Understanding the break-up phenomena in an orifice-valve high pressure homogenizer using spherical bacterial cells (Lactococcus lactis) as a model disruption indicator

Journal of Food Engineering 236 (2018) 60–71

Contents lists available at ScienceDirect

Journal of Food Engineering journal homepage: www.elsevier.com/locate/jfoodeng

Understanding the break-up phenomena in an orifice-valve high pressure homogenizer using spherical bacterial cells (Lactococcus lactis) as a model disruption indicator

T

Nicola Coccaroa, Giovanna Ferraria,b, Francesco Donsìa,∗ a b

Department of Industrial Engineering, University of Salerno, Italy ProdAl scarl, c/o University of Salerno, Italy

A R T I C LE I N FO

A B S T R A C T

Keywords: High pressure homogenization Orifice valve Lactococcus lactis Break-up Fluid dynamics Bacterial inactivation

The break-up phenomena occurring in a high pressure homogenizer equipped with an interchangeable orifice valve were investigated by measuring the inactivation of Lactococcus lactis. Data were collected at varying the orifice size (80, 100, and 150 μm), the operating pressure (100–200 MPa), the number of passes (1–10), and the fluid viscosity (2.5–7.9 mPa s, changed by adding 0–50 % wt PEG 200 to buffered peptone water) to identify the correlations of the fragmentation occurring in the valve with the main fluid dynamic phenomena (turbulence, elongational and shear stresses, and cavitation). In addition, also the effects of a purely shearing or ultrasound treatment on cell break-up were considered. The results show that the most intense break-up phenomena occur for the smallest orifice size, highest pressure, and lowest viscosity. However, at low viscosity, turbulence, together with the elongational stresses appear to be the controlling factors of cell break-up, whereas, at higher viscosities, the shear stresses become increasingly important. The occurrence of cavitation is only slightly affected by viscosity, and mainly depends on the velocities reached in the homogenization valve.

1. Introduction High-pressure homogenization (HPH) is an emerging technology, which has found wide application as a wet-milling, size-reduction process to produce fine suspensions of particles or droplets (Singh et al., 2016) or submicrometric emulsions (Gupta et al., 2016), as well as to disrupt cells, for the recovery of intracellular material or the reduction of the microbial load in liquid matrices (Diels and Michiels, 2006; Donsì et al., 2013, 2009a; 2009b; Maresca et al., 2011). An HPH unit is a relatively simple continuous system, where the process fluid is compressed in one or more pressure intensifiers, and forced through a specifically designed homogenization chamber. In the homogenization chamber, the pressure energy accumulated in the fluid is released in the passage through a narrow gap, resulting in the development of fluid velocities up to 200–400 m/s (Donsì et al., 2013; Floury et al., 2004a), with the generation of significant fluid mechanical stresses, such as elongational and shear stresses, turbulence and cavitation, which cause particle or cell disruption (Donsì et al., 2013, 2009a; Floury et al., 2004b). The disruption efficiency in HPH processes is mainly related to the operating pressure and the valve geometry, comprising the flow path ∗

and the gap size. Therefore, the research efforts in recent years have mainly focused on the development of new materials capable of withstanding pressure levels up to 350–400 MPa (Dumay et al., 2013), as well as on the proprietary design of valve geometries better finalized to optimizing the final product properties (Donsì et al., 2013, 2012; Lee and Norton, 2013). More specifically, the microbial inactivation by HPH occurs through a purely mechanical process of disruption of the cell walls, whose efficiency depends on the balance between the disruptive forces generated in the homogenization chamber and the cell wall resistance (Donsì et al., 2013, 2009a). However, the fundamentals of the process need to be better clarified. Previous studies from our research group, comparing the microbial inactivation in a piston-valve system with that achieved in an orificevalve system, showed that the inactivation levels in the piston valve were always significantly higher than in the orifice valve for Escherichia coli (Gram-negative), Lactobacillus delbrueckii (Gram-positive) and Saccharomyces cerevisiae (yeast). Therefore, it was hypothesized that the fluid dynamics conditions establishing in the piston valve are extremely favorable to microbial inactivation, probably also because of the direct interaction of the valve surfaces with the cell walls, which reduced the

Corresponding author.Department of Industrial Engineering, University of Salerno, via Giovanni Paolo II, 132, 84084, Fisciano, SA, Italy. E-mail address: [email protected] (F. Donsì).

https://doi.org/10.1016/j.jfoodeng.2018.05.011 Received 15 December 2017; Received in revised form 2 May 2018; Accepted 12 May 2018 Available online 14 May 2018 0260-8774/ © 2018 Elsevier Ltd. All rights reserved.

Journal of Food Engineering 236 (2018) 60–71

N. Coccaro et al.

List of symbols

Re Sh Tin Tout

aj bj BPW Ca dcell dvalve HPH j K kj Lc n N N0 P2

kinetic constant of cell break-up [−] kinetic constant of cell break-up [−] buffered peptone water capillary number [−] mean droplet size of the bacterial cells [m] orifice-valve diameter [m] High Pressure Homogenization treatment type (HPH, Sh, or US) cavitation number [−] kinetic constant of cell break-up [−] orifice length [m] HPH pass number [−] bacterial population [cfu/mL] initial bacterial population [cfu/mL] absolute pressure downstream of the homogenizing valve [MPa] Pv vapor pressure of water downstream of the homogenizing valve, at local temperature conditions [MPa] Ca capillary number [−] PEG 200 Poly(ethylene glycol) with an average molecular weight of 200 g/mol

US uvalve We WUS ΔPh

Reynolds number [−] shearing treatment inlet temperature to the homogenization valve [°C] discharge temperature from the homogenization valve [°C] ultrasound average fluid velocity in the valve orifice/gap [m/s] Weber number [−] volumetric power of the applied US treatment [W/mL] total pressure drop in the homogenization valve [MPa]

Greek symbols α γ η ϑj ϕj ρ σ τ

fluid acceleration in the valve orifice, uvalve/Lc [s−1] shear rate [s−1] dynamic viscosity of the process medium [Pa·s] treatment duration [as number of passes n, or s or min] intensity level of each treatment [Pa for HPH or Sh, or W/ mL for US] density of the process medium [kg/m3] cell wall resistance [N/m] shear stress [Pa]

fixed geometry, based on a disruption chamber with interchangeable orifices. Orifice valves are becoming of increasing interest in food processing, because of their capability to sustain ultra-high operating pressures (Mohan et al., 2016). Moreover, the proposed setup enables to minimize the alteration of the fluid distribution especially in the discharge region, as well as to use the same pressure intensification device and to apply similar rates of heat removal. As a model on-off disruption indicator, we propose the use of cocci bacteria (Lactococcus lactis), to enable the quantification of the occurrence of disruption on simply enumerating the survivor bacteria. In previous studies, it was shown that HPH treatment causes an on-off microbial disruption, without any measurable sub-lethal damage in Listeria innocua (Briñez et al., 2006c) and Escherichia coli (Briñez et al., 2006a): apparently, once the cell membrane is torn apart by the fluid-mechanical stresses generated in the homogenization valve, the microbial cells are not able to recover. In addition, L. lactis exhibit two additional advantages in the planned experiments: they are nearly spherical microorganisms, enabling to rule out the shape factors from the break-up process, and have an average size of 1 μm (Nomura et al., 2009; Yeung et al., 2016), which is significantly lower than the size of the smallest orifice valve used (80 μm), reducing the extent of interaction with the valve surfaces.

