Preparation and testing of MEMS-based samples for in situ heating and biasing in the TEM/STEM
T.A. Macgregor, W. A. Smith, T. P. Almeida, and I. MacLaren
SUPA School of Physics and Astronomy, University of Glasgow, Glasgow, United Kingdom
Tom Macgregor is a PhD candi-date working in the Materials and Condensed Matter Physics Group at the University of Glasgow. His current work involves inves-tigating the structure of high temperature dielectric systems at the nanoscale using electron microscopy. This process conduc-ing in situ experiments to observe the properties of these ceramics systems under temperature and field. He studied Chemistry at undergraduate level at the Uni-versity of York before completing a MSc degree in Materials Science and Engineering at the University of Leeds.
In recent years, MEMS systems have been developed that enable in-situ heating and electrical biasing in the electron microscope. This report describes recent advances in how to prepare insulating samples from bulk material for use with such MEMS chips. The method is described in detail, showing both concept and reality, and the resulting specimen is tested under vacuum up to temperatures of 300°C and fields of 0.8 MV/m and behaves as it should. This method will be of use for anyone wanting to prepare bulk materials or thin films for understanding how temperature and field affect the atomic structure of dielectrics with nanometre resolution.
Funding was provided from EPSRC grant nos.: EP/P015514/1 and EP/P013945/1, together with a studentship for TAM (EP/N509668/1). Many thanks to Drs. Steven Milne and Andrew Britton (Leeds) for providing materials and helpful discussions.
Tom Macgregor, School of Physics and Astronomy, University of Glasgow, Glasgow, G12 8QQ
Temperature Stable Relaxor Ferroelectrics
Dielectric ceramic materials have been the subject of continuous research since their discovery in the 1940s. The electrical properties of these materials change with respect to mechanical stress (piezoelectricity) or temperature (pyroelectricity), they are used intensively in a range of devices, including ultrasound generators and receivers, sensors, and actuators, but of especial interest to our work, capacitors. Due to the need for sensing and control electronics in extreme conditions, for example at high-temperature, underground oil sites; there is considerable interest in developing dielectric systems capable of operating at high temperatures ( > 300°C)[2, 3].
Dielectric materials are divided into several sub-classes according to key properties; this work concerns relaxor ferroelectrics, which exhibit both a remanent polarisation and broad dielectric permittivity with respect to temperature and frequency[4–7]. The diffuse response of these relaxor materials is believed to originate from multiple polar phases within the material coupling to each other and producing a dielectric response across a range of temperatures, and typically with a notable frequency dependence. These polar regions are believed to exist as nanoscale regions while the macroscale structure of the material is defined by a non-polar matrix. Such a model has been predicted via Monte Carlo simulations and confirmed experimentally by measurements of the refractive index, inferred reflectivity, NMR studies, and neutron scattering functions[8–11]. Perovskite systems, with the general structure ABO3, can demonstrate relaxor behaviour when multiple elements occupy the A or B sites, it is believed that atomic scale structural disorder leads to the diffuse dielectric response, however the structure-property relationship of these systems is not well understood[5, 12]. Temperature stable relaxor ferroelectrics have been suggested as an ideal candidate for extreme temperature applications due to their stable dielectric permittivity at high temperatures, low loss tangent value and lead- free composition, after reviewing a number of different compositions (1-x)(Ba0.8, Ca0.2)TiO3-xBi(Mg0.5, Ti0.5)O3 (BCT-BMT) was shown to have the best dielectric response[13–15]. The lack of toxic lead in this system also reduces the environmental impact of using this system compared to the current generation of materials.
To get a better understanding of how the polar nanoregions in such materials behave as a function of field and temperature, it is necessary to observe them using suitable techniques in Transmission Electron Microscopy or Scanning Transmission Electron Microscopy. That requires a method for making in-situ measurements. Recent advances have provided platforms for doing just this (and other in-situ measurements) using systems patterned on MEMS chips together with suitable holders, and a number of companies now provide such systems. This article presents a method for the preparation of dielectric samples for TEM/STEM investigation under field over a range of temperatures using a commercial MEMS chip support and a Focused Ion Beam (FIB) liftout procedure.
