Composition, microstructure, rheology, and meltdown behavior of commercial nondairy frozen desserts

Abstract

Abstract

The demand for nondairy and plant‐based products has increased, but there is still a need for more information about and improvement in these products, especially when it comes to frozen desserts. Similar to ice cream, which simultaneously is an emulsion, dispersion, and foam, nondairy frozen desserts also have a complex structure. As a starting point, 15 commercial nondairy frozen desserts, marketed as offering an ice cream‐like experience, were purchased and evaluated for compositional, physical, structural, rheological, and meltdown properties. The correlations between the parameters of these characteristics were also studied. The parameters with the greatest variability in percentage terms for each property after melt‐down were fat content ranging from 3.8% to 18.6%, surface tension from 26.4 to 45.4 mN·m⁻¹, fat destabilization from 3.6% to 94.4%, yield stress from 0.0 to 25.2 Pa, and final height from 0% to 77.3%. Interestingly, the maximum overrun found was about 82%, considered low compared to economy dairy ice creams. The fat content was the most significant parameter that influenced drip‐through rate, height change rate, and final height. Additionally, yield stress was the only rheological parameter that correlated with the meltdown behavior, which affected only one of them, the induction time. Therefore, in the meltdown profile, the lag phase was driven by yield stress, while the fast‐melting phase and the plateau phase were influenced by the fat content. Moreover, four samples presented uncommon drip‐though curves, which were visually recorded. This study offers a baseline understanding of the compositional, structural, rheological, and melting characteristics of commercial nondairy frozen desserts available in the United States.

Practical Application

The compositional, microstructural, rheological, and meltdown characteristics of nondairy frozen desserts available in the US market were evaluated. The information can be used for industry research and development, understanding the challenges, and advancing knowledge regarding nondairy frozen desserts.

1. INTRODUCTION

There has been an increasing global consumer interest in the adoption of a plant‐based diet, driven by considerations of health, sustainability, and ethics (Grossmann & McClements, 2021). Consequently, the consumption of plant‐based dairy alternatives is experiencing a significant increase in popularity, and it is anticipated that the global market for these products will achieve a value of USD 47.95 billion by the year 2028 (Fior Markets, 2021). Currently, many products have attained considerable economic success, including alternatives for meat and milk, which are produced from plant sources. According to Future Market Insights (2023), the worldwide market for nondairy frozen desserts is experiencing increasing appeal among young individuals in developed countries and other developing economies worldwide. However, the challenges associated with developing plant‐based products that meet the preferences and standards of consumers have presented significant difficulties. The main factor contributing to this phenomenon can be ascribed to the complex composition and structure of the original products, which presents difficulty in reproducing these attributes using plant‐based components (Grossmann & McClements, 2021).

Ice cream can be characterized as a colloidal food with a complex structure consisting of various particles (such as crystalline fat droplets, air bubbles, and ice crystals) dispersed inside a freeze‐concentrated aqueous solution. Typically, it is produced from a blend of milk, supplementary fat (such as cream or other fats), nonfat milk solids, emulsifiers, hydrocolloids, flavoring agents, and various other constituents (Goff, 1997). In contrast, the term “plant‐based or nondairy frozen desserts” generally denotes products that are entirely free of dairy, eggs, and other animal‐derived ingredients. These products can be manufactured utilizing similar processing techniques to those used for the production of dairy ice cream; however, the components utilized in nondairy frozen desserts differ in a variety of functional parameters (Goff & Hartel, 2013). Various types of plant‐based milk substitutes, such as cashew, pea, soy, or oat milk, as well as refined ingredients like flours, concentrates, or isolates, can be utilized for producing ice cream analogs (McClements & Grossmann, 2022).

Nevertheless, the production of plant‐based frozen desserts poses a technological challenge as it is inherently difficult to mimic the unique taste and structure provided by milk and dairy components (Pinto et al., 2012). The primary distinction between dairy ice creams and their nondairy substitutes is the utilization of plant proteins and fats instead of animal‐derived ingredients. Hence, it is essential to choose plant proteins that exhibit comparable emulsifying and water‐holding characteristics to those found in dairy proteins in order to maintain similar fat globule diameters and viscosity values within the ice cream mixture (McClements & Grossmann, 2022). Furthermore, the role of stabilizers and emulsifiers extends beyond their function in ice cream, as they must also contribute to viscosity, fat structure, and aeration properties. These functionalities are particularly important in the case of various plant‐based proteins, as their interfacial properties are not as favorable as those of milk protein. Furthermore, due to the absence of lactose, the addition of extra sugar (potentially up to 20%) is required to achieve freezing properties comparable to ice cream (Goff & Hartel, 2013).

As a result, there are a number of factors that affect the structural characteristics of plant‐based frozen desserts, but the interactions among them are not well understood or characterized, leading to products that fall short of consumer expectations. Also, a comprehensive analysis of commercial plant‐based frozen desserts in the United States has not been documented in the literature, addressing their compositional, microstructural, and rheological characteristics. Therefore, this study bridges a critical knowledge gap between academic research and the frozen dessert industry. In this context, the aim of this research was to survey and evaluate the rheological, structural, and melting features of commercial nondairy frozen desserts.

2. MATERIALS AND METHODS

2.1. Materials

A total of 15 different vanilla nondairy frozen dessert products (ingredient lists can be found in Table S1) were analyzed to evaluate their microstructural, rheological, and meltdown characteristics. Three containers of each product were bought from nearby supermarkets in Madison, Wisconsin. The commercial nondairy frozen dessert samples used in the study were randomly numbered to reduce analytical bias. Since nondairy frozen dessert products were purchased, formulations, production, and storage conditions prior to buying were unknown, which is a limitation of the nondairy frozen dessert samples. Following the purchase, the products were stored in a hardening freezer at a temperature of −28.9°C prior to analysis. The analyses were performed in triplicate, using three independent samples from the same product lot.

