«c SIMON LINDBERG, 2013 Cover: To the left is an image of a mono crystalline silicon solar cell , In the middle is an image of three samples of ...»
However the Ni-H bonds that were of particular interest have a higher excitation energy, the result from the measurements in the energy region were these peaks should appear is found in Figures 3.12 and 3.13. No peaks were found in this interval for the Mg2.0 NiH4 samples, therefor only the results from the Mg2.2 NiH4 and Mg2.4 NiH4 samples are included.
The plots in Figures 3.12 and 3.13 show no signs of Ni-H bonds in the amorphous samples, it is ﬁrst in the crystalline samples that the Ni-H bonds appear. No peaks from OH-bonds or other impurities are not present in any of the samples.
In Tables 3.10, 3.11 and 3.12 summaries of the obtained peaks compared to the literature data is listed for each composition Table 3.10: Comparison of obtained Raman-spectras for Mg2.0 NiH4 with reference .
Figure 3.15: The Raman spectra of oxidized amorphous Mg2.
4 NiH4 compared to the unmodiﬁed sample.
In Figure 3.14 the Ni-H bonds disappear in the oxidized samples, the spectra also looks more like the amorphous samples in the low energy region. But no OH-groups or other oxides are visible in the spectra. However in Figure 3.15 a new peak appears in the spectra from the oxidized sample, which seems to represent bonds not seen in any of the previous samples.
Finally one last attempt to try and detect hydroxide groups in the samples was performed, which can be found in Figure 3.16. The measurements were allowed to run for a full hour to be able to detect low amounts of OH-groups and focused only on the narrow region where peaks corresponding to OH-bonds are visible. To reduce the background noise from the substrate the thin ﬁlms were deposited on cleaned silicon substrates. Both measurements on crystalline and oxidized Mg2.4 NiH4 -samples were performed.
Figure 3.16: OH Region of the Raman spectra of Mg2.
4 NiH4 -thin ﬁlm deposited on silicon substrate.
In Figure 3.16 no difference can be seen between the clean silicon substrate and the thin ﬁlms which would indicate the presence of hydroxide groups.
4 Discussion Let’s start the discussion with the result from the sample depositions. All samples have a uniform and similar color as deposited, independent of magnesium content. However upon heating, the samples with excess magnesium content, Mg2.2 NiH4 and Mg2.4 NiH4, ﬁrst turn orange at 220C.
Before 220C no color change is seen. There is little difference in this orange color between the samples, yet no crystallization has occurred because the XRD data reveals that the samples are still amorphous. So what is the reason for this color change? One explanation could lie in the size of the nanocrystals; upon heating the nanocrystals could grow in size, thus increasing the transmission. However that should be visible in the Raman spectra, which is not the case in the results presented in this thesis and just a change in crystallite size shouldn’t change the magnitude of the band gap. The second explanation is that upon heating the crystal structure in some of the crystallites change from monoclinic to cubic, so the orange ﬁlm consists of a mix of monoclinic and cubic crystallites. This should explain why the band gap of the orange ﬁlms lies between the as deposited and crystallized samples. However to investigate this further TEM-measurements are suggested.
To achieve crystallisation of the thin ﬁlms the temperature was increased to 290C. At this temperature the Mg2.2 NiH4 and Mg2.4 NiH4 samples change permanently to the same bright yellow color. So what happened to the Mg2.0 NiH4 during annealing? Well not much, the samples might have turned slightly more transparent but the change is very subtle. However when synthesizing Mg2 NiH4 nanoparticles excess of magnesium is standard practice, so this behavior is not unexpected.
The XRD-pattern for the yellow samples ﬁts well with the reference pattern from nanoparticles with the cubic Mg2 NiH4 structure. However there are some discrepancies, most notable is the peak at 19 degrees which is absent in the reference pattern. There are also slight shifts in the location of the other peaks compared to the reference data and when compared to each other, which could indicate that the dimensions of the unit cells might differ slightly between the samples and the reference.
