*3.1. Blueberry Characteristics*

Blueberry size was selected within the diameter range of 14 to 15 mm. Assuming a spherical shape for the blueberry fruit, and, given a diameter of 14.5 mm (1.45 cm), its surface area ( *A* = <sup>4</sup>π*r*2) results in a unit area of 6.6 cm2.

Total Soluble Solids of the blueberries used as raw material for experimental procedures showed a characteristic quality indicator within the narrow enough range of 9.4 ± 0.7 ◦Brix.

#### *3.2. Busting Process of Nontreated Whole Blueberries*

The whole blueberry FD process was set as described in the methodology section.

A slide dataset that consists of images extracted from the photo sequence every 20 min was used to visually quantify the fraction of busted blueberries over time; this evolution is depicted in Figure 3. The busted blueberries are identified by red circles, reflecting the progression of the busting process, which starts with zero busted at 20 min and describes a saturation curve, achieving a maximum of 26 busted blueberries at 180 min.

**Figure 3.** Visual evaluation of the busting process. Inspection of photographs taken at di fferent times during the freeze-drying of whole nontreated blueberries.

Note that the dynamic of the busted blueberry percentage increased by up to 43% after 3 h of freeze-drying, and then it stayed constant (Figure 4). This means that the resistance to the mass flow of sublimated water is high at the beginning, and then it decreases as the number of busted blueberries increases. Then, a normal FD process should continue with increasing resistance as the dried layer gets thicker. The main process parameters and measurements of the whole blueberries FD, treatment 1 in Table 1, are depicted in Figure 5.

As expected, the percentage pressure increase (PIT) was low at the beginning of the process until around 2.5 h, in agreemen<sup>t</sup> with that shown in Figure 4. Then, it stayed approximately constant for 2.5 h, before decreasing at a moderate rate until 13 h, and finally, a very low decrease continued to ge<sup>t</sup> down to values under 10% after around 17 h. The product temperature also achieved the shelf temperature at approximately 15 h. Two important outcomes can be extracted from this result. First, normally, the resistance to vapor flow from the FD product is low at the beginning of the process, as no dried layer exists; however, in this case, resistance seems to have been high at the process initiation due to the high skin resistance. Then, as the busting process proceeds, the resistance goes down, before finally increasing again as the dried layer develops. Secondly, the results indicate that the primary drying time is in the range of 15 to 20 h. From the experiments carried out in triplicate, the end of the

primary drying time was assessed by the product of the thermocouple response and PIT, and there was good agreemen<sup>t</sup> between the two techniques, with the global primary drying time being 17 ± 0.9 h with a final moisture content of 14% ± 2%.

**Figure 4.** Busting process dynamics as evaluated from photographs taken during the freeze-drying process of whole nontreated blueberries.

**Figure 5.** Main processing variables and measurements during the freeze-drying of whole nontreated blueberries, PIT.

In order to quantify the effect of the blueberry skin on mass transfer, the MTM method was applied. It has been demonstrated that the computed value of the exponential expression of Equation (1) (Q; without considering t) should be equal to, or higher than, 0.2 (1/h) (Equation(2)) to ensure complete depiction of the exponential phase of the pressure rise in the product chamber. This phase reflects the influence of the resistance (RT) on the mass transfer phenomena. For a given value of RT, Equation (2) allows us to estimate the minimum number of fruit units N (number of blueberry fruits) that must be loaded in a particular freeze-drier system (characterized by its void volume V). A value of RT of approximately 3 (torr h cm<sup>2</sup>/g) has been reported to be acceptable to carry out this assessment [28]. To satisfy the above condition, the following set of values were considered:

R/MH2O = 3.461 (torr <sup>L</sup>/g-K);

N ≥ 60 units; A = 6.6 (cm<sup>2</sup>/unit); V = 32.5 (L); TS = 293 K; RT= 3.0 (torr h cm<sup>2</sup>/g).

These data were input into Equation (2), resulting in a Q value of 4.1 > 0.2.

