In this paper, we present a protocol to directly grow an epitaxial yet flexible lead zirconium titanate memory element on muscovite mica.
Flexible non-volatile memories have received much attention as they are applicable for portable smart electronic device in the future, relying on high-density data storage and low-power consumption capabilities. However, the high-quality oxide based nonvolatile memory on flexible substrates is often constrained by the material characteristics and the inevitable high-temperature fabrication process. In this paper, a protocol is proposed to directly grow an epitaxial yet flexible lead zirconium titanate memory element on muscovite mica. The versatile deposition technique and measurement method enable the fabrication of flexible yet single-crystalline non-volatile memory elements necessary for the next generation of smart devices.
The successful fabrication of flexible nonvolatile memory elements (NVME) plays a key role in exploiting the full potential of flexible electronics. NVME shall feature light weight, low cost, low-power consumption, fast speed and high storage density capabilities besides data storage, information processing and communication. Perovskite Pb (Zr,Ti)O3 (PZT) acts as a popular system for such applications considering its large polarization, fast polarization switching, high Curie temperature, low coercive field and high piezoelectric coefficient. In ferroelectric nonvolatile memories, an external voltage pulse can switch the two remnant polarizations between two stable directions, represented by '0' and '1'. It is non-volatile, and the write/read process can be completed within nanoseconds. NVME based on organic1,2,3,4,5,6 and inorganic7,8,9,10,11,12,13,14,15 ferroelectric materials have been attempted on flexible substrates. However, such integration is limited by not only the substrates' inability of high-temperature growth but also the degraded device performance, current leakage and electrical shorting due to their rougher surfaces. Despite promising results, alternate strategies like the thinning of substrate8 and the epitaxial layer transfer on a flexible substrate15 suffer restricted viability in view of the sophisticated multistep process, the unpredictability of transfer, and the limited applicability.
For the aforementioned reasons, it is critical to explore an appropriate substrate that is able to overcome limited thermal and operational stabilities of soft substrates to further advance flexible electronics. A natural muscovite mica (KAl2(AlSi3O10)(OH)2) substrate with unique features like atomically smooth surfaces, high thermal stability, chemical inertness, high transparency, mechanical flexibility, and compatibility with current fabrication methods can be used to effectively deal with these issues. More so, the two-dimensional layered structure of monoclinic mica supports van der Waals epitaxy, which mitigates lattice and thermal matching conditions, thereby significantly suppressing the substrate clamping effect. These advantages have been exploited in the direct growth of functional oxides16,17,18,19,20,21,22,23 on muscovite recently, in view of flexible device applications.
Herein, we describe a protocol to directly grow epitaxial yet flexible lead zirconium titanate (PZT) thin films on muscovite mica. This is achieved through a pulsed laser deposition process relying on the versatile properties of mica, resulting in van der Waals heteroepitaxy. Such fabricated structures retain all the superior properties of epitaxial PZT on rigid single crystalline substrates and exhibits excellent thermal and mechanical stabilities. This simple and reliable approach provides a technological advantage over multistep-transfer and substrate thinning strategies and facilitates the development of much-awaited flexible yet single-crystalline non-volatile memory elements prerequisite for next-generation smart devices with high performance.
1. Fabricating Flexible PZT Thin Films
2. Ferroelectric Characterization
The epitaxial PZT/SRO/CFO/mica thin films were deposited with the pulsed laser deposition technique as outlined in Step 1. Figure 1 shows the growth scheme and Figure 2 shows an actual flexible NVM element based on the PZT.
Mechanical stability is a crucial aspect of flexible device application. The macroscopic ferroelectric performance of the heterostructure against mechanical flexing was evaluated under both tensile and compressive bending. Figure 7a and 7b show the P-E and C-E hysteresis loops of the PZT capacitors under various compressive and tensile bending radii (R). Figure 7c shows constant Psat, Pr, Ec and capacitance values within experimental errors under different bending radius. The corresponding nominal strain values estimated by where η = tf /tS, χ = Yf /YS, Yf is Young's modulus of the PZT layer and YS is Young's modulus of the mica are also marked. These results suggest that the PZT thin film capacitor maintains stable electrical properties under mechanical constraints required for the flexible electronics device applications, which was also checked by Raman spectroscopy20.
