This supplementary information provides additional details for the measurement 3 techniques of immersion freezing of illite NX particles with S1.1. suspension techniques and 4 S1.2. dry-dispersed particle measurement techniques (both in alphabetical order as in Table 5 1). The discussions of measurement uncertainties of temperature and n s for each measurement 6 technique are also provided. We note that the uncertainty in frozen fraction (α) used in 7 calculating n s may not be adequate, since the sensitivity of Δα (an increase or a decrease in 8 frozen fraction) is much higher at high temperatures which unexceptionally coincide with a 9 low fraction of frozen illite NX. The BINARY setup is an optical freezing apparatus that makes use of the change in An immersion-freezing method was used to obtain INP temperature spectra for NX- 46illite clay, both when in bulk suspension and for size-selected particles. 47For the bulk clay, a 0.5 wt% suspension was made in 10 mM sodium phosphate buffer 48(at pH 8.7 to match the pH of the sample and to prevent flocculation, and filtered through a on a series of 20-fold dilutions to 3.1 x 10 -6 wt% in the same buffer. 52Polydisperse NX-illite particles were generated for size selection using the simple diameter-pore filtered). Filters were dried on foil in a particle-free, laminar flow cabinet, as 67were filter holder components after excess water was removed with a gas duster. 68After particle collection, the filter was transferred to a sterile, 50 mL Falcon Experimental uncertainties: The uncertainties for T and n s are ± 0.7 ˚C and ± 30%, 157 respectively. The n s uncertainty includes errors of the frozen fractions of drops, the specific 158 particle surface area, the particle masses per drop, and the drop sizes. fractions of drops, the specific particle surface area, the particle masses per drop, and the drop 225Experimental uncertainties: The accuracy of the thermistor embedded in the lower 226 aluminum block designed for the -40 < T < 0 °C and digitized with precision ±0.01°C. 227Repeatability of the temperature where 50% of pure water pico drops froze via homogeneous 228 nucleation was -35.7 °C ± 0.1 °C (n = 5, average diameter of drops ~86 µm). In comparison,

No abstract