2.2. Total solids, density, and total fat

The total solid contents of commercial nondairy frozen dessert samples were measured using a Microwave Moisture/Solids Analyzer (CEM Smart System 5; CEM Corporation) (Warren & Hartel, 2014). Total fat and protein contents were estimated from the label information of the samples. The density values (ρ _mix) of the samples were determined using Equation (1) based on the total solids (% total solids) and fat (% fat) (Hui et al., 2005), where “% Water” is the moisture content and “Wt. per liter of water” is water density:

$${\rho }_{\mathrm{mix}}\left(\frac{\mathrm{Wt}.}{\mathrm{L}\mathrm{mix}}\right)=\frac{(\mathrm{Wt}.\mathrm{per}\mathrm{liter}\mathrm{of}\mathrm{water})}{\frac{\%\mathrm{fat}}{100}\times 1.07527+\left(\frac{\%\mathrm{total}\mathrm{solids}}{100}-\frac{\%\mathrm{fat}}{100}\right)\times 0.6329+\frac{\%\mathrm{Water}}{100}}.$$ (1)

2.3. Overrun and freezing point

Overrun values of samples were calculated according to Equation (2) (Clarke, 2004; Warren & Hartel, 2014), where V _fd, the frozen dessert volume, and m ₁, the weight of the frozen dessert, were obtained from the label; ρ _mix is the density of the mix:

$$\mathrm{Overrun}=\left(\frac{V_{\mathrm{fd}}\times {\rho }_{\mathrm{mix}}}{m_1}-1\right)\times 100.$$ (2)

The freezing point was measured using an osmometer (3250 Model; Advanced Instruments). For the purpose of measuring the freezing point, 25 µL of melted product was pipetted into the tube. The osmolality measurement was recorded, and afterward, the freezing point was determined through the use of the standard curve, which was generated by establishing a correlation between osmolality and the corresponding freezing point, utilizing a known osmolality standard solution (Advanced Instruments).

2.4. Fat globule/particle size distribution

A Malvern Mastersizer 3000, employing laser light diffraction and scattering techniques, was used to evaluate the particle size distribution of melted commercial frozen desserts (Malvern Instruments Ltd.). Approximately two drops of melted frozen dessert, maintained at 4°C, were added to the cell to reach obscuration levels ranging from 13% to 15%. The dispersant refractive index (deionized water) was set at 1.33, with an absorbance of 0.01. The refractive indexes for the dispersed fat phase varied from 1.44 to 1.48, depending on the fat sources listed in the ingredient list: coconut, safflower, soybean, sunflower, corn, low‐erucic‐acid rapeseed oils, or cocoa butter.

Typically, fat destabilization is evaluated by comparing the individual fat globule peak from the frozen dessert mix's initial emulsion curve with the peak representing destabilized fat globules in the melted frozen dessert curve (Bolliger et al., 2000). Since the initial mixes were unavailable for this study, some assumptions, as detailed by Warren and Hartel (2014), were employed to estimate the extent of fat destabilization. Initially, to mitigate potential interference from unknown particles affecting laser light diffraction and scattering in the curves, the assumption was made that the equipment exclusively measured the fat dispersion phase, encompassing individual or clustered globules. Furthermore, fat destabilization is calculated by determining the ratio of the percent volume of the destabilized fat peak (generally starting around 5 µm) to the total percent volume of fat particles, as illustrated in Figure 1. This total volume consists of the initial emulsion peak (primarily ranging from 0.1 to 5 µm) and the destabilized fat peak (second peak).

FIGURE 1.

Illustration of a (A) light scattering curve and (B) microscopy image (scale bar = 100 µm) representing melted ice cream (sample 800). The initial emulsion, represented by fat/oil globules, shows a peak ranging from 0.2 to 1.5 µm. Additionally, the destabilized fat, characterized by the formation of fat globule clusters, exhibits two peaks within a range from 1.5 to 111 µm. Ingredient list of sample 800: water, sugar, coconut oil, sunflower oil, nonanimal whey protein, contain less than 2% of maltodextrin, natural flavor, vanilla bean seed, calcium potassium phosphate citrate, salt, disodium phosphate, carob bean gum, and mono and diglycerides.

2.5. Surface tension

Pendant drop tensiometry using the Kruss DSA 30R Drop Shape Analyzer (Kruss) and a needle with an outer diameter of 1.835 mm was carried out to assess the surface tension of the melted samples at the air–water interface. The samples were equilibrated to room temperature (22 ± 1°C) before measurement. A droplet was suspended for 1 h at the tip of the capillary needle within a sealed humid cuvette to reduce evaporation. The Young–Laplace equation, used by the ADVANCE software (Kruss) to analyze the droplet contour and determine surface tension, describes the relationship between the Laplace pressure (ΔP), the interfacial tension (σ), and the curvature radii (r ₁ and r ₂) of the droplet surface in a pendant drop setting:

$$\mathrm{\Delta }P=\sigma \times \left(\frac{1}{r_1}-\frac{1}{r_2}\right)$$ (1)

The droplet shape is analyzed using grayscale image processing. A shape factor, referred to as B, is iteratively adjusted until the calculated droplet shape aligns with the observed shape. The interfacial tension is determined using the density difference (Δρ), between the liquid sample and the air, and the value of B parameter from the iteration.