The well-saturated and symmetric polarization-electric field (P-E) hysteresis loops and the capacitance-electric field (C-E) with the "butterfly" curves of the heterostructure measured at 1 MHz and temperatures ranging from 25-175 °C for a new device are shown in Figure 8a and 8b, respectively. This ferroelectric capacitor exhibits constant saturation polarization (Psat), a remnant polarization (Pr), a coercive field (Ec) and capacitance in a wide temperature range as shown in Figure 8c. The heterostructure also maintains high retention and endurance at room temperature as well as at 100 °C20. These results imply that the PZT/mica heterostructure can have potential applications in high-temperature electronic devices.
A series of cyclability tests were carried out to validate PZT/mica heterostructures for practical applications. Figure 9 shows P-E loops before and after 1000 bending cycles in both tensile and compressive strain states. The P-E loops at different bending modes are displaced vertically for the sake of convenience. It is noteworthy that the heterostructure retains its ferroelectric behavior even after 1000 bending cycles at a bending radius of 5 mm irrespective of the nature bending strain.
Figure 1. The growth scheme of a flexible memory element on mica. Evacuate the chamber to a base pressure (~10-6 Torr) and raise the sample temperature to 590 °C. Adjust the oxygen pressure to 50 mTorr to grow the CFO. Increase the temperature to 680 °C and adjust the oxygen pressure to 100 mTorr to grow the SRO. Decrease the temperature to 650 °C and adjust the oxygen pressure to 100 mTorr to grow the PZT. Please click here to view a larger version of this figure.
Figure 2. The photograph of a flexible memory element on mica. The flexible memory element can be bent easily. Please click here to view a larger version of this figure.
Figure 3.The schematic configuration for measuring P-E hysteresis loops and C-E curve. Contact the SRO bottom electrode using one probe, while the other probe contacts the Pt top electrodes on the films to measure the electronic properties of the PZT thin film. Please click here to view a larger version of this figure.
Figure 4. Molds with different fixed bending radii (R). The bending mold designs were made/drawn using autoCAD and printed using a 3D printer. These molds of fixed bending radii (R) induce the reported compressive and tensile bending strains (R = ±12.5 mm, ±10.0 mm, ±7.5 mm, ±5.0 mm, ±2.5 mm, the positive (negative) sign corresponds to tensile (compressive) strain that the heterostructures undergo when mounted on them). Please click here to view a larger version of this figure.
Figure 5. Bending stage to perform bending cycles test. The heterostructure length (C) is measured by a rule in unbent state. For the bending cycle measurement, use a computer-aided home built bending setup. The bending stage consists of two arms with grooves to hold thin sheets. One arm is fixed while the other arm can be moved to bend a thin sheet with a stepper motor interfaced with the computer. Please click here to view a larger version of this figure.
Figure 6. Program protocol to perform bending tests. Use a computer-aided home built bending setup to control the movement of the motor. The setup allows the length of the sample to be bent by providing the displacement as small as 1 µm on the bending stage. One can set the bending radii (see 2.3.4) as well as perform bending cycles. Please click here to view a larger version of this figure.
Figure 7. Ferroelectric properties under different bending radii. Electric field dependence of (a) polarization and (b) capacitance under various tensile and compressive bending radii. (c) Saturation polarization (Psat), remnant polarization (Pr), coercive field (Ec) and capacitance as a function of bending radius. Corresponding strain values are also indicated (see text). This Figure has been modified with permission20. Please click here to view a larger version of this figure.
Figure 8. Ferroelectric properties under high temperature. Electric field dependence of (a) polarization and (b) capacitance at different temperatures. (c) Thermal evolution of Psat, Pr, Ec and capacitance. This Figure has been modified with permission20. Please click here to view a larger version of this figure.
