Ferroelectricity in simple binary ZrO2 and HfO2
Ferroelectricity in simple binary ZrO2 and HfO2 Supplementary Information J. Müller, et al. I. Methods and sample preparation Thin films of HfO2 and ZrO2 as well as of the HfO2-ZrO2 solid solution were deposited by thermal atomic layer deposition (ALD) on 300 mm Si substrates. Prior to dielectric deposition the substrates received a 10 nm chemical vapor deposited (CVD) TiN (TiCl4/NH3) bottom electrode in a single wafer reactor. The ALD process was based on the commercially available metal organic precursors tetrakis-(ethylmethylamino)-(hafnium/zirconium) (TEMAH and TEMAZ) with ozone. The ZrO2 and HfO2 content in the solid solution was defined by varying the cycle ratio of the precursors and monitored by inline X-ray photoelectron spectroscopy (XPS). Due to the nearly similar growth rates per cycle of the alkylamide precursors a fairly linear mixing of the oxides with ALD cycle ratio is observed (see Fig 1B, main trext). Controlled by the ALD super cycle number the thickness of the films was set to 9 nm, as confirmed by inline spectral ellipsometry and HR-TEM (see Fig. 1A, main text). Full crystallization of the as deposited amorphous Hf-rich and partially crystalline Zr-rich films occurred during the formation of the CVD-TiN top electrode at 500 °C. Grazing incidence X-ray diffraction (GIXRD) of all samples was measured on a Bruker D8 Discover diffractometer using Cu K-alpha radiation from a Cu tube operated at 40 kV / 40 mA. The incident angle was set to 0.55°. A
1
detailed structural analysis was performed on samples that were directly deposited on Si and for which the TiN top electrode had been wet chemically removed. This measure revealed characteristic diffractions of the oxides not identifiable when interfered with a cubic TiN diffraction pattern. Temperature dependent in-situ GI-XRD experiments were further conducted on crystalline samples that exhibited ferroelectricity. For this in-situ GI-XRD analysis, samples with and without TiN electrode were heated from room temperature to 800°C under constant nitrogen purging in a furnace covered with a hemispherical beryllium dome. During the cooling stage, the in situ GI-XRD measurements were performed. For electrical testing, planar metalinsulator-metal (MIM) capacitors were structured. PV hysteresis was characterized using an aixACCT TF Analyzer 2000 system and a custom built setup at a frequency of 1 kHz. The small signal CV characteristics of the MIM capacitors were determined with an Agilent 4284 LCR Meter. Typical measurement conditions were a frequency of 10 kHz and a 50 mV ac probing signal on device areas of 104 µm². Temperature dependent measurements down to 80 K were performed in a cryostat cooled with liquid nitrogen.
II. GI-XRD analysis of full HfO2-ZrO2 composition range As already descried in the main text, a size driven stabilization of the high temperature polymorphs in HfO2 and ZrO2 can be observed in thin films. This effect being strongest in ZrO2 leads to a tetragonal phase in pure ZrO2 and an increasing monoclinic phase with HfO2 admixture. The GI-XRD measurements depicted in Figure 1A show this transition in the films investigated in this work. The strongest reflections of the individual phases can be evaluated to estimate the phase fractions being presented at different ZrO2-HfO2 mixing ratios. The integral intensities of the monoclinic 111m can be weighed against the tetragonal 011t reflection to extract
2
the phase composition of the polymorphic mixture1. Even though the 011t and 111o reflections cannot be separated, a monoclinic fraction of the system relative to the tetragonal and orthorhombic phase can still be given. The results of this procedure are shown in the paper and are contrasted to the dielectric constant, which with the appearance of the monoclinic phase sharply decreases. III. Low temperature PV of pure ZrO2 Polarization hysteresis loops for pure ZrO2 were recorded at 80 K (Figure 1B). Due to the reduced leakage current at low temperatures excitation fields of over 4 MV/cm were applicable leading from sub loop hysteresis at low fields to almost saturated antiferroelectric-like hysteresis at high fields. As can be seen from the switching current in the lower part of Figure 1B, a slight polarity asymmetry is observed in the magnitude of the induced polarization.