2.6. Air cell size distribution

Air cell size was analyzed at −6°C inside a temperature‐controlled insulated glove box, following the method outlined by Chang and Hartel (2002). The images were obtained using an optical light microscope (Accu‐Scope 3000‐LED Microscope; Microscope Central). Approximately six images were taken to ensure a minimum count of 300 air cells per container. Subsequently, air cell analysis was performed using Image‐Pro Plus software (version 7.0; Media Cybernetics, Inc.), with the resulting data compiled in a custom‐written Microsoft Excel spreadsheet.

2.7. Ice crystal size distribution

Ice crystal size analysis was conducted at −15°C inside a temperature‐controlled insulated glove box, following the method outlined by Donhowe et al. (1991). Images were captured using an optical light microscope (Accu‐Scope 3000‐LED Microscope; Microscope Central). Approximately 10 images were taken to ensure a minimum count of 300 ice crystals per container. Subsequently, ice crystal images were traced. Measurements were performed using Image Pro Plus software (version 7.0; Media Cybernetics, Inc.), with the resulting data compiled in a custom‐written Microsoft Excel spreadsheet.

2.8. Rheological measurements

Rheometry was carried out using a rotational rheometer with a cup and bob geometry (DHR‐2; TA Instruments) at 0°C. The frozen dessert was melted at 4°C. The melted nondairy frozen dessert structure was stirred, and visible bubbles were removed during sampling of 23 mL. The shear rate from 0.001 to 100 s⁻¹ with 10 points per decade was applied to the sample for 10 min. Then, the shear rate from 100 to 0.001 s⁻¹ was applied to the sample for another 10 min. After that, the apparent viscosity was determined at a shear rate of 50 s⁻¹.

To evaluate the flow behavior of the melted commercial nondairy frozen products, an up and down flow ramp test was applied to the melted products at 0°C. As the up‐flow ramp disrupts the structure in melted ice cream (Freire et al., 2020), the breakdown of structure and release of remaining stress were also expected from the melted commercial nondairy desserts. Therefore, after the structure disruption, the down‐flow ramp provided a smooth curve to evaluate the flow behavior. The Herschel–Bulkley model revealed the highest coefficient of determination (R ²).

$$\sigma ={\sigma }_0+k\times {\dot{\gamma }}^n$$ (2)

Yield stress (σ _0HB), consistency coefficient (k _HB), and the flow behavior index (n _HB) were obtained.

2.9. Meltdown

Approximately 80 g of frozen dessert product was sliced into a cylindrical shape from the center of the container. The meltdown tests were carried out at room temperature (22.3 ± 0.3°C). The piece of the product was placed in the middle of a metal screen (with three holes per centimeter) held by a metal ring. A 1000‐mL beaker placed on a scale (PioneerTM; Ohaus) was used to collect the melted product through the screen. Meltdown parameters, weight, and height were recorded for 2 h. Images were also obtained for visual comparisons. The weight (g) of the melted product was plotted against time (min). The drip‐through rate was obtained from the slope of the linear part of the curve (Bolliger et al., 2000).

2.10. Statistical analysis

The data were processed using JMP statistical software (JMP Pro 17.0; SAS Inst.). One‐way ANOVA and Tukey's HSD tests (α < 0.05) were used to evaluate the means of the commercial products. Furthermore, Pearson's correlation was used to evaluate the linear correlations among pairs of variables. Total fat was included in Pearson's correlation analysis, whereas protein content was not included in any of the statistical analyses.

3. RESULTS AND DISCUSSION

3.1. Compositional and structural properties

The results of the compositional and structural properties of 15 different commercial plant‐based frozen desserts are given in Table 1. The variability in the characteristics can be attributed to the diverse compositions of individual frozen dessert formulations, as well as variances in processing parameters, both of which were unknown. For example, dasher speed, specific ingredients, levels, the type of freezer used, and storage conditions can contribute to the variation in features observed in nondairy frozen dessert products. Table 1 shows the average values of total solids, density, overrun, and freezing point of 15 nondairy frozen dessert samples and the significant differences between samples at the p < 0.05 level. Also, total fat and protein percentages were estimated using information from the label of the samples.

TABLE 1.

Compositional and structural properties of commercial nondairy frozen desserts.

