Abstract
We present the laboratory results of immersion freezing efficiencies of cellulose particles at supercooled temperature (T) conditions. Three types of chemically homogeneous cellulose samples are used as surrogates that represent supermicron and submicron ice nucleating plant structural polymers. These samples include micro-crystalline cellulose (MCC), fibrous cellulose (FC) and nano-crystalline cellulose (NCC). Our experimental data show that particles resembling the MCC lab particle occur also in the atmosphere. Our immersion freezing dataset includes data from various ice nucleation measurement techniques available at seventeen different institutions, including nine dry dispersion and eleven aqueous suspension techniques. With a total of twenty methods, we performed systematic accuracy and precision analysis of measurements from all twenty measurement techniques by evaluating T-binned (1 °C) data over a wide T range (−36 °C < T < −4 °C). Specifically, we inter-compared the geometric surface area-based ice nucleation active surface-site (INAS) density data derived from our measurements as a function of T, ns,geo(T). Additionally, we also compared the ns,geo(T) values and the freezing spectral slope parameter (Δlog(ns,geo) / ΔT) from our measurements to previous literature results. Results show that freezing efficiencies of NCC samples agree reasonably well, whereas the diversity for the other two samples spans for ~ 10 °C. Despite given uncertainties within each instrument technique, the overall trend of the ns,geo(T) spectrum traced by the T-binned average of measurements suggest that predominantly supermicron-sized (giant hereafter) cellulose particles (MCC and FC) generally act as more efficient ice-nucleating particles than NCC with about one order of magnitude higher ns,geo(T). Further, our results indicate significant diversity between dry and aqueous suspension measurement techniques. The ratios of the individual measurements (ns,ind) to the log average of ns,geo(T) range 0.6–1.4 across the examined T range. In general, the ratios of the log average of dry dispersion measurements are higher than those of aqueous suspension measurements. The observed discrepancy may be due to non-uniform active site density for different sizes and/or the alteration in physico-chemical properties of cellulose by liquid-suspending it. Unless otherwise defined, the cellulose system may not be an ideal calibrant. Given such a distinct difference between two subgroups of immersion freezing techniques, standardization of our methods, especially INP sampling and treatment, may be one approach to reduce the measurement diversity and valiability when we deal with a complex material like cellulose. A community-wide effort to identify specimen-specific limitations and characteristics of each technique, as well as consolidating the ns,geo(T) parameterization, is an alternative approach to achieve overall precise and accurate ice-nucleating particle measurements.
