**2. Fabrication and Experiment Setup**

According to previous studies, the AlScN films with Sc contents below 27% are prone to break down near the coercive field, and a distinct ferroelectric polarization occurs with a Sc content of more than 27%. As the Sc contents increase, the coercivity, saturation polarization and remanent polarization all decrease [8,11,14]. Therefore, we prepared AlScN thin films at a specific Sc content of 30% in this study. In the study, 200 nm thick *Al*0.7*Sc*0.3*N* films were deposited using a pulsed DC magnetron reactive sputtering (EVATEC CLUSTERLINE ® 200 MSQ) with a single 4-inch *Al*0.7*Sc*0.3*N* alloy target, as shown in Figure 2a. The films deposited in this way grow along the *c*-axis direction [17,18]. In order to apply the electric field across the film thickness direction, 100 nm Pt and 200 nm Mo were used as the bottom electrodes, respectively. Then, 100 nm Al was used as the top electrodes for a simplified process flow. The final device structure is shown in Figure 2b.

X-ray diffraction (XRD) was used to characterize the crystalline quality of *Al*0.7*Sc*0.3*N*. A comparison and evaluation of the number of abnormal orientation grains on the film surface was obtained from SEM images. Then, the dielectric properties were measured using Keysight B1500 to test the I−V and C−V curves of the samples. To characterize the ferroelectric polarization of the films in Pt and Mo, hysteresis tests with different polarization voltage and frequencies were performed using a Radiant Multiferroic II II tester. However, other components of the system, including electrodes, leads and interfaces, could dominate the electrical response rather than the intrinsic properties of the material of interest [19]. Therefore, PUND measurements were used to separate the different components of the electrical response of a ferroelectric film. In this measurement, a sequence of five pulses was introduced. The first pulse (pulse 1) flips the polarization of the sample to a defined state. The second pulse is in the opposite direction of the first pulse and *Vmax* is

maintained for one pulse width to ensure that the sample is saturated with polarization, at which point, the polarization value *P*<sup>1</sup> is recorded. After the second pulse, their is a wait of one pulse width and the second polarization value *P*<sup>2</sup> is recorded. After certain pulse delay, a third pulse is applied and the third polarization value *P*<sup>3</sup> is recorded at its end. Then, there is a wait of one pulse width to record the polarization value *P*<sup>4</sup> and subtract *P*<sup>1</sup> from *P*<sup>3</sup> to obtain *dP*, since *P*<sup>1</sup> contains both switching and non-switching components, whereas *P*<sup>3</sup> contains only non-switching components, so *dP* can represent the correct remanent polarization. Typically, *P*2-*P*<sup>4</sup> is written as *dP<sup>r</sup>* , since *P*<sup>2</sup> and *P*<sup>4</sup> are the polarization values recorded after waiting for a pulse width and losing a certain polarization, and *dP<sup>r</sup>* should be equal to *dP*. Pulse 4 and pulse 5 are similar to pulse 2 and pulse 3, only in the opposite direction, in order to obtain −*dP* and −*dP<sup>r</sup>* . Figure 3 shows the pulse sequence of the PUND test.

**Figure 2.** (**a**) Schematic diagram of magnetron sputtering deposition of *Al*0.7*Sc*0.3*N*. (**b**) Stacking schematic of Pt/Mo-AlScN-Al structure.

**Figure 3.** Schematic of the pulse sequence for the PUND test. The pulse intensity is *Vmax*, and each pulse maintains a pulse width. When a pulse ends, a pulse width plus pulse delay is waited for in order to input the next pulse. The polarization values are recorded twice for each pulse cycle.