All published studies about the application of the MTM method have used a cylindrical shape (vials in the pharmaceutical industry), where the mass transfer area (A) does not change as the FD proceeds (one-dimensional axial mass transfer). However, assuming a spherical geometry to represent the blueberry shape, area A does change over time. For this reason, in this study, the quotient (AT/RT), was taken as a regression variable, together with Pi and X (see Equation (1)). Experimental data of the pressure rise vs. time over 30 s were generated every 20 min. The pressure rise data were taken at a rate of 100 data/s and then averaged to generate a data set of 5 data points per second. These data were then regressed against Equation (1) to obtain (AT/RT). In order to show the evolution of the process, Figure 6 depicts the first 10 s of some of the pressure rise vs. time experimental data and their respective regression curves. As expected, the regressions showed neglected values for variable X (in the range of 0.001 to 0.0001), indicating a good seaming of the FD system (no air infiltration). Also, the change of Pi values was minor in comparison with the changes of (RT/AT), meaning that the latter is the main agen<sup>t</sup> of chamber pressure rise [28].

**Figure 6.** Fit of MTM equation to experimental pressure rise profiles for progressive times along the freeze-drying process of whole nontreated blueberries.

It can be seen that at the beginning (0 h 40 min), the pressure rise was low, and it kept increasing up to 2 h 20 min. Then, it started to decrease, approximating the pressure rise observed at the process initiation (10 h 00 min). The (AT/RT) values obtained by the regression analysis at different processing times, together with a tendency line, are shown in Figure 7.

**Figure 7.** Values of (AT/RT) along the freeze-drying time (•) and its corresponding tendency line (**—**) of whole nontreated blueberries.

This value started at a low value, which means that a high RT value has an important effect in the quotient (AT/RT), and it increased rapidly over time up to 4 h, as some of the blueberry fruits busted and RT decreased, in spite of the AT value that should have been decreasing during this period. This reveals that the main effect is that given by the important reduction of RT (busting process). If an average AT over time is assumed (for example, AT = 31.4 cm2), the resistance RT as a function of time can be approximated, and its result is shown in Figure 8, including a tendency line, where, contrary to a normal FD process, RT started at a high value and then decreased sharply before finally increasing again.

This observation reflects the presence of a surface barrier, which appears to be the blueberry skin, which, after being exposed to increasing vapor pressure for some time, cracks, and the effective resistance decreases. This event does not occur at the same time for all FD blueberry fruits, resulting in the sharp, but not instantaneous, RT reduction. This is in agreemen<sup>t</sup> with what was observed by Lu and Pikal [30], Pikal et al. [21], and Pikal [23], where some experiments using water solutions with different solutes showed initial values of RT significantly larger than zero, which suggests a surface barrier resulting from a different structure for the dried product near the surface, which was attributed to the formation of a high-resistance "skin" during the freezing step of the FD process, a conclusion that was supported by scanning electron microscopy. These authors, after performing various experiments with different compositions, suggested four types of resistance curves: Type I, with a linear dependence of RT on the dried layer thickness (or concomitant elapsed time); Type II. with a sharp decrease in RT after a short period of time, followed by a linear increase in RT with increasing time; and Type III and IV, with a saturation curvature toward the time axis, with Type III being less severe than Type IV. The bust process described in this study is clearly a Type II resistance curve.

**Figure 8.** Values of (RT) along the freeze-drying time (•) and its corresponding tendency line (**—**) of whole nontreated blueberries.

These results indicate that the reduction of the initial value of RT, considering a pretreatment to the blueberry skin, would enhance the water vapor flow at the beginning of the process, and consequently would reduce the primary drying time and/or increase the quality yield by bringing down the fraction of busted blueberry units at the end of the process.

#### *3.3. Evaluation of CO2 Laser Microperforation and Blueberries Cut in Half*

Experimental results were obtained for all five pretreated blueberries: 1, 3, 6, and 9 perforations and cut in half, treatments 1 to 6 in Table 1. Microperforations were made as described in the methodology, obtaining perforations in a square grid of 2.0 mm × 2.0 mm density, 0.5 mm diameter, and 1/3 of the blueberry diameter in depth, by utilizing 19% power (19 W). The FD system was set at the same conditions used for whole nontreated blueberries, and the same number of blueberry units were loaded in each experiment. The final moisture at the end of primary drying was also controlled and was within the range of 13% (95% CI (10%, 17%)) to 15% (95% CI (12%, 18%)), for all treatments. Figure 9 shows a blueberry with nine and three CO2 laser microperforations in a square arrangemen<sup>t</sup> with 2.5 mm spacing. It demonstrates that the CO2 laser microperforation process is minimally invasive with a negligible impact on quality characteristics.

