8. Summary and Outlook

Transcription

1 8. Summary and Outlook This thesis deals with the synthesis and evaluation of all-nitrogen coordinated complexes of the rare earths such ad Gd, Er, and Y bearing the amidinate and guanidinate class of ligands. Furthermore, their application in MOCVD experiments for the deposition of rare earth nitride thin films has been presented. In addition, first experiments with respect to the materials functional properties were performed showing its enormous potential for future technologies. In chapter 3 the synthesis, characterization, evaluation, and application of five different homoleptic gadolinium amidinate precursors is reported. All compounds have been analyzed in detail with respect to their structural, chemical, and physical properties. The influence of the ligand structure on the precursors properties is investigated and the benefits of slight modifications are presented. The asymmetric substituted compound Gd5, which is the first Gd complex with an asymmetric amidinate ligand system to be reported, exhibits the best thermal properties with respect to its thermal stability as well as its volatility and it was therefore tested in preliminary experiments for the MOCVD of GdN thin films under single source (SSP) conditions and as well with ammonia as reactive gas and additional N-source. Cubic rock-salt structured and essentially C- and O- free GdN thin films on Si(100) substrates were only obtained in combination with ammonia as reactive gas resulting in a preferred orientation along the (200) plane. The films deposited at 750 C under ammonia atmosphere exhibited a columnar morphology with Gd to N ratios close to stoichiometric GdN according to RBS and NRA measurements as well as X-ray diffraction. Further investigations with respect to the materials functional properties need to be performed. Especially the magnetic properties are interesting, as highly oriented GdN thin films were obtained, which might influence the paramagnetic behavior of the material. Nevertheless the data on the homologous series of alkyl-substituted, homoleptic tris-amidinates of gadolinium hold promise for a further development of this kind of compounds as precursors for RE nitride materials. In chapter 4, the comparison of custom-made homoleptic all-nitrogen coordinated precursors [Er{( i PrN) 2 CNMe 2 } 3 ] (Er1) and [Er{( i PrN) 2 CMe} 3 ] (Er2) with a commercially available precursor [Er( i PrCp) 3 ] (Er3) are shown on the example of ErN thin films. All three compounds have been characterized and evaluated with respect to an application in a MOCVD process. The compounds showed suitable thermal properties, 228

2 but it should be mentioned that the commercially available precursor is less thermally stable compared to the applied guanidinate and amidinate based precursor. The precursors Er1 Er3 were successfully applied in a custom-made MOCVD reactor. Precursor Er1 shows a high performance as a single-source precursor (SSP) yielding highly crystalline fcc ErN thin films with an preferred orientation in the ErN (200) plane. However, the films are nitrogen rich with an Er/N ratio of 0.5 having a carbon contamination of about 26 %. Upon using ammonia in the process, no crystalline ErN was obtained but XRD analysis proved the formation of crystalline erbium carbide. In contrast, the application of precursor Er2 in combination with ammonia resulted in thin films with a significantly reduced carbon contamination to a value of about 2 %. Furthermore, the films showed to be almost stoichiometric ErN with an Er/N ratio of 1.1. Interestingly, the crystalline orientation of the thin films can be tuned by choosing the proper precursor. Precursor Er1 shows preferred orientation along the ErN (200) plane while precursor Er2 prefers a growth along the (111) plane. The use of precursor Er3, as an example of a commercially available rare earth precursor, did not yield in ErN thin films at all. Moreover, instead of erbium nitride thin films, erbium oxide thin films were obtained, most probably due to gas-phase reactions of the precursor with the reactive gas ammonia, which could contain traces of moisture. The functional properties of the obtained ErN thin films were investigated in detail. SQUID magnetic measurements revealed the strong paramagnetic behavior of the thin films. Furthermore, Curie temperatures of about T c = 5 K were calculated based on these results, matching well with the reported values. UV-Vis measurements showed a transparency up to more than 90 % in the visible range and a strong absorption in the UV range making the material interesting for application in optical devices. Additionally, the transmission measurements allowed a rough calculation of the bandgap by using the Tauc approach. A value of about 3.5 ev for the direct optical bandgap was calculated, which is just slightly higher than the only one available literature data dated back to Overall, these results demonstrate the limited choice of precursors available with respect to the deposition of rare earth nitride (REN) thin films. There are no commercially available rare earth precursors that have the potential to be used for MOCVD of REN. The limitations of the commercially available precursors are successfully shown. Moreover it was presented that the custom made guanidinate and amidinate based precursors are two powerful precursors for the deposition of ErN thin films and till now are the only precursors that are efficiently applicable for the MOCVD of REN materials. These precursors yield clean and almost stoichiometric ErN thin films. 229

