Abstract
This study investigated the liquid leakage mechanism through microchannels in a flexible single-use packaging system composed of multilayer plastic film. Based on this study, a relationship between the maximum allowable leakage limit (MALL) and the loss of package integrity can be established under different use-case conditions. The MALL is defined as the greatest leak size that does not pose any risk to the product. A specifically designed liquid leak test was used to determine the leakage time, i.e., the time it takes for a package to show leakage. As a result, this method was able to determine the leak size for which no liquid leakage is observed after 30 days. This leak size varied between 2 µm and 10 µm and can be considered the MALL for liquid egress under different use-case conditions. This article also compared the MALL results of this liquid leak test with those of the microbial ingress test, showing a direct correlation between both tests. As test samples, an ethylene vinyl acetate multilayer film (300 µm thick) and a polyethylene multilayer film (400 µm thick) were cut into 50 mm patches. Before the patches were assembled in a filter holder to form a leak-tight seal, artificial leaks in sizes of 2 –25 µm were laser drilled into the center of each patch. The test units were filled aseptically with culture media and mounted vertically on the test setup. Various pressures were applied to each test unit to simulate the constraints that single-use systems may be subject to under real-world conditions. To detect the exact leakage time, electric circuits with timers were attached below each film patch. Microscopic investigations, including light microscopy and computed tomography, were used to interpret and understand the physics and geometries of the microchannels to explain any deviation from the expected results.
- Single-use system (SUS)
- Single-use system integrity (SUSI)
- Liquid leak testing
- Maximum allowable leakage limit (MALL)
- Microbial ingress testing
- Integrity testing
Introduction
Effective quality assurance in the pharmaceutical industry is essential to demonstrate the integrity of products throughout their life cycle. Because it is known that containers in general are not leak-free, the concept of integrity is defined by the parameter called maximum allowable leakage limit (MALL) for each product type and under certain application conditions. According to USP < 1207> (1), MALL is the greatest leak size tolerable that poses no risk to product safety and no or inconsequential impact on product quality.
Based on its definition, MALL will ensure the sterility of the system, preserve the product contents, and prevent any unwanted entry and exit of detrimental gases or other substances like liquid or microorganisms; therefore, the relevant physicochemical and microbiological specifications of the product will not change throughout the life cycle.
There are many different types of integrity test methods, which can be used to mitigate the risks of leaks, confirm the integrity, and determine the MALL. To choose the appropriate integrity test, it is important to know whether the leakage of a hazardous gas needs to be avoided, microbial ingress must not happen, or whether leakage of expensive material should be prevented. The current and previous studies conducted by the authors focused on microbial ingress and liquid leakage, whereas the passage of gases to and from the system was not mentioned.
Integrity tests of containers are generally classified into deterministic and nondeterministic tests. Although nondestructive, deterministic physical tests with sensitive techniques are capable of identifying the presence of leaks; a nondeterministic microbial integrity test is needed to determine the MALL for package configurations used to preserve sterility. It is therefore important to correlate microbial tests with physical tests, directly or indirectly, and to evaluate the capability of each method to detect a given leak.
In the category of rigid container-closure systems, Kirsch et al. (2) measured a MALL of <6 × 10-6 mbar L/s by helium mass spectrometry in the vacuum mode and correlated it with microbial ingress. This leakage rate corresponds to an orifice of a nominal diameter between 0.1 µm and 0.3 µm and to the probability of microbial ingress of <10%. For container systems made of flexible material, USP < 1207> (1) states that the risk of microbial ingress or liquid passage through leak paths has not yet been fully understood.
However, to the authors’ knowledge, the studies done by Moghimi and colleagues (3, 4) on flexible pouches can be a good starting point for investigating the integrity of flexible systems. In the first part of their study, a microchannel with a 15 µm diameter was predicted to be the MALL for the bacterial aerosol penetration test. In the second part of their study, published separately, they used the dye penetration test method to determine the integrity of flexible pouches and the MALL. As a result, they were able to ensure that the MALL for their flexible package was 15 µm, the same value obtained in their bacterial aerosol challenge test.
