Binary and Ternary Intermetallics for Advanced Interconnect Metallization

In the recent past, alternatives to Cu metallization have elicited much interest for advanced interconnect applications due to increasing intrinsic limitations of line resistance and reliability at line widths below 20 nm [1,2]. Elemental metals with short mean free path (MFP) such as Co, Mo [3] or Ru [4] were first proposed to replace Cu and to improve line resistance and reliability. The effort has been recently extended to include binary and ternary compounds [5, 6].<br/>Amongst binary compound metals, only certain ordered intermetallics show low bulk resistivities with values that are potentially of interest for interconnect applications. The study of binary metals for interconnect applications has therefore focused completely on intermetallic ordered compounds. <i>Ab initio</i> screening (by assessment of the ρ₀ × λ figure of merit together with calculations of the cohesive energy [7]) indicated that especially binary aluminides show promise for alternative metallization. The most promising materials include NiAl, CuAl, CuAl₂, RuAl and ScAl₃. Thin films of these aluminide intermetallics were then studied experimentally to assess their properties and their potential for interconnect metallization. Low thin film resistivities have been obtained for NiAl, CuAl, CuAl₂ and ScAl₃ [8, 9]. Despite different crystal structures and constituents, binary aluminide intermetallics share common properties and challenges such as crystalline order, point defects, composition control, composition uniformity, secondary phase formation, agglomeration, (interface) reactivity, or non-stochiometric surface oxidation. These common challenges will be discussed in detail, highlighted by multiple experimental examples. We show by atom probe tomography that stochiometric NiAl films are prone to local compositional variations on nm scales, which are difficult to detect with other techniques. For thicknesses below ~15nm, the thin film resistivity strongly increased, and the effect of annealing was reduced. Resistivity modelling found that the resistivity of thin NiAl films below ~ 15 nm was dominated by grain boundary scattering due to small grain sizes, with little impact of surface scattering. To mitigate this issue, the grain size for a given film thickness was increased by back thinning of thicker films (with larger grains) using ion beam etching (IBE) or chemical mechanical polishing (CMP). The resistivity of NiAl could be further optimized at small thicknesses by combining the back thinning experiments with epitaxial deposition: a resistivity as low as 11.5 µΩcm at 7.7 nm for epitaxial NiAl on Ge (100) has been demonstrated [10], which outperforms PVD Ru and Cu at comparable thicknesses and thermal budget. We discuss prospects of such approaches for integration.<br/>Finally, we discuss ternary compounds as replacement for Cu in advanced metallization schemes. We will discuss potential classes of materials (e.g. MAX materials) and their prospects. We introduce a materials complexity index for alternative metals to reflect the complexity of the material control. The index can also be used to develop mitigation activities to reduce complexity.<br/><br/>References<br/>[1] D. Gall, J. Appl. Phys., 127 (2020) 050901<br/>[2] C. Adelmann <i>et al.</i>, IEEE IITC (2014) 173<br/>[3] V. Founta <i>et al.</i>, Materialia, 24 (2022) 101511<br/>[4] L.G. Wen <i>et al.</i>, IEEE IITC (2016) 34<br/>[5] L. Chen <i>et al.</i>, Appl. Phys. Lett., 113 (2018) 183503<br/>[6] K. Sankaran <i>et al.</i>, Phys. Rev. Materials, 5 (2021) 056002<br/>[7] S. Dutta <i>et al.</i>, J. Appl. Phys., 122 (2017) 025107<br/>[8] L. Chen <i>et al.</i>, Appl. Phys. Lett., 113 (2018) 183503<br/>[9] J-Ph. Soulié <i>et al.</i>, IEEE IITC (2021) 1<br/>[10] J-Ph. Soulié <i>et al.</i>, IEEE IITC (2023) 1