Plasma-enhanced atomic layer deposition of binary metal oxides as gas barrier layers on polymers; English

Author
Maximilian Gebhard
Year
2018
Abstract & Cover

In this thesis, the deposition of inorganic oxides, namely Al2O3 , SiO2 and TiO2 by means of PEALD on the two polymers PET and PP was investigated regarding the application of these materials as GBLs for improved packaging materials. The dis cussed experiments and results demonstrate a contribution toward the understand ing of the fundamentals of the respective PEALD depositions, thereby comprising studies on precursor development and the in uence of precursor chemistry on the growth of GBLs on polymers, the combination of PECVD and PEALD as well as correlation of material properties toward barrier performance. In the  rst part, the growth of TiO2 during PEALD on PET was investigated in a comparative study, employing the commercial available precursor TMDAT and a newly developed mixed amino-alkoxide-Ti-precursor, TDMADT. The resulting out come and more details on the precursor synthesis, characterization and process development were published in 2016 and reproduced in chapter 2, page 59ff. These studies were performed based on the fact that the TDMAT precursor represents, among many bene cial properties such as volatility and reactivity, limited ther mal stability. It was shown earlier that the rational precursor development of Ti precursor compounds yields precursors with  ne-tuned thermal properties, mak ing these compounds promissing for both PEALD and MOCVD application. As the here investigated growth on polymers sets two additional limitations, i.e. the reactivity toward a polymer substrate and the sensitivity of these substrates toward heat treatment, the motivation for further systematic precursor development is on hand. The new precursor compound TDMADT was obtained from substituting one of the amide-ligands of TMDAT using dimethylamino-2-isopropoxide, thereby creating a  ve-fold coordinated Ti-complex with promising thermal properties and suf cient reactivity for PEALD processes. In the comparative study, an ef cient PEALD process using TDMADT, exhibiting a simliar GPC as TDMAT ( 0.9 Å cycle-1 vs. 1.0 Å cycle-1) was developed and the obtained thin  lms revealed high qual ity in terms of low roughness (rms = 0.2 nm) and nearly perfect stoichiometry as found from RBS/NRA and XPS studies. Employing in-situ QCM, the growth of the two processes, using either TMDAT or TDMADT, was monitored and from the relation of the frequency shifts and possible molecular fragments of the different precursors, it was found that the new precursor chemisorbs to the surface via cleav age of two amide-ligands, while the more stable and more bulky dimethyl-amino-2- propanolato ligand stays intact and faces, most likely, upwards with respect to the substrate surface plane. Furthermore, both TDMAT and TDMADT were used to de posit TiO2 thin  lms as GBLs on 23 µm PET foil with different thickness and the re spective coatings were investigated regarding their barrier performance in terms of OTR. For both compounds, thin  lms from applying 150 cycles (15 nm for TiO2 from TDMAT, 14 nm TiO2 from TDMADT) exhibit OTR values below 0.2 cm3 m−2 day−1 , thereby showing a drop in OTR by two orders of magnitude compared with the uncoated PET foil. A second major part of this thesis dealt with a combinatorial approach to fabri cate GBLs by means of PECVD and PEALD. In chapter 3, page 79ff., this approach was discussed for both a seeding and a capping route. While the approach to com bine PECVD and PEALD with respect to GBL application was investigated by several groups, which was outlined in the chapters 1 and 3, the here performed experiments were designed to investigate the in uence of ultrathin PEALD  lms (1 nm to 5 nm) on the overall barrier performance with respect to macro defect density and bar rier performance in terms of OTR. This approach is unique in terms of the chosen barrier layer thickness and the used substrates, which were of only 23 µm thick ness. From the seeding approach, in which SiO2 PEALD thin  lms of 1 nm to 5 nm thickness served as improved surface for subsequent PECVD growth of 15 nm SiOx , a signi cant impact on the latter measured defect density was found. A PEALD thin  lm of only 1 nm thickness decreased the defect density by 50%, while a 5 nm PEALD grown seeding layer caused a drop in OTR by two orders of magnitude, in dicating a strong in uence of the PEALD grown seeding layers on the overall barrier performance of the combined GBL materials. The capping approach was investigated for four different systems, consisting of the PECVD grown materials SiOxCyHz or SiOx and the PEALD grown materials Al2O3 or SiO2 . Similar to the seeding approach, the PECVD coating thickness (5 nm) was chosen to exhibit a poor barrier performance to allow the investiga tion of a potential in uence also from ultrathin PEALD thin  lms of only 1 nm to 5 nm thickness. PEALD grown cappings of Al2O3 on SiOxCyHz were found to result in a drastically reduced defect density, being ≤2 defects mm−2 . Furthermore, these cappings decreased the OTR substantial from > 62.0 cm3 m−2 day−1 down to 15.0 cm3 m−2 day−1 (1 nm Al2O3 , 6 nm GBL in total) and 7.8 cm3 m−2 day−1 (5 nm Al2O3 , 10 nm GBL in total). Highly encouraging results were found for PEALD grown Al2O3 cappings on PECVD grown SiOx , where an ultrathin capping of only 1 nm suppressed the formation of defects