3 But more effort needs to be invested to understand the different behavior of the guanidinate and amidinate class of precursors under MOCVD conditions. Finally concluding, it should be pointed out once more that the results presented in this chapter show the first example of MOCVD of ErN thin films, using tailored custom-made precursors, and furthermore first investigations with respect to functional properties were made giving a deeper insight into the still mostly undiscovered field of REN thin films. In chapter 5 the new material YN is presented. Three different precursors, namely [Y{( i PrN) 2 CNMe 2 } 3 ] (Y1), [Y{( i PrN) 2 CMe} 3 ] (Y2), and [Y{(EtN)( t BuN)CMe} 3 ] (Y3), have been synthesized and characterized in detail. Among them, precursor Y3 is a novel unsymmetrically substituted amidinate complex. In addition, their properties are compared to a commercially yttrium containing precursor Y4. All compounds have been successfully applied in a custom made MOCVD reactor using either the SSP approach or ammonia as reactive gas. The application of precursor Y1 in a single-source process yielded highly crystalline fcc YN thin films orientation along the YN(200) plane. The films had an Y/N ratio of 0.4 and a carbon contamination of about 27 %. Adding ammonia to the deposition process did not yield crystalline YN thin films but XRD analysis viewed the formation of crystalline yttrium carbide. This result was further proven by XPS. In contrast to this, the application of precursor Y2 as well as Y3 in combination with ammonia resulted in thin films with a drastically reduced carbon contamination to a value of about 1 %. Furthermore, the films showed to be yttrium rich YN with an Y/N ratio of 1.4. Interestingly, by choosing precursors Y2 and Y3 for the deposition of YN materials, the obtained thin films showed to be highly crystalline with a preferred orientation along the YN (111) plane. Furthermore, it should be noted that the YN showed to be less sensitive towards oxygen and moisture compared to the GdN thin films presented in chapter 3. The use of precursor Y4, as an example of a commercially available rare earth precursor, did not yield in YN thin films at all. Moreover, instead of yttrium nitride thin films, yttrium oxide thin films were obtained, most probably due to gas-phase reactions of the precursor with the reactive gas ammonia. SQUID magnetic measurements revealed the diamagnetic nature of the thin films, which was expected, as Y 3+ ions do not contain unpaired electrons that could cause paramagnetism. UV-Vis measurements showed a transparency up to more than 80 % in the visible range and a strong absorption in the UV range making the material interesting for application in optical devices. Additionally, the transmission measurements allowed a rough calculation of the bandgap by using the Tauc approach. A value of about 2.4 ev for the direct optical 230

