Rational Process Development Based On Physical Understanding And Holistic Linking: Effectively Enhancing Electrical Properties Of Magnetron Sputtered Doped ZnO Films

The steady development of high-performance materials,i.e.,effective improvement of properties,is one of the central challenges in materials studies.A rational approach to this development of enhanced properties has,in most cases,been difficult for those materials prepared by far-from-equilibrium processing methods(e.g.,vapor deposition at low temperatures,quenching from high temperatures,rapid solidification,etc.).Thus an empirically-based approach to process development — costly yet ineffective — is commonly adopted,especially for those films prepared by low-temperature vapor deposition.The goal of this study is to develop a physicsbased rational approach for process development and to improve effectively selected properties of some thin-film materials.To this aim,we have selected deliberately magnetron sputter deposition and used doped zinc oxides [i.e.,Ga-doped ZnO(GZO)and Al-doped ZnO(AZO)] as a model system,on the basis of its importance in thinfilm-based industries and their potential for such applications as flat-panel displays,photovoltaics,and smart windows.By systematically investigating magnetron sputtered GZO films,through physical understanding combined with holistic linking,we succeeded in developing a rational approach to process development and improving effectively their electrical properties.In particular,the whole process of our approach consisted of four steps: preliminary physics-based understanding –– holistic linking –– in-depth physical analysis –– sound prediction/validation.First of all,pattern identification techniques were used — based on thorough literature research and data analysis — to pick out,qualitatively,two key process parameters(i.e.,discharge voltage and substrate temperature).Subsequently,a single-parameter experimental approach was carried out to systematically examine the effects of these two process parameters,one by one,on the structure and electrical properties of GZO thin films.We found that controlling the defect generation process(i.e.,by decreasing the discharge voltage)was more important than enhancing the defect annihilation process(i.e.,by rising the substrate temperature).On the basis of these studies,we prepared 60 GZO samples,by varying systematically four process parameters(i.e.,discharge voltage,substrate temperature,thickness,and substrate position),developed a series of characterization and quantification techniques for hierarchical structural features,and effectively obtained quality data needed for holistic linking — that is,medium in size but include those key processing and structural features adequately.The artificial neural network method was then applied to these data to establish a “processing-structure-property” holistic link for these GZO films through retrospective reasoning,sequentially,from properties to structure and then to process.This holistic link captures the key structural descriptors [i.e.,(0002)lattice parameter and relative intensity of LO band)] accurately and processing descriptors(discharge voltage and substrate temperature,with the former being more important)that have prevented these GZO films from steady improvement of electrical properties.Furthermore,we rationalized the effects of the key processing descriptors,at the atomic scale,on the structural evolution of these GZO films,in terms of an “activation-relaxation” model and a crystal-chemical analysis of mobility of displaced atoms.It is found that insufficient relaxation — i.e.,Ga atoms displaced from regular tetrahedral lattice sites were unable to fully recover from the octahedron interstices — was the rate-limiting atomic step that restricts the improvement of electrical properties.Therefore,it is necessary to avoid “overactivation” for the Ga atoms — i.e.,by reducing discharge voltage to decrease the energy of negative oxygen ions and thus preventing the Ga atoms from being displaced permanently into octahedron interstices.The follow-up experiments,based on this prediction from the above atomistic understating,realized effective improvement of electrical properties.Particularly,the conductivity of our GZO films can be improved to ~3400 S/cm,which is ~27% higher than the highest value reported for the sputtered GZO films(<600 nm)with low Ga content(?2 at.%).This rational approach to process development enables us to generate a medium size of quality data,to capture essential structural and processing descriptors for holistic linking,to formulate physics-based atomistic understanding,and thus provide reliable predictions.The chain of “preliminary physics-based understanding –– holistic linking –– in-depth physical analysis –– sound prediction/validation” developed in this work can be extended to improve properties of other thin-film materials in a similarly effective way.