A viable technique for the formation of strontium titanate bicrystals at high pressure and fast heating rate via the spark plasma sintering apparatus is developed.
A spark plasma sintering apparatus was used as a novel method for diffusion bonding of two single crystals of strontium titanate to form bicrystals with one twist grain boundary. This apparatus utilizes high uniaxial pressure and a pulsed direct current for rapid consolidation of material. Diffusion bonding of strontium titanate bicrystals without fracture, in a spark plasma sintering apparatus, is possible at high pressures due to the unusual temperature dependent plasticity behavior of strontium titanate. We demonstrate a method for the successful formation of bicrystals at accelerated time scales and lower temperatures in a spark plasma sintering apparatus compared to bicrystals formed by conventional diffusion bonding parameters. Bond quality was verified by scanning electron microscopy. A clean and atomically abrupt interface containing no secondary phases was observed using transmission electron microscopy techniques. Local changes in bonding across the boundary was characterized by simultaneous scanning transmission electron microscopy and spatially resolved electron energy-loss spectroscopy.
Spark plasma sintering (SPS) is a technique in which application of high uniaxial pressure and pulsed direct current leads to the rapid densification of powder compacts1. This technique also leads to the successful formation of composite structures from various materials, including silicon nitride/silicon carbide, zirconium boride/silicon carbide, or silicon carbide, with no additional sintering aids required2,3,4,5. The synthesis of these composite structures by conventional hot-pressing had been challenging in the past. While application of a high uniaxial pressure and fast heating rate via the SPS technique enhances consolidation of powders and composites, the phenomenon causing this enhanced densification debated in the literature2,3,6,7. There also exists only limited information regarding the influence of electric fields on grain boundary formation and the resulting atomic structures of grain boundary cores8,9. These core structures determine the functional properties of SPS sintered materials, including electric breakdown of high voltage capacitors and the mechanical strength and toughness of ceramic oxides10. Therefore, understanding the fundamental grain boundary structure as a function of SPS processing parameters, such as applied current, is necessary for the manipulation of a material's overall physical properties. One method to systematically elucidate the fundamental physical mechanisms underpinning SPS is the formation of specific grain boundary structures, i.e., bicrystals. A bicrystal is created by manipulation of two single crystals, which are then diffusion bonded with specific misorientation angles11. This method provides a controlled way to investigate the fundamental grain boundary core structures as a function of processing parameters, dopant concentration, and impurity segregation12,13,14.
Diffusion bonding is dependent on four parameters: temperature, time, pressure, and bonding atmosphere15. Conventional diffusion bonding of strontium titanate (SrTiO3, STO) bicrystals typically occurs at a pressure below 1 MPa, within a temperature range of 1,400-1,500 °C, and time scales ranging from 3 to 20 hours13,14,16,17. In this study, bonding in a SPS apparatus is achieved at significantly lower temperature and time scales in comparison to conventional methods. For polycrystalline materials, reduced temperature and time scales via SPS significantly limits grain growth, thereby providing advantageous control of a material's properties through manipulation of its microstructure.
The SPS apparatus, for a 5×5 mm2 sample, exerts a minimum pressure of 140 MPa. Within the conventional diffusion bonding temperature range, Hutt et al. report instantaneous fracture of STO when the bonding pressure exceeds 10 MPa18. However, STO exhibits temperature dependent plasticity behavior, indicating bonding pressure can exceed 10 MPa at specific temperatures. Above 1,200 °C and below 700 °C, STO exhibits some ductility, at which stresses greater than 120 MPa can be applied without instantaneous fracture of the sample. Within the intermediate temperature range of 700-1,200 °C, STO is brittle and experiences instantaneous fracture at stresses greater than 10 MPa. At 800 °C, STO has minor deformability prior to fracture at stresses less than 200 MPa19,20,21. Hence, bonding temperatures for STO bicrystal formation via SPS apparatus must be selected according to the plasticity behavior of the material.
1. Sample Preparation of Single Crystal Strontium Titanate
NOTE: Single crystal STO is supplied with a (100) surface polished to a mirror finish.