FIB and MEMS Systems
FIB instruments allow the preparation of extremely thin (often <100 nm) specimens for use in TEM, usually by a lift-out procedure, whereby a sample region is protected with ion-beam deposited platinum, trenches are created to either side using ion beam sputtering, the sample is attached to a micromanipulator needle, cut away from the bulk material, and lifted out, and then attached to a support for final milling[17, 18].
One of the key benefits of FIB-based preparation methods is the ability to integrate them with the newer Micro Electrical Mechanical System (MEMS) in-situ specimen supports. MEMS systems consist of a single chip that combines assembly of electrical contacts and resistors with an integrated circuit, these chips are arranged so that a specimen can be placed across an electron transparent window in an arrangement that completes the circuit. Typically the surface of a MEMS devices will be coated with a silicon nitride membrane surrounding the contact points and electron transparent windows; this membrane allows for the heat to be generated and controlled right next to the sample all on this light, thin, and low thermal mass membrane. The use of local heating elements means that the power required to heat specimens prepared on MEMS devices is much lower for conventional TEM heating holders and the reduced power and tiny thermal mass means that temperature stability is excellent meaning that MEMS-based specimens can be imaged with minimal sample drift (allowing nano- or atomic-scale imaging at hundreds of degrees Celsius)[16, 20].
The MEMS chips used in this work are designed to be mounted into a purpose-built sample rod that transmits external signals to the sample from outside the microscope, thus enabling in situ heating and biasing experiments. There has been previous work on preparing samples for use with such sample supports using FIB procedures.
The preparation of samples for in-situ heating experiments has hitherto involved the use of pretilted holders inside the FIB chamber, which enables the milling of material at a low offset angle to the ion beam, greatly reducing the likelihood of specimen damage due to ion exposure[16, 21]. Pretilted holders have enabled the preparation of several types of specimen, including magnetic thin films such as FeRh, membranes of orientated silicon deposited with various metallic layers, and various atomised aluminium alloys[16, 18, 21, 22].
FIB-based preparation methods have previously been used to prepare sample for in-situ biasing experiments on other class of materials such as Pt/HfO2/TiN devices for use in resistive memory applications when exposed to range of applied electric fields. Similar preparation techniques have also been used to investigate the performance of LiCoO2 for use as a solid-state batteries, as well as doped perovskite materials that are used in photovoltaic cells[24, 25]. The goal of this project is to prepare specimens suitable for simultaneous heating and biasing,
The MEMS device used in this work was a DENSsolutions Lighting D9+ Nanochip configured for use in a JEOL ARM 200cF microscope, this system allows for controlled heating and biasing (including pre-programmed sequences) via a dedicated control hardware and software.
Concept of the Preparation Method
The main conceptual layout of the preparation procedure is shown in Figure 1. Prior to placing the specimen in the FIB chamber, the ceramic pellet was coated with amorphous carbon, a-C, using a sputter coater in order to prevent charging in the FIB. A normal liftout is then performed in Stage A and the sample is then temporarily attached to a comb- shaped holder[26, 27].
Figure 1. Schematic diagram of the principle stages in the sample preparation method with a) showing Stage A: initial liftout, b) Stage B: thinning on the temporary mounting; and c) Stage C: final mounting and final thinning on the MEMS chip.
Stage B consists of then thinning the sample, especially the back face, until the sample is approximately 500nm thick with the ion beam whilst on the support. The vacuum is then broken, the comb holder reoriented by 90°, and the MEMS chip placed on the sample holder too (attached with copper tape), the vacuum system is re-pumped, the needle is attached to the lamella, and this is then finally lifted off the comb support.
Stage C is then comprised of lifting the lamella away from the comb support, placing it onto the MEMS chip (on some pre-deposited platinum bars) and securing it by depositing platinum on both sides. Once the specimen was secured to the MEMS chip it was removed from the chamber again, remounted to a 45° stub, replacing this in the stage, re-pumping the microscope and doing final thinning of the front face at a shallow angle (around 7°) to the surface, including thinning away any protective platinum on the front surface of the specimen. The resulting sample attached to the inner contacts on the MEMS chip and capable of responding the heating and biasing signals in addition to being thin enough of analysis with analytical electron microscopy methods.
