Stability of the nanomartensitic phase in ultrathin Fe films on Cu(100)

The stability of the bcc-like ``nanomartensitic'' (NM) phase in epitaxial Fe films grown layer by layer on Cu(100) is characterized by variable-temperature scanning tunneling microscopy. While 3 monolayer (ML) films are found to be completely NM at least up to 340 K, films 4 and 5 ML thick are pseudomorphic fcc but show a transition to the NM phase induced by hydrogen adsorption. A statistical mechanical description of these transitions, in particular of the surface H distribution in the region of fcc-NM phase coexistence, is used to estimate the free-energy difference $\ensuremath{\Delta}{F}^{\text{fcc}\ensuremath{\rightarrow}\text{NM}}$ in dependence of thickness and temperature: The continuously increasing stability of the NM phase with decreasing thickness, qualitatively reflected in the decreasing ${\text{H}}_{2}$ doses required to stabilize it ($\ensuremath{\approx}100\text{ }\text{L}$, $\ensuremath{\approx}0.5\text{ }\text{L}$, and none for 5, 4, and 3 ML films), is explained by finite-thickness terms in the energy balance which are detectable already in 5 ML films but can overcome the fcc-stabilizing lattice mismatch terms only in films less than 4 ML thick. Furthermore it is found that the NM phase, like bulk bcc Fe, becomes more stable with decreasing temperature. However, below 200 K the relative stability is almost temperature independent, i.e., phase changes between fcc and NM can still be fully controlled by hydrogen adsorption but hardly by temperature variation. These results are discussed in terms of finite-thickness modifications to current zero-temperature and finite-temperature models of bulk Fe, in particular that by Hasegawa and Pettifor [H. Hasegawa and D. G. Pettifor, Phys. Rev. Lett. 50, 130 (1983)], which emphasizes the importance of the magnetic free energy for the phase stability of Fe.