In many cases, there is no appropriate test method to determine and measure liquid leakage, and the existing techniques are not sensitive enough, so a gas is used as a test medium instead of the liquid itself (5). Despite this, there are reasons that liquid leak test methods are required and important to determine the liquid leakage in the quality control of certain systems.
One reason is the need to detect the leak size in the systems filled with liquid; another is that it is not always easy to convert the gas leakage rate to a liquid leakage rate.
Because of the importance of liquid flow in microtubes and microchannels, extensive research has been conducted to investigate the different aspects of this topic (6⇓⇓⇓⇓⇓–12). In 1998, Mala and Li (7) studied the flow characteristics of water in microtubes of fused silica and stainless steel and with diameters of 50 µm to 254 µm. The calculated pressure drop of a liquid flowing through a long cylindrical pipe indicated a significant deviation from their experimental measurements. They proposed two reasons for this deviation: either the flow was not laminar or the pressure difference was because of the surface roughness effects on microtube flow. To interpret the experimental data, they used a roughness–viscosity model.
In 2005, Bahrami et al. (9) developed an analytical model to predict the increase in the pressure drop of laminar incompressible flow because of wall roughness in circular microtubes. The results of their model were in good agreement with the experimental data collected from other studies (7, 13⇓⇓–16).
The microtubes used in those studies were made of silicon or stainless steel and had a diameter ranging from 8 µm to 254 µm.
The other factor that needs to be considered is slip length. Usually, the no-slip boundary condition assumed in solving the Navier–Stokes equation is valid for macroscopic scales. However, because of the high confinement of the liquid at microscopic scales, there are several studies suggesting the possibility of slip at the wall for a simple liquid on a surface (17⇓⇓⇓⇓⇓⇓⇓⇓⇓–27). Among these studies, Tretheway and Meinhart (19) proved that slip happens only for the microchannels with hydrophobic properties. Choi et al. (20) also examined the slip effects of water flow in hydrophobic and hydrophilic microchannels of a depth of 1 µm and 2 µm. The value of the microchannel slip length for the flow of water over hydrophobic surfaces was approximately 30 nm. Choi et al.’s study also proved that the flow rate in hydrophobic microchannels was higher than that in hydrophilic ones, and that the effect of viscosity on the very low rate flow was negligible.
Grine and Bouzid (24) used an analytical-computational methodology in their study to predict the liquid micro- and nanoflows through nanoporous gaskets. To calculate the theoretical liquid leak size, they introduced the wall slip boundary conditions into the conventional Hagen–Poiseuille flow equation. However, as they did not consider the surface tension present at very low porosity and interfacial leaks occurring at low stress, the theoretical liquid leak was 40%–70% higher than their experimental leak measurements.
To investigate the effect of liquid surface tension on liquid flow through microtubes, one can refer to the study of Keller et al. (28). They measured the effects of leak size and liquid surface tension on the pressure required to initiate liquid flow. To match the theoretical and experimental results, they needed to add a correction factor to the equation used to calculate the initiation of liquid flow. This factor was required because of the hydrophilic nature of the nickel microtubes employed in their experiments to simulate leaks in the glass vials used in their test setup.
Although this approach seems to be valid for other materials, further factors may need to be considered for single-use systems (SUSs).
Because it is important to understand the liquid leakage mechanism through microchannels that are formed as leaks in a single-use plastic system, we used Keller’s approach as a basis and modified it according to the material of the system tested in the present study.
In the former recently published study (29), the authors presented the outcome of the microbial integrity test they performed to detect the MALL of SUSs under certain application conditions.
The objective of the current investigation was to understand the liquid leakage mechanisms of flexible materials employed in SUSs and to determine the MALL for these systems using the liquid leak method. Knowing that the presence of liquid leakage in the leak passageway is necessary in order for microbial ingress to occur, we studied the liquid leakage mechanism through the microchannel leaks in the present investigation
Finally, the results were compared with those we obtained in the microbial ingress test to draw conclusions.
Materials and Methods
In the following section, the material used for the study together with the test assembly, test procedure and test design and plan are explained in details.
