Intensification of atmospheric freeze drying for thin food slices with impinging jet

Abstract

Atmospheric freeze drying obviates the complexities and costs of maintaining a high vacuum for freeze drying. One of the main drawbacks of atmospheric freeze drying is the low sublimation rate, which is restricted by drying temperatures possible at ambient pressure to prevent food products from softening during dehydration. There is a strong need to intensify the process to reduce the total drying time. This study evaluated the feasibility of using impinging jets to enhance the mass transfer characteristics of the atmospheric freeze drying process. Atmospheric freeze drying experiments on thin lamb slices between −3 to −7 °C showed that the impinging jet configuration has a significant effect in improving the rate of mass transfer flux compared to the conventional cross-flow configuration. This was consistent across different thicknesses of the lamb slices, which would have presented different degrees of internal resistance to mass transfer. Given the amount of non-frozen water in the lamb slices at the drying temperature range evaluated, a scheduled switch from cold air atmospheric freeze drying to a mild hot air drying condition was explored. This strategy enhanced the removal of the remaining water content in the lamb slices, mainly the non-frozen water.

Keywords:

• Atmospheric freeze drying
• thin meat slices
• texture analysis
• drying kinetics
• impinging jet

1. Introduction

Atmospheric freeze drying (AFD) is convective dehydration utilizing cold, dry air. This process has been explored for drying various food products, examples being apples, cabbages, cods, eggplants, kiwi fruits, orange peels, potatoes, red peppers, and shrimps^[1–6] Compared to the conventional approach of freeze drying, AFD without a vacuum chamber and its auxiliary equipment, processing becomes simpler, lowering investment costs and significantly reducing energy costs. Wolff and Gibert^[7] investigated energy expenses in vacuum freeze-drying and AFD of potato slices. Based on their calculated results, a continuous-duty facility can save 35% of its energy consumption. Donsì et al.^[4] proposed a similar evaluation of energy requirements for drying shrimps and concluded that operating the freeze-drying process at atmospheric pressure reduced energy costs by around 30%.

Besides these positive aspects, one main drawback of AFD is its relatively slow drying rate. The absence of vacuum conditions in this process means that the maximum drying temperature has to be below the sample’s freezing point to preserve the product’s structural integrity.^[8,9] On the flip side, for vacuum freeze drying, the absence of water melting below the triple point means that the product temperature can be increased beyond the solid-vapor equilibrium temperature to enhance the drying rate. This is particularly toward the latter stages of drying provided the temperature is not increased beyond the softening temperature of the product. For food drying applications, the potential impact of freezing point depression will also need to be accounted for. Therefore, AFD is typically accomplished using cold air in the range of −3 °C to −10 °C, and this was proven to be a viable tradeoff between costs and final product quality.^[10] Even between this optimal range of the cold air temperature, the dehydration rate of AFD is relatively slow compared to conventional vacuum freeze drying. This is mainly because, in vacuum freeze drying, high temperatures may be employed: (a) between 10 °C and 35 °C for heat-sensitive products and (b) 50 °C or more for less-heat-sensitive products,^[11] which increases the vapor concentration differences between the material and the condenser that drive dehydration. To enhance the dehydration rate of AFD, several approaches have been proposed in the literature to overcome this limitation.

One approach is to enhance the heat transfer to the material with microwave or infrared. Eikevik et al.^[12] compared the drying kinetic and product quality of AFD of green peas with and without microwave assistance. The drying time was reduced by approximately 50% with the application of the microwave, and slight shrinkage on the dry product diameter (≤0.4 mm) indicated no significant quality reduction. Reyes et al.^[13] reported the study of using infrared radiation as an additional energy supply in AFD exhibits a statistically significant effect on the moisture content of the carrot particles throughout the whole drying process. Infrared radiation intensifies the sublimation process by providing a greater heat flux to the solid. A similar-designed study was carried out in drying blueberries,^[14] which further approved the application of infrared radiation accelerates the AFD drying process. Moreover, applying infrared radiation increased ascorbic acid and polyphenol in dried blueberries. Even though microwave and infrared radiation can speed up the drying process, the process should be carefully controlled due to the risk of overheating and product thawing with subsequent product quality degradation.^[15]