Figure 1 (A) GI-XRD experiments on TiN-based MIM capacitors reveal a suppression of the monoclinic phase in HfO2-ZrO2 solid solution with increasing ZrO2 content; (B) Polarization
3
hysteresis measurements of ZrO2 thin films at 80 K show a characteristic sub loop behavior at low fields and a saturated hysteresis at high fields.
The atomic layer deposited TiN-based MIM capacitors investigated in this work are manufactured quite similar to the conventional integration approach of this material when used as a DRAM storage node. Several research reports on this asymmetric behavior in the otherwise symmetrically designed stack have been published2–4. It is for instance assumed that due to the ozone based ALD processing the TiN bottom electrode is deteriorated by oxidation leading to differently behaved electrode interefaces. A direct impact on polarization switching is anticipated. A slight imprint was witnessed for all samples.
IV. Temperature dependence of dielectric properties
Additional temperature dependent PV and CV measurements were carried out to gain further insight on stability and succession of the observed dielectric nonlinearities in the HfO2-ZrO2 system. Figure 2 (A) reveals that for the Hf0.5Zr0.5O2 stoichiometry the ferroelectric phase is stable in a temperature range from 100 to 400 K. In the same temperature span the Zr-rich Hf0.3Zr0.7O2 films show a transition to a double-loop hysteresis, whereas the pure ZrO2 remains in this double-loop hysteresis starting from low temperatures. It has to be noted that due to the strongly increasing leakage current contributions at elevated temperatures, the PV hysteresis measurements, as well as the CV measurements cannot be driven into saturation. Hence, the smaller polarization observed with increasing temperature must be attributed to sub-loop behavior and cannot unambiguously be identified as a slowly approaching Curie temperature.
4
Figure 2 (A) Temperature dependent PV characteristics ranging from 100 to 400 K reveal a stable ferroelectric phase in Hf0.5Zr0.5O2, a transition to a double loop hysteresis in Hf0.3Zr0.7O2, as well as a stable double-loop hysteresis in ZrO2. (B) Increasing temperature dependence of the permittivity is observed from Hf0.5Zr0.5O2 over Hf0.3Zr0.7O2 towards ZrO2, suggesting the ZrO2 thin films being closest to Curie temperature. The increasing leakage and the technical setup limits further heating and therewith the characterization of the actual Curie temperature.
Since all three stoichiometries characterized in Fig. 2 (A) do not show a transition to a paraelectric phase in the temperature range investigated, no characteristic signature marking the Curie temperature in the temperature dependent permittivity is expected. This assumption is reflected in Fig. 2 (B); only weak temperature dependence of the permittivity is observed. Nevertheless an increasing dependency can be observed with increasing ZrO2 content, suggesting the pure ZrO2 being closest to an actual Curie temperature. This is in agreement with the observed succession of the phases described in the main text, as well as with quiet similar systems, such as PLZST5, showing a phenomenologically equally behaved FE-AFE-PE
5
transition. Assuming the high temperature successor of the tetragonal phase, the cubic phase, as the paraelectric phase, it can be predicted from literature6 that this phase will be stable in Zr-rich samples at much lower temperature than in Hf-rich samples. The paraelectric nature of the cubic phase has been suggested for highly stabilized Y-doped HfO2, that prior to reaching the cubic phase, meaning for lower doping concentrations showed ferroelectric behavior7. REFERENCES (1) Schmid, H. K. J. Am. Ceram. Soc. 1987, 70 (5), 367–376. (2) Weinreich, W.; Reiche, R.; Lemberger, M.; Jegert, G.; Müller, J.; Wilde, L.; Teichert, S.; Heitmann, J.; Erben, E.; Oberbeck, L.; Schröder, U.; Bauer, A. J.; Ryssel, H. Microelectronic Engineering 2009, 86 (7-9), 1826–1829. (3) Agaiby, R.; Hofmann, P.; Dayu Zhou; Kerber, M.; Heitmann, J.; Schroeder, U.; Erben, E.; Oberbeck, L. IEEE Electron Device Lett. 2009, 30 (4), 340–342. (4) Dayu Zhou; Schroeder, U.; Xu, J.; J. Heitmann; Jegert, G.; Weinreich, W.; Kerber, M.; Knebel, S.; E. Erben; Mikolajick, T. J. Appl. Phys. 2010, 108 (12), 124104. (5) Markowski, K.; Park, S.-E.; Yoshikawa, S.; Cross, L. E. J. Am. Ceram. Soc. 1996, 79 (12), 3297–3304. (6) Ruh, R.; Garrett, H. J.; Domagala, R. F.; Tallan, N. M. J. Am. Ceram. Soc. 1968, 51 (1), 23–28. (7) Müller, J.; Schröder, U.; Böscke, T. S.; Müller, I.; Böttger, U.; Wilde, L.; J. Sundqvist; M. Lemberger; Kücher, P.; Mikolajick, T.; Frey, L. J. Appl. Phys. 2011, 110 (11), 114113.