Test Assembly
The test samples used in the current study, as explained in our previously published paper (29), were an ethylene vinyl acetate (EVA) multilayer film (300 μm and 360 μm thick) and a polyethylene (PE) multilayer film (400 μm thick), the materials used in Flexboy, Celsius, and Flexsafe bags (Sartorius). Each sample was cut into 50 mm diameter patches. To simulate a pinhole leak, the patches were laser drilled in the center with different microhole sizes ranging from a nominal size of 2 μm to 25 µm. However, later on we showed that with this method, one can produce mostly conical microchannels instead of pinholes. In our study, the nominal size represented the size of the outlet of such cones. The leak size of laser–drilled holes was calibrated by Lenox Laser using flow measurement (1). Laser drilling was done by Lenox Laser.
A glass and polypropylene filter holder (Sartorius) held the film patch and was connected to the test setup by a silicone tube (ID: 1/4; OD: 1/8), which was sealed with parafilm to avoid any leakage (Figure 1A). To detect the exact leakage time, an electric circuit with a timer was attached below the film patch. As soon as liquid leakage occurred, the circuit closed and the timer began to record the respective leakage time.
(A) Multiunit setup for detection of liquid leakage. (B) Individual test module magnified to show patch placement with a zoomed image of a microchannel viewed under a microscope.
Test Procedure
The test setup we used consisted of six filter holders, each with a defined leak size, all attached to a stand (Figure 1B). Each run included a negative control, which was a nondefective film patch, to evaluate the integrity of the test setup.
The test filter holders, including the one with the negative control, were each filled with a model solution up to a graduated mark of 10 cm and pressurized to the desired test pressure at 25°C ± 2.5°C; the humidity was maintained between 40% relative humidity (Rh) and 80% Rh.
Two different model solutions were used for this experiment. Because of the importance of water for injection (WFI) in parenteral drugs, water was given first priority as the primary model solution. The ultrapure water employed in this test was produced by the arium Pro UF lab water system (Sartorius). Methylene blue dye (0.5% v/v) was added as a color indicator to the ultrapure water. The second model solution was tryptic soy broth (TSB), which is generally used in microbiology as a culture broth to grow aerobic bacteria. The TSB was utilized as purchased from the company (30 g/L, Merck).
The test pressures were chosen based on the different pressures to which each SUS may be exposed during its life cycle. To keep the pressure stable, a pressure controller was used to maintain the test pressure at the desired value during the experiment.
Liquid leakage was detected when the electric circuit for a test unit closed. At this point of detection, the connected timer recorded the leakage time. After liquid leakage was reported for a certain leak size, the test was run for the next smaller leak size available.
If no liquid leakage was detected after a maximum test duration of 30 days, the leak size was reported as the MALL for the respective combination of film material, applied pressure, and model solution.
Test Design and Plan
In order to design the number of experiments needed to be performed, all variables that could affect liquid leakage were considered. These variables included the film type and/or thickness, pressure, leak size, and surface tension of the liquid.
At the beginning, each film patch was tested three times at different pressures for a certain leak size. If leakage occurred, one smaller leak size was tested. This continued until no leakage occurred for the leak size under test. To investigate the effect of surface tension on liquid leakage, the entire procedures were repeated for the second liquid.
A total number of 54 film patches were tested within 18 runs. At this stage, multivariable data analysis indicated which parameters had minor and/or major influence.
For the rest of the experiments, design of experiments (DoE) was used to reduce the number of tests that needed to be performed and ensure that every possible combination of parameters was considered. Experiments were designed based on the interaction with the factors selected, that is, test pressure, model solution, leak size, thickness of the film along with the leakage time as the response. A maximum number of 2048 runs were obtained from DoE. This number of runs decreased to 20 in which three were replicate runs and 17 were unique runs using the D-optimal model (Table I).
