Tingfeng
Song
,
Raul
Solanas
,
Mengdi
Qian
,
Ignasi
Fina
* and
Florencio
Sánchez
*
Institut de Ciència de Materials de Barcelona (ICMAB-CSIC), Campus UAB, Bellaterra 08193, Barcelona, Spain. E-mail: ifina@icmab.es; fsanchez@icmab.es
First published on 7th December 2021
The ferroelectric phase of HfO2 is generally stabilized in polycrystalline films, which typically exhibit the highest polarization when deposited using low oxidizing conditions. In contrast, epitaxial films grown by pulsed laser deposition show low or suppressed polarization if a low oxygen pressure is used. Epitaxial films are essential to better understand physical properties, and obtaining films that have intrinsic polarization is of great importance. In order to advance towards this objective, we have carried out a systematic study of the epitaxial growth of Hf0.5Zr0.5O2 combining inert Ar gas with oxidizing O2 gas. This allows us to control the oxidizing conditions (through O2 partial pressure) and the energy of the pulsed laser deposition plasma (through the total pressure of O2 and Ar). A pressure of Ar high enough to significantly reduce plasma energy and that of O2 low enough to reduce oxidation conditions are found to allow a large increase in ferroelectric polarization up to about 30 μC cm−2, representing an increase of around 50% compared to films grown by conventional pulsed laser deposition. This simple growth process, with high impact in the development of ferroelectric HfO2, can be also beneficial in the growth of thin films of other materials by pulsed laser deposition.
Ferroelectric HfO2 epitaxial films are generally grown by pulsed laser deposition (PLD).13–17 The technique, unlike sputtering, does not require the use of Ar gas and an atmosphere of pure O2 is commonly used. Lyu et al.16,18 reported the growth window of ferroelectric epitaxial HZO by PLD. HZO crystallizes at temperature and oxygen pressure (PO2) greater than about 700 °C and 0.02 mbar, respectively.16 The amount of orthorhombic phase increases with PO2, the maximum PO2 ranges from 0.08 to 0.2 mbar (the highest pressure used in the experiments).16 The ferroelectric polarization exhibits the same dependence on PO2, confirming that a low oxidizing atmosphere during PLD suppresses the formation of the orthorhombic phase. Regarding leakage current, it is important to note that it decreases with increasing ferroelectric polarization. Therefore, the influence of oxidizing conditions in PLD is the opposite to that in chemical methods or sputtering. On the other hand, ferroelectric doped-HfO2 epitaxial films have been also prepared by solid phase epitaxy using sputtering, and the films deposited under pure Ar showed much better ferroelectric properties than the films deposited under a mixed Ar/O2 atmosphere.19 Therefore, epitaxial and polycrystalline films show the same dependence on oxidizing conditions when the same deposition method is used. Consequently, the inhibition of ferroelectric phase formation under low oxidizing conditions is specific to films grown by the PLD technique.
The oxygen pressure in a PLD process has an obvious effect on the oxidation of the films. It affects the amount of oxygen vacancies and the phase formed in the case of competing valence states (in FexOy films, for example). The oxygen content in the film will depend, besides the oxygen pressure in the PLD chamber, on the substrate temperature because of the high vapor pressure of oxygen. But the oxygen pressure during PLD is also critical because the material ablated with each laser pulse interacts with the oxygen gas.20,21 The ablated atoms constitute a very high-energy plasma, the so-called PLD plume, which propagates along the normal direction of the target. When the ablation process occurs in the presence of a background gas, the interaction with the gas atoms reduces the kinetic energy of the ablated species. As the background pressure increases, the plasma energy decreases. The plume scattering also causes a decrease in the growth rate and can produce off-stoichiometry in the case of multi-cation films.22–24 The kinetic energy of the emitted atoms, which also depends on the laser fluence, can reach tens of eV under low pressure conditions.25 The effect of the high-energy atoms on the film can be dramatic, reducing the crystallization26,27 and even causing self-sputtering in the film.28–30 On the other hand, if the ambient pressure is high enough to thermalize the PLD plasma, the crystallinity of the films also degrades.27 Therefore, the gas (pressure and composition) during PLD needs to be optimized. These effects must be considered, if a low oxygen pressure is used to decrease oxidation conditions, in order to favor the formation of a particular phase.
