broad for both laser fluences shown in Figure 2, with standard deviations of QD height equal to ~15% of the mean height at the higher fluence and ~30% at the lower fluence. The distribution of the larger QDs is comparable to those of QDs produced by strain induced growth methods. No direct comparison can be made for the smaller QDs, which cannot be produced by strain induced growth.
The dependence of QD size on laser fluence for 2 ps pulses is illustrated below, in Figure 13.
The steepness of the slope of this graph is again indicative of the exponential dependence of material removed from the target on laser fluence near the ablation threshold. This poses significant experimental difficulties for reducing the width of QD size distributions. Pulse to pulse energy fluctuations must be reduced, for example, to less than 5% to achieve the narrowest size distributions for QDs near 10 nm. This restriction is eased somewhat for extremely small QDs, as the slope of the graph in figure 12 flattens considerably at low fluences. The critical dependence of QD size on laser fluence also indicates that changing the spatial profile of the laser beam from a Gaussian to a flattop profile may significantly reduce QD size dispersions.
Fig. 13 Fluence dependence of the average height of GaAs QDs produced by ablation with 2 ps pulses in vacuum. Error bars indicate the standard deviation in the height from atomic force microscope measurements.
QD size is also affected by the location of the sampling site on the substrate. This is a direct consequence of the correlation between plume density, which typically follows a dependence (where is the angle from the normal to the target and n ranges from 2 – 20, depending on laser and target parameters), and QD size. QDs at the plume center are largest, while QDs at the outer edges of the plume are the smallest. This dependence has been mapped out for GaAs QDs produced with 2 ps pulses, as shown below in Figure 14.
Fig. 14 Dependence of GaAs QD size on distance from plume center. Horizontal error bars arise from the uncertainty of the precise position of the plume center on the substrate. Laser pulse width = 2 ps, laser fluence = 0.6 J / cm2. Target to substrate distance = 10 cm.
The QD size decreases significantly as the sampling site is moved away from the plume center. This has both positive and negative implications: simultaneous production of a variety of QD sizes in one experimental run is beneficial for basic research, as the size dependence of QD properties can easily be studied with such samples; but measures must be taken to restrict deposition to one portion of the plume for devices which require high QD homogeneity.
Ablation with a pair of pulses in rapid successions can affect QD size, due to changes arising in the target surface due to the initial pulse or due to interaction of the second pulse with the plume of materials ablated by the first pulse. We have begun construction of prism based ultrafast pulse shaper to investigate these effects, with the aim of optimizing the efficiency of QD production by ultrafast pulses. The active element in the pulse shaper is a liquid crystal spatial light modulator that is used for frequency domain pulse shaping. A genetic algorithm has been implemented to automate the search for the optimum pulse shape. The genetic algorithm uses the level of the broadband blackbody emission, which arises from QDs in the plume (atoms and ions emit at discrete wavelengths), as a fitness signal that it seeks to maximize through varying the signal levels to each pixel of the spatial light modulator. Preliminary results indicate that the blackbody emission signal is increased by a factor of four for a weaker first pulse followed within 5 ps by a stronger second pulse. AFM studies of the particle sizes are currently underway to assess the impact of the pulse shaping on QD mean size and size dispersion. Studies of the laser induced fluorescence of QDs in the plume are also being carried out to determine if the fluorescence signal can be used to provide a fitness function that distinguishes between large and small QDs.