Sample Preparation and Mounting
FIB-Based Sample Preparation
This work uses samples of BCT-BMT of a range of compositions (different x values) which were produced via solid state synthesis following the sintering processes detailed by Zeb et al. to produce sintered pellets of ceramic material[28, 29]. Focused ion beam liftout and mounting on a MEMS chip was performed using a FEI Nova Nanolab 200 DualBeam instrument.
Figure 2 shows a sequence of secondary electron images illustrating the FIB liftout process. This starts with the milling of trenches to extract the lamella from the bulk material prior to attaching it to the comb holder (Figure 2a and 2b). Two protective layers of platinum were applied to the surface of the specimen using the electron and ion beams respectively in order to reduce the amorphisation of surfaces under exposure to the ion beam. The trench milling was carried out using an automated series of exposures to produce the tapered trenches shown.
Figure 2. FIB-SEM secondary electron images of Stages A and B of the preparation process in practice: a) milling the trenches to isolate the lamella for liftout; b) the extracted lamella attached to the comb holder; c) the lamella after back face thinning; d) a view of the back face from the electron beam; e) and f) the specimen before and after front-face thinning and a final low voltage exposure.
Using the Ga FIB’s ion source, high and low current pulses were used to reduce the lamella thickness to approximately 500nm according to Stage B of the process. The thinning process produced an 8µm long lamella as shown in Figure 2c-d. The thinning process consisted of several two-minute exposures of 0.28µA were applied with the ion beam operating at 30kV with a 2° offset were used until the specimen was approximately 800nm thick; after that, a series of one minute pulses at 92pA and a 1.5° offset were used to remove an additional 200nm of material.
The final thinning stages were completed using a serial milling procedure, also termed a ‘cleaning cross section’ where the exposure proceeds in series of lines across the targeted area. The use of this process meant the milling process could be monitored by taking an image with the electron beam after each line of milling. Once the back face was suitably thin, a low voltage, low current (5kV, 96pA) exposure was applied with a 7° offset to the ion beam in order to remove any Ga+ ions implanted on the specimen surface.
Figure 2 e-f highlights the results of Stage C of the preparation process where the front face of the specimen was thinned after the specimen had been attached to the MEMS chip and it had been mounted on the 45° pre-tilted stub. In this configuration applying a 10° stage tilt places the chip at a 3° angle to the ion beam (which is incident at 52° in this specific instrument); this enables exposure of the front face of the specimen to ion beam while the rest of MEMS chip is unexposed and therefore not susceptible to damage.
Once the specimen was in this position using 30 second beam ion beam exposures (96pA 10kV) were used to thin the front face. As before the process of thinning the front face was also monitored with the electron beam using cleaning cross sections. A final two-minute, 5kV, 200 pA) exposure was applied at the end of the thinning process after the stage tilt was increased to 15° in order to complete the FIB based proceedure.
Preparation Refinements for Biasing Experiments
The preparation described above is suitable for in-situ heating experiments; unfortunately, the protective platinum at the top of the specimen forms a conductive path linking one contact to another as shown in Figure 3a, which would make the specimen short. Applying additional exposures with the ion beam removes the excess platinum from the top of the specimen as shown in Figure 3b and enables the application of an electrical field laterally across the specimen.
Figure 3. Schematic of specimen layouts for a) heating-only experiments; and b) simultaneous biasing experiments with electric field running laterally across the specimen.
In order to test this refinement of the procedure, a second specimen of BCT-BMT was prepared. In addition to the steps above, several additional 96pA cleaning cross sections were applied to the sample to remove the top bar of platinum from section of the lamella. Figure 4 shows the sample prior to being mounted to the MEMS chip (Figure 4a) as well as before and after front face thinning (Figure 4b and 4c respectively).
