Data from: Mapping ferroelectric fields reveals the origins of the coercivity distribution
Abstract
Better techniques for imaging ferroelectric polarization would aid the development of new ferroelectrics and the refinement of old ones. Here we show how scanning transmission electron microscope (STEM) electron beam-induced current (EBIC) imaging reveals ferroelectric polarization with obvious, simply interpretable contrast. Planar imaging of an entire ferroelectric hafnium zirconium oxide (Hf0.5Zr0.5O2, HZO) capacitor shows an EBIC response that is linearly related to the polarization determined in situ with the positive-up, negative-down (PUND) method. The contrast is easily calibrated in MV/cm. The underlying mechanism is magnification-independent, operating equally well on micrometer-sized devices and individual nanoscale domains. Coercive-field mapping reveals that individual domains are biased “positive” and “negative”, as opposed to being “easy” and “hard” to switch. The remanent background E-fields generating this bias can be isolated and mapped. Coupled with STEM’s native capabilities for structural identification, STEM EBIC imaging provides a revolutionary new tool for characterizing ferroelectric materials and devices.
README: Data from: Mapping ferroelectric fields reveals the origins of the coercivity distribution
This dataset: 10.5061/dryad.8w9ghx3v5
The corresponding article [1] is open-access in ACS Nano and can be found here: 10.1021/acsnano.4c04526.
This dataset contains raw data from nano-positive-up, negative-down (PUND) transport, and scanning transmission electron microscope (STEM) imaging experiments.
[1] Ho Leung Chan, Shelby S. Fields, Yueyun Chen, Tristan P. O’Neill, Megan K. Lenox, William A. Hubbard, Jon F. Ihlefeld, and Brian C. Regan, ACS Nano 2024 18 (31), 20380-20388 DOI: 10.1021/acsnano.4c04526
Description of the data and file structure
Data Structure
The data is organized into the following folders and subfolders:
|- Data/
| |- 20230703_EBIC/
| | |- 20230703_EBIC_1024/
| |- 20230703_PUND/
| |- 20230809_EBIC/
| | |- 20230809_EBIC_calibration_128/
| | |- 20230809_EBIC_calibration_256/
| |- 20230809_PUND/
| | |-20230809_PUND_in_manuscript/
The folder names all begin with a date formatted as YYYYMMDD, indicating the date that the data within them was acquired. Additionally, folders that have "PUND" in their names contain PUND data, and folders that have "EBIC" in their names contain STEM images (BF, ADF, and EBIC).
All filenames have a timestamp as well. PUND data filenames begin with a timestamp formatted as YYYYMMDD_HHMMSS. STEM images begin with a timestamp formatted as HHMMSS, however, the STEM image filename timestamps are all off by about five minutes because the computer that was used to save them does not have a connection to the internet. To match PUND data with the corresponding STEM EBIC data, add five minutes to the STEM EBIC image filename timestamp. The STEM image timestamp+5 minutes is the time that an image acquisition began, not the time that the acquisition ended.
STEM Image Data
All STEM images are .ser files. ImageJ can be used to open them with the plugin "TIA Reader", which can be found at https://imagej.net/ij/plugins/tia-reader.html. Images labeled Acquire 1, 2, 3, or 4 are BF, bottom electrode EBIC, ADF, or top electrode EBIC, respectively. The first and last few rows of every STEM image are "beam-blanked"; these measure the offsets in the transimpedance amplifiers (TIAs) used to collect EBICs. To convert the EBIC images to currents, multiply them by a scaling factor of 0.0033 pA.
Most images were acquired with both electrodes fixed at the same potential. The only exceptions to this are the calibration images: the bottom electrode was biased using a sequence of voltages from 0 V up to 1 V, then down to -1 V, and finally back to 0 V, with steps of 0.25 V. (The HZO is 20 nm thick, so the applied electric fields range from -0.5 MV/cm to 0.5 MV/cm.) One set of 128x128 px STEM calibration images was acquired at each bias voltage. Additionally, sets of 256x256 px STEM calibration images were acquired at 1 V and -1 V. The entire calibration procedure was performed once with the device maximally polarized up, and then again with the device maximally polarized down. The correspondence of EBIC to the electric field can be determined by the calibration procedure.
Nano-PUND Data
The raw data for each nano-PUND measurement is split into two files: a .bin (binary) file and a .xlsx (Microsoft Excel) file. The binary file contains the raw voltage waveform applied to the capacitor and the current response. It is saved in Python (v3.9.7) using NumPy's (v1.22.3) ndarray.tofile()
method as an array of signed 8-bit integers and can be read using NumPy's ndarray.fromfile()
method. The corresponding Excel file contains the parameters necessary to convert the numbers in the binary file to the measured voltages and currents.
The Excel file has the following headers:
- unix_time: The Unix timestamp of this data.
- channel: The channel of the digitizer used for measuring voltages during PUND measurements. Channel 0 corresponds to the PUND waveform voltage. Channel 1 corresponds to the measured device current (from the transimpedance amplifier).
- samples: The number of samples acquired. This will always be the same for both channels.
- x_increment: The time in seconds between each data point.
- y_increment: The scaling factor to apply to convert the saved signed 8-bit integer values in the corresponding .bin file to voltages.
- y_offset: The voltage offset of the digitizer. This should be added to the data after the y_increment has been applied.
- gain: The transimpedance of the TIA during the acquisition, in ohms. Though both rows contain a gain, only channel 1 was attached to the TIA, so the gain should only be used for that channel to convert the voltage measured by the digitizer to the current from the device.
- NDPU: For convenience. If this cell is 0, the voltage waveform from channel 0 is a PUND waveform. If it is 1, then it is an NDPU waveform (the opposite polarity of a PUND waveform). Inspection of the corresponding waveform in channel 0 also reveals whether it is a PUND or an NDPU waveform.
- pulse_delay: For convenience. The time in seconds between triangle pulses in the PUND waveform.
- plateau_time: For convenience. This is always 0. It indicates that we never used trapezoidal pulses for PUND, instead opting for triangular pulses.
- init_v: For convenience. The initialization voltage amplitude is used for PUND. This is always 7 V.
- init_t: For convenience. The rise time in seconds of the initialization pulses in our PUND waveforms. This is always 250 us. As described in the corresponding manuscript, the ramp rate is always the same, 28 kV/s.
- init_str: For convenience. This string provides the sign of the initialization pulses in the PUND waveform, e.g. a (1,0,1,0,-1,0,-1) is one PUND pulse.
- sec_period: For convenience. The rise time of the var pulses in our PUND waveforms; i.e. one triangle is 2*sec_period. This is NOT for the initialization of PUND pulses at the beginning of each waveform.
- var_str: For convenience. This is like the init_str but for the var pulses after the init pulses.
- temperature: This was hardcoded in our acquisition code always to represent room temperature (22 C). It is not a measured temperature, so it should not be relied upon.