Synthesis and characterization of transition metal complexes of malondialdehyde and related derivatives as potential chemical vapor deposition precursors

Chemical vapor deposition (CVD) is a process used especially in the microelectronics industry to deposit films onto surfaces by means of chemical reactions of a volatile precursors. Its principal advantage over physical vapor deposition and atomic layer deposition techniques is that it can coat hidden and high aspect-ratio features both quickly and conformally. Some of the best- studied metal-containing precursors for CVD are metal complexes of deprotonated β-diketones such as acetylacetone (Hacac = CH3C(=O)CH2C(=O)CH3). Complexes of acac and other β- diketonates are known for almost every metal in the periodic table. When used as CVD precursors, these compounds primarily afford films of metal oxides, but require relatively high decomposition temperatures (>250 °C), and – when used as single source precursors – tend to deposit films contaminated with carbon. In contrast, β-dicarbonylates (the parent class of β-diketonates) with one or more aldehyde groups are essentially unstudied as ligands for metal centers. For the anion of malondialdehyde (Hmda = HC(=O)CH2C(=O)H)), only complexes of six d- and f-block elements are known – Cr(mda)3, Pd(mda)2, PtMe3(mda), and two complexes each of yttrium, europium, and terbium; of these, only Cr(mda)3 had been fully characterized. One report of Pd(mda)2 noted its high volatility and that it deposited metallic films, apparently Pd0, at 125 °C; in contrast, the β-diketonate Pd(acac)2 undergoes thermolysis only at much higher temperatures of ca. 250 °C. This difference, which is consistent with the known ability of aldehydes to reduce late transition metals to zero- valent metals, indicates that complexes of mda and related aldehyde-containing β-dicarbonylate ligands may serve as useful CVD precursors. Many β-diketones are easy to handle and commercially available, but malondialdehyde is unstable and self-reactive; consequently it is usually prepared from its acetal and handled as its sodium salt. After the preparation of Na(mda) was optimized, we used this salt to prepare the new compound Cu(mda)2. Thermolytic studies revealed that Cu(mda)2 quantitatively decomposes by an acid-base pathway enabled by the presence of the aldehyde group, which explains the lower decomposition temperatures seen for Pd(mda)2. Similar acid-base pathways are a minor (< 10%) pathway for the thermolysis of metal β-diketonates, for which radical and redox pathways predominate. The EPR parameters for Cu(mda)2, which are essentially identical to those of Cu(acac)2, contradict an established correlation between ligand pKa and g-values. This aberration is due to the anomalously high acidity of H(mda); H- and CH3-groups have similar inductive properties, but H(mda) is 4.5 pKa units more acidic than H(acac). We attribute this behavior to the steric preorganization of β-diketones into a U-shape, which enables formation an intramolecular hydrogen bond that quenches its acidity. In contrast, H(mda) adopts an open-chain conformation in which the OH group is more acidic because it is not stabilized by internal hydrogen bonding. The binary complexes M(mda)2 of Mn, Fe, Co, and Ni are insoluble, involatile solids, and could not be isolated in pure form; in contrast, the corresponding β-diketonate analogues are soluble and volatile oligomers. Adducts of stoichiometry M(mda)2L2 (where M = Fe, Co, Ni and L = pyridine or 1⁄2 N,N,N′,N′-tetramethylethylenediamine) can be prepared and characterized, however. The tmed complexes are volatile and may be useful as CVD precursors. In the solid state, Mn(mda)2(tmed) adopts a structure in which one mda ligand per formula unit is non-chelating and instead assumes the open-chain structure seen in solution and for the sodium salts, but which is unprecedented in transition metal β-dicarbonylates. The insolubility of the binary mda complexes is likely due to the presence of similar bridging mda units. Attempts to make Pt(mda)2 result instead in the formation of Pt metal. The tris-chelate compounds M(mda)3 (where M = Al, V, Fe) were also prepared and characterized. Interestingly, the titanium analogue Ti(mda)3 can be observed by EPR spectroscopy at low temperatures but decomposes above -78 °C. Attempts to prepare Co(mda)3 instead afforded Co(mda)2. The thermal instability of the titanium(III), cobalt(III), and platinum(II) mda complexes suggests that the mda anion is unstable with respect to strong reductants and strong oxidants, so that M(mda)x complexes are isolable only if the metal center has no redox activity within a potential window ranging from about -0.009 V to 1.12 V vs. NHE. This limited redox stability is in contrast to the β-diketonates, which are stable toward a much wider range of redox potentials. Modification of the mda structure by substitution of electron-withdrawing or -donating groups should shift the stability window. Consequently, ten substituted analogs of mda were synthesized, each of which contains at least one aldehyde or aldimine group. Because one possible decomposition pathway for Cu(mda)2 leads to formation of an unstabilized carbene, we also synthesized analogs of mda in which a 1,2-hydrogen migration would allow this energetic fragment to rearrange into more stable products. Some of these mda analogues have previously been reported but their coordination chemistry is essentially unexplored. Depending on the nature of the substituents, these ligands are best made by one of three different preparation methods; 1) acetal formation and hydrolysis, 2) Claisen condensation, or 3) trifluoroacylation of enol ethers followed by nucleophilic attack. Fluorinated compounds could be prepared only by method 3 even when appropriate precursors were available for other methods. The copper(II) compounds of these mda analogues were prepared and characterized. Whereas most are bis(chelates) both in solution and in the solid state, the 2-Me-mda ligand exclusively exhibits the non-chelating binding mode, as seen in the crystal structure of Cu(2-Me- mda)2(tmed). The EPR spectra of the new complexes show modest agreement with the aforementioned pKa correlation. Preliminary investigations on the utility of Cu(1-CF3-2-Me- mda)2, the most promising of this class of compounds, as a CVD precursor have been carried out.U of I OnlyAuthor requested U of Illinois access only (OA after 2yrs) in Vireo ETD syste