New Chemistry for the Growth of First-row Transition Metal Films by Atomic Layer Deposition

Author
Joseph P. Klesko
Year
2015
Abstract & Cover

The stringent demands of microelectronic device fabrication favor the use of ALD for applications requiring conformal film growth with sub-nm thickness control. However, many existing ALD processes require high deposition temperatures, hazardous coreagents, or the use of plasma to facilitate film growth. These limitations can lead to the incorporation of contaminating elements, safety concerns for HVM, and loss of thickness control during the film growth process. Consequently, the research presented herein demonstrates new chemistry for the low-temperature thermal ALD of first-row transition metal films which meets the requisite criteria for integration into future microelectronic device fabrication processes. Chapter 1 offers a motivation for the research herein by outlining the challenges facing the continued miniaturization of microelectronic devices. Relevant examples of materials containing transition metal thin films are presented. Next, common techniques for vapor-phase film growth are introduced, including PVD, CVD, and ALD. Finally, the present status of first row transition metal precursors and vapor deposition processes is reviewed in detail. Chapter 2 introduces a class of bis(trimethylsilyl) six-membered rings as strongly reducing coreagents for ALD.347 Compounds 1 and 2 were used with TiCl4 for the growth of Ti metal films. The reaction is driven by the formation of strong Si-Cl bonds, affording ClMe3Si as a volatile byproduct with two equivalents of the resulting 8 π electron dianionic intermediate species coordinated to a Ti(IV) cation. Subsequent aromatization of these dianionic rings facilitates the reduction of Ti(IV) to Ti(0) by reducing the metal center by two electrons per ligand. Compound 2 was especially useful, supporting a growth rate of 0.06 Å/cycle within a temperature range of 110–240 °C. Films grown at < 180 °C were smooth by SEM and AFM analyses, while films grown at ≥ 180 °C were increasingly rough, with additional nanoparticulate growth on top of the films. All films readily oxidized upon exposure to air, producing a surface of TiO2 and a gradient layer of Ti suboxides.300,304,310 However, XPS analyses showed the definitive presence of Ti metal at the film/substrate interface. Importantly, the 3 keV argon ions used for sputtering are incapable of reducing Ti oxides to Ti metal, thus confirming reduction by a chemical process.300 Upon the basis of the reduction potential of toluene (E1/2 = −2.46 V), most metal ions in the periodic table should be reduced by the 2-methyl-1,4-cyclohexadienyl dianion,41 which results upon double desilylation of 1. This work constitutes the first report of a thermal Ti ALD process. Compound 2 was also used for the reduction of SbCl3 to Sb metal at 180 °C. After 1 min of sputtering, XPS analysis showed a film composition of 87.2% Sb as a mixture of Sb2O3 and Sb metal. Chapter 3 describes novel processes using 3 and formic acid for the ALD of high-purity Co metal films. Initially, a three-step approach, including 2 as a reducing coreagent, afforded 97.1% pure Co metal films at 180 °C. The discovery that 2 was unnecessary led to the development of a binary process with a growth rate of 0.95 Å/cycle within a 170–180 °C window. These films were very smooth by SEM and AFM analyses. Powder XRD confirmed the presence of Co metal, with an average crystallite size of 13.4 ± 3.0 nm. XPS analysis with continued sputtering resulted in ionizations that exactly matched the known values for Co metal after 2 min and a film composition consisting of 91.6% Co metal after 8 min. Films grown on Ru, Pd, Pt, and Cu substrates showed notably lower sheet resistivities compared to the bare substrates. Bulk resistivities of films grown on Ru from 165–220 °C were 13.4–19.0 μΩ∙cm, approaching that of bulk Co metal.312 Deposition data demonstrate that there is a small (15–16%) CVD component to the growth at 180 °C and also a nucleation period of ≤ 250 cycles for the initial nanoparticulate growth to coalesce into a continuous film. This CVD growth from the self decomposition of 3 may provide a seed layer for the ALD of Co metal. The approach described herein overcomes the limitations of previously-reported Co ALD processes, including low growth rates, 36,39 growth temperatures that are well above the precursor decomposition temperatures, 31,36,38,136–138 no demonstration of self-limited ALD growth, 143 and variable, high resistivities of the Co films. 136–138 Accordingly, three-step processes were subsequently explored for the growth of other first-row transition metal films using diazabutadienyl and alkoxide complexes with formic acid and 2 as coreagents. Chapter 4 explores the utility of 8–13 as ALD precursors that contain the redox non innocent 1,4-di-tert-butyl-1,3-diazabutadienyl ligand. The syntheses of these compounds employed a convenient metathesis approach, analogous to that reported for 3–7. 42 Proton NMR or magnetic moment data show the oxidation state of the ligand, and are reported for diamagnetic (9, 11) and paramagnetic (8, 10, 12–13) complexes, respectively. These compounds sublime at ≤ 125 °C/0.05 Torr with little or no residue. Compound 10 may be especially useful for ALD, as it has a fairly low melting point (135–137 °C) and a high thermal decomposition temperature (235 °C). A crystal structure for the novel 13 is also reported. Chapter 5 outlines processes for the ALD of Mn and Co borate films from 14 and 16, respectively, with ozone as the coreagent. These processes give distinct borate stoichiometries, demonstrating rare compositional control in ALD growth that is governed by the elements present in a single precursor. The 14/ozone process affords a growth rate of 0.18 Å/cycle within an ALD window of 300–350 °C. Films remained amorphous after annealing to 1100 °C under either an N2 or O2 atmosphere. While XPS analyses suggested a stoichiometry of MnBO2, TOF ERDA implied a more complex, oxygen-rich composition approximating Mn3(BO3)2. The data suggest that 14 chemisorbs on the surface of the growing film with an average loss of one Tp ligand per molecule of 14. Subsequent treatment with ozone then leads to Mn3(BO3)2 films. The analogous ALD process using 16/ozone affords a growth rate of 0.39 Å/cycle. XPS and TOF ERDA suggested a film stoichiometry of CoB2O4 that matched the 2:1 B/Co ratio of 16. The data imply that 16 physisorbs in a molecular fashion to the surface of the growing film, and is then efficiently transformed by ozone to CoB2O4. Films grown using 14 or 16 were non-conductive, and smooth by SEM and AFM analyses. These processes are the first examples of vapor-phase growth of first-row transition metal borate films. Film growth was also achieved using 15 or 17 with ozone at 325 °C. Finally, the thermal properties of 20–21 suggest that these compounds may be useful for ALD; crystal structures of both compounds are reported.

Source of Information
DigitalCommons@WayneState
University
Wayne State University
(Detroit, USA)
External Link
Read Thesis
LinkedIn
linkedin invite