Alternatively, instead of enhancing the heat transfer rate in the AFD process, which increases the driving force for dehydration, ultrasound assisted AFD is another potential technique to enhance the process’s heat and mass transfer coefficient. In addition to the ultrasonic waves disrupting the external boundary of the material, this technique may also enhance the water vapor transport within the porous dehydrated material. Carrión et al.^[16] evaluated the drying kinetics and influence on product quality of ultrasonically assisted AFD; the drying time was reduced by 58.5% and 74.2%, respectively, according to the power of ultrasound applied with no remarkable impact on product quality. Similar results were found in the ultrasound-assisted AFD of orange peels that the application of ultrasound led to an intensification of the drying process with a significantly shorter drying time compared to AFD without ultrasound assisted; moreover, ultrasound application enhanced the drying rate without reducing the functional properties of the fiber in orange peels.^[5] Santacatalina et al.^[17] modeled ultrasonic-assisted AFD of apple cubes under different experimental conditions to gain insight into the impacts of power ultrasound on AFD and quantify those impacts. Previous studies have shown that the application of power ultrasound dramatically improves water removal during drying. However, the application of ultrasound requires high power demands, which may limit its applicability. It is important to note that such hybrid drying methods are truly economical only for relatively high-cost materials and otherwise intractable biological products such as fruits and vegetables.^[18]

One potential simpler and cheaper approach to enhance the heat and mass transfer coefficient is using a fluidized bed for AFD. The premise of this approach is to enhance the heat and mass transfer coefficients for the process. Earlier attempts by Malecki et al.^[19] showed that this implementation did not significantly improve the drying rate. In order to make improvements in freeze-drying at atmospheric pressure, the drying process was conducted by immersion of frozen products in a fluidized bed of adsorbent held at a low temperature (below the product’s incipient melting point).^[20] The adsorbent particles play a double role as heat and mass transfer agents, decreasing external resistance, and freeze-drying requires less energy.^[7] Literature in AFD assisted by adsorbent in a fluidized bed is often carried out by immersing the food sample in the adsorbent or air passes through a fixed bed of adsorbent placed before or after the fluidized bed inlet or outlet.^[4,13] Using alumina or carbon as adsorbents can enhance the drying rate of immersed carrot samples, as reported;^[21] additionally, higher AFD rates were obtained with activated alumina than with activated carbon. In an AFD study utilizing fluidized beds of adsorbents,^[3] the heat and mass transfer coefficients were higher than those expected in vacuum operation. The drawbacks of this technique, however, include the difficulty in separating the food products from the adsorbent particles.^[7] There was also a limitation in the shape and size of the frozen food sample to facilitate fluidization, in which needle- or platelet-shaped samples were not suitable.^[22,23] Some potential food products are given here: okra, straw mushrooms and slice fruits, vegetables and meat e.g. Because to the rigorous movement in the fluidized bed, there was also difficulty in maintaining the structure of the product due to the high potential of attrition during drying.^[24,25] Mechanical damage to potato products as a consequence of the attrition in the fluidized bed was reported by Di Matteo et al.^[3] after evaluating different adsorbent materials.

In order to overcome these limitations while achieving high heat and mass transfer coefficients, one potential approach is to use impinging jets. The impingement drying process consists of a single jet or array of jets that impinge air or steam directly on the product’s surface at high velocity. This reduces the thickness of the thermal boundary layers and increases the heat and mass transfer rate. The heat transfer coefficients of impingement driers are typically five times higher than those of cross-circulation driers.^[26,27] Impinging jets are widely used in various hot air industrial drying operations involving rapid drying of continuous sheets form (e.g., textiles, tissue paper, photographic films, coated paper, and nonwovens) or relatively large, thin sheets (e.g., carpets, lumber, and veneer), or even granular products (e.g., coffee beans, nuts, cocoa beans, and rice).^[28,29] There was an attempt to impinging jet with the AFD process to dry granular probiotic encapsulates by Bórquez et al.^[30] with promising results. However, in their design, the probiotic encapsulates are actually moving in a rotating motion within the drying chamber. In our opinion, it may resemble more of a high-velocity rotating bed and may be vulnerable to the challenges highlighted earlier. Nevertheless, in their work, there was no need for any desiccant bed and certainly highlights the potential of high velocity to intensify the AFD process.

Therefore, in contrast to these past reports, this current work explores the potential of using an impinging jet technology to intensify the AFD process for a static thin food layer. The aim is for the process to achieve a high heat and mass transfer coefficient without the vulnerability to product quality highlighted earlier. The degree of drying rate increment achievable with the impinging jet on a static layer for atmospheric freeze drying will be quantified.

2. Materials & methods

2.1. Atmospheric freeze drying experimental setup

A schematic layout of the lab-scale AFD system is shown in Figure 1. A vortex tube was used to generate cold, dry air into the drying chamber, following the series of work by Rahman and Mujumdar.^{[20,25,31,32]} The vortex tube required input of compressed air, which was controlled with a mass flow controller (Model FMA-LP2612A, Omega) at the inlet. The control valve at the hot-air outlet regulated the separation of the inlet flow into hot and cold outlets, which was kept constant throughout the experiment. The compressed air was dehumidified via a 250 psig desiccant air dryer (W74D-4GN-NMN, IMI Norgren) and a compressed air filter (Model F11, IMI Norgren), and two desiccant bowls with desiccant beads, before entering the vortex tube (Model 3240, EXAIR Corporation).