Sample code	TS (%)	MP density (g·mL−1)	Fat (%)	Ptn (%)	FP (°C)	Overrun (%)	FD (%)	ST (mN·m−1)	AC (µm)	IC (µm)
540	42.2 ± 0.2a	1.078 ± 0.001g	18.6	<2	−3.34 ± 0.03f	82.3 ± 0.1a	89.1 ± 1.1a	31.5 ± 1.0c	32.9 ± 3.1abc	42.5 ± 3.4c
381	35.7 ± 0.5ef	1.116 ± 0.002b	6.1	1	−2.75 ± 0.01cd	77.8 ± 0.4b	93.0 ± 0.6a	42.5 ± 1.1ab	29.2 ± 1.3bc	54.1 ± 2.3bc
900	35.9 ± 0.6ef	1.098 ± 0.002de	9.5	1	−2.32 ± 0.10a	65.0 ± 0.4h	18.5 ± 2.4f	40.2 ± 1.0b	29.9 ± 3.1bc	42.1 ± 5.7c
767	37.2 ± 0.7cde	1.097 ± 0.003def	10.8	<1	−2.93 ± 0.11de	69.7 ± 0.5de	3.6 ± 1.2g	43.9 ± 0.7ab	42.8 ± 3.9a	49.1 ± 7.5c
516	35.5 ± 0.6ef	1.128 ± 0.002a	3.8	2	−2.44 ± 0.11ab	71.0 ± 0.4d	75.8 ± 4.9bc	45.4 ± 1.7a	42.0 ± 3.7a	55.6 ± 4.9c
849	37.3 ± 0.5cde	1.110 ± 0.002bc	8.7	2	−2.61 ± 0.09bc	68.3 ± 0.4ef	69.6 ± 8.7c	42.0 ± 1.1ab	38.9 ± 3.5ab	43.5 ± 2.6c
732	40.4 ± 0.9ab	1.100 ± 0.004d	13.0	1	−3.30 ± 0.12f	73.5 ± 0.6c	89.4 ± 0.9a	27.7 ± 0.4cd	33.6 ± 1.2abc	51.9 ± 4.6c
465	32.1 ± 0.3g	1.059 ± 0.001h	14.0	2	−2.74 ± 0.07cd	67.0 ± 0.2fg	93.3 ± 0.2a	39.4 ± 2.8b	40.2 ± 3.9ab	89.5 ± 18.0a
670	39.2 ± 0.4bc	1.117 ± 0.002b	8.8	1	−2.91 ± 0.08de	55.9 ± 0.3j	86.3 ± 0.8ab	28.9 ± 2.7cd	35.3 ± 4.3abc	52.1 ± 2.1c
238	36.3 ± 0.6e	1.089 ± 0.002f	11.7	<1	−2.25 ± 0.06a	66.7 ± 0.4g	32.0 ± 0.8e	40.4 ± 1.9b	33.8 ± 3.6abc	43.1 ± 1.2c
489	36.7 ± 0.6de	1.095 ± 0.002def	10.8	1	−2.34 ± 0.03ab	69.4 ± 0.4e	45.7 ± 5.1d	28.8 ± 1.8cd	25.2 ± 1.6c	42.0 ± 3.0c
800	38.8 ± 0.8bcd	1.078 ± 0.003g	15.8	3	−2.29 ± 0.05a	49.2 ± 0.5k	90.7 ± 1.7a	42.9 ± 1.3ab	35.7 ± 7.5abc	41.1 ± 4.0c
533	36.6 ± 0.6e	1.104 ± 0.002cd	9.2	2	−2.24 ± 0.12a	59.7 ± 0.4i	12.5 ± 1.0fg	39.5 ± 0.7b	40.6 ± 1.5ab	45.9 ± 2.7c
364	39.6 ± 0.6b	1.089 ± 0.002f	14.4	1	−3.17 ± 0.04ef	54.7 ± 0.4j	94.4 ± 0.3a	30.8 ± 1.9cd	35.4 ± 6.5abc	49.6 ± 1.0c
510	34.0 ± 1.5fg	1.090 ± 0.006ef	9.5	1	−2.74 ± 0.16cd	81.0 ± 1.1a	36.3 ± 6.9de	26.4 ± 1.1d	32.5 ± 4.9abc	71.3 ± 6.9ab

Note: Fat and Ptn—fat and protein content estimated from the label; TS—total solids; MP density—melted product density; FP—freezing point; FD—fat destabilization; ST—surface tension; AC—mean air cell size; IC—mean ice crystal size. Means not sharing a common letter within a column are significantly different (p < 0.05).

The fats used in nondairy frozen desserts are usually highly saturated to provide the desired structure for the product (Goff & Hartel, 2013). Also, it is crucial to use plant fats/oils that have similar crystallization behavior to milk fat to mimic the structure of dairy ice cream (McClements & Grossmann, 2022). In the present study, there was a great variation in the fat content of the commercial samples, ranging from 3.8% to 18.6%, with a mean of 11.0% ± 3.8%. Similarly, Warren and Hartel (2014) indicated that the fat percentage of commercial dairy ice cream samples showed a significant variation between 0.01% and 14.3%. On the other hand, the fat content of nondairy frozen desserts was reported to be between 10% and 12% (Beegum et al., 2022; Hasan et al., 2023).

The addition of protein to dairy ice cream enhances its nutritional value and also significantly affects the structural attributes (Segall & Goff, 1999). As shown in Table 1, most of the commercial samples had a protein content of around 1%–2%, except one sample that contained 3% total protein. In general, this is lower than the protein content of regular dairy ice cream, which is between 2.5% and 4% (Goff & Hartel, 2013). The studies of Beegum et al. (2022) and Ghaderi et al. (2021) reported an average protein content between 3% and 4% in nondairy frozen desserts.

Surface tension arises from the cohesive forces between molecules in the same phase (VanWees et al., 2022). The type and amount of component adsorbed at the water–air interface determine the surface tension, which directly influences the air cell distribution and stability (Goff & Hartel, 2013). The surface tension of melted frozen desserts varied between 26.4 and 45.4 mN·m ^{− 1}, with an average of 36.7 ± 6.8 mN·m ^{− 1}. For comparison, a dairy frozen dessert (containing about 18% sugar, 11.4% milk solids nonfat, 7% hydrogenated vegetable oil, 0.5% blend of stabilizer and emulsifier, 0.1% vanilla powder, and 0.03% salt) has a surface tension around 47 mN·m ^{− 1} (Abbasi & Saeedabadian, 2015). The wide range in this study is due to the presence of surface‐active components at higher or lower levels in the formulations. These components are amphiphilic, including mainly emulsifiers and proteins, anchoring the hydrophilic part in the serum phase and hydrophobic part in the air phase (Goff & Hartel, 2013). More studies are necessary to understand the presence of surface‐active components in nondairy frozen desserts.

The total solid content of the samples plays an important role in providing a smoother texture and firmer body to the end product (Goff & Hartel, 2013). As demonstrated in Table 1, there is a significant difference in the total solid values among samples, between 32.1% and 42.2%, which is quite similar to commercial dairy ice cream samples with a range from 31.1% to 42.6% (Warren & Hartel, 2014). In general, nondairy frozen dessert samples with high total solids had higher fat content when compared to samples with lower total solids. The densities of nondairy frozen desserts were found to be significantly different among samples, ranging between 1.05 and 1.12 g·mL⁻¹. There was an inverse relationship between fat content and density (r = −0.8307, p < 0.0001), where the density of the mix decreased with the increased fat content, as previously reported by Goff and Hartel (2013).