6
1
detailed structural analysis was performed on samples that were directly deposited on Si and for which the TiN top electrode had been wet chemically removed. This measure revealed characteristic diffractions of the oxides not identifiable when interfered with a cubic TiN diffraction pattern. Temperature dependent in-situ GI-XRD experiments were further conducted on crystalline samples that exhibited ferroelectricity. For this in-situ GI-XRD analysis, samples with and without TiN electrode were heated from room temperature to 800°C under constant nitrogen purging in a furnace covered with a hemispherical beryllium dome. During the cooling stage, the in situ GI-XRD measurements were performed. For electrical testing, planar metalinsulator-metal (MIM) capacitors were structured. PV hysteresis was characterized using an aixACCT TF Analyzer 2000 system and a custom built setup at a frequency of 1 kHz. The small signal CV characteristics of the MIM capacitors were determined with an Agilent 4284 LCR Meter. Typical measurement conditions were a frequency of 10 kHz and a 50 mV ac probing signal on device areas of 104 µm². Temperature dependent measurements down to 80 K were performed in a cryostat cooled with liquid nitrogen.
II. GI-XRD analysis of full HfO2-ZrO2 composition range As already descried in the main text, a size driven stabilization of the high temperature polymorphs in HfO2 and ZrO2 can be observed in thin films. This effect being strongest in ZrO2 leads to a tetragonal phase in pure ZrO2 and an increasing monoclinic phase with HfO2 admixture. The GI-XRD measurements depicted in Figure 1A show this transition in the films investigated in this work. The strongest reflections of the individual phases can be evaluated to estimate the phase fractions being presented at different ZrO2-HfO2 mixing ratios. The integral intensities of the monoclinic 111m can be weighed against the tetragonal 011t reflection to extract
2
the phase composition of the polymorphic mixture1. Even though the 011t and 111o reflections cannot be separated, a monoclinic fraction of the system relative to the tetragonal and orthorhombic phase can still be given. The results of this procedure are shown in the paper and are contrasted to the dielectric constant, which with the appearance of the monoclinic phase sharply decreases. III. Low temperature PV of pure ZrO2 Polarization hysteresis loops for pure ZrO2 were recorded at 80 K (Figure 1B). Due to the reduced leakage current at low temperatures excitation fields of over 4 MV/cm were applicable leading from sub loop hysteresis at low fields to almost saturated antiferroelectric-like hysteresis at high fields. As can be seen from the switching current in the lower part of Figure 1B, a slight polarity asymmetry is observed in the magnitude of the induced polarization.
Figure 1 (A) GI-XRD experiments on TiN-based MIM capacitors reveal a suppression of the monoclinic phase in HfO2-ZrO2 solid solution with increasing ZrO2 content; (B) Polarization
3
hysteresis measurements of ZrO2 thin films at 80 K show a characteristic sub loop behavior at low fields and a saturated hysteresis at high fields.
The atomic layer deposited TiN-based MIM capacitors investigated in this work are manufactured quite similar to the conventional integration approach of this material when used as a DRAM storage node. Several research reports on this asymmetric behavior in the otherwise symmetrically designed stack have been published2–4. It is for instance assumed that due to the ozone based ALD processing the TiN bottom electrode is deteriorated by oxidation leading to differently behaved electrode interefaces. A direct impact on polarization switching is anticipated. A slight imprint was witnessed for all samples.