Figure 1. Schematic layout of the AFD system.

The drying chamber (37 mm ID and 27 mm high) consisted of a vertical PVC tube fitted with a wire mesh to hold the sample. The cold-dry air impinges onto the frozen sample from the bottom with this experimental setup. Cold stream air velocity was measured at the outlet of the drying chamber with an anemometer (EA3000, Technoline, accuracy: 0.028 m/s). T-type thermocouple wire and humidity sensor were inserted in the drying chamber to measure the temperature (T1) and relative humidity (RH1). The data were monitored and collected using a National Instrument compact data acquisition devise (NI cDAQ-9174) as the interface between the computer and the external sensors.

2.2. Sample preparation

Fresh lamb meat (hind leg) samples were purchased from the local market in Auckland, New Zealand and the outer surface fat was skimmed. The bone was removed, and the meat was wrapped tightly into a roll and wrapped with food cling wrap, then frozen at −18 °C for cutting. The frozen meat was sliced into three different thicknesses (1, 2, and 3 mm) by an electrical food slicer (Model SL4382, Haoluck, China) and cut into approximately 25 mm × 25 mm size. The lamb slices were spread in one layer on baking paper and then stored at −18 °C before drying. The range of the sample weight is 1 mm: 0.500–0.550 g; 2 mm: 0.700–0.750 g; 3 mm: 0.800–0.850 g.

2.3. Atmospheric freeze drying experimental procedures

Table 1 shows the experimental matrix undertaken. It is important to note that the temperature of the cold air generated by the vortex tube system may vary slightly from day to day as it is affected by the temperature of the compressed air from the main system. Within the variability of the system, the drying temperatures used in the experiments varied between −3 °C and −7 °C. From preliminary experiments, it was not practical or possible to have a tighter temperature control with the system used. Therefore, there was no attempt to deliberately vary the drying temperature in the experiments, and all the experimental runs were interpreted as the same temperature within the variability observed.

Table 1. Summary of the experimental parameters.

Test parameters		Control parameters
Thickness	Drying configuration	Air Temperature	Air Humidity	Air velocity
1mm 2 mm 3 mm	Cross-flow (CF) Impinging jet (IPJ)	−3˚C to −7˚C	≤ 10%	2.4 m/s to 3.0 m/s

In order to fully ascertain the intensification of the impinging jet flow, a comparison was made to the cross flow scenario. Figure 2 illustrates how this was generated in the experiments. For the impinging jet (IPJ) experiments, the frozen meat slices were placed on the mesh of the drying chamber. Cold air then impinges the frozen meat slices from the bottom. Cross-flow to the frozen meat slices was then generated by clipping and suspending the frozen slices so that the cold air from the bottom flows across the vertically hung meat slices.

Figure 2. The meat sample displayed in the drying chamber under: (a) impinging jet (IPJ-AFD) and (b) cross-flow (CF-AFD) configuration.

The frozen meat slice was weighed before placement into the pre-conditioned (to steady state) drying chamber. Only one slice was loaded into the drying chamber at each run. Throughout the drying process, the weight of the meat slices was measured (moisture analyzer: XM 50, Precisa, precision of 0.001 g) at intervals of 2 h by quickly taking them out from the drying chamber before replacing them back into the chamber. At the end of the experiment, the moisture content of the dehydrated meat slices was determined. This allowed the calculation of the bone-dry solid mass of the meat slices for drying kinetics calculation. The drying curves were expressed by converting the experimentally measured moisture content data sets from the drying runs of lamb slices to the moisture ratio (MR), which was relative to the initial moisture content of 70%. Water activity (a_w) was also determined for the dried product using a water activity meter (CWA-1, Coffee Tech).

2.4. Combined impinging jet-assisted atmospheric freeze-drying (IPJ-AFD) and hot-air drying (HAD)

In the latter part of this manuscript, a combination drying process consisting of initial IPJ-AFD, followed by HAD will be discussed. Lamb slices were dried by IPJ-AFD for 4 h first and the moisture content of samples reached approximately 30%, corresponding to the moisture ratio of 0.4–0.45, which is the moment at which it was estimated that the frozen free water was completely removed. The lamb slices, which reached the desired moisture content for the next drying stage, were transferred under a hot air gun (D00398, Duratool) with a maximum power of 2000 W for further HAD. The HAD processing parameters were set as an air temperature of 50 °C and an airflow rate of 6.94 × 10⁻⁵ m³/s. For data consistency, the distance between the heat gun and the samples was kept constant for each run through the HAD process. The mass of samples in the HAD process was measured by manually weighing at intervals of 15 min until no mass change. Such a drastic drying temperature switch from freeze drying (albeit vacuum freeze drying) to hot air has been reported before in the literature.^[33] In that reference cited, the hot air temperature used was 60 °C in a cabinet crossflow condition.

