Dairy samples (DS) were made using unflavored dairy ice cream mix and adhered to regular ice cream overrun requirements for the federal ice cream standard [2]. The unflavored dairy mix recipe included the following: milk, cream, sugar, nonfat dry milk, dry sweet whey, guar gum, mono and diglycerides, cellulose gum, carrageenan, and locust bean gum. Non-dairy samples (NDS) frozen dessert used unflavored Meadowvale dairy-free dessert mix, which contains coconut cream, sugar, water, tapioca syrup solids, and locust bean gum. To standardize dairy and NDS, the samples were both frozen in a 6-quart batch freezer (Model CB-350, Emery Thompson Middleby Inc., USA) for 7 minutes at 165 rpm, so as not to exceed 100% overrun. Samples were then placed in a –28.8°C freezer for 48 hours to allow static freezing to occur.

DS and NDS were separated into 4 different storage temperatures: –28.8°C (control), –17.7°C, –15.0°C, and –12.2°C. These temperatures were selected as treatments to replicate the manufacturer’s freezer temperature (–28.8°C) as well as varying temperatures reaching –12.2°C, which is Unilever’s goal temperature for commercial freezers. One cup from each sample group contained a thermocouple to measure internal temperature during storage.

The proximate composition and nutritional information were taken from the supplier’s label information.

The percent overrun was calculated using the following formula.

%   o v e r r u n = w e i g h t   o f   m i x   u s e d − w e i g h t   o f   i c e   c r e a m w e i g h t   o f   i c e   c r e a m × 100

While scooping samples into 118.3 mL cups, 6 samples from each set (temperature and sample type) were weighed to 35 g. This standardized weight was used as an initial measurement to determine weight loss over time. Samples were measured again after 12 weeks.

Samples for SEM analysis were first placed in a –28.8°C freezer to ensure products were frozen solid. The freeze dryer (Model HRC 100, Harvest Right Inc., USA) was pre-frozen for 30 minutes to allow the chamber to reach –28.8°C, to prevent any loss of structural integrity during sample transfer. Samples were placed on the same tray in the center of the freeze dryer. At 400 mTorr and –34.4°C, the drying cycle slowly increased the internal chamber temperature. Internal chamber temperature reached –5.7°C to sublimate ice, and then pressure slowly increased during the final drying stage. Samples were fully freeze-dried after 24 hours. Freeze-dried samples were kept in airtight containers until analyzed with Electron Microscopy.

Ice crystal pore size was analyzed using a PC-based variable-pressure SEM (Model S-3400N, Hitachi, Japan). Samples were transported to the Clemson University Microscopy Laboratory and removed from airtight containers one at a time for analysis. A laboratory spatula was used to cut samples in half and remove a small interior piece. The interior piece was mounted on the SEM stage and adjusted for height. Sample heights ranged between 3 mm and 5 mm.

The SEM was placed in variable-pressure vacuum mode (VP-SEM), which operates at greater pressures to allow higher resolution imaging of non-conductive samples, including food products. The samples were tested at 30 Pa with an accelerating voltage level of 20 kV. Images were captured using the Quartz personal computer imaging (PCI) software. Using the Quartz PCI measuring tool, pore size was determined by measuring the diameter of ice crystal pores shown in each micrograph.

A total of twelve individuals with prior experience in dairy product evaluation, either through successful completion of a University Evaluation of Dairy Products course, were initially recruited for the study. Due to scheduling conflicts that prevented full participation, four individuals were excluded, resulting in a final panel of eight participants. These eight panelists underwent a structured training program consisting of three sessions focused on the assessment of “smoothness” as a sensory attribute. During training, panelists were presented with reference samples representing distinct levels of smoothness and grittiness, and were trained to accurately distinguish and rate these textural differences. Participants ranged in age from 19 to 31 years. Prior to beginning the study, all participants provided written informed consent and completed both a standardized sensory panel script and an allergen disclosure form, in accordance with institutional ethical guidelines (IRB2020-285).