2. Bicrystal Formation via Spark Plasma Sintering Apparatus
NOTE: For 5×5 mm2 crystal use a 30 mm diameter graphite die. If a die with a diameter smaller than 30 mm is used, the bicrystal catastrophically fractures during bonding. Optimal die size as well as pressure exerted by the SPS apparatus is highly dependent on the size of the crystals.
3. Sample Preparation of Bicrystal for Electron Beam Imaging
4. Cleaning the FIB Copper Grid
NOTE: Improper cleaning of the FIB grid can lead to carbon contamination of the lamella in the TEM.
5. Preparation of Transmission Electron Microscopy (TEM) Lamella via Focused Ion Beam (FIB) Apparatus
NOTE: All parameters used in FIB preparation are typed or selected from a drop down menu in the FIB apparatus software.
Bonding temperature, time, and misorientation angle were all altered to determine optimum parameters needed for the maximum possible bonded interface fraction of the STO bicrystal (Table 1). The interface was considered 'bonded' when the grain boundary was not visible during SEM imaging (Figure 2a). A 'non-bonded' interface was exhibited when a dark image contrast or voids were present at the boundary location (Figure 2b). Dark image contrast signified colloidal graphite from the FIB mounting procedure had diffused between the two STO crystals due to capillary effects. This non-bonded interface is observed at the edges of the bicrystal, while the bonded interface is observed at the center of the bicrystal. The change in bonding behavior from the edge to the center of the bicrystals formed by SPS apparatus is also seen in bicrystals formed via hot pressing techniques6.
Micro-crack formation in the bulk of the bicrystal is observed in all samples. For successfully bonded bicrystals, micro-crack formation does not interfere with the bonded interface. At a bonding temperature of 1,200 °C, extensive micro-cracking occurs, leading to brittle failure of the bicrystal during annealing. Therefore, bonding temperature in the SPS apparatus was kept below 800 °C to prevent catastrophic fracture.
For bicrystals with 0° misorientation angle at bonding temperatures of 600 °C and 700 °C, a 95% successfully bonded interface was obtained. An increase in the misorientation angle of the bicrystal to 44.4° requires a bonding temperature of 800 °C and bonding time of 90 minutes to achieve a 45.8% successfully bonded interface. High resolution TEM (HRTEM) and high angle annular dark field scanning transmission electron microscopy (HAADF-STEM) micrographs of this sample reveal an atomically abrupt grain boundary structure with no intergranular film or secondary phases present (Figure 3). Spatially resolved electron energy loss spectroscopy (EELS) recorded directly from the interface shows a reduction in the crystal field splitting of the Ti L3 and L2 edges as well as a reduction in the O K edge intensity when compared to the bulk (Figure 4).
Bicrystal | Twist Orientation (°) | Bond Temperature (°C) | Bond Time (min) | Anneal Temperature (°C) | Anneal Time (h) | Bonded Interface ± 0.3 (%) |
A | 0* | 1200 | 15 | 1,200 | 16 | — |
B | 0 | 600 | 90 | 1,200 | 100 | 92.3 |
C | 0 | 700 | 90 | 1,200 | 100 | 99.7 |
D | 4.3 ± 0.3 | 800 | 20 | 1,200 | 50 | 79.2 |
E | 45* | 700 | 60 | 1,200 | 140 | 1.3 |
F | 46.1 ± 0.5 | 800 | 20 | 1,200 | 140 | 35.4 |
G | 44.4 ± 0.1 | 800 | 90 | 1,200 | 140 | 45.8 |
Table 1. Bonding Parameters and Subsequent Interface Fraction of Bicrystals. SPS apparatus diffusion bonding parameters and subsequent bonded interface fractions of STO bicrystals. A pulsed bias voltage of ~4 V and direct current of ~550 A is applied for all experiments. Bonded interface fractions were calculated from an average grain boundary length of (1.5 ± 0.4) mm. These regions were taken from varying areas within the bicrystal.
Figure 1. Spark Plasma Sintering Apparatus. Set up for the spark plasma sintering apparatus. Pressure is applied perpendicular to the bicrystal interface. Please click here to view a larger version of this figure.