effect of the differences in cell shape and wall resistance (Donsì et al., 2013). Moreover, the comparison of the microbial inactivations achieved by different homogenization valve assemblies, even when equipping the same HPH machine, is hindered by the variation in the temperature reached by the processing medium due to the different heat dissipation rates occurring in each assembly (Cavender and Kerr, 2011). In fact, a significant part (40–60%) of the pressure energy is dissipated as frictional heat (Cortés-Muñoz et al., 2009), causing an instantaneous temperature increase in the product of 0.15–0.20 °C/MPa (Donsì et al., 2013), which might significantly affect the extent of microbial inactivation, if not rapidly removed by heat exchange devices, not only through the direct thermal inactivation, but also because of the changes occurring in fluid viscosity and vapor pressure, which influence the intensity of the fluid-mechanical stresses and cavitation. In addition, in the case of the emulsification process, it was reported that the drop fragmentation is strongly correlated to the valve geometry and in particular to the volume over which the energy dissipates (Lee et al., 2014), which is, often, in the discharge region where a jet forms (Innings and Trägårdh, 2007). However, in the discharge region also recoalescence phenomena take place, which are controlled by the extent of adsorption of the emulsifier molecules at the newly formed interfaces (Håkansson et al., 2009a), making the comparison of different valve geometries even more complex. Previous results have shown that the emulsification process is more efficient in the orifice valve than in the piston valve (Donsì et al., 2012; Stang et al., 2001), and that the differences between the two valves diminish when using fast-adsorbing emulsifiers, supporting the hypothesis that the recoalescence phenomena, in the case of food oil-inwater emulsions, play a dominating role (Donsì et al., 2012). More recently, it was shown that the performance of a Microfluidizer chamber (impinging jets microchannels) is better than a piston valve system for the production of O/W emulsions (Lee and Norton, 2013), whereas for the production of W/O emulsions, characterized by a significantly higher viscosity of the continuous phase, the performances of the two homogenization valves are comparable (Lee et al., 2014), suggesting a strong role of the viscosities of the continuous and dispersed phases. This work aims at investigating the fundamentals of the break-up phenomena occurring in an HPH disruption chamber, by using a simple

2. Materials and methods 2.1. Bacterial viability tests 2.1.1. Bacterial suspension Lactococcus lactis is a bacterial species of nearly spherical shape, as resulting from several observations by scanning electron microscopy (Nomura et al., 2009; Yeung et al., 2016), with an average size comprised between 1 and 2 μm (Nomura et al., 2009; Yeung et al., 2016). The Lactococcus lactis subsp. cremoris (ATCC® 19257™) bacterial culture was grown in brain heart infusion broth (BHIB, Oxoid, UK) to the stationary phase (approximately 108 CFU/mL) in an aerated incubator (Function Line 7000 incubator (Heraeus Instruments, Germany)) for 24 h at 30 °C. Subsequently, the culture was diluted with buffered peptone water (BPW, Oxoid, UK) to obtain a working suspension at a final concentration of 104 CFU/mL. The viscosity of the working suspension (buffered peptone water 61

Journal of Food Engineering 236 (2018) 60–71

N. Coccaro et al.

with the bacterial cells at 104 CFU/mL) was modified by the addition of 20 % wt or 50 % wt of PEG 200 (Poly(ethylene glycol), Sigma-Aldrich, US) to the BPW prior to dilution, causing an increase in viscosity. The viscosity values, measured in the range 5–50 °C, and reported in Table S.1 in the Supplementary Material, were comprised between 3.1 and 2.2 mPa s for buffered peptone water without PEG 200 addition (η1), between 4.5 and 2.4 mPa s for 20 % wt of PEG 200 addition (η2), and between 10.7 and 3.7 mPa s for 50 % wt of PEG 200 addition (η3). The viscosity measurements were carried out using a strain-controlled rheometer (AR 2000, TA Instruments, Ltd., Crawley, UK), at temperatures from 5 to 50 °C, which is a range wide enough to encompass the homogenization inlet temperature (5 °C) and maximum discharge temperature (39 °C). The viscosity was measured over shear rates in the range of 10–1000 s−1 with a conical concentric cylinder geometry (stator inner diameter 15 mm, rotor outer diameter 28 mm, cylinder immersed height 42 mm, cone angle 2, gap size was set to 1 mm). The values of the apparent viscosity η were extrapolated when a constant value was reached at increasing the shear rate, at the different temperature values.

2.2. Experimental 2.2.1. HPH treatment The working bacterial suspensions were processed by high pressure homogenization (HPH), using an in-house developed system, equipped with an air-driven Haskel pump model DXHF-683 (EGAR S.r.l., Milano, Italy), and a disruption valve module (model WS1973, Maximator JET GmbH, Schweinfurt, Germany) with zirconia orifice disks with diameters of 80, 100, and 150 μm. The orifice geometry is shown in Fig. 1a and c, and is illustrated with the help of a section view (Fig. 1b). Considering that the system was operated with a single pressure intensifier, a pulsed flow was generated, at an oscillating pressure value. Therefore, the indicated operating gauge pressure and flow rate values (Table S2 in the Supplementary Material, for the different operating conditions) correspond to average values during operation. The typical range of variation for pressure was ± 20 MPa. The HPH treatments consisted of multiple (1–10) passes at three pressure levels, which depended on the orifice used. For 80 and 100 μm orifices, the gauge pressures of 100, 150, and 200 MPa were investigated, whereas for the 150 μm orifice, the gauge pressures of 100, 150, and 180 MPa were tested, because of the limitations of the test rig at the higher flow rates occurring with the larger orifice. After each pass, the suspension was cooled at 5 °C in a tube-in-tube heat exchanger, mounted immediately downstream of the orifice valve, to minimize the thermal effects associated with the frictional heating (0.17 ± 0.1 °C/MPa in the system used). Two additional cooling systems were added immediately upstream and downstream of the pump, in order ensure an inlet temperature to the homogenization valve of 5 °C. The schematics of the HPH unit is shown in Fig. 1d. The HPH system was fully characterized by preliminary measurements of flow rates as a function of the used orifice disk and operating pressure.

2.1.2. Viable cell count The level of bacterial inactivation was determined by the microdilution agar plate method. A sample of 1 mL was withdrawn from treated and control bacterial suspensions, serially diluted and inoculated in Petri plates containing sterile BHI agar. After incubation at 25 °C for 36–48 h, the colonies were enumerated to determine the viable bacterial population, which was reported as CFU/mL (colony forming units per mL of sample). For each test, the level of inactivation, log(N/N0), where N is the surviving colony count after treatment, and N0 the initial colony count, were evaluated. The detection limit of the viable cell count was 1 CFU/mL. The bacterial samples, withdrawn at different processing conditions from the specific treatments described in Section 2.2, were placed in sterile falcon tubes and placed in ice to minimize bacterial growth prior to plating, which always started within 1 h from the HPH treatment. For each condition, independent duplicate experiments were conducted, and each of them was plated in duplicate. Results are, hence, reported as mean values of four measurements ± standard deviations.

2.2.2. Ultrasonication The bacterial suspensions were also treated by ultrasonication (US), to single out the effect of cavitation on the inactivation process from the effect of the other fluid mechanical stresses. The sonicator UP400S (Hielscher Ultrasonics GmbH, Germany), (400 W, 24 kHz) equipped with the sonotrode H3 made of titanium and with a tip diameter of Fig. 1. Pictures and schematics of the exchangeable orifice disks used in the homogenization valve assembly. The upper panel shows the orifice disk for a fluid flow from left to right: (a) picture of the orifice disk casing from the high pressure side, (b) section view of the orifice disk casing, with the indication of the zirconia orifice disk (units in mm), and (c) picture of the orifice disk casing from the low pressure side/fluid jet side. The lower panel (d) shows the schematics of the high pressure homogenization unit, highlighting the pressure intensification system, the pressure indicator (PI), the orifice homogenization valve as well as the heat exchangers used before and after pressure intensification, and after homogenization.