Samples were removed from freezers and placed in a refrigerator until the internal temperature reached –12.2°C, defined by the International Dairy Foods Association as the proper temperature for texture and body analysis [12]. (International Dairy Foods Association, Storing and Handling, 2023). Samples stored at –12.2°C were removed from the freezer and immediately placed on the tray. Internal temperature was measured using a thermocouple placed directly in the center of each sample.

Lids from sample cups were removed prior to paneling to prevent participants from knowing which sample was associated with which temperature. For the evaluation form, samples were given a unique code for the participants to use. Two repetitions were completed by participants, and samples were rearranged for each repetition, to ensure panelists would not associate placement with and sample. Participants were given one tray at a time with a spoon, a cup of water, and a sensory evaluation form.

Participants ranked samples on a 5-point hedonic scale (Figure 1), starting at 1, for extremely smooth, to 5, for extremely gritty and rough (1 = extremely smooth, 2 = very smooth, 3 = moderately smooth, 4 = very gritty, 5 = extremely gritty). The term “smoothness” was used to describe creaminess and iciness as an all-encompassing term to describe both DS and NDS, and to ensure panelists were not confusing dairy judging terminology (“icy”) with this descriptive analysis ranking scale. This term was also used to prevent participants from attempting to compare the formulations (dairy vs. non-dairy).

The experiment was replicated 2 times using two different batches of dairy and non-dairy frozen desserts prepared in different time periods. The main effects of desert type (dairy or non-dairy), storage temperature (–12.2°C, –15.0°C, –17.8°C, and –28.9°C), storage time (4, 8, and 12 weeks) as well as the two and three interactions were tested using the analysis of variance for their significant effect on ice crystal pore size, sensory grittiness, and weight loss. When the main effect or interaction had a significant effect (p < 0.05), means were separated using Tukey’s test or the pdiff command in the Statistical Analysis System (SAS).

Nutrition information for each formulation based on the supplier is shown in Table 1, and the overrun for samples ranged between 72.5% and 76.6%, as shown in Table 2.

Samples from each storage time and temperature were analyzed using SEM. Two pictures were taken from each sample and then measured using the Quartz PCI measuring tool. We hypothesized that ice cream stored at –15.0°C and –12.2°C would see significant differences in ice crystal size and morphology regardless of sample type due to the temperatures’ proximity to recrystallization temperatures. Dairy frozen dessert showed significant differences in ice crystal pore size based on storage time and temperature (Table 3). Pore sizes for the lower temperatures (–28.9°C and –17.8°C) gradually increased in size between weeks 0 and 4 and then showed a greater increase between weeks 4 and 8. Samples stored at –28.9°C had pore sizes significantly smaller than those at other storage temperatures. As hypothesized, samples at lower temperatures were more resistant to recrystallization since samples were held at temperatures low enough to minimize ice cream thawing. Lower temperatures (–15°C and –12.2°C) had ice crystals grow significantly between weeks 0 and 4. The larger increase in pore size may be due to the drastic temperature change between initial static freezing (–28.9°C) and storage temperatures. On average, DS and NDS did not change significantly after 8 weeks, regardless of storage temperature (Table 3 and Table 4).

All samples increased in pore size; however, there was more homogeneity in pore size (smaller range in size) and shape at the lowest temperature (–28.9°C). Lower storage temperatures caused pores to take on sharp, uneven shapes. Over time, NDS did not increase in pore size as much as DS. NDS tended to stay consistent, with the exception being samples in weeks 8 and 12 stored at –12.2°C. These samples contained the highest average pore size for all NDS. Goff et al. (2003) [13] and Donhowe et al. (1991) [14] reported that crystals over 50 μm in diameter are sensed as grainy, whereas finer crystals in the 30–40 μm diameter range create a desired creaminess sensation. Average pore sizes above 50 μm are bolded in Tables 3 and 4, reflecting the effect of higher storage temperature on projected frozen dessert sensory texture.

The most representative micrographs from DS stored at –28.9°C and –12.2°C for 12 weeks are shown in Figures 2A and 2B, and NDS stored at –28.9°C and –12.2°C for 12 weeks are shown in Figures 2C and 2D. These micrographs reflect the difference in pore size in both DS and NDS after 12 weeks of storage at –28.9°C versus –12.2°C.