Figure 4. FIB-SEM secondary electron images of a second specimen prepared for biasing experiments with a) the specimen mounted for the back-face thinning; and b) and c) showing the specimen on the chip before and after front face thinning, respectively.
Testing of the Sample under Heating and Biasing
The specimen mounted on the MEMS chip was loaded into the DENSsolutions sample specialised sample rod and the appropriate contacts made. Our previous paper has demonstrated that the procedure generates a sample of suitable quality for high quality STEM imaging and EELS mapping[30, 31]. For the purposes of the present work, the main focus was to prove that a sample prepared by the full procedure is suitable for heating and biasing investigations. For this reason, it was placed into a turbo-pumped vacuum station for testing. Firstly, it was connected to an electrically grounded interconnect unit (ICU) which was in turn connected to a temperature control unit and source measuring unit (SMU) [Keithley 2450 SourceMeter]. The temperature controller and SMU are used to send heating and biasing signals to the specimen. A second serial connector on the rod allows for specimen to be tilted about the β axis, although this was not actually needed in the current test. The heating control is controlled using the Impulse software provided by DENSsolutions. Figure 5 is a schematic showing how all the different controllers are connected to allow for simultaneous heating and biasing. A USB interface was used to connect the SMU and heating unit to the same laptop in order to control all the variables within the DENSsolutions Impulse software interface.
Figure 5. A schematic showing the connections between temperature control unit and source measuring needs to carry out simultaneous heating and biasing. An interconnect unit is used to send signals to the appropriate resistors on the MEMS chip.
The temperature of the specimen was controlled using the Impulse software provided by DENSsolutions. Figures 6 a) and b) show the results of two separate tests. The top graphs show the instructed and actual temperature-time profiles, with the difference between the two shown in the second row of graphs. There is an understandable temperature lag (of order 20°C) on heating and cooling, but otherwise, the match between set and actual temperature is exceptionally good. In each test, a programmed voltage profile was applied (third row of graphs) and the current was measured. No statistically significant change in current was registered in any case on a change of voltage, either at 300°C or room temperature – the reading of ~ 1.5-2 pA is probably a zero offset on the meter rather than a real current, and this suggests that these samples are suitably insulating for field biasing experiments. The first test increased the voltages in steps up to 1V while the second test increased this to 8V. That is, the sample withstood up to 0.8 MV/m (8 kV/cm), which is significant, although higher voltages might have to be applied to match those applied in determining P-E loops in the thesis work of Zeb.
Figure 6 .Profiles of signals applied to and measured from the sample as a function of time during the in-situ test for two separate experiments on the same sample, a) and b). The graphs are from top to bottom: set and actual temperatures; difference between set and actual temperatures (ΔT); applied voltage; and measured current in pA, with significant noise and probably a small zero offset.
All this demonstrates that the samples fabricated by this process are suitable for (scanning) transmission electron microscopy and can cope with reasonable temperatures and fields, and that there is little or no current flow (which could have been an issue with surface damage or platinum re-deposition). The next stage of this work will now be to combine such preparation with careful electron microscopy to actually observe structural alterations in polar nanoregions as a consequence of varying temperature and electric field.
Summary and Conclusions
This work presents an improved procedure for preparing samples of dielectric material for in situ experiments in a (S)TEM. This method has the benefits of using established and easily automated steps, like trench milling, to prepare the initial lamella. It is suitable for a variety of different materials, and not just the thin films used in the predecessor method to this.
The method demonstrates the advantage of temporary mounting to a conventional TEM FIB specimen holder for the first part of thinning since this means that the back face thinning can be done at a wider of tilt angles prior to attaching it the chip. This then provides more flexibility in positioning on the chip, and denser ion-beam deposited platinum can be used for attachment, resulting in a more robust and reliable bonding compared to deposition using the electron beam.