62

Journal of Food Engineering 236 (2018) 60–71

N. Coccaro et al.

3 mm, was used at an amplitude ranging from 60 to 100% (maximum amplitude of 210 μm, acoustic power density of 460 W/cm2). 30 mL of samples in a falcon tube, placed in an ice bath to avoid temperature rise, were subjected to treatments of duration variable between 30 s and 5 min at volumetric powers of 8.0, 10.7 and 13.3 W/mL. The current knowledge on cavitation phenomena during HPH treatments enables to identify the possibility of its occurrence, but not to estimate its intensity. Therefore, the treatment conditions of ultrasonication were selected on the basis of the available US unit, investigating a wide range of volumetric power, in order to capture the effect of the main operating parameters.

ranging between 1.3 and 7.5 Pa for η1, between 1.7 and 10.2 Pa for η2, and between 3.4 and 20.1 Pa for η3. In the case of shearing, it was not possible to reproduce in the shear cell the extreme shearing conditions occurring in the HPH valve. Through the use of the capillary number (Ca), presented in details in Section 4, it was possible to estimate that the intensity of the shear stresses occurring on bacterial cells was about 2 order of magnitudes lower in the shear cell than in the HPH valve. Therefore, the main goal of the shearing experiments is only to observe the dependency of bacterial inactivation on the main operating parameters, when shearing alone is applied.

2.2.3. Shearing The effect of shearing (Sh) on the inactivation of the bacterial suspension was singled out from the other fluid mechanical stresses by conducting a specific treatment in a shear cell with a conical concentric cylinder geometry (stator inner diameter 15 mm, rotor outer diameter 28 mm, cylinder immersed height 42 mm, cone angle 2°, gap size was set to 1 mm) of a strain-controlled rheometer (AR 2000, TA Instruments, Ltd., Crawley, UK). 19 mL of samples were poured in the preliminarily sterilized cell and were subjected to treatments of duration between 1 and 10 min and temperature of 20 °C (where η1 = 2.5 mPa s, η2 = 3.4 mPa s, η3 = 6.7 mPa s, as shown in Supplementary Material), at shear velocities of 500, 1000 and 3000 s−1, corresponding to shear stresses

2.3. Data fitting 2.3.1. HPH inactivation N The bacterial inactivation curves (log N ), as a function of the in0 tensity of HPH, US or Sh treatments were fitted by an empirical power law equation (Weibull model), reported in equation (1), previously validated for the inactivation by HPH of E. coli, L. delbrueckii and S. cerevisiae (Donsì et al., 2013).

log

N a b = −kj⋅ϕj j⋅ϑ jj N0

(1)

In equation (1), the variables ϕj and ϑj represents respectively the intensity level of each treatment j and the corresponding treatment

Fig. 2. Survivor fraction of L. lactis as a function of number of HPH passes at different pressure levels (parametric in each panel) and an inlet temperature of 5 °C, for different orifice diameters (80 μm, 100 μm, and 150 μm) and for different suspension viscosities (η1 = 3.1 mPa s, η2 = 4.5 mPa s, η3 = 10.7 mPa s, evaluated at 5 °C). Solid lines represent the fitting curves of the experimental data (symbols) using equation (1). 63

Journal of Food Engineering 236 (2018) 60–71

N. Coccaro et al.

mechanistic fitting equation (1), for the description of the inactivation levels attained as a simultaneous function of the gauge pressure ΔPh and number of passes n. The fitting is always good, as shown by R2 values. Moreover, also the pressure levels required to attain a 90% inactivation of L. lactis are reported for a different number of passes (Table 1). The data of Table 1 suggest that the differences in the kinetics of inactivation between the 80 and the 100 μm orifices are negligible at low viscosity (η1) and become more substantial as the viscosity is increased (η2 and η3). In contrast, the differences between the 150 μm orifice and the smaller ones are more significant at low viscosity than at higher ones. In fact, a higher inactivation is attained for aHPH, bHPH → 1 (linear dependence of log N / N0 on ΔPh and n), while a worse performance, with weaker dependence on ΔPh and n, occurs when aHPH, bHPH → 0. However, the analysis of the single kinetic parameters might generate some confusion in the reader, due to simultaneous fluctuations of aHPH and bHPH. Therefore, the ΔP0.9 values for different number of passes are reported to help visualizing the inactivation levels predicted by the model for the different orifices and viscosities. The data of Table 1 show that the orifice valve with the smallest diameter, 80 μm, at the operating pressure of 200 MPa after 10 passes is able to completely kill the bacterial cells (bacterial count below the sensitivity limits), whereas at increasing the orifice diameter and at reducing the pressure level, the process is less effective because the fluid velocity is reduced and with it the intensity of turbulence and elongational stresses.

duration. For HPH, ϕHPH = ΔPh , that is the total pressure drop occurring in the valve, and ϑHPH = n , that is the number of HPH passes. For US, ϕUS = WUS , that is the volumetric power of the applied treatment, and ϑUS = tUS , that is the treatment time. For shearing, ϕSh = τSh , that is the shear stress of the treatment, and ϑSh = tSh , that is the treatment time. The kinetic parameters kj , aj , and bj have all a physical meaning. In particular, kj indicates the intrinsic disruption ability of the homogenization valve, of the US equipment or of the shearing treatment: it represents the theoretical log-cycle inactivation corresponding to a single pass at 1 MPa for HPH, to a US treatment of 1 s at 13.3 W/mL, and to a shearing treatment of 1 s at a shear stress of 1 Pa, respectively. Instead, aj and bj represent the kinetic constants of bacterial inactivation, and therefore indicate the sensitivity of the microorganism species to the operating conditions of intensity (ϕj ) and duration (ϑj ), respectively. The fitting models were applied to the set of data comprising varying treatment intensity and treatment times, whereas the suspensions of different viscosity were fitted as different sets of data. 3. Results 3.1. Effect of HPH orifice size The effect of the fluid dynamics conditions established in the homogenization valves on the break-up efficiency is investigated by determining the extent of inactivation of L. lactis at varying the diameter of the orifice valve (between 80 and 150 μm) and the viscosity of the bacterial suspension (η1, η2, η3). The results are reported in Fig. 2. The different panels of Fig. 2, which are all plotted against the same scales on x- and y-axis, give a visual overview of the effect of the main parameters involved: the degree of inactivation, which corresponds to the efficiency of cell break-up, increases (a) with the number of HPH passes, (b) with increasing the operating pressure, (c) with decreasing the orifice size, and (d) with decreasing the viscosity. In particular, the dependence of the kinetics of bacterial inactivation on the number of passes and pressure level is well known, and has been discussed in details by several authors (Diels and Michiels, 2006; Donsì et al., 2013, 2009a), whereas the effect of orifice size and fluid viscosity has not been explicitly investigated before. When analyzed in terms of the maximum inactivation level achieved (10 passes at 200 MPa for 80 and 100 μm orifices, and at 180 MPa for 150 μm orifice), the data of Fig. 2 show that, in the case of a fluid of viscosity η1, the observed inactivation is of 4.8 log cycles, 4.2 and 3.2 log cycles, respectively, in the case of a fluid of viscosity η2, it is of 4.3, 3.4 and 3.0 log cycles, respectively, while in the case of a fluid of viscosity η3 it is of 2.7, 2.4 and 2.4 log cycles respectively for the 80, 100, and 150 μm orifices. However, the maximum inactivation level does not take into account the kinetics of inactivation. Therefore, the data of Fig. 2 are compared in Table 1 in terms of the kinetic parameters of the

