Elsevier

Applied Surface Science

Volume 273, 15 May 2013, Pages 415-425
Applied Surface Science

Charge transfer and band bending at Au/Pb(Zr0.2Ti0.8)O3 interfaces investigated by photoelectron spectroscopy

https://doi.org/10.1016/j.apsusc.2013.02.056Get rights and content

Abstract

The growth of gold layers on Pb(Zr,Ti)O3 (PZT) deposited on SrTiO3 is investigated by X-ray photoelectron spectroscopy in the Au thickness range 2–100 Å. Two phases are identified, with compositions close to nominal PZT. The ‘standard’ phase is represented by all binding energies (Pb 4f, Ti 2p, Zr 3d, O 1s) sensibly equal to the nominal values for PZT, whereas the ‘charged’ phase exhibits all core levels are shifted by ∼1 eV toward higher binding energies. By taking into account also scanning probe microscopy images together with recent photoemission results, the ‘charged’ phase belongs to P(+) regions of PZT, whereas the ‘normal’ phase corresponds to regions with no net ferroelectric polarization perpendicular to the surface. Au deposition proceeds in a band bending of ΦPZT  ΦAu  0.4–0.5 eV for both phases, identified as similar shifts toward higher binding energies of all Pb, Ti, Zr, O core levels with Au deposition. The Au 4f core level exhibits also an unusually low binding energy component 1 eV below the ‘nominal’ Au 4f binding energy position (metal Au). This implies the existence of negatively charged gold, or electron transfer from PZT to Au, although the ‘normal’ PZT phase have a higher work function, as it is derived from the band bending. Most probably this charge transfer occurs toward Au nanoparticles, which have even higher ionization energies. High resolution transmission electron microscopy evidenced the formation of such isolated nanoparticles.

Highlights

► First follow-up by XPS, with detailed curve fitting, of the interface Au/PZT. ► Nearly perfect composition transfer from the target to the PZT film by pulsed laser deposition. ► A 0.94 eV binding energy difference between P(+) and P(0) areas of the PZT surface. ► Evidence of 0.45 eV band bending toward higher binding energies with Au deposition.

Introduction

Lead zirconate-titanate Pb(Zr,Ti)O3 (PZT) is a material widely used nowadays [1]. Applications include piezoelectricity, pyroelectricity, ferroelectric capacitors and random access memories (FERAMs). For most applications the active material PZT is contacted with metals in order to apply the necessary voltages. However, to date interfaces formed by metals with PZT are not yet well understood at the atomic level, including phenomena such as surface disruption, charge accumulation and band bending, oxygen depletion, etc. There is a general agreement of the scientific community that surface layers can have quite different composition with respect to bulk [2], [3] and also that the early stages of metal deposition proceed via complicated processes, involving sometimes formation of undesirable compounds by uptaking the inherent (mainly carbon) contamination of the PZT layer [4], [5].

Many properties useful for application are strongly dependent on the Schottky barrier formed at the interfaces between ferroelectric semiconductors and metals [6], [7]. For instance, in Ref. [7] it was demonstrated that the short circuit photocurrent strongly depends on the work function of the metal deposited on PZT, being orders of magnitude more elevated when the work function of the metal (Pt: 5.65 eV [8]) exceeds that of PZT (5.1 eV [9]). Therefore, the investigation of the barrier formed at metal/ferroelectric interface is of prime importance.

Although the bulk atomic structure, polarization and ferroelectric domains are nowadays known with high accuracy [10], complicate depolarization phenomena, including lattice relaxation [11], [12] and strong depolarization fields [12], [13] occur at ferroelectric surfaces. These effects surely affect the band bending of free ferroelectric surfaces. Therefore, obtaining reliable experimental data allowing one to quantify the surface band bending is crucial.

Metal deposition induces further band bending whose curvature depends on the difference between the work function of the metal and that of the semiconductor (Fig. 1). The Schottky barrier height is influenced by charge transfer induced by the ferroelectric surface states via the S factor, such that for electrons Φbi = SM  ΦS) + S  χS), where S = 1 corresponds to the pure Schottky case (no charge transfer, just band alignment), whereas S = 0 correpond to the Bardeen case, where the charge transferred fully compensates the interface band bending [14]. For holes, Φbi = SS  ΦM) + (χS  ΦS + Eg), where ΦM, ΦS are work functions for the metal and for the semiconductor, χS the semiconductor's electronic affinity, and Eg the bandgap. The S factor can be expressed by using the surface density of interface states NS, the penetration of these states into the semiconductor δ and the dielectric constant of the semiconductor ∈: S−1 = 1 + e2NSδ/∈ and S is found to lie between 0.31 for PbTiO3 and 0.4 for PbZrO3 [14].