2.5. Texture measurements

The textural characteristic of dried lamb slices was measured using a texture analyzer (CT3, Brookfield) with a load cell of 10 kg and analyzed with the TexturePro CT software. The lamb slice samples were compressed by a cylindrical-shaped probe of diameter 12.7 mm (smaller than the size of the slices), and the pre-test-speed, test-speed, and post-test-speed were set at 2.0 mm/s, 0.3 mm/s, and 0.3 mm/s, respectively. The dried lamb slices were firstly rehydrated by immersing in distilled water at 25 °C for 10 min.

It was difficult to measure the thickness of the rehydrated thin slices precisely because of the uneven surface and softness. Moreover, measurement by a caliper may not be so reliable due to the potential of excessive measuring force on the soft, thin meat slices. Therefore, this did not allow the use of a constant strain setting which was typical for texture measurements and comparison. To overcome this limitation, the rehydrated slices were compressed for each measurement until the maximum load of the equipment (10 kg weight) was reached. As a simplifying assumption, the compression distance required to reach this maximum load was taken as the maximum thickness of the slices. This assumed maximum thickness was then used to back-calculate the strain values in generating the force-strain curve. In the subsequent analysis, following past literature on textural profile analysis for meat products,^[34] the force-strain curve was analyzed up to 80% strain. This is illustrated in the Appendix (Figure A1). All the measurements were undertaken in triplicates.

3. Results & discussion

3.1. Comparison of the drying curves between impinging jet (IPJ-AFD) and cross-flow (CF-AFD) configurations

The effects of different sample thicknesses (1, 2, and 3 mm) and airflow conditions on the drying kinetics of lamb slices are shown in Figures 3 and 4, respectively. All the drying curves exhibited a constant rate of drying between the first four to six hours of drying. This is then followed by a short falling rate period before reaching the equilibrium condition. This trend of the drying curve was found to be consistent with other related studies on AFD of cooked beef,^[8] orange peels,^[5] peas,^[35] potatoes,^[20] and red pepper.^[1] From the results in Figure 3, the constant drying rate for the 1 mm sample thickness under both airflow conditions (IPJ-AFD and CF-AFD) was consistently higher when compared with the 2 mm and 3 mm samples. The higher drying rate of the lamb slice for 1 mm might be due to the thinner dry sample layer decreasing the internal resistance of water vapor. The distance for moisture moves from the ice-vapor interface to the surface is the nearest in 1 mm samples, allowing moisture diffusion to be faster. As Di Matteo et al.^[3] reported, a higher drying rate was observed with decreased slice thickness. For the experiments reported here, it appears that even with the internal resistance, a constant rate of drying was observed for the thin slices in the first four to six hours. According to the literature data in Table 2, lamb’s unfreezable water content is 30 to 40% at the drying temperature used in the present study. This corresponded to the moisture ratio of 0.4–0.45 in Figures 3 and 4, and further supports that the unfreezable water content in the thin lamb slice already started decreasing when the AFD process entered the falling rate period. Most likely, type I and type II of the unfreezable water would have been removed at this stage of dehydration.^[40] For simplicity, we will refer to these as bound water for the remainder of this communication. Therefore, the short falling rate period may be attributed more to the impact of bound water than water vapor transport resistance within the porous frozen meat structure. This is an indication that the idealized receding evaporation front visualization of the freeze drying process may not be applicable for such thin meat structures. An unexpected observation from Figure 3 was that consistent for both IPJ and CF, the drying rate of 1 mm samples tapered off at a higher moisture ratio equilibrium when compared to the 2 mm and 3 mm slices. This result might be due to the change in product structure during desorption that causes the difference in equilibrium moisture content. The changed material structure during drying can cause significant variations in the moisture-binding ability of the solid.^[41]

Figure 3. Drying kinetics of lamb slices at different sample thicknesses (1, 2, and 3 mm) under (a) impinging jet-assisted atmospheric freeze-drying (IPJ-AFD) and (b) conventional cross-flow atmospheric freeze-drying (CF-AFD).

Figure 4. The effect of airflow conditions on drying kinetics of lamb slices thicknesses at (a) 1 mm, (b) 2 mm, and (c) 3 mm.