3.2. Effect of temperature Especially at high operating pressure, the effect of the temperature reached in the homogenization valve might become significant on bacterial inactivation, hindering the effect of the fluid mechanical stresses on cell disruption. Therefore, the temperature threshold for a purely thermal inactivation treatment is determined under shearing, at a shear rate of 200 s−1, through the measurement of bacterial inactivation at different temperatures as a function of time. The results are reported in Fig. S.1 of the Supplementary Material. The shearing conditions used were selected to maintain the bacterial suspension (in BPW) well mixed, at the same time being sure that it did not significantly affect the bacterial inactivation: the shearing treatments at 200 s−1 at 20 °C caused only a very marginal inactivation (< 0.2 log-cycle), as described in details in section 3.3. Fig. S.1 clearly shows that at 70 °C L. lactis population is significantly reduced (4.6 log-cycle reduction) already after 1 min and is almost completely inactivated after 6 min (5.4 log-cycle reduction). At 60 °C, a gradual increase of the inactivation level is observed over time, with 1.3 log-cycle reduction after 1 min and 4.8 log-cycle reduction after 10 min. At 50 °C, only a marginal bacterial inactivation is observed, corresponding to 0.5 log-cycles after 1 min and 0.8 log-cycles after 10 min. At 40 °C, the bacterial inactivation is always negligible

Table 1 Kinetic parameters of the fitting curves of the experimental data of Fig. 2, according to equation (1), together with the corresponding coefficients of variation R2. In addition, the estimated gauge pressure levels required to attain a level of inactivation of 90% (ΔP90) are reported as a function of the number of passes n. Fluid viscosity

Orifice diameter (μm)

kHPH

aHPH

bHPH

R2

ΔP0.9 (n = 1) MPa

ΔP0.9 (n = 5) MPa

ΔP0.9 (n = 10) MPa

η1

80 100 150

8.79·10−2 8.93·10−2 1.66·10−2

0.583 0.566 0.858

0.406 0.401 0.392

0.962 0.961 0.926

65 75 127

21 55 91

13 48 79

η2

80 100 150

9.71·10−3 9.83·10−3 4.47·10−2

1.073 0.987 0.651

0.209 0.300 0.349

0.979 0.942 0.983

71 108 148

23 66 110

14 54 97

η3

80 100 150

6.59·10−3 1.64·10−3 3.30·10−3

1.037 1.284 1.081

0.212 0.235 0.419

0.971 0.946 0.974

118 119 197

57 50 106

41 34 81

64

Journal of Food Engineering 236 (2018) 60–71

N. Coccaro et al.

these experiments should be intended only as indications of the dependence of bacterial disruption on shear stresses. In fact, also the sublethally injured cells were neglected in these experiments, which are likely to occur under these conditions. These results, with the inactivation level increasing with the applied shear stresses, exhibit contrasting trends when compared with the HPH treatments, where the level of inactivation instead increases at increasing the treatment intensity and at decreasing the medium viscosity, and therefore at decreasing the shear stresses in the homogenization valve. In Fig. 4, the effect of changing the temperature at the highest shear rate (3000 s−1) is shown, for the suspension of viscosity η1. With respect to the inactivation reached after 20 min at 20 °C (0.38 log cycles), an increase in temperature causes an increase in the bacterial inactivation, which reaches 0.44 log cycles at 30 °C, 0.52 log cycles at 40 °C and 0.79 log cycles at 50 °C.

(< 0.2 log-cycle reduction). Considering that, during the HPH treatment, the maximum temperature level attained immediately downstream of the homogenization valve is of 39 °C at 200 MPa (0.17 °C/MPa is the measured temperature increase immediately downstream of the valve), and the residence time in the section comprised between the homogenization valve and the heat exchanger is < 5 s, it can be safely assumed that the bacterial inactivation due to a purely thermal effect can be neglected, with respect to the fluid mechanical stress. Obviously, the inherent heating occurring during HPH might play an indirect role, especially for what concerns the abrupt changes in the fluidity of the cell membrane, over the scale of milliseconds, affecting its resistance to the fluid mechanical stresses (Donsì et al., 2013), which has been neglected in this study. In addition, also the formation of hot spots during HPH was neglected, because the used test rig enabled only the measurement of the average temperature in the valve discharge section. The temperature increase in the fluid is due to the dissipation of the pressure energy in the homogenization valve, that many authors correlated to the dissipation of the turbulent kinetic energy of the jet (Håkansson et al., 2011), because this latter term can be predicted via computational fluid dynamics (Becker et al., 2014) or characterized by particle image velocimetry (Håkansson et al., 2011), whereas direct experimental measurements of the former are not feasible in the HPH valve (Becker et al., 2014). The literature results have clearly highlighted the inhomogeneous distribution of the dissipation of the turbulent kinetic energy within homogenization valves of different geometries (Becker et al., 2014; Håkansson et al., 2011, 2009b; Kelemen et al., 2015), hence suggesting also the formation of hot spots. 3.3. Effect of shear stress During the high pressure homogenization process, the shear stress is reckoned to play an important role on the resulting microbial inactivation, together with the elongational stresses, the turbulence and the cavitation (Donsì et al., 2013). The effect of the shear stresses is singled out in a specifically designed experiment, where the bacterial suspensions of different viscosities (η1, η2, η3) are treated at different shear rates (500, 1000 and 3000 s−1) at 20 °C, which corresponds, as reported in Section 2.2.3 to shear stresses ranging between 1.3 and 7.5 Pa for η1, between 1.7 and 10.2 Pa for η2, and between 3.4 and 20.1 Pa for η3. The results, reported in Fig. 3, show that the bacterial inactivation increases with the shear stress, which is increased either by increasing the shear rate (from 500 to 3000 s−1) or the viscosity (from η1 to η3). In particular, when increasing the shear rate from 500 to 3000 s−1, the bacterial inactivation after 10 min increases from 0.17 to 0.36 log cycles at the lowest viscosity (η1), from 0.24 to 0.37 at intermediate viscosity (η2), and from 0.44 to 0.68 at the highest viscosity (η3). In general, these inactivation levels, which are all < 1 log cycle, can be considered negligible. However, the experimental data of Fig. 3 suggest a consistent trend as a function of the processing conditions, which was analyzed in Table 2 in terms of kinetic parameters (equation (1)) for the fitting of the data, together with the estimated treatment times required to achieve an inactivation level of 50% at the different shear rates tested. Because of the low inactivation level reached by the shearing treatment, a 50% inactivation level was selected, instead of the 90% value used in Tables 1 and 3, to avoid model predictions outside of the range of the experimental data used in the fitting. Table 2 highlights, as easily expected, that, under the investigated conditions, the bacterial inactivation increases with increasing the shear stresses, either through a viscosity increase or a shear rate increase. Similarly to the thermal treatments, the selected treatment times were significantly longer than the characteristic residence times in the homogenization valve, because it was not possible to replicate the same scale of intensity of shear stresses during HPH in the used shear cell, and longer times were needed to obtain measurable bacterial inactivations. Therefore,

Fig. 3. Survivor fraction of L. lactis as a function of treatment time at different shear rates (parametric in each panel) and at 20 °C for different suspension viscosities (η1 = 2.5 mPa s, η2 = 3.4 mPa s, η3 = 6.7 mPa s, evaluated at 20 °C). Solid lines represent the fitting curves of the experimental data (symbols) using equation (1). 65

Journal of Food Engineering 236 (2018) 60–71

N. Coccaro et al.

Table 2 Kinetic parameters of the fitting curves of the experimental data of Fig. 3, according to equation (1), together with the corresponding coefficients of variation R2. In addition, the estimated treatment times required to attain a level of inactivation of 50% (tSh,0.5) are reported for the different shear rates γ investigated. Fluid viscosity

ksh

ash

bsh

R2

tSh,0.5 (γ = 500 s−1) min

tSh,0.5 (γ = 1000 s−1) min

tSh,0.5 (γ = 3000 s−1) min

η1 η2 η3

2.41·10−4 6.19·10−3 1.95·10−2

0.437 0.244 0.232

0.531 0.355 0.245

0.973 0.986 0.990

29.6 15.8 1.8

16.8 9.8 0.9

6.8 4.6 0.3

work, thus further affecting cavitation. To simplify the analysis of the effect of the processing conditions on bacterial inactivation, also for the ultrasound treatment the kinetic parameters (equation (1)) for the fitting of the data of Fig. 5, are reported in Table 3, together with the estimated treatment times required to achieve an inactivation level of 90% at the different volumetric powers tested. Table 3 clearly highlights the dependence on the

Fig. 4. Survivor fraction of L. lactis as a function of treatment time at different temperatures (in BPW, with viscosity (η1 = 2.2–3.1 mPa s, sheared at 3000 s−1)). Solid lines represent the fitting curves of the experimental data (symbols) using equation (1).