An experiment with the specimen under vacuum conditions confirms that specimens prepared this manner can be used in simultaneous heating and biasing experiments and rapid temperature and field changes do no damage to the sample. Moreover, the sample seems to be properly insulating meaning that structural changes in the dielectric as a result of field and/or temperature can be studied without any loss of field due to a leakage current. This preparation method enables the in-depth studies of dielectric materials in future work and will lead to advances in the understanding their behaviour. This advance will now be used to develop an understanding of the nanoscale dielectric response of relaxor ferroelectrics and allied materials, including investigating different compositions of BCT-BMT.
 W. D. Kingery, ms New York, and L. V. Sydney Toronto D Kingery, Introduction to Ceramics, 2nd ed. London: Wiley, 1976.
 M.-J. Pan and C. A. Randall, “A brief introduction to ceramic capacitors,” IEEE Electrical Insulation Magazine, vol. 26, no. 3, pp. 44–50, May 2010.
 D. Shaddock, “High-temperature electronics and applications,” Chip.Scale Rev., vol. 19, no. 4, pp. 46–50, 2015.
 G. Smolensky, “Ferroelectrics with diffuse phase transition,” Ferroelectrics, vol. 53, no. 1, pp. 129–135, Apr. 1984.
 L. E. Cross, “Relaxor ferroelectrics,” Ferroelectrics, vol. 76, no. 1, pp. 241–267, Dec. 1987.
 A. F. Devonshire, “Theory of ferroelectrics,” Advances in Physics, vol. 3, no. 10, pp. 85–130, Apr. 1954.
 A. J. Moulson and J. M. Herbert, Electroceramics, 2nd ed. Chichester, UK: John Wiley & Sons, Ltd, 2003.
 G. Burns and F. H. Dacol, “Glassy polarization behavior in ferroelectric compounds Pb(Mg1/3Nb2/3)O3 and Pb(Zn1/3Nb2/3)O3,” Solid State Communications, vol. 48, no. 10, pp. 853–856, Dec. 1983.
 R. Blinc, V. V. Laguta, B. Zalar, and J. Banys, “Polar nanoclusters in relaxors,” Journal of Materials Science, vol. 41, no. 1, pp. 27–30, Jan. 2006.
 G. Xu, J. Wen, C. Stock, and P. M. Gehring, “Phase instability induced by polar nanoregions in a relaxor ferroelectric system,” Nature Materials, vol. 7, no. 7, pp. 562–566, Jul. 2008.
 I. Levin et al., “Local structure in BaTi O3-BiSc O3 dipole glasses,” Physical Review B, 2016.
 R. GM, R. A. Cowley, S. N. Gvasaliya, S. G. Lushnikov, B. Roessli, and G. M. Rotaru, “Relaxing with relaxors: a review of relaxor ferroelectrics,” Adv Phys, vol. 60, no. 2, pp. 229–327, Apr. 2011.
 A. Zeb and S. J. Milne, “High temperature dielectric ceramics: a review of temperature-stable high-permittivity perovskites,” Journal of Materials Science: Materials in Electronics, vol. 26, no. 12, pp. 9243–9255, Dec. 2015.
 A. Zeb and S. J. Milne, “Stability of High-Temperature Dielectric Properties for (1 − x)Ba0.8Ca0.2}TiO3 - xBi(Mg0.5,Ti0.5)O3 Ceramics,” Journal of the American Ceramic Society, vol. 96, no. 9, pp. 2887–2892, Sep. 2013.
 T. Roncal-Herrero, J. Harrington, A. Zeb, S. J. Milne, and A. P. Brown, “Nanoscale compositional segregation and suppression of polar coupling in a relaxor ferroelectric,” Acta Materialia, vol. 158, pp. 422–429, Oct. 2018.
 S. Vijayan, J. R. Jinschek, S. Kujawa, J. Greiser, and M. Aindow, “Focused Ion Beam Preparation of Specimens for Micro-Electro-Mechanical System-based Transmission Electron Microscopy Heating Experiments,” Microscopy and Microanalysis, vol. 23, no. 04, pp. 708–716, Aug. 2017.
 I. MacLaren, M. Nord, C. Jiao, and E. Yücelen, “Liftout of High-Quality Thin Sections of a Perovskite Oxide Thin Film Using a Xenon Plasma Focused Ion Beam Microscope,” Microscopy and Microanalysis, vol. 25, no. 1, pp. 115–118, Feb. 2019.