IV. Temperature dependence of dielectric properties
Additional temperature dependent PV and CV measurements were carried out to gain further insight on stability and succession of the observed dielectric nonlinearities in the HfO2-ZrO2 system. Figure 2 (A) reveals that for the Hf0.5Zr0.5O2 stoichiometry the ferroelectric phase is stable in a temperature range from 100 to 400 K. In the same temperature span the Zr-rich Hf0.3Zr0.7O2 films show a transition to a double-loop hysteresis, whereas the pure ZrO2 remains in this double-loop hysteresis starting from low temperatures. It has to be noted that due to the strongly increasing leakage current contributions at elevated temperatures, the PV hysteresis measurements, as well as the CV measurements cannot be driven into saturation. Hence, the smaller polarization observed with increasing temperature must be attributed to sub-loop behavior and cannot unambiguously be identified as a slowly approaching Curie temperature.
4
Figure 2 (A) Temperature dependent PV characteristics ranging from 100 to 400 K reveal a stable ferroelectric phase in Hf0.5Zr0.5O2, a transition to a double loop hysteresis in Hf0.3Zr0.7O2, as well as a stable double-loop hysteresis in ZrO2. (B) Increasing temperature dependence of the permittivity is observed from Hf0.5Zr0.5O2 over Hf0.3Zr0.7O2 towards ZrO2, suggesting the ZrO2 thin films being closest to Curie temperature. The increasing leakage and the technical setup limits further heating and therewith the characterization of the actual Curie temperature.
Since all three stoichiometries characterized in Fig. 2 (A) do not show a transition to a paraelectric phase in the temperature range investigated, no characteristic signature marking the Curie temperature in the temperature dependent permittivity is expected. This assumption is reflected in Fig. 2 (B); only weak temperature dependence of the permittivity is observed. Nevertheless an increasing dependency can be observed with increasing ZrO2 content, suggesting the pure ZrO2 being closest to an actual Curie temperature. This is in agreement with the observed succession of the phases described in the main text, as well as with quiet similar systems, such as PLZST5, showing a phenomenologically equally behaved FE-AFE-PE
5
transition. Assuming the high temperature successor of the tetragonal phase, the cubic phase, as the paraelectric phase, it can be predicted from literature6 that this phase will be stable in Zr-rich samples at much lower temperature than in Hf-rich samples. The paraelectric nature of the cubic phase has been suggested for highly stabilized Y-doped HfO2, that prior to reaching the cubic phase, meaning for lower doping concentrations showed ferroelectric behavior7. REFERENCES (1) Schmid, H. K. J. Am. Ceram. Soc. 1987, 70 (5), 367–376. (2) Weinreich, W.; Reiche, R.; Lemberger, M.; Jegert, G.; Müller, J.; Wilde, L.; Teichert, S.; Heitmann, J.; Erben, E.; Oberbeck, L.; Schröder, U.; Bauer, A. J.; Ryssel, H. Microelectronic Engineering 2009, 86 (7-9), 1826–1829. (3) Agaiby, R.; Hofmann, P.; Dayu Zhou; Kerber, M.; Heitmann, J.; Schroeder, U.; Erben, E.; Oberbeck, L. IEEE Electron Device Lett. 2009, 30 (4), 340–342. (4) Dayu Zhou; Schroeder, U.; Xu, J.; J. Heitmann; Jegert, G.; Weinreich, W.; Kerber, M.; Knebel, S.; E. Erben; Mikolajick, T. J. Appl. Phys. 2010, 108 (12), 124104. (5) Markowski, K.; Park, S.-E.; Yoshikawa, S.; Cross, L. E. J. Am. Ceram. Soc. 1996, 79 (12), 3297–3304. (6) Ruh, R.; Garrett, H. J.; Domagala, R. F.; Tallan, N. M. J. Am. Ceram. Soc. 1968, 51 (1), 23–28. (7) Müller, J.; Schröder, U.; Böscke, T. S.; Müller, I.; Böttger, U.; Wilde, L.; J. Sundqvist; M. Lemberger; Kücher, P.; Mikolajick, T.; Frey, L. J. Appl. Phys. 2011, 110 (11), 114113.
6