To sum up the deposition of the thin ﬁlms we can conclude that uniform samples were successfully synthesized. Upon heating, the samples with excess magnesium ﬁrst turned into an amorphous, orange phase at 220C and at 290C the same samples crystallized. The Mg2.0 NiH4 sample didn’t change signiﬁcantly upon heating.
One interesting phenomenon that was discovered in the annealing process was that some samples changed color when heated. The crystalline samples are yellow in room temperature while at higher temperature they gradually turned to a darker red color. This color change was seen as a decrease in the transmission spectra, while the reﬂection spectra only changed marginally, leading to a narrowing of the band gap. This thermochromic effect has not previously been found in litterature for Mg2 NiH4.
Narrowing of the band gap is seen in silicon as well at higher temperatures, only in the magnitude of a couple of meV . So what could lie behind this big change? To investigate if this could be due to a structure change XRD measurements were done on a heated sample and compared to the result at room temperature. In the diffraction pattern from the heated sample the most striking difference is a broad peak at 17 degrees, however the nature of this peak was hard to identify. The remaining peaks were only slightly shifted and cannot explain the big change.
Unfortunately there was a considerable background from the sample holder, which coincided with the peaks from the samples, so the intensities of the peaks could not be investigated.
Are there any other mechanisms that could explain the thermochromic effect? It could be linked to desorption of hydrogen, however the effect is reversible so it does not seem probable. Instead this could be due to changes in the material which are not detectable with XRD. The position of hydrogen in the material cannot be investigated directly with XRD so it is possible that the hydrogen atoms changes position upon heating which might lead to a change in the Mg-H interaction which is crucial to the band gap. To investigate this further it would be interesting to use a hydrogen sensitive technique to investigate the shape of the NiH4 complexes, also temperature dependent Raman measurement would be interesting because if the shape of the Ni-H bonds affects the band gap this would be seen in the Raman spectra. Not only heating is interesting, spectrophotometry measurements on cooled samples would be of interest to see if the band gap changes.
Moving on to the electrical measurements conducted in the thesis, the Mg2.4 NiH4 samples shows very high resistivities, matching that of intrinsic silicon, while the as deposited Mg2.0 NiH4 has a signiﬁcantly lower resistivity, which then is drastically increased upon heating.
These results differs signiﬁcantly from the results obtained by Enache et al, they describe Mg2.0 NiH4 as a heavily doped semiconductor with low electrical resistivity in the range of 0.1 ohm*cm.
What could be the reason behind this big difference? First of all one must consider that their samples were slightly different with a palladium capping layer to prevent oxidizing the sample and to enable hydrogen loading after deposition. It is not impossible that this capping layer could affect the electrical properties of the material. They also detect large amounts of oxygen present in their sample but this does not seem to affect conductivity. Could the reason behind these relatively high resistivities also explain the strange behavior of the Hall-measurements and inability to determine the sign of several important properties? During the course of this thesis several attempts were done with different Hall-devices to try and measure important physical properties such as mobility and majority charge carriers of the material which would give vital knowledge about the material. Such experiments have been performed before in the previously mentioned paper by Enache , where Mg2 NiH4 thin ﬁlms were investigated with reasonable results.
So what could be the reason behind the high resistivities in the Mg2.4 NiH4 samples? The ﬁrst thing to discuss is the extra 20% magnesium atoms that do not ﬁt into the crystal structure they can’t disappear and must be ordered in the structure some way. A sign of this could be the unidentiﬁed peak at 19 degrees in the GIXRD diffraction patterns. This could peak be assigned to metallic magnesium nanoparticles or other magnesium compounds present in the samples.
However more peaks should be present if there were a substantial amount of foreign substances in the samples. It is also worth noting that the peak at 19 degrees disappears in the XRD pattern, which could indicate that this peak is related to a more long ranging structure since the planes of higher indices also ’disappear’ in the XRD pattern. However both metallic magnesium nanoparticles and unwanted magnesium compounds would probably increase the resistivities in the samples.