The freezing point of frozen desserts is dependent on the concentration of the soluble components and changes according to the composition (Goff & Hartel, 2013). In the present study, the freezing point of commercial nondairy frozen dessert samples ranged from −3.3 to −2.2°C with a mean of −2.7 ± 0.4°C. Furthermore, nondairy frozen dessert samples with high total solids were found to have lower freezing points (r = −0.5386, p = 0.0383). In contrast, Ghaderi et al. (2021) reported a freezing point of −0.88°C for soymilk‐based frozen desserts, which is notably higher than the freezing points observed for the commercial nondairy frozen desserts analyzed in the present study. Since the direct effect of sweeteners and nonfat milk solids on the freezing point has been reported previously, the considerable variation in the characteristics can be attributed to the diverse compositions of each sample (Goff & Hartel, 2013).

Overrun refers to the quantitative measurement of air incorporated into frozen desserts during the manufacturing process. Strict management of overrun is crucial in ice cream production, as it has a direct impact on the yield of final products (Goff & Hartel, 2013). In the present study, the overrun values of commercial nondairy frozen dessert samples were calculated from density and found to be between 49.2% and 82.3% with a mean of 67.4% ± 9.5%. Previous research on commercial ice creams (Warren & Hartel, 2014) indicated that the overrun values of the samples ranged from 21.7% to 119%. Also, they reported that ice creams with higher fat content on the market typically have lower overrun values to obtain the desired quality. However, the same trend was not observed in commercial nondairy frozen desserts used in the present study.

In ice cream products, partial coalescence is the primary mechanism of fat destabilization during dynamic freezing, where partially crystalline fat globules collide and share liquid fat due to shear forces from the dasher in the freezer (Goff & Hartel, 2013; Méndez‐Velasco & Goff, 2012; Sofjan & Hartel, 2004). This partial coalescence can create a tridimensional network of fat globules influenced by formulation and process. Fat flocculation has also been reported to occur in plant‐based frozen desserts (Cheng et al., 2016; Ng et al., 2023). Here, given the unknown ingredients and uncontrolled process conditions of the commercial products, it is assumed that the fat/oil globules of the initial emulsion create a new structure in the final product similar to partially coalesced fat structures during dynamic freezing. These clusters resulted in fat destabilization that ranged from 3.6% to 94.4%, with an average of 62.0% ± 33.5% (Table 1). This range is wider compared to commercial ice creams observed in the study from Warren and Hartel (2014), which was between 2.6% and 55.3% fat destabilization. Figure 2 illustrates the fat destabilization range of this study using samples 381 and 767. Sample 381 presented low fat destabilization, and sample 767 showed high fat destabilization (Table 1). No linear relationship was observed between overrun and fat destabilization (Table 2), aligning with a study of ice cream by Liu et al. (2022). In contrast, a positive linear correlation between those parameters has been identified in other studies (Segall & Goff, 2002; Sofjan & Hartel, 2004; Warren & Hartel, 2018). This difference may be attributed to variations in formulations and unspecified manufacturing processes used in the production of the commercial nondairy frozen desserts of this survey.

FIGURE 2.

Microscopy images (scale bar = 100 µm) illustrating the fat globule size distribution for samples 381 and 767. Ingredient lists: 381—water, sugar, maltodextrin, almonds, expeller safflower oil, potato starch, cellulose gel, sunflower lecithin, guar gum, locust bean gum, sea salt, vanilla extract, vanilla bean seeds, cellulose gum, and natural flavor; 767—oat milk (filtered water whole oat flour), organic cane sugar, organic coconut oil, organic tapioca syrup, cane sugar, cocoa, pea protein, apple juice concentrate (for color), chocolate liquor, natural flavor, guar gum, locust bean gum, sea salt, sunflower lecithin, and xanthan gum.

TABLE 2.

Rheology parameters of commercial nondairy frozen desserts.

Sample code	$\mathrm{Visc}{\mathrm{o}}_{0^{\circ }\mathrm{C}\mathrm{at}50{\mathrm{s}}^{-1}}$	σ 0HB (Pa)	k HB (Pa·s n )	n HB
540	0.299 ± 0.02cdef	0.7 ± 0.1c	0.96 ± 0.07bcd	0.71 ± 0.00cdef
381	0.082 ± 0.006fg	0.0 ± 0.0c	0.17 ± 0.03cd	0.83 ± 0.04bc
900	0.443 ± 0.018bc	0.2 ± 0.0c	1.27 ± 0.09bcd	0.74 ± 0.01cdef
767	0.331 ± 0.029cde	0.2 ± 0.3c	0.97 ± 0.15bcd	0.73 ± 0.02cdef
516	0.073 ± 0.008fg	0.0 ± 0.0c	0.14 ± 0.01cd	0.84 ± 0.01bc
849	0.423 ± 0.137bc	1.9 ± 2.0bc	1.42 ± 0.55bc	0.66 ± 0.04def
732	0.621 ± 0.221b	6.5 ± 2.8b	1.13 ± 0.30bcd	0.80 ± 0.04bcd
465	0.018 ± 0.002g	0.1 ± 0.1c	0.02 ± 0.00d	1.01 ± 0.03a
670	1.614 ± 0.070a	25.2 ± 4.6a	5.92 ± 1.36a	0.62 ± 0.07ef
238	0.307 ± 0.147cdef	1.8 ± 2.3bc	0.96 ± 0.60bcd	0.70 ± 0.06cdef
489	0.265 ± 0.029cdef	0.5 ± 0.4c	1.28 ± 0.19bcd	0.60 ± 0.05f
800	0.168 ± 0.053defg	0.2 ± 0.2c	0.48 ± 0.13cd	0.74 ± 0.03cdef
533	0.204 ± 0.037cdefg	0.2 ± 0.1c	1.99 ± 0.66b	0.43 ± 0.04g
364	0.131 ± 0.007efg	1.3 ± 0.5c	0.19 ± 0.15cd	0.94 ± 0.15ab
510	0.380 ± 0.010bcd	1.3 ± 0.3c	0.89 ± 0.03bcd	0.77 ± 0.02cde