Table 2. Literature values of unfreezable water contents in various meat products.

Material	Total water content, %	Unfreezable water content, %	Temperature, °C	Reference
Beef (semimembranous muscle)	76.0	49.18	−5	^[^36^]
Beef	74.5	32	−4	^[^37^]
Beef (sirloin)	75.0	30	−0.9	^[^38^]
Lamb (leg)	71.0	40	−1.1
Pork (leg)	75.1	27	−1.03
Turkey (breast)	74.0	15	−1.2
Lean lamb loin	64.9	36	−3.33	^[^39^]

Figure 4 replots the data from Figure 3 to directly compare IPJ-AFD and CF-AFD at different slice thicknesses. While the final moisture content of IPJ-AFD dried lamb slices was consistently slightly lower than CF-AFD dried lamb slices for thicknesses of 1, 2, and 3 mm, the drying curves were nearly identical. This was very unexpected as it indicated that the use of impinging jet has no intensification potential whatsoever, apart from a higher potential to decrease the unfreezable bound moisture content of lamb slices. An impinging jet is expected to increase the external heat and mass transfer rate and should be more effective in removing surface or unbound moisture (Mujumdar, 2014). However, from Figure 4, there was no enhancement in the constant rate period, where the internal resistance of mass transfer would have been relatively small. This unexpected finding indicated the effect of the impinging jet on the AFD drying kinetics observed may need to be evaluated from a flux perspective.

3.2. Evaluation of the dehydration fluxes

In order to analyze the dehydration fluxes, the effective surface area available for dehydration will need to be ascertained. Figure 5(a), (b) illustrates the different flow configurations evaluated in this study. Neglecting the four side surfaces of these thin meat slices, the effective mass transfer surface area of lamb slices in the CF-AFD configuration involved both the front and the back surfaces equally. This is because both surfaces are exposed fully to the cross-flow of the cold, dry air. As shown in Figure 5(c), the front and the back surfaces are expected to have equal mass transfer rates delineated by the size of the arrows. In contrast, for the IPJ-AFD, due to the way the experiment was set up, only one surface side (bottom) is fully exposed to the airflow; the top surface does not have direct impinging air contact. The expected differences in mass transfer rate are shown in Figure 5(d). The mass transfer rate is expected to be higher at the bottom surface facing the inlet because of the immediate contact with the air. On the other hand, the mass transfer rate is expected to be relatively low at the top surface facing the outlet, which is in the vicinity of the wake or vortex-shedding region when the drying air passes across the sample.

Figure 5. Schematic of the drying surfaces and corresponding heat and mass transfer analysis: (a) and (c) for CF-AFD configuration; (b), (d), and (e) for IPJ-AFD configuration.

There are numerous works in the literature reporting the local heat transfer coefficients for geometries with similar impinging jet-wake surfaces: 2D rectangular,^[42,43] cylindrical,^[44] and 3D geometries.^[45–47] Adopting the model reported by Ateeque et al.^[45] for a cuboid immersed in a convective airflow, Table 3 presents quantitatively the heat and mass transfer coefficients for the surface of the meat slices, facing and in the wake of the convective airflow, used in the IPJ-AFD experiments. The values were calculated via the following equations for inlet velocities used in this study in a range of 2.4 m/s to 3.0 m/s, corresponding to the environmental temperature fluctuation observed in the experiments. (1) $$\mathit{\text{Nu}}=0.8185{\mathit{\text{Re}}}^{0.5427}\left(\text{surface facing the convective airflow}\right)$$(1) (2) $$\mathit{\text{Nu}}=0.1182{\mathit{\text{Re}}}^{0.6026}\left(\text{surface at the wake of the convective airflow}\right)$$(2)

Table 3. Variations of minimum, maximum, and average values for heat and mass transfer coefficient of IPJ configuration in this work, calculated using the local heat transfer correlation by Ateeque et al. ^[45]and converted to the mass transfer coefficient using the Equation (1)(1) $$\mathit{\text{Nu}}=0.8185{\mathit{\text{Re}}}^{0.5427}\left(\text{surface facing the convective airflow}\right)$$(1) .

	Heat transfer coefficient (W/m^2.K)			Mass transfer coefficient (m/s)
Surfaces	Min.	Max.	Average	Min.	Max.	Average
Facing the convective airflow	71.2093	79.9320	75.4190	0.0614	0.0689	0.0636
Wake of the airflow	16.9108	19.2260	18.0244	0.0146	0.0166	0.0152

The maximum heat transfer coefficient was obtained with the maximum inlet velocity (3.0 m/s). The average heat transfer coefficient for each surface is calculated by taking an average of all the runs’ inlet velocities (2.6 m/s) considered in this work. The average heat and mass transfer coefficients obtained at the surface facing the inlet are significantly higher than at the surface facing the outlet for the reason discussed earlier.