completely. Consequently, the defect den sity was found to be zero macro defects mm−2 for both the 1 nm and 5 nm capping. While SiO2 cappings on SiOxCyHz showed a less strongly pronounced impact on the defect density with 156 macro defects mm−2 for the 5 nm, the barrier perfor mance found to be affected stronger, as OTR dropped from >15.0 cm3 m−2 day−1 to 0.5 cm3 m−2 day−1 . Regardng barrier performance, similar results were found for SiO2 capping on SiOx , where a 5 nm cappings caused a stronger decreased OTR of 38.6 cm3 m−2 day−1 . The defect density for SiO2 cappings was found to be strongly decreased compared to the bare PECVD grown SiOx thin  lms on PET, with values of 23 macro defects mm−2 (1 nm SiO2 , 6 nm GBL in total) and 1 macro defect mm−2 (5 nm SiO2 , 10 nm GBL in total). Here, it was further observed that for >90% of the detected defects, macro defect formation takes place preferentially at sites contam inated with dust, while larger areas without surface features exhibit only negligible amounts of detectable defects, indicating the superior properties of thin  lms from PEALD with respect to GBL applications. In chapter 4, the fabrication of multilayers for a potential GBL application was in vestigated with respect to stackings of Al2O3 and SiO2 on PP and their mechanical properties in terms of residual stress. The idea followed the one from the seed ing/capping approach (chapter 3) in terms of using a polymer substrate exhibiting rather poor barrier performance against oxygen gas and to investigate the in u ence from thin coatings, deposited by PEALD. In these studies, both the in uence from precursor chemistry on the growth of Al2O3 thin  lms and a potential in u ence on  lm growth of Al2O3 and SiO2 from an oxygen plasma pre-treatment of the polymer were investigated. For these studies, the recently reported and novel Al-precursor [3-(dimethylamino)propyl]dimethyl aluminum(III) (DMAD) was com pared with TMA to grow Al2O3 via PEALD. From in-situ QCM studies employing functionalized QCM crystals, the growth of Al2O3 was monitored and could be ex plored with respect to the growth modes in dependency of the applied Al-precursor. For this, QCM crystals were coated with a top-layer of spin-coated PP (scPP), thereby providing an excellent model layer for industrially applied PP foil. From the inves tigations using the scPP-QCM crystals, a  rst indication of different growth modes for the two Al-precursors TMA and DMAD was obtained. While the growth of Al2O3 on scPP using TMA was decreased in terms of lower amounts of deposited mass per cycle than found for a reference deposition on an uncoated QCM crystal, the DMAD process revealed similar MGPCs for both crystals. It was further observed that the  rst cycles of the TMA process most likely comprise a rather strong contribution from etching, which could explain the hindered growth and hints toward a more diffusive interface between the scPP and the inorganic oxide. For DMAD, such etching contribution was not found, thus forming, most likely, a more abrupt inter face with the substrate. Al2O3 thin  lms from DMAD of 10 nm thickness exhibited no gas barrier when applied on biaxial oriented PP (BOPP), while Al2O3 from TMA improved the OTR by a factor of 3.6. For SiO2 of the same thickness, an improve ment of the barrier performance by the factor 23.3 was found, and as for TMA, QCM hints here also toward a more diffusive interface as etching during the  rst cycles must be taken into account. Considering the  ndings from in-situ QCM experi ments, a more diffusive interface could be bene cial in terms of improved barrier performance, which could be explained by a better sticking of the inorganic thin  lm on the substrate, due to a gradual increase of the inorganic component. Using Si-sensor chips, the residual stress of the binary oxides was investigated and SiO2 thin  lms exhibited compressive stress over a broad thickness range with a lower value of −250 MPa for thicknesses ≥45 nm. Al2O3 thin  lms from TMA showed tensile stress with a lower value of 125 MPa for  lms ≥45 nm. In contrast, Al2O3 from DMAD showed a transition in residual stress from compressive (≤30 nm) to tensile (≥60 nm), with a 45 nm sample exhibiting residual stress close to 0 MPa. This change in residual stress for thicker  lms could be explained with formation of crystallites, inducing additional stress for thicker  lms. As GI-XRD revealed no evidence for crystallization, these potential crystallites must be nanoscopic. To in vestigate the in uence from precursor choice on the properties of multilayers of Al2O3 and SiO2 , dyads of 5 nm SiO2 and 5 nm Al2O3 , either grown from TMA or DMAD, were investigated regarding their barrier performance and residual stress. Here, the combination of SiO2 /Al2O3 (DMAD) exhibited a barrier improvement fac tor of 22.4 and a residual  lm stress of 114 MPa (tensile), while the combination with Al2O3 from TMA showed an improvement factor of only 4.8 and a residual stress of −113 MPa (compressive). These  ndings are encouraging, as the barrier perfor mance seems to be dependent to a major degree on the SiO2 coating while the Al2O3 coating can be used to tailor the residual stress of GBL materials, which is of high importance for the encapsulation of  exible display using OLED technology.

Source of Information
Anjana --> Riikka
University
Ruhr University Bochum
(Bochum, Germany)
External Link
Read Thesis
LinkedIn
linkedin invite