A method of reducing the plasma energy without increasing the film oxidation is introducing an inert gas during the PLD process. PLD under an inert gas was scarcely done in the past.28,29,31,32 Recently, it has been used successfully to obtain transparent conducting SrVO3 films without spurious phases.33,34 An inert gas can be also necessary to deposit oxide films on highly reactive substrates. For example, pure Ar gas was used to grow HZO films on bare Si.35 Here, we combine O2 and Ar as ambient gas to grow HZO ferroelectric films by PLD under low oxidation conditions and reduced plasma energy. This allows extending the growth window of epitaxial HZO to lower oxygen pressure without epitaxy degradation caused by an excessively energetic plasma. We show that a low plasma energy allows the increase of ferroelectric polarization in epitaxial HZO films by more than 50% with respect to equivalent films prepared by conventional PLD. The polarization values match well with the theoretically calculated polarization for the ferroelectric Pca21 orthorhombic phase of HfO2. Low plasma energy processes could be also used to prepare conventional (perovskites) or unconventional (as ε-Fe2O3 or Al1−xScxN) ferroelectrics with improved properties.
Fig. 1 XRD θ–2θ scans of films deposited under a mixed Ar/O2 atmosphere. (a) Fixed PO2 = 0.01 mbar and varying PAr. The scan corresponding to PAr = 0 mbar and PO2 = 0.01 mbar has been reported in ref. 16. (b) Fixed PAr = 0.05 mbar and varying PO2. (c) Fixed PAr = 0.1 mbar and varying PO2. |
The intensity of the o-(111) reflection allows the quantification of the dependence of the amount of orthorhombic phase on the O2 and Ar partial pressure (Fig. 2). There are no significant differences if normalization is performed with the STO(002) substrate peak (Fig. S6, ESI†). IHZO(111)/ILSMO(002) increases with PO2 (Fig. 2a), with little additional effect of the Ar pressure when PO2 is high. The crystallization is very low in films grown under a pure oxygen pressure lower than 0.05 mbar, but in the presence of additional Ar the stabilization of the orthorhombic phase is greatly enhanced. The amount of orthorhombic phase is slightly underestimated in the two PAr = 0.2 mbar films due to their lower thickness, but the equivalent graphs normalized to the film thickness exhibit the same relation (Fig. S7, ESI†). The dependence of IHZO(111)/ILSMO(002) on PAr (Fig. 2b) evidences that, under a high PAr pressure (0.1 or 0.2 mbar), the amount of orthorhombic phase only depends slightly on PO2. In brief, the presence of Ar notably enhances the stabilization of the o-phase for a low PO2 (up to about 0.05 mbar) and does not cause a significant effect for a higher PO2.