Figure 2. Typical Interfaces Found in SPS Apparatus Formed Bicrystals. Nominal 45° twist bicrystal formed at a temperature of 800 °C for 90 minutes. (a) SEM image of defined 'bonded' interface, the grain boundary location is inferred by the presence of a faceted void, and (b) SEM image of the defined 'non-bonded' interface. Spherical beads observed in images are residual silica from polishing. Please click here to view a larger version of this figure.
Figure 3. High Resolution Imaging of Bicrystal Grain Boundary. Boundary of nominal 45° twist bicrystal formed at a temperature of 800 °C for 90 minutes recorded in <100> zone axis with an edge-on orientation for the interface plane. (a) HRTEM image, (b) HRSTEM DF image, and (c) structure model composed of two single crystals, one in <100> and one in <110> zone axis orientation with a {001} interface plane. Deviations of the experimental imaging data from the projected structure model represent changes of the interface configuration compared to the ideal single-crystal atom positions. Please click here to view a larger version of this figure.
Figure 4. Structure and Chemistry of Bicrystal Grain Boundary. Near edge fine structure of (a) the Ti L2,3 edge and (b) the O K edge taken at the boundary and the bulk of 45° twist bicrystal formed at a temperature of 800 °C for 90 minutes. Please click here to view a larger version of this figure.
Figure 5. FIB Milling at Bicrystal Grain Boundary. HRTEM images of FIB TEM lamella of (a) sample with grain boundary parallel to the ion beam and (b) sample with grain boundary perpendicular to the ion beam. Please click here to view a larger version of this figure.
The bonding temperature of 1,200 °C was chosen to maximize diffusion as small changes in temperature can greatly impact the kinetics of all diffusion bonding mechanisms. A temperature of 1,200 °C is outside the brittle-ductile transition temperature range of STO. However, the sample underwent brittle fracture at this temperature. The catastrophic failure of the STO bicrystal was not unexpected as STO has ~ 0.5% ductility at 1,200 °C. Also, the sample was held at a pressure of 140 MPa throughout the heating process and STO transitions through its brittle stage during this heating process where it has 0% ductility21. Thus successful diffusion bonding of single crystals via the SPS apparatus necessitates an in-depth understanding of the temperature dependent plasticity behavior of a material.
Two bicrystals were formed using a bonding temperature of 800 °C and a bonding time of 20 minutes. The bicrystal with a nominal misorientation of 4˚ was annealed for 50 hours and exhibited a bonded interface fraction 2.2x greater than the bicrystal with nominal misorientation of 45°, annealed for 140 hours. Annealing times longer than 50 hours did not reveal any significant improvement of the diffusion bonding quality. Annealing temperature and time were selected according to previous work in which the bicrystals were formed in a high vacuum furnace, similar to the SPS apparatus11. To ascertain if there was an impact of the annealing process on the diffusion bonding of these bicrystals, diffusion bond lengths were calculated and found to be 0.27 nm at a temperature of 1,200 °C for 140 hours23. The selected annealing parameters, therefore, only had limited impact on the diffusion bonding of the bicrystal. This analysis was further supported when the diffusion bonding of a bicrystal with a 45° twist misorientation was not successful for the same parameters utilized during annealing.
While the selected annealing parameters did not significantly impact the diffusion bonding, the misorientation angle does have a pronounced effect. High-angle misorientation angles create a greater structural mismatch between the two half-crystals, which hinders cross diffusion and decreases interface bonding. For high misorientation angles, bonding temperature and time must be increased in order to have a larger successfully bonded interface fraction.
Bicrystal formation via SPS processing techniques occurs at significantly accelerated times scales and moderate temperature ranges compared to the conventional hot press methods. This difference in processing parameters between conventional diffusion bonding and the SPS method is also seen in the formation of powder compacts and composites. As mentioned above, bicrystals formed via conventional diffusion bonding methods are formed at temperatures greater than 1,400 °C with times scales ranging from 3 to 20 hours11,24. Using a SPS instrument, diffusion bonding occurs at a temperature of 800 °C with times scales ranging from 20 to 90 minutes. The SPS technique is then useful for the rapid diffusion bonding of bicrystals compared to conventional methods. Bicrystal formation by the SPS instrument also allows for experimental observation of the mass transfer mechanisms at a grain boundary with a selected misorientation. This experimental observation will provide more insight into the mechanisms underpinning the SPS technique.
