The effect of concentration and post-deposition annealing on silica coated germanium quantum dot thin films grown by vertical deposition

2.1 Materials and methods

All materials were obtained commercially and were not further purified. Silica coated Ge QDs (Ge@SiO₂) were prepared by previously reported methods [18, 19]. Quartz slides (75×25×1 mm³) were obtained from Chem Glass. Indium tin oxide (ITO) coated glass slides (75×25×1 mm³, 8–12 Ω/sq surface resistivity) came from Sigma Aldrich. The n-type silicon wafers were provided by Natcore Technology, Inc. (NXTV) with additional heavy phosphorus doping on the back to create a very conductive surface ready for the Au metal as a contact [19]. The wafers were passivated with silica to ensure the dopants remained in the wafers before use. The wafers were cut into 2×2 cm² pieces for experimental use. The oxide layer needed to be removed from each piece of wafer before thin films could be deposited, and the p-type wafers needed the excess aluminum paste to be removed as well to stop the aluminum from contaminating the particles during deposition. The wafers were cleaned in buffered oxide etchant for 3 minutes to remove the native oxide layer and the excess aluminum paste. The pieces were then cleaned with DI H₂O and plasma cleaned for 10 seconds to put a very thin oxide layer of about 10 nm back on the surface of the wafer. Without the thin oxide layer, the surface is too hydrophobic to allow the silica coated particles to come near the surface to create the thin film.

Characterization of the arrays was performed with a FEI Quanta 400 ESEM FEG scanning emission microscope equipped with an EDAX energy dispersive spectroscope. X-ray diffraction was performed on a Rigaku D/Max Ultima II configured with a vertical theta/theta goniometer, Cu-K _α radiation, graphite monochromator, and scintillation counter. Plasma cleaning was performed with a Plasma Cleaner 1020 equipped with Ar:O₂ (95 : 5) gas solution.

The efficiency of the QD/Si cells were calculated from the I-V curves detected via Keithley 2420 and 2425 High Current SourceMeter with an Oriel Model 81190 Solar simulator equipped with a Xenon lamp, including light intensity feedback control [19]. The intensity of incident light was 100 mW/cm². The solar cells were kept at a fixed distance of 6 inches from the light source for optimal conditions. The cells were attached to two leads, one connected to the back contact and the other connected to the busbar of the front contact [20]. This setup allowed for the measurement of the current produced by the cell with change in voltage.

2.2 Vertical deposition

To control the array, a 4 cm² silicon wafer was placed vertically in a centrifuge tube with the coated QD solution (12 mL) k completely submerged. Water was used as the solvent to best control the film. The solution with the wafer was then dried in a fume hood at the solvent evaporation rate at room temperature. Not increasing the speed of evaporation with heat or vacuum allowed the particles to align themselves at the very top of the meniscus while the solvent dried slowly (Fig. 1) [10–13]. The solution with the wafer was sonicated briefly every couple of hours to ensure that the particles were evenly dispersed in the sample and not settling over time. In the case of Si wafer substrates, once the wafers are coated with the particles, a back contact of gold is added by sputtering. The front contact (gold or silver) is sputtered using a mask to create fingers across the surface without completely covering it with metal [20]. Both contacts are 100 nm thick, and the metal used depends on the identity of the dopant of the wafer and the presence of a QD containing layer [19].

To determine if the amount of material in the solution controlled the thickness of the thin film, several samples were made with varying concentrations of nanoparticles (NPs). Ge@SiO₂ NPs varied from 0.094–1.5 M, and P-Ge@SiO₂ NPs varied from 0.034–0.63 M. These concentrations were diluted with DI H₂O to keep everything consistent for comparison, and they were treated identically to the regular vertical deposition samples.

Annealing studies were performed on these samples. A set of four wafers was prepared with the same concentration of SiO₂@Ge NPs (0.375 M) to create four very similar thin films of the same material at the same concentration. These wafers were then each annealed at a different temperature for comparison. One was not annealed at all as a control sample. The second wafer was annealed at 200°C under an inert atmosphere of argon in a sealed system in a tube furnace for 1 hr. The third wafer was treated similarly at 400°C, and the fourth wafer was treated similarly at 600°C. These wafers were allowed to rest for 24 hr. after annealing to become reacclimated with the natural humidity of the atmosphere [21–24].

