Teaching 'failed metals' new tricks
Jul 30, 2019
Related publication: Scheiderer P., Schmitt M., Gabel J., Zapf M., Stübinger M., Schütz P., Dudy L., Schlueter C., Lee T. L., Sing M. & Claessen R. Tailoring Materials for Mottronics: Excess Oxygen Doping of a Prototypical Mott Insulator. Adv. Mater. 30, 1706708 (2018). DOI: 10.1002/adma.201706708
Publication keywords: Electronic phase transitions; Mottronics; Photoelectron spectroscopy; Thin films; Transition metal oxides
A team of researchers from the Universität Würzburg in Germany, and Diamond Light Source, investigated the electronic properties of LaTiO3 (LTO) thin films, which tend to over-oxidise. This presents a challenge for the preparation of the thin films, however, it also allows for chemical doping by excess oxygen once control of the oxygen content has been gained. The team used an in situ vacuum suitcase supplied by beamline I09 to prevent exposure of the samples to air, and then used a combination of soft and hard X-ray photoelectron spectroscopy to monitor the oxygen stoichiometry via the titanium valence, and valence band photoemission to determine the electronic phase.
Based on the spectroscopic characterisation of the thin films, they developed a method to control the oxygen stoichiometry, and demonstrated that stoichiometric LTO films are insulating and that oxygen excess doping induces the filling controlled Mott transition into the metallic phase. This work identified a suitable channel material for Mottronics, characterised the Mott insulator-to-metal transition in LTO, and characterised a way to tune the electronic phase.
- Figure 1: Sketch of the electronic phase diagram of Mott materials including some prototypical d1 materials. U is the on-site repulsion and t is the hopping term.
The strong electron correlations in transition metal oxides induce a rich electronic phase diagram that waits to be harnessed in the development of future electronic devices, often referred to as Mottronics. One of the most prominent potential applications of this field is the Mott transistor that uses the electronic phase transition between the Mott insulator phase (OFF-state) and the correlated metal phase (ON-state) as a switch.
Mott insulators are ’failed metals‘ in the sense that their charge carrier densities correspond to a half-filled narrow band (usually of d character) which, however, leads to strong on-site Coulomb interactions and thus localised electrons. As sketched in the phase diagram1 in Fig. 1, tuning the band filling away from the half-fill (d1) condition by, e.g., an external electric field, may trigger a Mott transition (MT) into a metallic phase, rendering all previously localised electrons mobile. Therefore, a Mott transistor exhibits a large ON/OFF ratio and fast switching speed, and due to the high carrier concentration, miniaturisation of devices below the current limits of conventional semiconductors is possible. The downside of these high carrier densities are the large electric fields required to induce a sizable change of the band-filling. A solution to this problem has recently been suggested based on dynamical mean field theory simulations2,3. In Mott insulating materials that are already close to the phase transition, the electronic phase is highly sensitive to external stimuli including moderate electric fields. In order to reach such an operating point, well characterised Mott materials are required, and methods to adjust their properties within the electronic phase diagram need to be developed, both of which is the scope of this work.
Good candidates as Mott materials are the 3d1 perovskite series: SrVO3, CaVO3, LaTiO3 and YTiO3, which range from the metallic vanadates to the insulating titanates (Fig. 1) and exhibit a bandwidth controlled MT. From this series, the insulating compound closest to the phase transition, LaTiO3, is selected and oxygen doping is demonstrated to tune the electronic properties in LaTiO3+x thin films all the way from being Mott insulating to metallic. In view of future device designs the samples are fabricated as thin films by pulsed laser deposition (PLD).
The electronic properties can be best investigated by photoelectron spectroscopy (PES), which allows us to monitor the titanium valence with the binding energy of the Ti 2p core level, and the spectral weights of the Ti 3d states near the Fermi level. Taking advantage of the wide photon energy range uniquely available at beamline I09, that covers both soft and hard X-rays, we examine the Ti valence near the surface and in the bulk of the thin films and probe both angle-integrated and -resolved valence electronic structures of LaTiO3+x at the Ti L resonance.
- Figure 2: (a) The titanium valence is monitored by core-level photoelectron spectroscopy of the Ti 2p line. The absence of any sizable Ti4+ signal in the spectrum from the LaTiO3 film (blue line) verifies the correct oxygen stoichiometric; (b) Resonant photoemission at the Ti L absorption threshold allows to investigate the Ti 3d valence band states. Whereas the stoichiometric sample exhibits a gapped spectrum, the oxygen excess doped thin film is metallic as indicated by the quasiparticle peak (QP) and the Fermi cutoff; (c) Fermi surface measured by resonant SX-ARPES at the Ti L absorption threshold depicted together with a tight binding model for a filling of d1.