The local mass transfer coefficients ($h_m)$ were then determined using the Chilton-Colburn analogy between the thermal and concentration boundary layers, with Equation (3)(3) $$h_m=h\left(\frac{D{\mathit{\text{Le}}}^{\frac{1}{3}}}{k}\right),$$(3) below.^[45,46] (3) $$h_m=h\left(\frac{D{\mathit{\text{Le}}}^{\frac{1}{3}}}{k}\right),$$(3)

Where $\mathit{\text{Le}} (=\frac{\alpha }{D})$ is the Lewis number representing a measure of the relative thermal and concentration boundary layer thicknesses, $h_m$ is the surface mass transfer coefficient (m/s), $h$ is the surface heat transfer coefficient (W/m²K), $D$ is moisture diffusivity (m²/s), $\alpha$ is the thermal diffusivity (m²/s), $k$ is the thermal conductivity of the air (W/(m.K)).

The variation between the average $h_m$ values from Table 3 showed that the mass transfer coefficient at the top wake side was about 24% that of the impinging surface facing the inlet. On the basis that mass transfer occurred over the entire surface area on both sides of the sample slice, the mass transfer rate of the wake surface is about 24% that of the impinging surface at the same driving force. This phenomenon can be demonstrated in Figure 5(d), with the length of the arrow indicating the two surfaces’ relative mass transfer rate. Discussing this from another perspective, should the mass flux be the same on both sides, the wake surface will make a smaller effective surface area. This alternate visualization is shown in Figure 5(e), in which the arrow on both sides of the sample slice is the same length. However, the surface area where mass transfer occurs became smaller, only about 24% of the entire surface of the wake region. This quantified effective surface area of the wake including the surface area facing the convective air stream, was then used to calculate the practical mass flux of the sample in the IPJ-AFD configuration.

A comparison of the calculated mass flux between IPJ-AFD and CF-AFD at different sample thicknesses is shown in Figure 6. Similar trends of the curve were observed across all thicknesses. The initial value of mass flux of the IPJ-AFD was around 0.00021, 0.00030, and 0.00035 g/cm²min for sample thickness at 1, 2, and 3 mm, respectively, which was almost 1.5 times higher than the mass flux of the CF-AFD. The results showed that impinging jets consistently increase the mass flux for all tested thicknesses. This result is in agreement with the hypothesis of this study that the application of IPJ in AFD has the potential to improve drying efficiency. A significantly higher mass flux was observed at the initial drying stage of IPJ-AFD and CF-AFD due to the sublimation starting at the sample’s surface with relatively low internal resistance to mass transfer. The mass flux in both configurations decreases as the drying progresses, which is caused by increased internal resistance to the mass transfer of water vapor. Li et al.^[48] also discussed similar observations in the AFD of apple cubes as the mass flux decreased with the migration of the ice-vapor interface. As discussed in section 3.1, non-frozen bound water may have a bigger influence on the falling rate behavior, requiring high-intensity drying potential for dehydration, than the resistance due to the internal water vapor transport within the porous slices.

Figure 6. Variation of mass flux under IPJ-AFD and CF-AFD configuration at different sample thicknesses: (a) 1 mm, (b) 2 mm, (c) 3 mm.

The analysis here brings out an important point for future impinging jet atmospheric freeze dryer design for thin food slices. At the velocity evaluated in this work, the impinging configuration could only increase the initial mass transfer flux by 38–51%. With this modest increment in the mass transfer flux, the impact of the available surface area for dehydration on the overall process intensification seems to be more significant. In fact, the larger effective surface area of dehydration of the cross-flow configuration offset the benefit of the impinging jet configuration. This aspect will need to be kept in mind in the evaluation of cross-flow and impinging jet atmospheric freeze dryer designs.

It is noteworthy that the present study only illustrates the experiment results based on a thin substrate with minimized internal resistance. IPJ-AFD application may not be superior to the conventional CF-AFD in the case of using a thicker product as the path length of moisture migration from the inside of the sample to the surface is much longer, resulting in a greater internal mass transfer resistance. The present study revealed that the AFD process with IPJ assistance can potentially intensify the drying process’s efficiency with a thin product. To further develop a more befitting drying process for accelerating the drying rate, the IPJ-AFD process requires additional modifications in the design of drying procedures.