3.1 Vertical deposition

We have previously deposited films of nanoparticles of silica coated Si QDs (Si@SiO₂ NPs) with an average particle size of 150 nm [17] by suspending substrates vertically in DI water and the solution was evaporated on its own time at room temperature. Although the films appeared to be glassy, a closer view showed that there were gaps between the particles. This indicated that this was not an optimized array [17]. The average particle center to particle center distance was 170±30 nm, consistent with a porous film. Other solvents were previously tested with this method [27], but none were found to be better than pure water. Furthermore, significantly smaller Ge@SiO₂ NPs have been produced and are required for solar cell applications [18, 19]. Thus, it is expected that the packing of the particles and overall topology of the thin film may alter.

Figure 2a shows the SEM image of a Ge@SiO₂ film grown from 1.5 M solution in DI H₂O. As may be seen by comparison with the results of an analogous growth of larger Ge@SiO₂ NPs (Fig. 2b), the film is more densely packed. However, the film is actually made of large plaques of particles rather than a continuous layer (Fig. 3). Such a cracked morphology is typically due to stresses upon solvent evaporation; however, it is interesting that this was not observed for the larger Si@SiO₂ particles. Presumably, lower particle^…particle forces in the latter film mean that a lower density structure is formed (Fig. 2b). Obviously the plaque like structure is not appropriate for a device since metallization causes a short through the film, see below, Furthermore, not only are the plaques separated from each other in the plane of the substrate, but in some cases, the plaques are even separated and peeled up from the silicon wafer surface (Fig. 3). This separation also causes an issue with achieving higher efficiencies because whatever current the thin film generates cannot be transferred to the wafer to complete the circuit. Due to these issues, concentration and annealing studies were done to determine if these problems could be combatted with slight modifications.

In order to determine if the concentration of the particle solution affected the uniformity as series of the films were grown with both Ge@SiO₂ and P-doped Ge QD derivatives (P-Ge@SiO₂). The concentrations of the as prepared (see Experimental) Ge@SiO₂ (1.5 mol/dm³) and P-Ge@SiO₂ (0.63 mol/dm³) NP solutions were determined from weight measurements after drying. This difference in NP concentration is most likely due to the change in QD surface with the presence of phosphorus [19]. A series of standard dilutions (¹/₂, ¹/₄, ¹/₈, and ¹/₁₆ of the original solutions) were prepared to determine if there is a correlation between concentration and film thickness and/or continuity.

Figure 4 shows top and cross sectional views of films grown from different concentration solutions of Ge@SiO₂ NPs. Regardless of the concentration of the particles, the packing between the particles remains constant and relatively close, resulting in a dense film. However, as the concentration decreased, the uniformity of the film decreased. More importantly, in most of the films, regardless of concentration, plaques of particles were still formed with the exception of the lowest concentration of Ge@SiO₂ NPs (0.094 mol/dm³). The non-uniform nature of the films may be a consequence of the slow rate of film growth (days). In order for the NPs not to precipitate out of solution through aggregation, the reaction must be sonicated. This sonication may result in the damage of the grown film and the loss of fragments to the solution.

The graph in Fig. 5 shows that little to no correlation between concentration and film thickness of Ge@SiO₂ particles, except for the lowest concentration. It is worth noting that it is only at the lowest and highest concentrations that uniform films (small standard deviation in thickness) are formed. We have previously observed that too rapid an evaporation rate can cause the surface of the film show wave-like features [17]. This is not the case here, instead there is an inhomogeneous film growth; however, the NP^…NP interaction compared to the NP^…solvent interaction has been discussed as a controlling factor in film uniformity [8]. Although there is an obviously smaller film thickness with the lowest concentration of particles, the film also covered much less of the wafer. Higher concentrations could not be studied since even at 1.5 mol/dm³ the particles have a hard time staying insolution.

The same experiments were performed with the phosphorus-doped quantum dots coated with silica (P-Ge@SiO₂). The base solution of these particles is lower than that of the undoped analogs; however, a similar trend is observed for the films grown. SEM images of the top and cross sectional views for films grown from each concentration are shown in Fig. 6. It is clear that these are very similar to the undoped films with plaques of close packed particles that are not touching each other or sometimes not touching the wafer. The variation of film thickness with the concentration of P-Ge@SiO₂ NP solution is shown in Fig. 7. Here there is the more expected trend, but again the variation in film thickness within a sample is wide.