Fig. 2 Intensity of the o-(111) reflection, normalized to that of the LSMO(002) peak, IHZO(111)/ILSMO(002), as a function of PO2 (a) and PAr (b). In (a) PAr is 0 mbar (black squares), 0.05 mbar (red circles), 0.1 mbar (blue up-pointing triangles), and 0.2 mbar (green down-pointing triangles). Data corresponding to PAr = 0 mbar are reported in ref. 16. In (b) PO2 is 0.01 mbar (black squares), 0.05 mbar (red circles), and 0.1 mbar (blue up-pointing triangles). |
The out-of-plane lattice parameter associated with the HZO(111) reflection, dHZO(111), was determined from the peak position. The dependences of dHZO(111) on PO2 and PAr are shown in Fig. 3a and b, respectively. The dHZO(111) value expands by decreasing PO2 (Fig. 3a). On the other hand, dHZO(111) of the films deposited at PO2 below around 0.05 mbar depends on PAr, with dHZO(111) less expanded for a high Ar pressure. The effect of PAr can be directly visualized in Fig. 3b. The lattice parameter of the films deposited under PO2 = 0.01 mbar (black squares) decrease with increasing PAr up to 0.1 mbar. The plasma is thermalized for a higher PAr and dHZO(111) does not change with PAr or PO2. To rationalize the intriguing dependence of dHZO(111) on PAr and PO2 shown in Fig. 3a and b, two causes of cell expansion have to be considered. On one hand, a higher number of oxygen vacancies is expected as PO2 is lower. On the other hand, other point defects can be formed if the PLD plasma has a high energy, which happens when the total pressure, PAr + PO2, is low. Indeed, deposition under high-energy PLD plasma (low PAr and low PO2) reduces strongly the film crystallization (Fig. 1 and 2). Thus, a high oxygen pressure is required to avoid film degradation if PAr is low. This implies that low oxidation conditions cannot be used to grow HfO2 films when the films are grown using a pure O2 atmosphere. Thus, conventional PLD does not permit growth conditions closer to those that result in the largest ferroelectric polarization when ALD or sputtering is used (low oxidizing conditions).11 However, as demonstrated here, the use of inert Ar gas to reduce the PLD plasma energy allows the stabilization of the orthorhombic phase with around one order of magnitude lower PO2 (0.01 mbar) than the optimal pressure (0.1 mbar) in conventional PLD (Fig. 2) and importantly reduces the number of defects (signaled by the d(111) expansion, Fig. 3).
Fig. 3 Out-of-plane lattice parameter, dHZO(111), as a function of PO2 (a) and PAr (b). In (a) PAr is 0 mbar (black squares, data reported in ref. 16), 0.05 mbar (red circles), 0.1 mbar (blue up-pointing triangles), and 0.2 mbar (green down-pointing triangles). In (b) PO2 is 0.01 mbar (black squares), 0.05 mbar (red circles), and 0.1 mbar (blue up-pointing triangles). |
Fig. 4a shows the dependence of the leakage current at 1 V on PO2, for fixed PAr = 0 mbar (black squares), PAr = 0.05 mbar (red circles) and PAr = 0.1 mbar (blue triangles). The corresponding leakage–voltage curves can be seen in Fig. S8 (ESI†). The leakage of the films deposited without Ar gas (PAr = 0 mbar) increases with PO2, from around 2 × 10−7 A cm−2 (PO2 = 0.01 mbar) to 3 × 10−4 A cm−2 (PO2 = 0.2 mbar).16 The films deposited under a mixed O2/Ar atmosphere and a very low PO2 in the 2 × 10−3–0.01 mbar range are very insulating too, with a leakage current of about 2 × 10−7 A cm−2. In contrast, the film grown under a high total pressure (PAr = 0.1 mbar and PO2 = 0.1 mbar) is much more conducting. The effect of the total pressure is evident in Fig. 4b, which shows the leakage current at 1 V as a function of PAr for fixed PO2 = 0.01 mbar (black squares), PO2 = 0.05 mbar (red circles) and PO2 = 0.1 mbar (blue triangles). The leakage current of the three PAr = 0.05 mbar samples is about 3 × 10−7 A cm−2, without a significant effect of PO2. However, in the case of the samples grown under a higher PAr, the leakage is very high when the total pressure (PAr + PO2) is about 0.2 mbar. Overall, Fig. 4 indicates that a PLD plasma thermalized by a high ambient pressure causes a strong increase in the film conductivity, whereas oxygen vacancies are not the main cause of leakage.