To discuss the other peaks in the GIXRD pattern, one can notice that they are shifted slightly compared to the literature data. This difference could indicate that the unit cell in the thin ﬁlm didn’t have exactly the same side length as in the powder form, which would indicate that there might be some kind of strain in the thin ﬁlms. Also there are some small differences in the diffraction patterns from Mg2.2 NiH4 and Mg2.4 NiH4 which indicates that the change in unit cell parameter is depending on the amount of extra magnesium added in the process.
The resistivity measurements on the Mg2.0 NiH4 samples does indicate that the extra magnesium atoms actually increases the resistivity in the samples. The as deposited sample has a resistivity that is much lower that that of the as deposited Mg2.4 NiH4 samples which indicates that the scattering defects in the sample is much lower with lower magnesium content. However the resistivity are still a factor 103 higher than in previous results. Since the Mg2.0 NiH4 samples were amorphous no XRD-pattern could be recorded for these samples. The results from the annealed Mg2.0 NiH4 samples must also be mentioned, the low resistivity only applies for the as deposited samples, when annealed the resistivity increases drastically. The reason why this occurs could be due to a change in crystallite size upon heating, which should be seen in the Raman spectra as sharpening of the peaks which is demonstrated in Raman spectra on silicon samples with different crystallite sizes . Another possibility is that oxides are formed on the surface, creating an insulating layer, but no peaks from such a layer could be detected which might indicate that it’s very thin.
To sum up the discussion so far about the connection between the Hall-measurements, resistivity measurements, XRD-result and impurities there is a big difference in resistivity between the as deposited Mg2.4 NiH4 and Mg2.0 NiH4 samples, which could be explained by the presence of extra magnesium atoms. Still however the resitivities are much higher than expected from previous papers even with the correct amount of magnesium. Could these magnesium atoms also explain the strange Hall-measurements? Maybe, if the number of charge carriers is too small and the scattering defects are too many then they will be recombined very quickly, thus reducing the amount of charge carriers and the current, making the Hall-current too weak to be measured. Hall-measurements were performed on Mg2.0 NiH4 samples as well with the same result as Mg2.4 NiH4, so this indicate that there are other impurities in the material as well.
It could also be that there are impurities in the sample that originate from the deposition which affects the conductivity without acting as a dopant. For example a small portion of air is still in the load lock when the carrier is transferred from the load lock to the processing chamber.
Consequently oxygen, carbon dioxide and other residual gases are most probably present in the processing chamber during the deposition and will be incorporated into the ﬁlms and bind with the nickel or magnesium atoms thus affecting the structure and physical properties of the material. During the autumn, experiments with Rutherford Backscattering Spectrometry (RBS) measurements were performed to try and measure the amount of impurities in the deposited ﬁlms. The result showed that carbon was present to a large extent in the samples and thus can be assumed to affect the conductivity properties. These data will be presented later in a paper by Raskovic-Lovre.
Another possible explanation of the presence of impurities in the sample could lie in the targets used in the sputtering process. The purity and composition of these were discussed previously in the thesis, it could be so that there are impurities such as silicon, carbon and other metal atoms which are sputtered from the targets which also affects the compositions. A quick EDS measurement was performed on the ’Mg2 Ni’-target and the small splinter that was analyzed showed that the composition is only ’Mg1.55 Ni’ and that other unwanted atomic species also was present in the target.
To summarize the discussion regarding the properties of the sputtering device, one could say that the sputtering device used in this thesis was a black box, with many unknown parameters which affects the composition of the deposited samples. It might not seem important if a few oxygen atoms go into the samples or not, but when discussing impurities in semiconductors it is important to keep in mind that a very small amount could drastically change the conduction properties. In silicon wafers for solar cell production a purity of 99.9999% is required for solar cell grade silicon. For the magnesium and ’Mg2 Ni’-targets the purity is only guaranteed to 99.99% which together with the presence of unwanted gases in the processing chamber might indicate that the purities of the synthesized ﬁlms are too low to be suitable for semiconductor production.