Micro-cracks in the bulk STO bicrystal prevented conventional mechanical TEM lamella preparation. The mechanical thinning process led to fracture of the TEM lamella due to micro-cracks spreading throughout the bulk. Therefore, FIB preparation of the TEM lamella was used. Conventional FIB lift-out of the lamella, in which the grain boundary is parallel to the ion beam, led to preferential milling along the interface plane (Figure 5a). The FIB preparation technique was subsequently modified. First, the thickness of the initial protective layer of carbon and tungsten was selected so, at the end of the lift-out step, the protective layer was milled away. If the protective layer was too thick and remained throughout the thinning process, re-deposition of tungsten occurred and obscured TEM analysis. Secondly, after the TEM lamella was attached to the tip of the micromanipulator, the micromanipulator was rotated by 180°. This rotation caused the grain boundary to become perpendicular to the ion beam, hence preventing preferential thinning (Figure 5b). Lastly, after the lift-out process, protective layers of carbon and tungsten were deposited onto the surface of the newly oriented TEM lamella. These modifications to the conventional FIB preparation technique led to a clean TEM lamella enabling atomically resolution imaging by HAADF-STEM.
HRTEM and HAADF-STEM micrographs for the bicrystal bonded at 800 °C for 90 minutes with a nominal 45° twist grain boundary show an atomically resolved grain boundary structure with no secondary phases. Spatially resolved EELS reveal changes in the Ti coordination within the grain boundary cores, indicating an increase in the oxygen vacancy concentration compared to the bulk. These results are consistent with reports in the literature for low-angle twist grain boundaries25. Further analysis of these experiments is described elsewhere26.
In this study, STO bicrystals were successfully synthesized for the first time using a SPS apparatus. Bicrystals with twist orientations of 0°, 4°, and 45° were formed at high pressure with moderate bonding temperatures and time scales compared to those parameters found in conventional bonding. Formation of bicrystals via the SPS apparatus provides an opportunity to quantitatively determine the impact of electric field as well as heating rate on selected grain boundary core structures.
The authors have nothing to disclose.
LH gratefully acknowledges financial support by an US National Science Foundation Graduate Research Fellowship under Grant No. 1148897. Electron microscopy characterization and SPS processing at UC Davis was financially supported by a University of California Laboratory Fee award (#12-LR-238313). Work at the Molecular Foundry was supported by the Office of Science, Office of Basic Energy Sciences, of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231.
Strontium titanate single crystal (100) | MTI Corporation | STOa101005S1-JP | |
Buffered oxide etch, hyrofluoric acid 6:1 | JT Baker | MBI 1178-03 | |
Scanning electron microscope (SEM) | FEI | Model: 430 NanoSEM | |
SPS apparatus | Sumitomo Coal Mining Co | Model: Dr. Sinter 5000 SPS Apparatus | |
High Temperature Furnace | Thermolyne | Model: 41600 | |
Ultrasonic Cleaner | Bransonic | Model: 221 | |
Mechanical polisher | Allied High Tech Products | 15-2100-TEM | |
Diamond lapping film | 3M | 660XV | 1 um to 9 um Grit Size |
Diamond lapping film | 3M | 661X | 0.5 um to 0.1 um Grit Size |
Colloidal silica | Allied High Tech Products | 180-20000 | .05 um Grit Size |
Sputter coater | QuorumTech | Model: Q150RES | |
Focused ion beam (FIB) instrument | FEI | Model: Scios dual-beamed focused ion beam (FIB) instrument | |
Nanomill TEM specimen preparation system | Fischione Instruments | Model: 1040 | |
Transmission electron microscope (TEM) | JEOL | Model: JEM2500 SE | |
Scanning transmission electron microscope (STEM) | FEI | Model: TEAM 0.5 |