3.3. Further intensification with scheduled hot air drying

When the moisture content of the material is relatively high at the beginning of drying, most of the water exists as free, non-bounded water. As drying progresses, the amount of free water decreases and the unfreezable bound water gradually becomes higher in proportion; therefore, it is necessary to enhance the drying rate to achieve a final product with relatively lower water activity for preservation.

As described earlier, the unfreezable water content in lean lamb meat ranges from 30% to 40%. As shown in Figure 3, the drying rate entered the falling rate period after the moisture content was reduced to 30% (at corresponding MR of 0.4–0.45), where most of the free water was removed, and the decrement in the unfreezable bound water content was initiated along with drying progressed. Therefore, during the falling rate period, dehydration is mainly dominated by the cold air removal of non-frozen bound water rather than the sublimation of solid water. Therefore, there may not be a need to limit the drying temperature to ensure that the water in the food slices remains frozen. AFD processes at a sub-zero temperature in the first step of drying to preserve the product structure with minimal shrinkage. The unfreezable bound water content can be further decreased by higher temperature evaporation to enhance the drying rate and reduce the overall drying time.

To investigate this strategy, hot air drying (HAD) was then scheduled after 4 h of IPJ-AFD when the total moisture content dropped to about 30% (MR of 0.4–0.45). Kumar et al.^[33] reported that the removal of 30% moisture content through sublimation is the minimum requirement in reducing shrinkage and energy and improving rehydration properties in case of combination drying involving freeze-drying followed by HAD. The present work evaluated a combination drying consisting of IPJ-AFD followed by HAD to remove the unfreezable bound water in lamb slices. The drying process for the combination drying IPJ-AFD + HAD and IPJ-AFD alone was visualized with drying curves based on the moisture ratio and drying time shown in Figure 7. In the combined drying method consisting of IPJ-AFD followed by HAD, moisture ratios reduced gradually with increasing drying time in the first stage IPJ-AFD due to the slow sublimation process, resulting in a slower rate of removing moisture. The moisture ratio reduced drastically upon switching to the second stage HAD due to the higher drying temperature, which enhanced the heat and mass transfer rate and resulted in a corresponding higher drying rate and decreased drying time. The unfreezable bound water was rapidly evaporated from the product by hot air. The application of HAD also resulted in a shorter total drying time with much lower moisture content in the final product for storage preservation compared to IPI-AFD alone. Ee et al.^[49] reported similar findings on applying hybrid drying that consists of freeze-drying followed by HAD to dry kedondong fruit with an enhanced drying rate. The relationship between the drying method and the water activity (a_w) parameter was also assessed. The water activity of dried meat slices is shown in Figure 8. The combination drying approach of the IPJ-AFD resulted in obtaining water activity with a much lower value than in the case of the IPJ-AFD alone for all tested thickness levels. According to the results of this section, an additional thermal drying procedure applied after the sublimation has enhanced more water to be removed and is a reasonable choice to reduce the total drying time of the AFD process after the removal of most free water by sublimation. This combination method has already been studied in food drying, particularly in fruits and vegetables.^[33,49–51] Results obtained in these studies have shown that applying a higher drying temperature may cause a negative impact on the final product’s texture in accordance with increased hardness. However, there are few studies on its application in meat drying. An evaluation of the dried product texture in the present study is immensely important to investigate the feasibility of applying this combination drying system to AFD and will be reported in the next section.

Figure 7. Drying kinetics of lamb slices dried by IPJ-AFD followed by hot air drying as compared to IPJ-AFD alone for sample thicknesses at: (a) 1 mm, (b) 2 mm, and (c) 3 mm.

Figure 8. The water activity of the 1, 2, and 3 mm thin lamb slices undergoing IPJ-AFD or IPJ-AFD + HAD.

3.4. Texture analysis

The appearances of lamb slices prepared under different drying conditions are compared in Figure 9. It can be seen that the colors of the original samples are well maintained in both the applied single-stage AFD operations. The combination drying process samples exhibited browning in color and slight shrinkage compared to the CF-AFD and IPJ-AFD conditions. The evaluation of the dried food product texture is essential to characterize the quality of the dried food product. In the present study, the force-strain curves of the rehydrated product are shown in Figures 10 and 11. Changes in texture quality is certainly the main problem associated with dehydrated meats and is often difficult to explain due to significant variations in results.^[52,53] Freeze-dried meat is somewhat lower in tenderness and juiciness in comparison with fresh meat.^[54–56] Harper and Tappel^[57] also reported poor rehydration observed in freeze-dried chicken pieces, and they mentioned the dryness in the texture of freeze-dried meat is the major problem in improving their quality. According to observations in Figure 10, the rehydrated samples after IPJ-AFD consistently exhibited higher hardness than the raw samples for all sample thicknesses. The increase in hardness may be due to changes in meat texture that caused losses in the water binding capacity. This capacity depends on the ability of the muscle proteins, especially of the actin and myosin, to form a network gel that allows water to be firmly bound.^[58] Connell^[59] also reported increases in fiber toughness, which are manifestations of changes in the actomyosin complex, such as protein cross-linkages and denaturation.

