853
(Invited) Initial State of Graphene Growth on Ge(001) Surfaces

Wednesday, October 14, 2015: 15:00
105-B (Phoenix Convention Center)
J. Dabrowski, G. Lippert (IHP), and G. Lupina (IHP)
To realize any of the envisioned applications of graphene one needs a material tailored to that particular use. Microelectronics will likely require the highest quality graphene deposited inexpensively on large areas. As graphene electronics is expected to complement the existing technologies, the ideal deposition should be compatible with the mainstream Si technology; in particular, it should be possible to grow graphene directly on CMOS compatible substrates. Though large area graphene can be grown with high quality on Cu or on Ni, for fabrication of electronic devices transfer of graphene to the target substrate must then follow. Transfer is not a preferred option in microelectronic manufacturing.

Direct growth of graphene on Si is hindered by high reactivity of Si against C. In contrast to Si, Ge does not form stable carbide and Ge is CMOS compatible. Moreover, graphene on Ge is directly useful in vertical transistors [2]. Graphene can be grown on Ge from CH4, but the process is slow. In addition, much higher growth temperatures are needed for CVD graphene on Ge than on hexagonal BN, indicating that the formation of C‑Ge bonds plays a major role in the former process.

We have demonstrated that graphene can be grown from atomic beam on Ge(001)/Si(001) wafers (Fig. a-b, [1]). Like for the growth from CH4, the quality improves when Ge begins to melt. This again highlights the importance of the C-Ge interaction. Here, we discuss results of ab initio density functional theory calculations for the interaction between C and Ge during the process of deposition of C atoms and/or of small hydrocarbon molecules, during graphene nucleation, and during the initial stage of graphene growth.

For example, ab initio calculations reveal that although carbon does not form a stable carbide phase on Ge, the chemical interaction between Ge and C atoms is strong. First, in contrast to deposition of atomic carbon on substrates such as hexagonal BN or graphene itself, the adsorption energy of C atoms on Ge is high enough to prevent their re-emission to vacuum. This differentiates Ge from mica and from oxide surfaces as well: on these substrates, the chemical interaction with atomic C leads to the formation of CO and, consequently, to carbon and oxygen loss. Carbon deposited on Ge(001) readily ejects Ge from surface dimers (Fig. c‑f); the reverse process is possible as well. Optionally, C interstitialcies may diffuse directly under the surface. But does not enter deeper into the bulk of Ge (Fig. g). Interaction between immobilized (in CGe dimers) and mobile (subsurface) carbon produces various species, ranging from Ge atoms (ejected from GeGe and CGe dimers) through mobile C chains (dimers, trimers, and longer) up to C clusters buit around a CGe seed or around a Ge dimer vacancy. Reactions between these species result in the formation of small graphene molecules. Some of them are attached along the whole edge to the substrate, while other may stand up (Fig.d). Mobile Ge tends to agglomerate at graphene edge and is likely to remain between the graphene molecules as these attach one to another. Further deposition produces better graphene on top of this higly defected interfacial layer.

The major difference to this in a CVD process is that Ge emission and subsurface diffusion of carbon are suppressed. In this report, we discuss such processes and the consequences for growth optimization strategies.

[1] G. Lippert et al., Carbon 75, 104 (2014).

[2] W. Mehr et al., IEEE EDL. 33, 691 (2012).