Dimethylether is a non-toxic liquefied gas, which is projected to become one of the fundamental chemical feedstock in the future. Dimethylether can be produced from syngas via a two-step (indirect) process that involves synthesis of methanol by hydrogenation of CO/CO2 over a copper based catalyst and subsequent dehydration of methanol to DME over an acidic catalyst. Alternatively, DME can be produced in an one-step (direct) process using a hybrid (bifunctional) catalyst system that permits both methanol synthesis and dehydration in a single process unit. In the present research work the production of DME has been studied by applying both the indirect and direct processes. Firstly, the methanol synthesis and methanol dehydration reactions involved in the indirect process have been studied separately. Afterwards, these two reactions have been combined in the direct DME production process by using a hybrid catalyst comprising a methanol synthesis and a methanol dehydration component. The methanol synthesis by CO2 hydrogenation has been investigated over commercial and home-made CuO/ZnO/Al2O3 catalysts with the aim to identify optimal experimental conditions (CO2:H2 ratio, flow rate, temperature) that could be then used in the direct conversion of CO2/H2 mixtures into methanol/DME. Obtained results reveal that the conversion of CO2 and the yields of reaction products (CH3OH and CO) increase when the concentration of H2 in the feed and the reaction contact time are increased. It was found that both Cu+/Cu0 species are important for the conversion of CO2/H2, although the presence of Cuo seems to be more important for selectivity/yield of methanol. The stability of the CuO/ZnO/Al2O3 catalyst has been also investigated. It was observed that the main reason for the deactivation of catalyst is the water produced via the methanol synthesis and reverse water gas shift reactions. However, the catalytic activity and products selectivity were recovered slowly to their original levels after applying a regeneration procedure, indicating that deactivation by water is reversible. The dehydration of methanol to dimethylether (DME) has been investigated over a range of catalysts including alumina, silica-alumina and zeolites with different physicochemical characteristics. The effects of temperature and the presence of water vapour in the feed on catalytic performance have been studied in detail. The reactivity of catalysts has been evaluated by determining the reaction rates per gram of catalyst per acid site (total: Brönsted+Lewis) and per Brönsted/Lewis mole ratio. In addition, the reaction mechanism has been investigated over a selected catalyst, with the use of transient-MS and in situ DRIFTS techniques. Results obtained for alumina catalysts show that the catalytic activity and selectivity are determined to a large extent by the textural properties, degree of crystallinity and total amount of acid sites of catalysts. In particular, the methanol conversion curve shifts toward lower reaction temperatures with an increase of specific surface area. However, the enhanced catalytic activity of high-SSA samples cannot be attributed solely to the higher amount of surface acid sites, implying that the reaction rate is determined to a large extent from other parameters, such as textural properties and degree of crystallinity. Results of mechanistic studies indicate that interaction of methanol with the Al2O3 surface results in the formation of two kinds of methoxy groups of different adsorption strength. Evidence is provided that DME evolution is associated with methoxy species that are weakly adsorbed on the Al2O3 surface, whereas more strongly held species decompose to yield surface formate and, eventually, CH4 and CO in the gas phase. Results obtained over zeolite catalysts show that catalytic performance depends on the topology of zeolites due to differences in micropore structure and Si/Al ratio as well as on the number, strength and nature of active acid sites. The activity of zeolite catalysts for the methanol dehydration to DME follows the order ZSM-5 > Ferrierite > Mordenite ~ Beta ~ USY > H-Y. The strong Brönsted acid sites of ZSM-5 zeolites with relatively high Si/Al ratio represent the most active sites in methanol dehydration to DME reaction. However, the