We have previously demonstrated [19] that both the Ge@SiO₂ and P-Ge@SiO₂ NP thin films deposited on n-type Si wafers (see Experimental) allow for the fabrication of low efficiency (as a consequence of film uniformity) solar cells of the type defined as Ag|Ge@SiO₂|n-Si|Au shown in Fig. 8. Ge@SiO₂ and P-Ge@SiO₂ NP thin film layers were deposited on n-type wafers with silver front and gold back contacts. The I/V curves were measured for these cells (see Fig. 9). As may be seen from Table 1, the as synthesized concentrations used previously [19] were not determined to be the optimal concentration for either Ge@SiO₂ or the P-doped homologs. Despite each type of NP having a different as-synthesized concentration (Ge@SiO₂ = 1.5 mol/dm³ and P-Ge@SiO₂ = 0.63 mol/dm³) the best devices were fabricated using films with a quarter of the natural concentration. The difference in the as-synthesized concentrations is not due to the pH of the QD solutions; as both reagent solutions are at pH 3.9. It can be observed that the incorporation of PCl₃ into the reaction mixture used for synthesizing the QDs (a toluene solution of GeCl₄ and LiAlH₄ in the presence of tetraoctylammonium bromide) slows the initiation of the reaction down (hours versus minutes) and the resulting material is a lot less concentrated. Although these two sets of particles are at two very different concentrations, the relative natural concentration ratio of their highest efficiencies is the same as seen in Fig. 10.

The plot in Fig. 10 raises two interesting points. First, for both QDs the best devices are produced using films grown from 1/4 of highest concentration; even though the absolute values are different. We propose this is related to solutions at some value below the saturation point, and therefore the deposition rate is more controllable. At lower concentrations presumably the quality of the films is too variable for a suitable cell. This is most likely the reason for the second observation, that the values of devices are produced using films grown from 1/8 appear worse than the more dilute solutions. Given the potential of short-circuiting the cells with incomplete coverage, this is most probably an anomaly.

3.2 Annealing studies

In an effort to densify the films, we have investigated thermal anneal in a stepwise manner. Based upon prior results with silica particles [21, 22] we first annealed the samples to 200°C, which has been reported to remove the residual water between particles without giving them enough energy to move closer [21, 22]. Heating to 400°C has been reported to aid in packing the thin film better [21]. Finally it has been reported that heating to 600°C results in the initiation of melting of the particles resulting in a change of shape. The latter would be considered sintering rather than annealing but could still be considered useful since the quantum dots are randomly placed in the silica particles [18] and removing any space between the particles could create a quantum dot impregnated glass [21–25].

SEM images of the top and cross-sectional images of Ge@SiO₂ films grown from 0.38 mol/dm³ solution after annealing at various temperatures is shown in Fig. 11. There was not much of a difference between any of the samples to be able to definitively say that the temperature has a real effect on the films as seen in the graph of Fig. 12. However, it is worthy to note that in the 600°C film there were a few areas where particles could not be discerned, and it appeared as though there were solid pieces of material instead of aggregates of NPs, as seen in Fig. 11d.

XRD analysis of the films also showed that there is little change in the their crystallinity upon annealing. The XRD data shows silicon (002) at 33.2°, while the silver from the front contacts and aluminum from the sample holder showed a broader peak due to the closeness of the peaks with Ag (111) at 38.2° and Al (111) at 38.5°. Regardless, these annealed films were studied for efficiency improvements, and the results are shown in Table 2. Two sets of samples were run. First, samples were annealed after deposition, but before the front and back contacts were deposited, i.e., Ge@SiO₂|n-Si. Second, thermal annealing was undertaken after the front and back contact were added, i.e., Ag|Ge@SiO₂|n-Si|Au. The data reveals that irrespective of at what stage the annealing was performed, heating to 400°C gave the most enhancement in the cell efficiency. However, the increase in efficiency for the cell with contacts is only 28% over the as deposited film, while the cell without contacts efficiency increases 100% indicating a stronger argument for annealing the solar cells before placing the contacts on them.