Furthermore, I09 provides a UHV suitcase that allows in situ transfer of samples from the growth chamber in Würzburg to the end station at the beamline, which is crucial for the investigations of LaTi3+O3 thin films since the material tends to over-oxidize towards LaTi4+O3.5, corresponding to a reduction of the valence band filling from 3d1 to 3d0. In air the films oxidise in an uncontrolled way impeding any investigations of the dependence of the electronic phase on the oxygen content. Even when the samples are transferred in vacuum, the Ti oxidation state may deviate from its nominal value due to other sources of excess oxygen supplied by, e.g., the substrate, during the growth process, a major challenge that is overcome by a novel growth strategy developed in this study, as discussed in more detail in a related publication4. Therefore, the first experimental task is to verify the stoichiometry of the samples, which can be best characterised by analysing the Ti 2p core level. The comparison of the corresponding spectrum with data obtained from the tetravalent reference material SrTiO3 (Fig. 2a) shows that no sizable amount of Ti4+ ions is present in the LaTiO3 thin films, thereby confirming the intended oxygen stoichiometry and band filling.
The electronic phase, i.e. Mott insulator or correlated metal, is investigated by measuring the Ti 3d valence states using resonant photoelectron spectroscopy (ResPES) at the Ti L absorption threshold. The resonant conditions strongly and selectively enhance the Ti 3d signal, providing a clear view of the valence states. The resulting spectrum of the stoichiometric sample (blue line in Fig. 2b) exhibits a single Ti 3d feature around 1 eV identified as the lower Hubbard band (LHB) and an energy gap at the Fermi level (see the inset in Fig. 2b). Note that, for non-interacting electrons, the configuration of one electron per unit cell would result in a conduction band that crosses the Fermi level. Instead, due to the strong on-site Coulomb interactions between the 3d electrons in a half-filled band, an LHB is formed in LaTiO3 and the material is in the Mott insulating phase.
Having demonstrated the Mott insulating phase in stoichiometric samples, the next challenge is to tune the electronic properties in order to reach the suggested operating point for a Mott transistor. As mentioned above, LaTiO3 thin films tend to over-oxidise. This behaviour can be harnessed to reduce the band filling, since every excess oxygen ion accepts two electrons from the Ti 3d states, and thereby drive the Mott transition (see Fig. 1). The effectiveness of this method in p doping and in triggering the Mott transition into a metallic phase is again tested by ResPES. The red spectrum in Fig. 2b shows that, indeed, the LaTiO3+x sample fabricated in higher oxygen pressure exhibits a Fermi cutoff with a quasiparticle peak (QP) characteristic for the correlated metal phase.
The momentum-dependence of the QP spectral weight is further investigated by soft X-ray angular resolved PES (SX-ARPES, Fig. 2c). The corresponding Fermi surface matches approximately the tight binding model of the related correlated metal SrVO3 (Fig. 1). This confirms three important statements about the Mott transition in LaTiO3+x that have been made above. First, the valence electrons form dispersing bands indicative of the metallic phase. Second, the required doping level to reach the metallic phase is rather low, i.e. of the order of only a few percent, since the Luttinger volume of the Fermi surface is still very close to one electron per unit cell (d1). Third, this large Fermi surface shows that indeed all 3d electrons are rendered mobile in the metallic phase and the electronic transport is not related to the doping level as in semiconductor materials.
These findings are not only important steps in material research on the prototypical correlated compound LaTiO3, they also demonstrate that it is a promising channel material for future Mott transistor devices.
References:
- Imada M. et al. Metal-insulator transitions. Rev. Mod. Phys. 70, 1039–1263 (1998). DOI: 10.1103/RevModPhys.70.1039
- Mazza G. et al. Field-Driven Mott Gap Collapse and Resistive Switch in Correlated Insulators. Phys. Rev. Lett. 117, 176401 (2016). DOI: 10.1103/ PhysRevLett.117.176401
- Zhong Z. et al. Electronics with Correlated Oxides: SrVO3/SrTiO3as a Mott Transistor. Phys. Rev. Lett. 114, 246401 (2015). DOI: 10.1103/ PhysRevLett.114.246401
- Scheiderer P. et al. Tailoring Materials for Mottronics: Excess Oxygen Doping of a Prototypical Mott Insulator. Adv. Mater. 30, 1706708 (2018). DOI: 10.1002/adma.201706708
Funding acknowledgement:
The authors are grateful for funding support from the Deutsche Forschungsgemeinschaft (FOR 1162 and SFB 1170 “ToCoTronics”) and acknowledge Diamond Light Source for time on beamline I09 under proposals SI14106, SI15200, SI15856, and NT18372. The authors would also like to thank D. McCue for his superb technical support at the I09 beamline, and also acknowledge fruitful discussions with M. Bibes and R. Aeschlimann.
Corresponding author:
Philipp Scheiderer, Physikalisches Institut and Röntgen Center for Complex Material Systems (RCCM), Universität Würzburg, [email protected]
Diamond Light Source is the UK's national synchrotron science facility, located at the Harwell Science and Innovation Campus in Oxfordshire.
Diamond Light Source Ltd
Diamond House
Harwell Science & Innovation Campus
Didcot
Oxfordshire
OX11 0DE
Copyright © Diamond Light Source. Diamond Light Source® and the Diamond logo are registered trademarks of Diamond Light Source Ltd
Registered in England and Wales at Diamond House, Harwell Science and Innovation Campus, Didcot, Oxfordshire, OX11 0DE, United Kingdom. Company number: 4375679. VAT number: 287 461 957. Economic Operators Registration and Identification (EORI) number: GB287461957003.