4 bandgap and 1.8 ev for an indirect optical bandgap were calculated. The value obtained for the direct bandgap is in good agreement with the literature. For the indirect bandgap, as no experimental data are available, it is difficult to quote whether these results are reasonable or not. It should be pointed out, that the results presented in this chapter show the first example of MOCVD of YN thin films using tailored custom-made precursors. The functional properties presented in this chapter, are a contribution to the research activities on the still unexplored topic of MOCVD of REN thin films. The deposition of YN via MOCVD offers the possibility for further research activities. YN is a nonmagnetic semiconductor that e.g. could be used as a matrix for doped materials like manganese or other rare earth atoms, offering the possibility to apply these materials in spintronic devices. Furthermore, the use of YN as a buffer layer between silicon substrates and rare earth nitride thin films could favor the deposition of epitaxial rare earth nitride thin films. The lattice mismatch of YN towards GdN (3 %) and ErN (1 %) is very small, which is a good precondition for the growth of epitaxial layers. In chapter 6 AlN as an alternative capping material is presented. AlN was deposited using a simple aluminum amide precursor in combination with ammonia. The benefit of AlN compared to e.g. copper is its transparency, offering new possibilities for characterizing the underlying materials. Amorphous AlN thin films were obtained on top of GdN. SEM analysis shows the ability of AlN to act as a protective capping layer. Densely packed thin films were obtained and a distinct interface is observed. XPS analysis shows that the oxygen contamination within the GdN thin film is in the expected region for the custom build reactor applied in this thesis and that no AlN diffused inside the GdN layer lie it was the case for a copper capping layer. Also the presence of a distinct and sharp interface was proven. AlN is a suitable capping layer for rare earth nitride thin films and has a high potential for further investigations. However, more effort needs to be devoted to completely control the growth of AlN on rare earth nitrides. In chapter 7 two rare examples of heteroleptic tetra coordinated rare earth complexes are presented. This is one of the rare examples of rare earth metal centers having the coordination number four without any solvent molecule coordinating to the metal center. The compounds were fully characterized and their single-crystal structure is reported, showing a distorted tetragonal coordination of the metal center. Furthermore, the compounds have been analyzed with respect to their application in as precursors for MOCVD experiments. TGA measurements were performed showing multiple step 231

5 decomposition with high residual masses. The compounds do not seem to have a sufficient thermal stability for MOCVD application. However, both compounds contribute to the organometallic chemistry of the rare earths giving additional information on the metal-ligand interplay. Altogether, this thesis is a significant contribution to the field of MOCVD of rare earth nitride thin films combined with precursor development for this kind of materials. As the rare earth nitrides are almost unexplored, the results presented in this thesis are very valuable and give a good foundation for future research activities. However, there are still challenges in this field leaving space for further activities. The biggest challenge is the oxygen content within the deposited thin films. Although protective capping layers are used, the oxygen content is typically in the order of 5 10 at%. This is most probably attributed to the fact that a custom build MOCVD reactor is applied for the deposition process. The transfer of the process from the custom build reactor to a commercially available system, which shows a higher level of leak tightness, could help to deposit pure and oxygen free REN thin films. In combination with a sufficient capping material, like e.g. the AlN resented in this thesis or GaN as an alternative capping material, this could be a powerful combination bringing the REN technology one step closer to prototype device manufacturing. Another challenge that needs to be approached is the investigation of the reaction mechanism for thin film growth. Till date it is not clear, why the amidinate and the guanidinate class of precursors show a significant difference in their film growth behavior, although the differences in the ligand structure are small. One possibility to gain information on the growth mechanism could be the use of in-situ monitoring using a mass-spectrometer attached to the MOCD reactor. The collected fragments could give a hint towards the decomposition and/or reaction mechanism and the resulting byproducts. The growth of epitaxial REN thin films is still a great issue due to the absence of substrates with a low mismatch between the substrate and the REN thin film. Partially this challenge was addressed by using materials like AlN or GaN as buffer layers, which reduce the lattice mismatch between the substrate and the REN layer. The growth of epitaxial or single crystalline RENs is interesting with respect to device manufacturing. Epitaxial films usually do not suffer from major impurities and defects and thus their functional properties can precisely be adjusted. However, epitaxial REN thin films have not been realized using MOCVD. 232

6 Another scope for the future might be the fabrication of diluted magnetic semiconductors (DMS) like e.g. GaGdN or AlGdN using MOCVD. These materials combine the semiconducting properties of the group-iii nitrides with the unique magnetic properties of the RENs. As this thesis has shown, the fabrication of both materials independently is possible using the custom build system, thus the next step could be the first attempts of the deposition of these alloy thin films using MOCVD, as it is not reported yet. 233