**Figure 9.** Appearance of blueberry fruit with nine (**a**) and three (**b**) CO2 laser microperforations.

The PIT and product temperature response were used to evaluate the effect of pretreatment on the primary drying time. Each test was conducted in triplicate to determine the average and standard deviation of the primary drying time. Figure 10 depicts the PIT for the blueberries that were whole and nontreated, cut in half, and microperforated nine times. It is interesting to note that for both cut-in-half and nine-times-perforated blueberries, the initial PIT value started higher than that of the whole nontreated fruit and decreased over time. This is due to the weakened pretreated skin or the area without skin being available to sublimation.

**Figure 10.** Pressure Increase Test (PIT) applied to blueberry freeze-drying to estimate primary drying time as affected by CO2 laser microperforations and cut-in-half treatments.

If the pressure increase was equal to or less than 10%, then the primary drying time was considered to be over.

Table 2 shows the results of the primary drying time for the most significant treatments (treatments 1, 6, and 2 in Table 1).

**Table 2.** Primary drying time of blueberry freeze-drying as affected by CO2 laser microperforations and cut-in-half treatments.


It can be seen that there was an important primary drying time reduction of 60.6% when blueberries were cut in half, as compared with that of whole nontreated blueberries. However, this option is associated with the loss of the important quality characteristic of maintaining the fruits' original shape. On the other hand, and not less importantly, it can be observed that primary drying time reduced by approximately 23.5% when the blueberry skin was pretreated with nine microperforations, as compared with that of whole nontreated blueberries. This result is explained by the fact that the nine-microperforation pretreatment avoids the initial high resistance (RT) at the beginning of the process, mitigating the busting process. This means that, in addition to primary time reduction, CO2 laser microperforation pretreatment can improve the final product quality. It can be seen from Figure 11 that the percentage of nonbusted blueberries significantly increased as the number of perforations per fruit increased, which is statistically sustained by a positive value of the covariance (covariance = +0.634). In addition, the average value of the percentage ofnonbusted blueberries with nine microperforations was 86%, significantly higher than 47% for those with no microperforations. This is supported by their corresponding 95% CIs:

Zero microperforations: average 47%; 95% CI (33%, 61%); Nine microperforations: average 86%; 95% CI (73%, 99%).

**Figure 11.** Experimental and statistical analysis of the nonbusted percentage (%) as affected by CO2 laser microperforations.

Figure 12 also shows that pretreated blueberries are significantly better than those that can be purchased in a local market.

Similar to the analysis made to the whole blueberry process, pressure rise data were regressed against Equation (1) to obtain values of (AT/RT) to finally assess the effect of two pretreatments (nine perforations and cut in half) on RT behavior during the FD process and compare them with that of the whole nontreated blueberry process. Figures 13 and 14 depict the resistance behavior of cut-in-half and nine-perforation treatments over time, respectively. In both cases, it can be observed that there is a sort of lag period for the RT value, which can be attributed to the fact that the skin resistance (Rsk) is still significantly higher than that of the dried layer, with this lag period being longer for the nine-perforation pretreatment than that of the cut-in-half treatment (about 3 h difference). After the lag period, both of them behave like the normal FD process.

**Figure 12.** Differences between freeze-dried blueberries as affected by pretreatment and comparison with those that can be found in local markets.

**Figure 13.** Values of (RT) along the freeze-drying time (•) and its corresponding tendency line (**—**) of cut-in-half treated blueberries.

**Figure 14.** Values of (RT) along the freeze-drying time (•) and its corresponding tendency line (**—**) of nine CO2 laser microperforations treated blueberries.

In order to compare all three treatments—whole nontreated, nine-perforation, and cut-in-half blueberries—the tendency lines were represented in the same graph (Figure 15). This figure summarizes the phenomenological effect of the skin and pretreatments on the blueberry freeze-drying, an effect that is reflected by the RT behavior during FD processing. According to the classification made by Pikal et al. [21], the whole nontreated blueberry FD process, as mentioned, would follow a Type II resistance curve; the cut-in-half blueberries would follow a Type IV process with a severe curvature toward the time axis; and finally, the nine-perforation blueberries would exhibit Type III behavior, with a less severe saturation curvature toward the time axis than that of Type IV. These outcomes phenomenologically describe the effect of the skin and pretreatments on the mass transfer process, which is mainly reflected by the evolution of RT throughout the FD processing time.

**Figure 15.** Comparison of mass-transfer resistance profiles as affected by the main treatments.