Note: σ _0HB, k _HB, and n _HB are yield stress, consistency coefficient, and the flow behavior index obtained from Herschel–Bulkley model parameters at 0°C, respectively. Means not sharing a common letter within a column are significantly different (p < 0.05).

The air cell size distribution can affect properties such as texture and meltdown rate (Sofjan & Hartel, 2004; VanWees et al., 2020). The average air cell size in commercial nondairy products ranged from 25.2 to 42.8 µm, with an overall mean size of 35.2 ± 5.0 µm (Table 1). This range was close to commercial ice creams, between 17.1% and 39.5 µm, as reported by Warren and Hartel (2014). Additionally, air cell size and fat destabilization did not exhibit any correlation (Table 1). In contrast, in dairy ice cream, Warren and Hartel (2018) found an inverse linear correlation. They suggested that formulation and process conditions, such as type and concentration of emulsifiers and high shear forces in dynamic freezing, had a positive impact on the creation of smaller air cells. Furthermore, no relationship was observed between air cells and overrun in this survey. In contrast, Sofjan and Hartel (2004) and Warren and Hartel (2018) found an inverse correlation between air cells and overrun in dairy ice cream. Studies are needed to understand those relationships in nondairy frozen desserts.

Ice crystals are the main microstructural component in ice cream, influencing its rheology, meltdown, and among other properties. The ice crystal size average of the commercial nondairy frozen desserts varied between 41.1 and 89.5 µm, with a mean size of 51.6 ± 13.1 µm. The ice crystal size average for commercial ice creams is 48.1 µm, ranging from 26.3 to 67.1 µm (Warren & Hartel, 2014). A negative relationship was observed between ice crystal size and total solids (Table 1), but no correlation was found with freezing point. An inverse relationship between ice crystal size and total solids corroborated previous ice cream studies (Donhowe et al., 1991; Flores & Goff, 1999).

3.2. Rheological measurements

The serum phase viscosity can influence various characteristics of nondairy frozen desserts, including sensory, texture, and meltdown properties. The apparent viscosity of the melted nondairy frozen desserts at 50 s⁻¹ varied between 0.02 and 1.61 Pa·s, with an average of 0.36 ± 0.38 Pa·s. Ghaderi et al. (2021) reported a higher viscosity in soy‐based ice creams compared to conventional and sesame milk‐based ice creams. Freire et al. (2020) reported that high levels of stabilizer, overrun, and polysorbate 80 led to melted ice creams high in residual viscosity, a measurement from the creep test at low shear rates. Furthermore, the study reported that the apparent viscosity at 50 s⁻¹ of ice cream mixes ranged between 0.02 and 0.29 Pa·s.

For the most part, ice cream mix and melted ice creams are non‐Newtonian fluids, exhibiting shear thinning behavior. They are usually fit to the Herschel–Bulkley rheological model, quantified by a yield stress, consistency coefficient, and flow behavior index. The Herschel–Bulkley parameters of the melted nondairy frozen desserts are presented in Table 2.

Yield stress (σ _0HB) denotes the force needed to initiate the flow in the material, which may be correlated to scoopability and dipping ability (Briggs et al., 1996), consistency, texture, melting, and other properties. The σ _0HB value of melted nondairy frozen desserts varied between 0.0 and 25.2 Pa, with a mean of 2.7 ± 6.4 Pa. Javidi et al. (2016) reported yield stress varying from 1.9 to 4.2 Pa in low‐fat ice cream mixes, and Kurt and Atalar (2018) reported yield stress varying from 0.1 to 0.6 Pa in ice cream mixes made with quince seed. Moreover, Freire et al. (2020) reported yield stress varying from 2.2 to 53.8 Pa in melted ice creams, which were closer to the results in this study.

The consistency coefficient (k _HB) and apparent viscosity values provide different information for non‐Newtonian fluids, but both affect the physical properties, such as melting behavior, texture, and body, of ice cream and other frozen desserts. The k _HB value of melted nondairy frozen desserts varied between 0.02 and 5.92 Pa·s ⁿ , with an average of 1.19 ± 1.42 Pa·s ⁿ . Higher values of consistency coefficient were observed in soy milk‐based ice creams compared to conventional and sesame milk‐based ice creams (Ghaderi et al., 2021).

The n _HB value of melted nondairy frozen desserts varied between 0.43 and 1.01, with a mean of 0.74 ± 0.14. Interestingly, sample 465, which was a Newtonian fluid with n _HB of 1.01, did not contain a stabilizer in its ingredient list. This explains its Newtonian behavior since stabilizers are the main hydrocolloids used to change flow behavior in frozen desserts. All other samples present showed values lower than 1.00, that is, shear thinning behavior, which is a typical behavior in frozen dessert mixes. Ghaderi et al. (2021) also reported pseudoplastic behavior for plant‐based ice cream with soymilk and sesame milk. The shear thinning behavior is more noticeable when stabilizer levels increase in an ice cream mix (Freire, 2020).