To this metal/semiconductor band bending, the bending due to ferroelectric polarization has to be added (Fig. 1c). The net result is either an increase, either a reduction of the band bending. Based on data from Ref. [6], one may infer that a polarization oriented from the ferroelectric toward the metal film acts like a reverse (forward) bias for the Schottky diode formed at the interface if the semiconductor is p- (n-) doped. Therefore, the surface charging is always (+) in the former case, whereas it may be (+) or (−) in the latter case.

The band bending induced by the polarization may be expressed as e Pδ/∈, where e is the elementary charge, P is the static polarization, δ is the distance between the surface and the polarization charge sheet, and ∈ = 0r is the dielectric constant of the ferroelectric [15]. Therefore, a method to extract this ferroelectric band bending was proposed [16], implying ferroelectric hysteresis loops, CV and IV characteristics measurements. Typical values obtained for epitaxial PbZr0.2Ti0.8O3 are P = 95 μC/cm2 [17], δ = 3.0 nm, ∈r = 180 [16]. This implies a maximum downoward band curvature of 1.8 eV induced by the ferroelectric film only.

A more immediate method to quantify surface charge transfer is X-ray photoelectron spectroscopy (XPS), since: (i) surface charging will directly be reflected in the kinetic energy of the detected photoelectron; (ii) the method is surface sensitive, the probing depth being in the range of 2–3 nm, owing to the photoelectron inelastic mean free path [18]. Usually, metal–semiconductor band bending spatial ranges are on the order of (∈kBT/N)1/2/e, 16–160 nm at room temperature (T) for PZT (∈r  180) with the density of ionized impurities N = 1018–1016 cm−3. Note also that for larger doping levels the extent of the band bending approaches the surface sensitivity range for XPS: therefore, an appropriate simulation of the XPS peaks should yield the band bending also in this case.

The main problem is that XPS evaluation of simple surface charging and/or band bending may be hindered by interface reactions [19], [20], [21], [22], [23], [24]. For instance, Pb segregation was detected in several metal/PZT interfaces [19], [20]; interfaces with indium tin oxides (ITO) [21] or with RuO2 [22] were found to have a complicated surface chemistry. A procedure for derivation of ionization potentials was proposed in Ref. [23]. However, none of the Refs. [19], [20], [21], [22], [23] identified a real effect of the ferroelectric polarization. Only recently, the same group published outstanding results where all core levels of BaTiO3 (BTO) were found to vary as function on the polarization voltage applied in situ on the sample [24]. A difference of about 1 eV in binding energy was found between P(+) and P(−) poled samples. No such result was published to our knowledge on PZT, although the expected effect should be bigger, owing to the larger polarization of PZT (in the range of 100 μC/cm2) as compared with BTO (in the range of 25 μC/cm2).

Other complementary methods to investigate the surface charge and/or band bending we should mention are: (i) Scanning Tunneling Spectroscopy (STS): for example, the interface barrier potential at a free BiFeO3 surface was found to vary between 0.62 eV and 1.0 eV when the sample was poled upwards P(+) [25]; (ii) Mirror electron microscopy (MEM) investigations on piezoresponse force microscopy (PFM) written domains exhibited a relative low difference between P(+) and P(−) written domains for the transition between the MEM regime (electrons completely deflected) to the low energy electron microscopy (LEEM) regime (electrons penetrating into the sample) [26].

The aim of this work is to pursue these efforts in order to elucidate the separate effects of surface reactivity, segregation, charging and subsequent band bending for Au/PZT interfaces. First of all, none of the Refs. [19], [20], [21], [22], [23], [24] analyzed the formation of this interface, although it should present a particular interest, since the work function of Au is quite close to that of PZT (5.1 eV) [8]. Therefore, metal/semiconductor band bending should be considerably smaller ΦM  ΦS  0 and eventually observed band bending are related to the ferroelectric polarization only. According to Fig. 1, any doping will result in a band bending. Secondly, the data will be analyzed by appropriate simulations (’deconvolutions’) with components of several binding energies in order to identify all the surface reaction products. Therefore, surface stoichiometries will be derived. Additionally, the evolution of these components with the metal layer thickness will provide information about the continuity of the layer, surface disruption and intermixing, as in most cases of metal/semiconductor studies [27], [28], [29], [30]. All this information will be supplemented with piezoresponse force microscopy (PFM) data used to quantify the relative percentage of regions of the PZT surface exhibiting a non-vanishing out-of-plane polarization or of regions which exhibit either in-plane polarization, either ferroelectric dead layers, and with high resolution transmission electron microscopy (HRTEM) data, used to confirm the continuity/discontinuity assertions from the XPS data.