overall reactivity of the ZSM-5 zeolites is also affected by the balance of the Brönsted to Lewis acid sites. The activity of Beta and USY zeolites is determined by both Lewis and Brönsted acid sites. The moderate/low reactivity of Ferrierite, Mordenite and H-Y zeolite are determined by the abundant Brönsted acid sites of relatively weak/moderate strength. The direct CO2 hydrogenation to methanol/DME has been investigated using admixed catalysts comprising a methanol synthesis (commercial copper based catalyst: CZA1) and a methanol dehydration component (different alumia/zeolite catalysts: γ-Al2O3, ZSM-5, W/γ-Al2O3, USY(6), Ferrierite(10)). It has been revealed that the conversion of CO2 is always lower than the corresponding equilibrium values predicted by thermodynamics, indicating operation in the kinetic regime. The nature of the methanol dehydration component of the admixed catalysts was found to be important for both CO2 conversion and methanol dehydration. In particular, DME selectivity/yield, depends strongly on the nature of acid sites (both Lewis and Brönsted) as well as the textural (meso/macro porosity) and topological properties of methanol dehydration component of the admixed catalysts. The yield of DME obtained at a temperature of 250oC decreases following the order CZA1/ZSM-5, CZA1/USY(6) > CZA1/Ferrierite(10) > CZA1/ W/γ-Al2O3 >> CZA1/γ-Al2O3. The long-term stability experiments conducted over selected bifunctional catalytic systems revealed that the catalysts deactivate with time-on-stream, mainly due to water produced via methanol synthesis, methanol dehydration and reverse water gas shift reactions. In case of the CZA1/ZSM-5 admixed catalyst the catalytic activity and products selectivity were almost recovered after regeneration indicating that deactivation by water is reversible. / --

Στη παρούσα Διατριβή διερευνήθηκε η χρήση υλικών υψηλής διηλεκτρικής σταθεράς (high-k) ως οξειδίων ελέγχου σε διατάξεις παγίδευσης φορτίου τύπου MONOS (Μetal-Οxide-Νitride-Οxide-Silicon). Τα οξείδια που εξετάστηκαν ήταν το HfO2, τo ZrO2 και το Al2O3. Η ανάπτυξή τους πραγματοποιήθηκε με χρήση της μεθόδου εναπόθεσης ατομικού στρώματος (ALD). Οι ιδιότητες των δομών μνήμης μελετήθηκαν συναρτήσει: (α) των πρόδρομων μορίων της εναπόθεσης για τα HfO2 και ZrO2, (β) του οξειδωτικού μέσου της εναπόθεσης για την περίπτωση του Al2O3 και (γ) της επακόλουθης ανόπτησης. Η ηλεκτρική συμπεριφορά των δομών εξετάστηκε με την κατασκευή πυκνωτών τύπου MOS. Τα υμένια του HfO2 αναπτύχθηκαν επί διστρωματικής στοίβας SiO2/Si3N4 με (α) αλκυλαμίδιο του χαφνίου (ΤΕΜΑΗ) και Ο3 στους 275 oC, και (β) κυκλοπενταδιενύλιο του χαφνίου (HfD-04) και Ο3 στους 350 οC. Ομοίως, τα υμένια του ZrO2 αναπτύχθηκαν επί διστρωματικής στοίβας SiO2/Si3N4 με: (α) αλκυλαμίδιο του ζιρκονίου (ΤΕΜΑΖ) και Ο3 στους 275 oC και (β) κυκλοπενταδιενύλιο του ζιρκονίου (ZrD-04) με Ο3 στους 350 oC. Ο δομικός χαρακτηρισμός, για το HfO2, φανέρωσε πως η ύπαρξη ή όχι κρυσταλλικού χαρακτήρα και η σύσταση του οξειδίου εξαρτάται τόσο από το πρόδρομο μόριο αλλά και από την ανόπτηση (600 οC, 2 min). Αντίθετα, το ZrO2 έχει σε κάθε περίπτωση κρυσταλλικότητα. Τα ηλεκτρικά χαρακτηριστικά των πυκνωτών Si/SiO2/Si3N4/high-k/Pt, δείχνουν ότι οι δομές έχουν ικανοποιητική συμπεριφορά ως στοιχεία μνήμης αφού όλες οι ιδιότητες πληρούν τις βασικές προϋποθέσεις ως στοιχεία μνήμης, παρά την ανυπαρξία ενεργειακού φραγμού μεταξύ στρώματος παγίδευσης και οξειδίου ελέγχου. Η ικανότητα παγίδευσης και η επίδοση των δομών με HfO2 και ZrO2 δεν διαφοροποιούνται σημαντικά με χρήση διαφορετικού πρόδρομου μορίου ή με την ανόπτηση. Ο έλεγχος όμως της αντοχής των δομών σε επαναλαμβανόμενους παλμούς εγγραφής/διαγραφής αναδεικνύει ότι αμφότερες οι δομές που ανεπτύχθησαν με βάση το κυκλοπενταδιενύλιο έχουν μειωμένη αντοχή ηλεκτρικής καταπόνησης. Τo Al2O3 αναπτύχθηκε χρησιμοποιώντας το μόριο ΤΜΑ και ως οξειδωτικό μέσο: (α) H2O, (β) O3 και (γ) Plasma Ο2 (μέθοδος PE-ALD) σε συνδυασμό με ΤΜΑ. Οι δομές στην αρχική κατάσταση, χωρίς ανόπτηση, χαρακτηρίζονται από ισχυρό ρεύμα έγχυσης ηλεκτρονίων από την πύλη (υπό αρνητικές τάσεις) περιορίζοντας την ικανότητα φόρτισης και την επίδοση διαγραφής. Η ανόπτηση σε φούρνο και αδρανές περιβάλλον (850 ή 1050 oC, 15 min) προκάλεσε σημαντική βελτίωση των ηλεκτρικών χαρακτηριστικών των δομών λόγω του σημαντικού