Figure 9. Ferroelectric properties after bending cycles. P-E hysteresis loops under tensile and compressive bending radius of 5 mm before and after 10 to 1000 bending cycles. This Figure has been modified with permission20. Please click here to view a larger version of this figure.
The key step in the fabrication of ferroelectric elements lies in the use of a clean and even/flat substrate surface. Though freshly cleaved mica surface is atomically smooth, it is necessary to pay attention to preventing surfaces from suffering visible splintering, split layers, cracks, inclusions, etc. After deposition of the PZT layer, the sample was cooled under a high oxygen pressure (200-500 Torr) to reduce the oxygen vacancies. Ex situ top platinum electrodes were deposited via a predefined mesh to form many Pt/PZT/SRO capacitor elements. To carry out bending tests, the sample was attached to a piece of paper of similar dimensions to enable an easy transfer of the sample between different molds. The molds used to mechanically strain the sample under compressive or tensile states were printed by a 3D printer. During the cycling tests, both ends of the sample were held firmly so as to avoid slipping of mica layers.
However, the inherent small area of uniformity of the PLD technique restricts its applicability in large scale production. The process to choose a good piece of mica without cracks is also time-consuming. Mica cannot be stretched and compressed, and accordingly the devices grown on mica cannot be stretched or compressed, too. A lot of materials grown on mica need a buffer layer to get a good-quality film, which increases the complexity for the production process. These inherent issues restrict the development of flexible devices. Thus, it is necessary to understand in detail the mechanisms governing the nucleation and growth during van der Waals epitaxy and electronic coupling across van der Waals-heterointerfaces in order to circumvent this these issues.
Currently employed strategies for realizing flexible NVME include use of a polymer substrate, thinning of substrate or epitaxial-transfer technique. Though polymer substrates exhibit excellent mechanical compliance, their low temperature stability affects the device performance in a negative way. Also, thinning substrate8 or epitaxial growth and subsequent transfer on flexible polymer substrate15 involves a multistep tedious process. The van der Waals epitaxy involving mica22,23 not only lessens the lattice and thermal matching conditions but also relieves the substrate clamping effect, beneficial for realizing epitaxial systems with performance metrics comparable to single crystalline bulk counterparts as reflected in PZT/mica. Furthermore, 2D layered mica substrate gives the advantage of realizing free-standing-like memory elements that maintain robust ferroelectric behavior against mechanical and thermal constraints. The PZT/mica system possesses the best performance among all the flexible memory elements to date20, which circumvents the issues of various approaches stated above.
The transparency of mica can be exploited to achieve transparent NVME. Due to the nature of (quasi) van der Waals epitaxy, the material database can be expanded beyond the limited material combinations inherent to conventional epitaxy. It is anticipated that van der Waals epitaxy on mica will trigger substantial research interest in the design and development of next generation of flexible electronic devices.
The authors have nothing to disclose.
This work was supported by National Natural Science Foundation of China (Grant Nos. 11402221 and 11502224), the State Key Laboratory of Intense Pulsed Radiation simulation and effect (SKLIPR1513) and the Hunan Provincial Key Research and Development Plan (No. 2016WK2014).
Equipment | |||
hot plate | Polish | P-20 | |
PLD system | PVD products | PLD 5000 | |
Ferroelectric test system | Radiant Technologies Precisions workstations | RT66A | |
Semiconductor device analyzer | Agilent | B1500A | |
Bending molds | home-made | Machined teflon material | |
Bending stage | home-built | Labview interfaced setup which provides a prescise displacemnt as small as 1 micrometer | |
Sputtering system | Beijing Elaborate | ETD-3000 | |
Materials | |||
mica(001) sheets | Nilaco corporation | 990066 | |
conductive silver paint | Ted Pella, INC | No.16033 | |
CoFe2O4 target | Kurt J.Lesker | ||
SrRuO3 target | Kurt J.Lesker | ||
PbZr0.2Ti0.8O3 target | Kurt J.Lesker | ||
Pt target | Hefei Ke jing |