It was hypothesized that the NDS would have smaller, more consistent ice crystal size and shape due to the higher fat content. Partially coalesced fat globules obstruct the growth of ice crystals, thus reducing the rate of recrystallization and lowering grittiness. Additionally, fat plays a role in stabilizing small, unstable ice crystals and air cells, which can prevent the formation of unfavorable, large ice crystals [15].

Sensory panelists ranked grittiness on a 5-point hedonic scale, with 1 being least gritty and 5 being most gritty. Increases in pore size and morphology diversity are experienced during the tasting of ice cream. Ice cream with sharper ice crystals and sizes > 50 μm is sensed as “icy” or “gritty” on the palate [10, 11]. On average, DS stored at –12.2°C and –15°C were grittier by week 12 (3.94 and 3.88, respectively) than DS stored at –17.8°C and –28.9°C [2.88 and 2.6, respectively (Figure 3)]. This parallels the pore size results related to the 50 μm threshold reported previously [10, 11]. Both dairy and non-dairy ice creams increased in grittiness over 12 weeks; however, dairy ice cream ranked significantly more gritty than NDS (Figures 3 and 4). With the exception of the highest storage temperature (–12.2°C), NDS’ grittiness did not increase significantly over 12 weeks between –28.9°C, –17.8°C, and –15.0°C. Grittiness for NDS all averaged between 1.63 and 3.13, with no week significantly grittier than the others. The difference in formulation and the presence of higher fat in the non-dairy sample may be responsible for less noticeable signs of recrystallization.

The relationship between pore size and sensory panel score, shown for DS in Figure 5 and NDS in Figure 6, had a general positive correlation for both dairy and NDS. As storage temperature increased (Figures 3 and 4), average pore size and grittiness also increased. While the trend was observed in both types of samples, the correlation was more prevalent in DS. The NDS had a smaller range of pore sizes and 2 outlier values (52 μm pore size, 3.1 grittiness, and 55 μm pore size, 1.6 grittiness) that reduced the R² of the relationship between pore size and grittiness score. Also, the pore size of NDS had less of an effect on grittiness than it did for DS.

Weight loss was calculated over the 12-week storage period, and storage temperature had a significant impact on weight loss, regardless of the type of frozen dessert (Table 5). There was significantly lower weight loss for both dairy and NDS stored at –28.9°C compared to desserts stored at other storage temperatures. Shrinkage and weight loss of frozen desserts during storage are the result of collapsed, discrete air pockets. Recrystallization of partially melted ice crystals and loss of air phase volume were found to be proportional to storage temperature. Higher storage temperatures allow for partial melting and the formation of unfavorably large ice crystals, leading to increased weight loss [16]. The relationship between storage temperature and both weight loss and pore size reflects a similar positive stepwise relationship, with both pore size and weight loss increasing with increased storage temperature.

Further research looking at the relationship of milkfat and plant-based fat content to ice crystal formation would be useful in determining the role of types of fats in preserving frozen desserts. Milkfat has been studied extensively; however, as more non-dairy frozen desserts enter the market, plant-based fats could be important in determining the best formulations. Additionally, current research looking at the performance of reformulated frozen desserts may provide useful information on how these products may withstand higher storage temperatures. The commercial freezer storage target of –12.2°C would not be practical, given the significant structural changes that occur in the product. However, increasing storage to –17.8°C or –15°C (depending on the length of storage) would reduce energy usage with minimal effect on product quality. The conclusion from this study is that increasing freezer storage temperature from –28.9 to –17.8°C had little effect on pore size, coarseness, and weight loss. It is supported by previous recommendations for deep-freezing to minimize recrystallization [24]. Storage of frozen desserts at –15.0°C is possible with minimal loss in quality if consumed in the first 4 weeks of storage. The study also calls for further investigation into the protective role of plant-based fats for frozen dessert quality. This is a subject underexplored in contrast to milkfat, which is consistent with the direction of current food science research, given the demand for vegan and lactose-free products.