**2. Materials and Methods**

Zr54.46Al9.9Ni4.95Cu29.7Pd0.99 BMG was prepared through arc melting under a highpurity argon atmosphere. To achieve homogeneity of composition, the master alloys were melted at least four times. Then, a BMG rod with a diameter of 6 mm was produced by copper mold suction casting and cut into a length of 27 mm for HRRF. The wear experiment was performed using an MMW-1A pin-on-disk apparatus in open air under dry conditions and with a 3.5% NaCl solution. BMG samples with a size of ϕ4 × 13 mm were machined as wear pins to be rubbed against a rotating steel disk of micro-hardness (HV) 728. In the pin-on-disk test, the normal load was 40 N while sliding at a speed of 0.13 m/s, and the sliding duration was 1800 s. The surface microstructure of the samples was recorded using a JSM-6490LV scanning electron microscope (SEM) with energy dispersive X-ray spectrometry (EDS) and a Cypher S atomic force microscope (AFM). All the pin surfaces of samples and disk surfaces were polished using diamond paste. Under each given condition, three samples were measured to ensure the reliability of the data. The glass structure of alloys was confirmed by an X-ray diffractometer (XRD) with Cu Ka radiation. After a wear experiment, each specimen was cleaned ultrasonically. The weight change of each sample before and after the friction tests was determined using an AUY 120 balance with a precision of 0.0001 g. The wear rate was calculated as [20].

$$\mathcal{W} = \frac{V}{F} \times \mathcal{S} \tag{1}$$

where *V* is the wear volume loss of the sample, *F* is the applied normal load, and *S* is the total sliding distance.

The nano-indentations were carried out over a square area of 48 <sup>×</sup> <sup>48</sup> <sup>µ</sup>m<sup>2</sup> for the as-cast, and samples were treated with a Berkovich diamond tip. The constant depth was 300 nm. Each nano-indentations were penetrated at the same depth (300 nm) at a constant loading rate of 0.05 s−<sup>1</sup> , and the spacing between adjacent indentations was 6 µm.

The HRRF was introduced to manipulate the microstructural heterogeneity of BMGs. As reported in our previous study [19], this fabrication process consists of three steps, heating up to the supercooled liquid region through fast Joule heating, squeezing the supercooled liquid into a copper mold cavity under the preset load, and then rapidly cooling it down to room temperature in the copper mold.