Figure 9. Appearance of the dried lamb slices.

Figure 10. Comparison of hardness values between different drying processes at lamb slices thickness of: (a) 1 mm, (b) 2 mm, and (c) 3 mm.

Figure 11. Variation of hardness values of 1, 2, and 3 mm rehydrated lamb slices undergoing (a) IPJ-AFD: impinging jet-assisted atmospheric freeze-drying, (b) IPJ-AFD + HA: combination drying consisting of impinging jet-assisted atmospheric freeze-drying followed by hot air drying, and (c) fresh slices.

Lamb slices undergoing the combination drying process consisting of IPJ-AFD followed by HAD exhibited a slight increase in the hardness when compared to the IPJ-AFD samples. (Figure 10). This is consistent with the texture analysis of products obtained by the drying process with the application of hot air.^{[50,51,60–63]} HAD caused dried products to have extremely low moisture content by thoroughly removing water. The increase in air temperature accelerates the rate of moisture elimination, causing profound physical and chemical alterations on the food surface, thus forming a hard, impenetrable surface layer.^[64] Surface hardening and slight shrinkage might have been the main reasons for IPJ-AFD + HAD samples having a denser structure and higher hardness values. Nevertheless, from visual inspection, the rehydrated IPJ-AFD + HAD samples were still soft to the touch with more significant hardening on the edges of the slices. It is interesting to note that for the 2 mm slices, the texture profile of the rehydrated IPJ-AFD and IPJ-AFD + HAD overlapped (Figure 10(b)), indicating similar texture behavior. Therefore, these are evidence that a combination of scheduled impinging jet-assisted AFD and subsequent hot air drying may achieve relative retention of the textural quality of dried products while shortening the overall drying time.

Replotting the data in Figure 10 to compare the impact of slice thickness on the texture revealed that the fresh 1 mm lamb slices seemed to have a slightly harder texture when compared to the 2 mm and 3 mm slices. This differences seemed to be magnified after IPJ-AFD treatment and rehydration. A similar trend was also observed for the IPJ-AFD + HAD rehydrated slices, albeit with overlap between the 2 mm and 3 mm slices. Babić et al.^[52] also reported in their study that the rehydrated freeze-dried thin samples were tougher and harder than thick ones. At the moment, it is unclear why the slice thickness affects the texture profile of the fresh and rehydrated slices.

4. Conclusion

The present study aimed to intensify the drying efficiency of the atmospheric freeze drying process by applying an impinging jet dry cold air flow. The mass transfer flux of the impinging jet AFD process was about 38–51% higher than the conventional cross-flow AFD process at the initial drying stage when the internal resistance to mass transfer was relatively low. From the current experiments, the impact of the available surface area for dehydration on the drying rate was very significant. The higher surface area available for the cross-flow configuration in the current experimental setup offsets the benefit of the intensification offered by the impinging jet, leading to similar overall drying kinetics. This is an important aspect to be kept in mind for the future design of impinging jet atmospheric freeze dryers for thin food slices. Based on the amount of non-frozen bound water in the meat slices, a combination of cold air sublimation followed by hot air dehydration strategy was evaluated to further intensify the process. The results of this combination drying process evidenced the potential of developing the IPJ-AFD into a hybrid drying technology to balance the drying efficiency and dried products’ textural quality retention. In terms of drying efficiency, the combinational drying process greatly reduced the overall drying time. The water activities in the dried product were approximately halved when a second stage of thermal drying was applied to the process. The effect of drying techniques on the quality of the product was also analyzed in terms of their textural properties. The present results demonstrated that a single-stage IPJ-AFD has less alteration in the textural quality of the dried product compared to the combined drying process. However, IPJ-AFD + HAD 2 mm samples showed textural quality results comparable to IPJ-AFD alone. This suggests that the sublimation process applied at the first stage of drying can be a promising method to retain the product structure effectively with less shrinkage. Optimization is needed for this combinational drying process to improve the final quality of the product further.

Acknowledgments

Martin Ryder’s help in constructing the atmospheric freeze drying experimental setup is greatly appreciated. Special thank you to Jenn Kang for calibrating and in commissioning the experimental setup.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Declaration of interest statement

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.

Appendix

Figure A1. Analysis of force-strain curves. The same approach analyzed the IPJ-AFD + HAD and the raw lamb slices for all tested thicknesses.