Fig. 4 (a) Leakage current at 1 V as a function of PO2 for PAr = 0 mbar (black squares, data reported in ref. 16), PAr = 0.05 mbar (red circles) and PAr = 0.1 mbar (blue triangles). (b) Leakage current at 1 V as a function of PAr for PO2 = 0.01 mbar (black squares), 0.05 mbar (red circles), and 0.1 mbar (blue up-pointing triangles). |
The ferroelectric polarization loops of the films grown under fixed PAr = 0.05 mbar and varying PO2 are shown in Fig. 5a. The PO2 = 0.01 mbar and 0.02 mbar films have low remanent polarization (Pr) values of 8.3 and 17.1 μC cm−2, respectively. A slightly higher PO2, 0.05 mbar, results in a high increase of polarization, with Pr = 32 μC cm−2. Further increase of PO2 does not influence significantly the polarization loops, and the Pr of the PO2 = 0.1 mbar film is slightly lower, 30 μC cm−2. The loops of the series of films grown under fixed PAr = 0.1 mbar are shown in Fig. 5b. In agreement with the presence of the XRD o-HZO(111) reflection in the PO2 = 2 × 10−3 and PO2 = 5 × 10−3 mbar films, these films exhibit hysteretic polarization loops, with Pr = 13.8 and 18.2 μC cm−2, respectively. The Pr of the films increases with increasing PO2, being 27.8 μC cm−2 in the PO2 = 0.05 mbar film. The high leakage of the PO2 = 0.1 mbar film did not allow the measurement of a loop. The dependence of Pr on PO2 is summarized in Fig. 5c for the fixed PAr = 0 mbar (black squares), PAr = 0.05 mbar (red circles), PAr = 0.1 mbar (blue up-pointing triangles) and PAr = 0.2 mbar (green down-pointing triangle). As described above, the Pr in the PAr = 0.05 mbar and PAr = 0.1 mbar films increases with PO2. The same PO2 dependence is observed in films grown using conventional PLD (PAr = 0 mbar).16Fig. 5c evidences a huge increase in Pr in the films deposited under a mixed Ar/O2 atmosphere. The highest Pr in the PAr = 0 mbar series is 20.5 μC cm−2 (PO2 = 0.1 mbar film), while it is 32 μC cm−2 and 27.8 μC cm−2 in the PAr = 0.05 mbar and PAr = 0.1 mbar series, respectively, in both cases in films grown under PO2 = 0.05 mbar. Deposition under a higher PAr, 0.2 mbar, results in Pr reduction. The second benefit of using a mixed Ar/O2 atmosphere is that the films grown under very low PO2 (5 × 10−3 mbar in the PAr = 0.1 mbar series) show high ferroelectric polarization and, as shown in Fig. 4, low leakage, while PO2 above 0.02 mbar is needed in conventional PLD. The color map of Pr as a function of PAr and PO2 (Fig. 5d) evidences graphically that the optimal conditions to maximize Pr are PAr = 0.05–0.1 mbar and PO2 around 0.05 mbar. It has to be noted that the measured polarization corresponds to a projection of the ferroelectric dipoles since the films are (111)-oriented. The highest measured Pr = 32 μC cm−2 corresponds to a polarization of about 55 μC cm−2, matching well the value of spontaneous polarization calculated for ferroelectric HfO2.36,37
Fig. 5 Ferroelectric polarization loops of the films grown (a) under fixed PAr = 0.05 mbar and varying PO2 and (b) under fixed PAr = 0.1 mbar and varying PO2. (c) Remanent polarization as a function of PO2, for fixed PAr = 0 mbar (black squares, data reported in ref. 16), PAr = 0.05 mbar (red circles), PAr = 0.1 mbar (blue up-pointing triangles), and 0.2 mbar (green down-pointing triangle). (d) Color map of Pr as a function of PAr and PO2. The color map has been constructed after interpolating the data of panel (c) along the PAr and PO2 axes using cubic splines. |
Footnote |
† Electronic supplementary information (ESI) available: Sketches of the PAr and PO2 partial pressures used to grow the films. XRD θ–2θ scans of the series of films. Simulation of Laue oscillations. Thickness and growth rate as a function of PAr. XRD pole figures. Intensity of the o-(111) reflection, normalized to that of the STO(002) peak. The normalized o-(111) reflection. Leakage–voltage curves of films. See DOI: 10.1039/d1tc05387f |
This journal is © The Royal Society of Chemistry 2022 |