 T. P. Almeida et al., “Preparation of high-quality planar FeRh thin films for in situ TEM investigations,” Journal of Physics: Conference Series, vol. 903, p. 012022, Oct. 2017.
 H. H. Pérez Garza et al., “MEMS-based system for in-situ biasing and heating solutions inside the TEM,” in European Microscopy Congress 2016: Proceedings, Weinheim, Germany: Wiley-VCH Verlag GmbH & Co. KGaA, 2016, pp. 237–238.
 S. Cai, C. Gu, Y. Wei, M. Gu, X. Pan, and P. Wang, “Development of in situ optical–electrical MEMS platform for semiconductor characterization,” Ultramicroscopy, vol. 194, no. June, pp. 57–63, Nov. 2018.
 S. Vijayan, B. A. Bedard, M. A. Gleason, H. R. Leonard, D. L. Cote, and M. Aindow, “Studies of thermally activated processes in gas-atomized Al alloy powders: in situ STEM heating experiments on FIB-cut cross sections,” Journal of Materials Science, vol. 54, no. 13, pp. 9921–9932, Jul. 2019.
 T. P. Almeida et al., “Quantitative TEM imaging of the magnetostructural and phase transitions in FeRh thin film systems Supplementary information This PDF file includes: Fast Fourier Transform smoothing of DPC images Micromagnetic Modelling.”
 A. Zintler et al., “FIB based fabrication of an operative Pt/HfO2/TiN device for resistive switching inside a transmission electron microscope,” Ultramicroscopy, vol. 181, pp. 144–149, Oct. 2017.
 Y. Gong et al., “In Situ Atomic-Scale Observation of Electrochemical Delithiation Induced Structure Evolution of LiCoO 2 Cathode in a Working All-Solid-State Battery,” Journal of the American Chemical Society, vol. 139, no. 12, pp. 4274–4277, Mar. 2017.
 Q. Jeangros et al., “In Situ TEM Analysis of Organic–Inorganic Metal-Halide Perovskite Solar Cells under Electrical Bias,” Nano Letters, vol. 16, no. 11, pp. 7013–7018, Nov. 2016.
 C. A. Volkert and A. M. Minor, “Focused Ion Beam Microscopy and Micromachining,” MRS Bulletin, vol. 32, no. 05, pp. 389–399, May 2007.
 R. M. Langford and A. K. Petford-Long, “Preparation of transmission electron microscopy cross-section specimens using focused ion beam milling,” Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films, vol. 19, no. 5, pp. 2186–2193, Sep. 2001.
 A. Zeb and S. J. Milne, “Temperature-stable dielectric properties from −20°C to 430°C in the system BaTiO3–Bi(Mg0.5,Zr0.5)O3,” Journal of the European Ceramic Society, vol. 34, no. 13, pp. 3159–3166, Nov. 2014.
 A. Zeb and S. J. Milne, “Dielectric Properties of Ba0.8 Ca0.2 TiO3 -Bi(Mg0.5 ,Ti0.5 )O3 -NaNbO3 Ceramics,” Journal of the American Ceramic Society, vol. 96, no. 12, pp. 3701–3703, Dec. 2013.
 I. MacLaren, R. Villaurrutia, B. Schaffer, L. Houben, and A. Peláiz-Barranco, “Atomic-Scale Imaging and Quantification of Electrical Polarisation in Incommensurate Antiferroelectric Lanthanum-Doped Lead Zirconate Titanate,” Advanced Functional Materials, vol. 22, no. 2, pp. 261–266, Jan. 2012.
 I. MacLaren and Q. M. Ramasse, “Aberration-corrected scanning transmission electron microscopy for atomic-resolution studies of functional oxides,” International Materials Reviews, vol. 59, no. 3, pp. 115–131, Apr. 2014.
 A. Zeb, “Lead-free Dielectric and Piezoelectric Ceramics,” PhD ThesisUniversity of Leeds, 2015.