περιορισμού του παραπάνω φαινομένου. Μετά το στάδιο της ανόπτησης οι συνδυασμοί ΤΜΑ/Η2Ο και ΤΜΑ/Plasma Ο2 έχουν καλύτερες χαρακτηριστικές σε σχέση με αυτές του συνδυασμού ΤΜΑ/Ο3. Το φαινόμενο της διαρροής ηλεκτρονίων από την πύλη αποδίδεται στη μεγάλη συγκέντρωση και χωρική κατανομή του υδρογόνου στο υμένιο υψηλής διηλεκτρικής σταθεράς. Τέλος, διερευνήθηκε η τροποποίηση των ιδιοτήτων μνήμης των δομών με εμφύτευση ιόντων αζώτου χαμηλής ενέργειας και υψηλής δόσης στο Al2O3 και επακόλουθη ανόπτηση υψηλής θερμοκρασίας. Η παρουσία αζώτου στο υμένιο καθώς και ο χημικός δεσμός του εμφυτευμένου αζώτου είναι συνάρτηση της θερμοκρασίας ανόπτησης. Επομένως, οι ιδιότητες μνήμης εξαρτώνται από τη μορφή σύνδεσης και την συγκέντρωση του εμφυτευμένου αζώτου στο τροποποιημένο Al2O3. Η υψηλή θερμοκρασία ανόπτησης (1050 οC, 15 min) φαίνεται να αποφέρει δομές με τις καλύτερες ιδιότητες μνήμης. / This thesis studies the functionality of high-k oxides as blocking oxide layers in SONOS type charge-trap memory devices. The oxide materials that were examined were the HfO2, the ZrO2 and the Al2O3. All these blocking oxide layers were deposited by atomic layer deposition technique (ALD). The electrical performance of the trilayer stacks was examined using Pt-gate MOS-type capacitors. The properties of the memory structures were examined as a function of: (a) precursor chemistry of HfO2 and ZrO2 deposition, (b) the deposition oxidizing agent in the case of Al2O3 and (c) subsequent high temperature annealing steps. The HfO2 films were deposited on SiO2/Si3N4 bilayer stacks using: (a) hafnium alkylamide (TEMAH) and O3 at 275 oC, and (b) hafnium cyclopentadienyl (HfD-04) and O3 at 350 oC. Similarly the ZrO2 films were deposited by (a) zirconium alkylamide (TEMAZ) and O3 at 275 oC, and (b) zirconium cyclopentadienyl (ZrD-04) and O3 at 350 oC The structural characterization of the HfO2 showed that the crystallinity of the deposited high-k material depends on the precursor choice and the post deposition annealing step (600 °C, 2 min). On the contrary ZrO2 is deposited in a crystalline phase independent of the deposition conditions and the choice of the precursors. The electrical characterization of Si/SiO2/Si3N4/high-k/Pt capacitors showed that all fabricated structures operate well as memory elements, despite the absence of an energy barrier between the trapping layer and control oxide. The trapping efficiency and the performance of structures with HfO2 or ZrO2 blocking layers do not revealed a dependence upon the precursor chemistry. However, endurance testing using continuous write/erase pulses showed that both structures deposited by cyclopentadienyl precursors cannot sustain the resulting electrical stress. The Al2O3 layers were deposited using the TMA molecule while three different oxidizing agents were used: (a) H2O, (b) O3 and (c) oxygen plasma. Electrical testing of the resulting Pt-gate trilayer capacitors showed that in the deposited condition all three samples were characterized by gate electrode induced electron leakage currents in the negative bias regime, which completely masked the substrate hole injection effects. This effect limits the performance and the functionality of the memory stacks. After a high temperature annealing step (850 or 1050 oC, 15 min) this leakage current is reduced significantly and the stacks can function as memory elements. The results point to suggest that after annealing the best performance is exhibited by the TMA/H2O and TMA/Plasma O2 samples. The effect of gate induced electron leakage current is attributed to hydrogen related contamination, which has been verified by ToF-ERDA in depth profile measurements, at least for the case of TMA/H2O samples. The modification of the memory properties of the SiO2/Si3N4/Al2O3 stacks was also investigated using low energy and high fluence nitrogen implantation into Al2O3 layer. The concentration and the chemical bonding of the implanted nitrogen is a function of annealing temperature. The memory properties of the stack depend therefore on the chemical bonding and the concentration of the remaining nitrogen in the modified Al2O3. The high temperature annealing (1050 oC, 15 min) appears to provide the structures with improved memory properties in terms of retention and fast erase performance.