The temperatures of 20 °C and 30 °C, where the minimum level of inactivation is observed, corresponds to the temperature range for the optimal growth of the microorganism (25–28 °C). Interestingly, when the temperature is decreased to 5 °C, for short treatment times (1–4 min), the level of inactivation is higher than observed at the temperatures comprised between 20 and 40 °C, and the kinetics appear to be different. First of all, the reduction in temperature causes an increase in viscosity (which goes from 2.5 mPa s at 20 °C to 3.1 mPa s at 5 °C) and hence in shear stresses (from 7.5 Pa at 20 °C to 9.2 Pa at 5 °C). However, it is also possible that, at low temperature, the cell membrane becomes less fluid and hence less resistant to the shear stresses. More detailed studies are needed to elucidate this issue. 3.4. Effect of cavitation The effect of the cavitation, which is also considered to contribute to the microbial inactivation during HPH (Donsì et al., 2013), is singled out from the other fluid dynamic effects by treating the bacterial suspensions of different viscosities by ultrasonication at different volumetric power (8.0, 10.7 and 13.3 W/mL) at 5 °C. The results are shown in Fig. 5. In general, it can be observed that the rate of bacterial inactivation as a function of time increases with increasing the volumetric power and, only slightly, with decreasing the viscosity. In particular, when increasing the volumetric power from 8.0 to 13.3 W/mL, the bacterial inactivation after 5 min increases from 0.87 to 1.15 log cycles at the lowest viscosity (η1), from 0.89 to 1.10 at intermediate viscosity (η2), and from 0.77 to 1.03 at the highest viscosity (η3). The maximum inactivation level is generally low when compared with the HPH treatment, even after 5 min of treatment. Moreover, it must be remarked that the dependence on viscosity is weak, and exhibits an opposite trend with respect to what observed for shearing. However, it is important to notice that the PEG addition causes not only a change in viscosity, but also the vapor pressure, not considered in this

Fig. 5. Survivor fraction of L. lactis as a function of number of time at different ultrasound volumetric power (parametric in each panel) for different suspension viscosities (η1 = 3.1 mPa s, η2 = 4.5 mPa s, η3 = 10.7 mPa s, evaluated at 5 °C). Solid lines represent the fitting curves of the experimental data (symbols) using equation (1). 66

Journal of Food Engineering 236 (2018) 60–71

N. Coccaro et al.

Table 3 Kinetic parameters of the fitting curves of the experimental data of Fig. 5, according to equation (1), together with the corresponding coefficients of variation R2. In addition, the estimated treatment times required to attain a level of inactivation of 90% (tUS,0.9) are reported as a function of volumetric power. Fluid viscosity

kUS

aUS

bUS

R2

tUS,0.9 (WUS = 8.0 W/mL) min

tUS,0.9 (WUS = 10.7 W/mL) min

tUS,0.9 (WUS = 13.3 W/mL) min

η1 η2 η3

1.42·10−1 2.65·10−1 4.01·10−3

0.567 0.360 0.833

0.388 0.268 0.702

0.990 0.988 0.973

7.4 8.7 8.3

4.8 5.9 5.9

3.5 4.4 4.5

SYTO9 and propidium iodide staining. More recent studies confirmed that, when plate counted on selective media, E. coli, treated by HPH using a throttling valve, did not show any evidence of sub-lethal injuries (De Lamo-Castellví et al., 2013). However, the same authors also highlighted that the treatment caused damages to the membrane integrity of survival cells, as highlighted by flow cytometry and epifluorescent microscopy. The results of the present study, especially those reported in Fig. 2, owing to the pronounced tailing of the curves, suggest that the occurrence of sub-lethal injuries can be excluded, or at least can be considered less important than the progressive selection of the most resistant species. However, in the case of the other treatments investigated (shearing and ultrasounds), sub-lethal injuries might have led to an over-estimation of the inactivation level, due to neglecting the possible recovery of injured cells. A preliminary analysis of the experimental data reported should consider the following evidence:

viscosity of the suspension at all the investigated treatment conditions. The data of Table 3 show that the treatment time required to achieve a 90% inactivation decreases as the volumetric power increases and the viscosity decreased, ranging from 3.5 min at 13.3 W/mL and lowest viscosity (η1) to more than 8 min at 8.0 W/mL and higher viscosities (η2 and η3). These results follow what observed during HPH treatments, with increased inactivation at increasing intensity and at decreasing viscosity.

4. Discussion The dependence of microbial inactivation by HPH on the number of passes and pressure level is well known and has been discussed in details by several authors (Diels and Michiels, 2006; Donsì et al., 2013, 2009a). However, the effect of the fluid dynamics, as affected by valve size and fluid viscosity, has not been systematically investigated for microbial inactivation, and only partly for emulsification processes. In particular, the previous studies on microorganism inactivation by HPH are difficult to rationalize, because of the influence of the shape factors and cell wall resistances of the different microorganisms tested (Diels and Michiels, 2006; Donsì et al., 2013, 2009a; Maresca et al., 2011), as well as of the diversified operating conditions and HPH equipment. The experiments described in Section 3 have been designed to minimize the effect of temperature, at least on direct thermal inactivation, by efficient cooling immediately upstream and downstream of the homogenization valve. In addition, the use of a simple interchangeable orifice valve enables to rule out any difference previously observed in both microbial inactivation (Donsì et al., 2013) and emulsification efficiency (Donsì et al., 2012; Lee and Norton, 2013), due to the difference in equipment characteristics, such as pumping systems and heat dispersion efficiencies. Moreover, the use of an orifice valve geometry also rules out the complexity deriving from the large shearing stresses deriving from impinging or colliding jets generated in microfluidizer or piston valves. In addition, dealing with living systems adds another level of complexity. First of all, it is well known that microorganisms, and hence L. lactis, exhibit a distribution of resistances to physical and thermal treatments, which might especially affect the results of experiments where multiple treatments are carried out on the same population, because of the progressive selection of the most resistant species (Peleg and Cole, 1998). This is, for example, observed in the case of the tailing survival curves for multiple HPH passes (Donsì et al., 2013, 2009a). Secondly, it must also be taken into account that the bacteria might be only sub-lethally injured, making the subsequent passes more effective. However, previous studies tended to exclude sub-lethal injuries during HPH treatments: the fact that L. innocua (Briñez et al., 2006c) and E. coli (Briñez et al., 2006b, 2006d) cells, treated by HPH in milk or orange juice, did not exhibit any significant difference in the lethality values when grown on non-selective and selective media, suggesting that eventually injured cells did not have any capability to recover. The absence of any sub-lethal injuries was observed also for Yersinia enterocolitica and Staphylococcus aureus (Diels et al., 2005a, 2005b; Wuytack et al., 2002), through the study of the sensitization to lysozyme, nisin, and lactoperoxidase enzyme of treated cells, as well as through the evaluation of the cytoplasmic membrane damage, using