Section snippets

Experimental

Single crystal 200 nm thick PZT layers are grown by pulsed laser deposition (PLD) on SrTiO3 with a SrRuO3 buffer layer. The PLD setup (Surface) operates with KrF radiation (248 nm wavelength), 0.7 J × 20 ns pulses, repetition rate 5 Hz, laser fluence 2 J/cm2. During the PZT growth, the substrate was heated at 575 °C and the partial O2 pressure was 0.2 mbar. Under these preparation conditions, there is a general agreement that the preferred polarization of the PZT layers is P(+) [6], [7], [17], [31]. The

Results

The first remark is that all photoemission data obtained did not need any energy correction, with the C 1s peak from the inherent contamination located at 286.60 ± 0.02 eV. Fig. 2 presents all photoemission data. Fig. 2(a) presents the valence band region, used to estimate the position of the valence band maximum (VBM) with respect to the Fermi energy EF  EVBM = 1.38 ± 0.02 eV. At the same time, onsets of the Au 5d and 6s bands are clearly seen, together with the progressive attenuation of the Pb 5d

Discussions

Fig. 3 presents the evolution of individual integral amplitudes for all components derived from XPS ‘deconvolutions’ (Fig. 3(a) presents amplitudes resulted from Fig. 2(b), Fig. 3(b) amplitudes from Fig. 2(c), and so on). A first remark is that the intensities are not dropping to zero, although, with an inelastic mean free path in the range of 1 (or 2) nm and a gold thickness of 10 nm, the peaks should have decreased to exp(−10) or exp(−5)  4 × 10−5 (or 7 × 10−3) of their intial amplitude after Au

Conclusions

This work presented the first complete assessment by photoelectron spectroscopy of the reactivity and band bending at Au/PZT interfaces. In addition to already reported data on similar systems, here a careful investigation was performed by ‘deconvolution’ of core level spectra, allowing one to trace the origin of each component and the separate influence of ferroelectric polarization, surface chemistry and band bending. This is an example where surface contamination helped the understanding of

Acknowledgements

This work was financed by the Romanian UEFISCDI Agency under contract No. PCCE-3/2012.

References (47)

  • L. Pintilie et al.

    Ferroelectric Schottky diode behavior from a SrRuO3-Pb(Zr0.2Ti0.8)O3-Ta structure

    Physical Review B

    (2010)
  • L. Pintilie et al.

    Interface controlled photovoltaic effect in epitaxial Pb(Zr,Ti)O3 films with tetragonal structure

    Journal of Applied Physics

    (2011)
  • J.F. Scott

    Ferroelectric Memories

    (2000)
  • C.L. Jia et al.

    Atomic-scale study of electric dipoles near charged and uncharged domain walls in ferroelectric films

    Nature Materials

    (2008)
  • C.L. Jia et al.

    Unit-cell scale mapping of ferroelectricity and tetragonality in epitaxial ultrathin ferroelectric films

    Nature Materials

    (2007)
  • C.R. Li et al.

    Dependence of lattice constant on layer thickness of Pb(Zr0.52Ti0.48)O3 thin films,

    Journal of Materials Science Letters

    (1998)
  • X. Lu et al.

    Space charge effect in ultrathin ferroelectric films

    Journal of Applied Physics

    (2012)
  • J. Robertson et al.

    Schottky barrier heights of tantalum oxide, barium strontium titanate, lead titanate, and strontium bismuth tantalate

    Applied Physics Letters

    (1999)
  • L. Pintilie et al.

    Metal-ferroelectric-metal heterostructures with Schottky contacts. I. Influence of the ferroelectric properties

    Journal of Applied Physics

    (2005)
  • L. Pintilie et al.

    Metal-ferroelectric-metal structures with Schottky contacts. II. Analysis of the experimental current-voltage and capacitance-voltage characteristics of Pb(Zr,Ti)O3 thin films,

    Journal of Applied Physics

    (2005)
  • I. Vrejoiu et al.

    Functional perovskites – from epitaxial films to nanostructured arrays

    Advanced Functional Materials

    (2008)
  • S. Hüfner

    Photoelectron Spectroscopy: Principles and Applications

    (2003)
  • Cited by (50)

    • Nanoscopic correlations from curve fitting of photoelectron spectromicroscopy data cubes of lead zirconate titanate films

      2022, Results in Physics
      Citation Excerpt :

      Photoelectron spectromicroscopy becomes henceforth a valid tool to assess non-invasively the polarization state of a free ferroelectric surface. On the other hand, the size of ferroelectric domains is quite often below 1 μm [2,16,22], while sources of X-ray radiation with such smalls spot are not often available. Hence, even in photoelectron spectromicroscopy in many cases an individual spectrum will not contain a single component which may be attributed to a well-defined (single) polarization state.

    View all citing articles on Scopus
    View full text