3.3. Meltdown properties

A typical ice cream meltdown profile commonly shows a sigmoid curve, encompassing three distinct phases—a lag phase, a fast‐melting phase, and a plateau phase—representing different stages of the ice cream meltdown test (Wu et al., 2019). The lag phase concludes when the first drop reaches the beaker, indicating the induction time (IT). The rate of ice cream meltdown is characterized by the slope of the fast‐melting phase (Koxholt et al., 2001; Wu et al., 2019), defined as the drip‐through rate (DT) in this study. The plateau phase is reached when the melting process decelerates, reaching a static state, denoted by the percentage of the final drip‐through weight (Wu et al., 2019). Here, the final drip‐through weight (FDW) or the plateau is assumed to be reached at a maximum of 2 h. Additionally, the height change rate (HCR) and final height (FH) were evaluated according to Wu et al. (2019).

Representative drip‐through curves of commercial nondairy frozen desserts are illustrated in Figure 3A. The IT value, the time when the first drop falls, of nondairy frozen desserts varied between 15.3 and 60.4 min, with a mean of 32.4 ± 13.3 min. This matches the IT between 16 and 25 min for coconut milk‐based ice cream (Góral et al., 2018). On the other hand, values of induction time between 14 and 55 min were found in ice creams (Wu et al., 2019). Additionally, although no significant correlation was found between IT and other parameters (Table S2), a slight trend was found between IT and apparent viscosity. As can be observed in Tables 2 and 3, in general, as apparent viscosity increased, IT values increased. Wu et al. (2019) reported that ice cream mixes with higher viscosity led to higher induction times in their respective ice creams.

FIGURE 3.

Drip‐through curves from the meltdown test: (A) Typical curves with a fast and slow melting of nondairy frozen desserts, samples 767 and 489, respectively and (B) samples 670, 238, and 540 with uncommon meltdown behavior.

TABLE 3.

Meltdown parameters of commercial nondairy frozen desserts.

Sample code	IT (min)	DT (g·min−1)	FDW (%)	HCR (cm·min−1)	FH (%)
540	25.1 ± 3.0ef	0.12 ± 0.02d	54.8 ± 4.02cde	0.010 ± 0.005abc	30.2 ± 5.4defgh
381	24.8 ± 0.3ef	1.40 ± 0.01a	86.0 ± 1.00ab	0.019 ± 0.002abc	10.8 ± 2.5ghi
900	27.1 ± 0.9ef	1.65 ± 0.06a	96.7 ± 2.05a	0.022 ± 0.014ab	0.0i
767	27.1 ± 0.9ef	1.75 ± 0.06a	98.7 ± 0.20a	0.025 ± 0.005a	0.0i
516	15.3 ± 1.2f	1.65 ± 0.36a	83.7 ± 5.41abc	0.021 ± 0.010abc	8.9 ± 10.2hi
849	41.3 ± 1.8bcd	0.21 ± 0.25cd	19.1 ± 22.31fg	0.022 ± 0.007ab	38.4 ± 10.1cdef
732	30.5 ± 9.8de	0.14 ± 0.01d	13.0 ± 0.84fg	0.009 ± 0.008abc	75.1 ± 15.0ab
465	26.0 ± 2.1ef	0.11 ± 0.02d	15.0 ± 2.04fg	0.005 ± 0.002bc	71.7 ± 6.4ab
670	51.9 ± 2.7ab	1.26 ± 0.35ab	57.6 ± 9.03bcd	0.017 ± 0.005abc	37.5 ± 17.0cdefg
238	42.0 ± 2.3bcd	0.50 ± 0.09cd	36.7 ± 17.18def	0.011 ± 0.005abc	49.4 ± 9.2bcde
489	60.4 ± 5.0a	0.03 ± 0.00d	2.3 ± 0.06g	0.003 ± 0.001c	77.3 ± 4.6a
800	45.5 ± 4.7bc	0.16 ± 0.00cd	13.4 ± 1.11fg	0.008 ± 0.001abc	55.3 ± 13.3abcd
533	17.7 ± 5.9f	1.27 ± 0.33a	65.9 ± 16.40bcd	0.020 ± 0.005abc	15.9 ± 9.5fghi
364	33.9 ± 0.7cde	0.35 ± 0.00cd	25.8 ± 0.49efg	0.010 ± 0.004abc	58.0 ± 3.1abc
510	17.0 ± 17.0f	0.71 ± 0.29bc	69.8 ± 15.47abc	0.021 ± 0.004abc	24.4 ± 8.7efghi

Note: DT is drip‐through rate; IT is induction time; FDW is final drip‐through weight; HCR is height change rate; FH is final height of melted ice cream. Means not sharing a common letter within a column are significantly different (p < 0.05).

The DT of nondairy frozen desserts varied between 0.03 and 1.75 g·min⁻¹, with a mean rate of 0.75 ± 0.66 g·min⁻¹. These values are close to the range 0.13–1.88 g·min⁻¹ and average 1.07 g·min⁻¹ found by Warren and Hartel (2014) in commercial ice creams. A negative correlation was found between DT and total fat. The effects of the increased fat content on reducing the meltdown rate corroborate with other studies evaluating dairy ice cream (Roland et al., 1999; Warren & Hartel, 2014). Additionally, a slight inverse trend was found between DT and fat destabilization (Tables 1 and 3). As the ice melts, the tridimensional network of fat globules offers a barrier to the diluted serum phase to drain through the lamellae. This correlation has been well reported in other studies (Koxholt et al., 2001; Segall & Goff, 2002; Warren & Hartel, 2018; Wu et al., 2019).