− The level of inactivation achieved by HPH increases when the fluid viscosity decreases (Fig. 2); − The level of inactivation achieved by simple shearing increases when the fluid viscosity increases (Fig. 3); − During ultrasound treatments, the effect of fluid viscosity on the inactivation level is negligible (Fig. 5). Therefore, it is likely that neither shear stresses alone nor cavitation alone play a dominating role in cell or particle break-up under the HPH conditions investigated in this work. However, the data of Fig. 2 bear also other important pieces of information, associated with the changes in the fluid dynamics following the variation of orifice size, fluid viscosity, and operating pressure, hence calling for a more comprehensive analysis. Therefore, the L. lactis inactivation data have been analyzed by comparing the achieved inactivation level with the values of the dimensionless numbers of Reynolds, Weber, Capillary and Cavitation attained in the orifice valve, similarly to what done in a previous work (Donsì et al., 2013). Most previous studies, in fact, use global mean effective values of the breakup forces, since local measurements are difficult (Håkansson et al., 2012), as a consequence of the steep local gradients and wide distributions of fluid velocity and pressure in the homogenization valves. In order to calculate these dimensionless numbers, the following measurements and assumptions were made: − The temperature considered is an average between the measured inlet temperature Tin and the estimated exit temperature Tout (Tout = Tin + 0.17 · ΔPh), where 0.17 is the experimentally measured rate of temperature increase as a function of pressure (°C/MPa). − The flow rates have been measured as a function of pressure for the different tested fluids and using the different orifices (see Table S1 of Supplementary Material). − The orifice length Lc of the orifice valve assembly is measured to be 0.5 mm. − The viscosity and density were measured as a function of temperature (see Table S.1 of Supplementary Material). − The fluid velocity in the valve orifice uvalve is calculated by dividing the measured flow rate by the orifice cross section, calculated from 67

Journal of Food Engineering 236 (2018) 60–71

N. Coccaro et al.

The intensity of the elongational stresses on bacterial cells, as assessed through We, strongly depends on the used orifice and the operating pressure. In particular, We increases at decreasing the orifice size and at increasing pressure, which is in agreement with the observed bacterial inactivation, whereas it is essentially unaffected by the fluid viscosity. Therefore, it can be assumed that, especially for the dependence on the orifice size, the elongational stresses are likely to play a significant role in the break-up phenomena. In contrast, the shear stresses do not appear to play a significant role, at least at the lower fluid viscosities. In fact, the Ca values decidedly increase when increasing the fluid viscosity, which is in strong disagreement with the observed reduction in bacterial inactivation. However, at the highest viscosity, when the Ca values increase and the Re values decreases, it is likely that the role played by the shear stresses become more important, which is in agreement with the experimental inactivation data (Ca increases when decreasing the orifice diameter). The K number attains the highest values at the largest orifice size (150 μm) and at lower pressures because it increases at decreasing the fluid velocity. For the same reason, no significant differences can be observed between the 100 μm and 80 μm orifices, with only minor differences in uvalve (see Table S1 of Supplementary Material). According to previous studies, cavitation is likely to occur at K ≤ 1 (Shirgaonkar et al., 1998), and, therefore, the probability of cavitation phenomena increases for smaller orifice sizes and at increasing operating pressures, which is in agreement with the observed bacterial inactivation. However, the extent of cavitation, estimated on the basis of K values, is not affected by the changes in viscosity, which is in agreement with the observed scarce dependence of the lethality of US treatments on fluid viscosity (Fig. 5). The analysis of the data of Fig. 6, therefore, suggests that the breakup mechanism cannot be explained by a single fluid dynamic phenomenon. However, it is possible to make a few considerations:

the nominal orifice size. − The acceleration of the fluid α is calculated as uvalve/Lc. − The L. lactis cells were considered as spheres, with a diameter of 1 μm, as a reasonable approximation based on experimental measurements (Kokkinosa et al., 1998). − The cell mechanical resistance σ is set at 10 N/m. Considering that experimental data are not available, it has been decided to use the same order of magnitude of what previously proposed for S. cerevisiae (Kleinig and Middelberg, 1996). In principle, the dependence on temperature should be included, but the data of Fig. 4 suggest that it is small in the temperature range under investigation (below 0.5 log cycles), being mainly related to the changes in membrane fluidity. Despite the last two assumptions might introduce some inaccuracies in the actual values of the dimensionless numbers, they do not affect the analysis of how each dimensionless number varies when varying the main operating parameters investigated. The effect of turbulence achieved in the orifice valve has been assessed through the Reynolds number (Re):

ρu valve d valve η

Re =

(2)

In principle, the contribution of turbulence to break-up phenomena becomes important during the interaction of the jet generated in the orifice and the surrounding fluid (discharge section) (Lee et al., 2014). However, the differences between the velocity of the jet (uvalve) and the mean velocity of the fluid in the discharge zone are such (about 2 order of magnitudes difference, see Table S.1 of Supplementary Material) that only a negligible error is introduced by neglecting the velocity in the discharge zone in computing the Re number. The elongational stresses acting on cells during the passage through the orifice valve have been assessed through the Weber number (We):

We =

3 ρ α 2 dcell 4σ

− The effect of the turbulence is definitely dominating at low fluid viscosity when Re attains the highest values. Moreover, the dependency of Re on viscosity is the only one, among the considered dimensionless numbers, that correlates well with the dependency of bacterial inactivation on viscosity. − However, the intensity of turbulence, in the tested configuration, appears to depend only marginally on the orifice size, whereas the bacterial inactivation is significantly dependent on it, with the same trends observed for We, Ca and K. − Therefore, it can be assumed that at low viscosity, together with turbulence, also the elongational stresses (We) play a significant role. − In contrast, at higher viscosity values, the shear stresses (Ca) become increasingly important. − Cavitation (K) is substantially unaffected by viscosity and increases at higher velocities in the valve gap.

(3)

The shear stresses acting on the cells have instead been assessed through the modified Capillary number (Ca):

Ca =

ηαdcell σ

(4)

The occurrence of cavitation, following the difference between the pressure upstream and downstream of the valve and the vapor pressure of the fluid, has been assessed through the Cavitation number (K):

K=

P2 − Pv 1 2 ρu valve 2

(5)

A detailed description of the sources and physical meaning of these dimensionless numbers can be found in our previous work (Donsì et al., 2013). The values of the dimensionless numbers Re, We, Ca and K are reported in Fig. 6 as a function of operating pressure, orifice size, and fluid viscosity. The highest Re values are reached for the largest orifice size of 150 μm, while the lowest values are reached for the 80 μm orifice, because, when increasing the orifice size, the effect of the increase of flow rate becomes dominant (see Table S1 of Supplementary Material). It must be remarked that the differences in Re when varying the orifice size are smaller than when changing the operative pressure and the process fluid viscosity. In particular, Re decreases of about one order of magnitude when the viscosity is increased by switching the fluids containing different concentrations of PEG 200, and therefore from viscosity values of η1 up to η3. When comparing the values of Re with the levels of bacterial inactivation, we notice that the dependence on viscosity is in agreement with the variation in Re, whereas it is not with the orifice size.

The apparent contradiction between the data of Fig. 2 and Fig. 3, where observed bacterial inactivation decreases (Fig. 2) or increases (Fig. 3) when increasing the fluid viscosity and therefore the shear stresses, can be explained by the fact that the experiments of Fig. 3 are carried out at Ca values ranging between 4.6·10−7 and 3.2·10−7 at 3000 s−1 for temperatures of 5 and 50 °C, respectively, which are significantly lower than the Ca values estimated during the HPH treatments (3.5·10−5 - 2.7·10−4, depending on fluid viscosity). The proposed analysis can be at least partly extended to the breakup phenomena during emulsification by HPH, considering that, typically, the mean droplet size of primary emulsions, prepared by high shear mixing, is in the micrometric size range (Donsì et al., 2010; Mahfoudhi et al., 2014). However, in this case, not only the O/W interfacial tension is important, which has been taken into account through the cell resistance parameter σ, but also the dispersed phase viscosity, whose increase might extend the time for droplet deformation 68

Journal of Food Engineering 236 (2018) 60–71

N. Coccaro et al.

In particular, the break-up mechanism in a homogenization valve can be correlated to the valve geometry and the volume over which the energy dissipates (Lee et al., 2014), which is, in most of the cases, the jet formed at the exit of the gap (Innings and Trägårdh, 2007). In particular, it is likely that the majority of the break-up phenomena takes place in the regions of highest shearing forces occurring where the jet and the surrounding fluid exhibit the highest velocity differences (Lee et al., 2014), which, in the orifice valve geometry, can be fairly estimated through the Re number. Therefore, our findings are in agreement with the above indications in the case of low viscosity fluids. However, in the case of W/O emulsions, and therefore for a

beyond the residence time within the region of energy dissipation (Lee et al., 2013). Seminal studies on the emulsification processes demonstrated that the fragmentation of emulsion droplets in an HPH valve is mainly caused by turbulence and/or cavitation (Walstra, 1993). However, previous experimental results for the production of O/W nanoemulsions showed that, since droplet size was independent of viscosity ratio between the dispersed and the continuous phase, the break-up phenomena are dominated by turbulence (Lee and Norton, 2013), similarly to what we observed for the correlation between bacterial inactivation and Re number at low viscosity.

Fig. 6. Main dimensionless numbers (Reynolds, Weber, Capillary and Cavitation number) as a function of HPH pressure for different orifice diameters (parametric in each panel), for different suspension viscosities (η1, η2, η3, evaluated at an average temperature between inlet and outlet, with the values reported in Table S.2 of the Supplementary Material). Solid lines are reported to guide the reader's eye. 69

Journal of Food Engineering 236 (2018) 60–71

N. Coccaro et al.

continuous phase with high viscosity, the flow is likely to become elongational when break-up occurs rather than turbulent (Lee et al., 2014), which is instead in agreement with our observations that the turbulence effects decrease at increasing the process fluid viscosity.

Microbiol. 101, 281–291. https://doi.org/10.1016/j.ijfoodmicro.2004.11.011. Diels, A.M.J., De Taeye, J., Michiels, C.W., 2005b. Sensitisation of Escherichia coli to antibacterial peptides and enzymes by high-pressure homogenisation. Int. J. Food Microbiol. 105, 165–175. https://doi.org/10.1016/j.ijfoodmicro.2005.04.027. Diels, A.M.J., Michiels, C.W., 2006. High-pressure homogenization as a non-thermal technique for the inactivation of microorganisms. Crit. Rev. Microbiol. 32, 201–216. https://doi.org/10.1080/10408410601023516. Donsì, F., Annunziata, M., Ferrari, G., 2013. Microbial inactivation by high pressure homogenization: effect of the disruption valve geometry. J. Food Eng. 115, 362–370. https://doi.org/10.1016/j.jfoodeng.2012.10.046. Donsì, F., Ferrari, G., Lenza, E., Maresca, P., 2009a. Main factors regulating microbial inactivation by high-pressure homogenization: operating parameters and scale of operation. Chem. Eng. Sci. 64, 520–532. https://doi.org/10.1016/j.ces.2008.10.002. Donsì, F., Ferrari, G., Maresca, P., 2009b. High-pressure homogenization for food sanitization. Global Issues in Food Science and Technology. https://doi.org/10.1016/ B978-0-12-374124-0.00019-3. Donsì, F., Senatore, B., Huang, Q., Ferrari, G., 2010. Development of novel pea proteinbased nanoemulsions for delivery of nutraceuticals. J. Agric. Food Chem. 58, 10653–10660. https://doi.org/10.1021/jf101804g. Donsì, F., Sessa, M., Ferrari, G., 2012. Effect of emulsifier type and disruption chamber geometry on the fabrication of food nanoemulsions by high pressure homogenization. Ind. Eng. Chem. Res. 51, 7606–7618. https://doi.org/10.1021/ie2017898. Dumay, E., Chevalier-Lucia, D., Picart-Palmade, L., Benzaria, A., Gràcia-Julià, A., Blayo, C., 2013. Technological aspects and potential applications of (ultra) high-pressure homogenisation. Trends Food Sci. Technol. https://doi.org/10.1016/j.tifs.2012.03. 005. Floury, J., Bellettre, J., Legrand, J., Desrumaux, A., 2004a. Analysis of a new type of high pressure homogeniser. A study of the flow pattern. Chem. Eng. Sci. 59, 843–853. http://dx.doi.org/10.1016/j.ces.2003.11.017. Floury, J., Legrand, J., Desrumaux, A., 2004b. Analysis of a new type of high pressure homogeniser. Part B. study of droplet break-up and recoalescence phenomena. Chem. Eng. Sci. 59, 1285–1294. https://doi.org/10.1016/j.ces.2003.11.025. Gupta, A., Eral, H.B., Hatton, T.A., Doyle, P.S., 2016. Nanoemulsions: formation, properties and applications. Soft Matter 12. https://doi.org/10.1039/c5sm02958a. Håkansson, A., Fuchs, L., Innings, F., Revstedt, J., Trägårdh, C., Bergenståhl, B., 2011. High resolution experimental measurement of turbulent flow field in a high pressure homogenizer model and its implications on turbulent drop fragmentation. Chem. Eng. Sci. 66, 1790–1801. https://doi.org/10.1016/j.ces.2011.01.026. Håkansson, A., Innings, F., Revstedt, J., Trägårdh, C., Bergenståhl, B., 2012. Estimation of turbulent fragmenting forces in a high-pressure homogenizer from computational fluid dynamics. Chem. Eng. Sci. 75, 309–317. https://doi.org/10.1016/j.ces.2012.03. 045. Håkansson, A., Trägårdh, C., Bergenståhl, B., 2009a. Studying the effects of adsorption, recoalescence and fragmentation in a high pressure homogenizer using a dynamic simulation model. Food Hydrocolloids 23, 1177–1183. https://doi.org/10.1016/j. foodhyd.2008.10.003. Håkansson, A., Trägårdh, C., Bergenståhl, B., 2009b. Dynamic simulation of emulsion formation in a high pressure homogenizer. Chem. Eng. Sci. 64, 2915–2925. https:// doi.org/10.1016/j.ces.2009.03.034. Innings, F., Trägårdh, C., 2007. Analysis of the flow field in a high-pressure homogenizer. Exp. Therm. Fluid Sci. 32, 345–354. https://doi.org/10.1016/j.expthermflusci.2007. 04.007. Kelemen, K., Gepperth, S., Koch, R., Bauer, H.J., Schuchmann, H.P., 2015. On the visualization of droplet deformation and breakup during high-pressure homogenization. Microfluid. Nanofluidics 19, 1139–1158. https://doi.org/10.1007/s10404-0151631-z. Kleinig, A.R., Middelberg, A.P.J., 1996. The correlation of cell disruption with homogenizer valve pressure gradient determined by computational fluid dynamics. Chem. Eng. Sci. 51, 5103–5110. https://doi.org/10.1016/s0009-2509(96)00354-5. Kokkinosa, A., Fasseas, C., Eliopoulos, E., Kalantzopoulos, G., 1998. Cell size of various lactic acid bacteria as determined by scanning electron microscope and image analysis. Lait 78, 491–500. https://doi.org/10.1051/lait:1998546. Lee, L., Hancocks, R., Noble, I., Norton, I.T., 2014. Production of water-in-oil nanoemulsions using high pressure homogenisation: a study on droplet break-up. J. Food Eng. 131, 33–37. https://doi.org/10.1016/j.jfoodeng.2014.01.024. Lee, L., Niknafs, N., Hancocks, R., Norton, I., 2013. Emulsification: mechanistic understanding. Trends Food Sci. Technol. https://doi.org/10.1016/j.tifs.2012.08.006. Lee, L., Norton, I.T., 2013. Comparing droplet breakup for a high-pressure valve homogeniser and a Microfluidizer for the potential production of food-grade nanoemulsions. J. Food Eng. 114, 158–163. https://doi.org/10.1016/j.jfoodeng.2012.08.009. Mahfoudhi, N., Sessa, M., Chouaibi, M., Ferrari, G., Donsì, F., Hamdi, S., 2014. Assessment of emulsifying ability of almond gum in comparison with gum Arabic using response surface methodology. Food Hydrocolloids 37. https://doi.org/10. 1016/j.foodhyd.2013.10.009. Maresca, P., Donsì, F., Ferrari, G., 2011. Application of a multi-pass high-pressure homogenization treatment for the pasteurization of fruit juices. J. Food Eng. 104, 364–372. https://doi.org/10.1016/j.jfoodeng.2010.12.030. Mohan, M.S., Ye, R., Harte, F., 2016. Initial study on high pressure jet processing using a modified waterjet on physicochemical and rennet coagulation properties of pasteurized skim milk. Int. Dairy J. 55, 52–58. https://doi.org/10.1016/j.idairyj.2015. 12.004. Nomura, T., Narahara, H., Tokumoto, H., Konishi, Y., 2009. The role of microbial surface properties and extracellular polymer in Lactococcus Lactis aggregation. Adv. Powder Technol. 20, 537–541. https://doi.org/10.1016/j.apt.2009.07.003. Peleg, M., Cole, M.B., 1998. Reinterpretation of microbial survival curves. Crit. Rev. Food Sci. Nutr. 38, 353–380. https://doi.org/10.1080/10408699891274246.