The structural change rate, such as height changes, in ice cream and other frozen dessert products observed during the meltdown test contributes to a different perspective on the collapse of its structure (Wu et al., 2019). The HCR varied between 0.003 and 0.025 cm·min⁻¹, with a mean of 0.015 ± 0.007 cm·min⁻¹. An HCR around between 0.007 and 0.042 cm·min⁻¹ in ice creams was reported by Wu et al. (2019), which is close to the range in this study. An inverse correlation was found between HCR and fat content. Additionally, HCR and fat destabilization showed an inverse trend (Tables 1 and 3). Fat structure, including its level and destabilization, directly contributes to the structural stability of the commercial nondairy frozen desserts, leading to a slower change in the body during the meltdown. An inverse correlation between fat destabilization and HCR was also found by Wu et al. (2019).

The FDW of nondairy frozen desserts varied between 2% and 99%, with a mean of 49% ± 33%. Additionally, an inverse trend between FDW and fat destabilization was found (Tables 1 and 3). Values from 15% to 98% were found in ice creams, close to the values in this study, and an inverse correlation between fat destabilization and final drip through weight was also found by Wu et al. (2019). As the ice cream melts, the diluted serum flows through the lamellae, and the remaining structured foam rearranges preventing further drainage.

As the commercial product melts, it begins to collapse, but in some cases, the remaining fat globule network and air cell structures rearrange and eventually form a structure that prevents further collapse of the matrix. The FH of melted ice cream varied between 0% and 77%, with a mean of 37% ± 27%. Wu et al. (2019) found final height values between 0% and 60% in ice creams, which are similar to this study. Direct correlations were found with fat content. Additionally, a direct trend was found between FH and fat destabilization. The fat structure directly contributed to the final structural stability of the commercial nondairy frozen desserts after the meltdown test, leading to higher body retention. A similar correlation was also found by Wu et al. (2019) in ice creams.

Lastly, three drip‐though curves presented atypical melting behavior, as illustrated in Figure 3B. Samples 238 and 670 showed at least one collapse of a larger piece/drop of the product, while sample 540 presented a higher drip‐through toward the end of the 2‐h test. Sample 540 presented the highest overrun in this study (Table 1). On the other hand, sample 670 was highlighted as having the highest rheological parameters, namely apparent viscosity, yield stress, and consistency coefficient, whereas sample 238 showed a high freezing point and no emulsifier in its ingredient list (Table 1). This behavior was also visually identified, and the samples showed at least 30% FH. It was not possible to identify a common pattern in these three samples through the variables evaluated in this study. It is worth highlighting that, overall, low protein levels and a maximum overrun of 82% were observed in the commercial samples in this study. The low protein levels may indicate poor formability, leading to an unstable foam system. Therefore, more studies are needed to understand what led to these samples showing this atypical melting behavior.

This study provides a foundational understanding of the compositional, structural, rheological, and melting properties of commercial nondairy frozen desserts in the United States. These insights are critical for both academic researchers and industry professionals aiming to improve the quality and sensory attributes of nondairy frozen desserts. Furthermore, the findings can guide the development of new formulations that leverage plant‐based ingredients to achieve an ice cream‐like experience while addressing consumer expectations for texture, flavor, and melting behavior. This work also provides a benchmark for future studies, enabling comparisons and fostering innovation in this segment of the frozen dessert market.

4. CONCLUSIONS

This study highlighted the significant impact of diverse formulations and ingredient sources on the compositional, microstructural, rheological, and melting properties of commercial nondairy frozen desserts. Additionally, the absence of regulations regarding product identity results in a lack of a minimum nutrient density for nondairy frozen desserts in the market, mainly related to protein level. It is worth highlighting that most of the commercial nondairy frozen desserts showed a low protein content compared to dairy frozen desserts and other studies with nondairy frozen desserts. Additionally, the maximum overrun of 82% may be an indication of poor foam stability since normal economy ice cream in the market has overrun nearing 100%, with some nonstandard dairy‐based frozen desserts achieving well over this level. The values of other compositional parameters, such as fat content and total solids, were within a range similar to ice cream. In the same way, structural elements, such as mean air cell and ice crystal size values, were also close to those found in commercial ice creams. Regarding physical properties, some samples presented a density below 1.100 g·mL⁻¹, which indicates an unbalanced formulation with a high fat and/or low protein content. Despite samples showing freezing point values above and below −2.5°C, the variation was still in a reasonable range. Although some samples displayed a normal or typical melting pattern, a few samples showed instability during the meltdown test. Therefore, except for the protein content and anomalous behavior of some samples, most commercial nondairy frozen desserts corresponded to ice cream. This study hopes to shed light on commercial nondairy frozen desserts to improve their quality, whether compositional, microstructural, physical, or meltdown properties.

AUTHOR CONTRIBUTIONS

Dieyckson O. Freire: Conceptualization; investigation; formal analysis; visualization; writing—original draft. Didem Sözeri Atik: Investigation; formal analysis; visualization; writing—original draft. Payton M. Gladem: Investigation; formal analysis. Scott A. Rankin: Conceptualization; supervision; writing—review and editing. Richard W. Hartel: Conceptualization; supervision; writing—review and editing.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

Supporting information

Table S1. Ingredient lists for commercial non‐dairy frozen desserts.

Table S2. Matrix of p‐values and their respective correlation coefficients showing the linear relationship strength between each pair of variables.

Freire, D. O. , Atik, D. S. , Gladem, P. M. , Rankin, S. A. , & Hartel, R. W. (2025). Composition, microstructure, rheology, and meltdown behavior of commercial nondairy frozen desserts. Journal of Food Science, 90, e17633. 10.1111/1750-3841.17633