5. Conclusions The fragmentation phenomena in a high pressure homogenizer depend on the intensity of the fluid dynamic stresses developing in the homogenization valve and in its discharge section. In this study, we have investigated the inactivation of Lactococcus lactis as a model breakup indicator, using interchangeable-orifice valves of different sizes (80 μm, 100 μm and 150 μm) within the same experimental setup, in order to remove the variability factor of the fragmentation phenomena unconnected to the valve itself (i.e. pressure intensifier, rate of heat dissipation). By varying the operating pressures and the fluid viscosities, the extent of inactivation, and hence of cell break-up, has been compared to the fluid-dynamics developing in the valve and in the discharge region, through the main dimensionless numbers Re, We, Ca and K. The results show that the most intense break-up phenomena occur for the smallest orifice size (80 μm), highest pressure (200 MPa) and lowest viscosity (buffered peptone water). However, the variation of inactivation at varying the main operating parameters suggests that at low viscosity the effect of the turbulence is dominating, together with the elongational stresses. In contrast, at higher viscosity values, the shear stresses become increasingly important. The effect of cavitation is, instead, unaffected by viscosity, and depends on the velocities reached in the homogenization valve. Further studies are needed to clarify the effect of the changing cell resistance with temperature, as it appears that higher membrane fluidity at higher temperatures, or increasing membrane rigidity at lower temperatures, might favor the cell break-up. Appendix A. Supplementary data Supplementary data related to this article can be found at http://dx. doi.org/10.1016/j.jfoodeng.2018.05.011. References Becker, P.J., Puel, F., Dubbelboer, A., Janssen, J., Sheibat-Othman, N., 2014. Coupled population balance-CFD simulation of droplet breakup in a high pressure homogenizer. Comput. Chem. Eng. 68, 140–150. https://doi.org/10.1016/j. compchemeng.2014.05.014. Briñez, W.J., Roig-Sagués, A.X., Hernández-Herrero, M.M., Guamis-López, B., 2006a. Inactivation of two strains of Escherichia coli inoculated into whole and skim milk by ultrahigh-pressure homogenisation. Lait 86, 241–249. https://doi.org/10.1051/ lait:2006006. Briñez, W.J., Roig-Sagués, A.X., Hernández-Herrero, M.M., Guamis-López, B., 2006b. Inactivation of two strains of Escherichia coli inoculated into whole and skim milk by ultrahigh-pressure homogenisation. Lait 86, 241–249. https://doi.org/10.1051/ lait:2006006. Briñez, W.J., Roig-Sagués, A.X., Hernández Herrero, M.M., Guamis, B., 2006c. Inactivation of Listeria innocua in milk and orange juice by ultrahigh-pressure homogenization. J. Food Protect. 69, 86–92. Briñez, W.J., Roig-Sagués, A.X., Hernández Herrero, M.M., Guamis López, B., 2006d. Inactivation by ultrahigh-pressure homogenization of Escherichia coli strains inoculated into orange juice. J. Food Protect. 69, 984–989. https://doi.org/10.4315/ 0362-028X-69.5.984. Cavender, G.A., Kerr, W.L., 2011. Inactivation of vegetative cells by continuous highpressure processing: new insights on the contribution of thermal effects and release device. J. Food Sci. 76. https://doi.org/10.1111/j.1750-3841.2011.02325.x. Cortés-Muñoz, M., Chevalier-Lucia, D., Dumay, E., 2009. Characteristics of submicron emulsions prepared by ultra-high pressure homogenisation: effect of chilled or frozen storage. Food Hydrocolloids 23, 640–654. https://doi.org/10.1016/j.foodhyd.2008. 07.023. De Lamo-Castellví, S., Toledo, R., Frank, J.F., 2013. Observation of injured E. coli population resulting from the application of high-pressure throttling treatments. J. Food Sci. 78. https://doi.org/10.1111/1750-3841.12074. Diels, A.M.J., Callewaert, L., Wuytack, E.Y., Masschalck, B., Michiels, C.W., 2005a. Inactivation of Escherichia coli by high-pressure homogenisation is influenced by fluid viscosity but not by water activity and product composition. Int. J. Food

70

Journal of Food Engineering 236 (2018) 60–71

N. Coccaro et al.

Walstra, P., 1993. Principles of emulsion formation. Chem. Eng. Sci. 48, 333–349. https://doi.org/10.1016/0009-2509(93)80021-H. Wuytack, E.Y., Diels, A.M.J., Michiels, C.W., 2002. Bacterial inactivation by high-pressure homogenisation and high hydrostatic pressure. Int. J. Food Microbiol. 77, 205–212. https://doi.org/10.1016/S0168-1605(02)00054-5. Yeung, T.W., Arroyo-Maya, I.J., McClements, D.J., Sela, D.A., 2016. Microencapsulation of probiotics in hydrogel particles: enhancing Lactococcus lactis subsp. cremoris LM0230 viability using calcium alginate beads. Food Funct 7, 1797–1804. https:// doi.org/10.1039/C5FO00801H.

Shirgaonkar, I.Z., Lothe, R.R., Pandit, A.B., 1998. Comments on the mechanism of microbial cell disruption in high-pressure and high-speed devices. Biotechnol. Prog. 14, 657–660. https://doi.org/10.1021/bp980052g. Singh, S.K., Vaidya, Y., Gulati, M., Bhattacharya, S., Garg, V., Pandey, N.K., 2016. Nanosuspension: principles, perspectives and practices. Curr. Drug Deliv. 13. https://doi.org/10.2174/1567201813666160101120452. Stang, M., Schuchmann, H., Schubert, H., 2001. Emulsification in high-pressure homogenizers. Eng. Life Sci. 1 (151). https://doi.org/10.1002/1618-2863(200110)1 4 < 151::AID-ELSC151 > 3.0.CO;2-D.

71