Number of Trials Planned Empirically and with DoE per Pressure for PE and EVA Filmsa
Although a low number of tests can be always questioned, liquid leakage through a microchannel under various pressure differences is a physical phenomenon, not a probabilistic one. In the current study, the experiments were done to confirm the hypothesis that theoretical studies have already proposed. Therefore, repetition of the experiment for a certain leak size will not increase the level of accuracy.
Theory
As the SUS under study was made of several layers with a total thickness of about 300 µm to 400 µm, any leak in the film material was not a simple pinhole but a microchannel; at best, this leak will have a cylindrical or conical shape. Therefore, understanding the dynamics of liquid leakage is important.
Once a microchannel leak occurs inside the film wall, a threshold pressure is required to initiate flow of the liquid into the microchannel of a certain diameter. This threshold pressure must be high enough to overcome the Laplace pressure, a pressure difference caused by surface tension of a liquid droplet formed at the liquid–air interface. Therefore, the threshold pressure needs to be determined in the first step of this liquid leakage study. Equation 1 shows the relation between the threshold pressure and the liquid surface tension:
(1)
where σ is the surface tension of the liquid and r is the radius of the droplet.1 The variable
ρgL shows the hydrostatic pressure of the column of liquid, where 
is the density of the liquid, g the acceleration due to gravity, and L the height of the liquid above the film or above the leak. P0 is given as an absolute pressure. As a result, the atmospheric pressure Pa must be subtracted as it both surrounds the system under study and is present inside this system.
In the previous studies done by Keller et al. (28) and Gibney (30), they needed to consider an empirical factor,
, to compare the theoretical and experimental results. This factor differs depending on whether the system under study is hydrophobic or hydrophilic.
Once eq 1 is found to hold true for the certain leak size, this is a specific indication that liquid flows into the microchannel. The volumetric flow rate can be calculated using the Hagen–Poiseuille equation:
(2)
This equation relates the effect of the pressure drop
) of the incompressible liquid and Newtonian fluid flowing through a long cylindrical pipe, the radius (R), and the length of the pipe (l), as well as the viscosity of the fluid (μ), to the volumetric rate of flow of liquids.
The second step in the liquid leakage study was to determine the leakage time, which is, by definition, the time required for the first drop to form and exit from the microchannel. There are already several studies that explain how to calculate the drop size and drop formation time for different capillary systems. The very first drop size measurements can be traced back to the studies conducted by Tate (31), Rayleigh (32), and Harkins and Brown (33), who investigated drop sizes based on the mass of falling droplets under quasi-static conditions.
However, this method does not apply to the situation in which it takes less than 1 s for a droplet to form (34). There are some other approaches that more accurately calculate the shape of a droplet and the time needed to form this droplet, yet these methods are very complex.
The new approach introduced by Hummel et al. (35) estimates the properties of falling droplets using a simple analytical model. They calculated the volume of falling droplets from a cylindrical glass capillary based on force balance and velocity. However, as our system was not a capillary per se with a defined outer radius and also had the flexible properties of a polymer instead of being rigid, it was not fully possible to use the formula introduced by Hummel et al.
Therefore, an estimation based on the experiment performed for large microchannels was used to measure the volume of one drop formed at the end of a small microchannel ranging in size from 2 µm to 30 µm; for more information, see the Appendix.
Results and Discussion
A total of 120 film patches were evaluated in different experimental runs of liquid leak testing.
Table II lists the results of a liquid leak test performed with PE and EVA film patches that had a leak size between 2 µm and 25 µm and were subjected to a pressure of 70 mbar, 150 mbar, and 300 mbar. Water and TSB were each used as model solutions.
Number of Tested Samples Showing Liquid Leakage at Certain Leak Sizes Compared with the Total Number of Tested Samples for PE and EVA Films at Different Pressures and for Water and TSB as Model Solutions; Minimum Leakage Time for Each Leak Sizea
As seen in Table II, for each pressure applied, there was a leak size below which no signs of liquid leakage were detected.
Therefore, to check the consistency of these results, we proceeded to compare them with the appropriate theory in the following section. Yet besides using theoretical assumptions, it was essential for us to examine how microscale geometry can affect the outcome of liquid leak tests to interpret the results correctly.
Therefore, the results are described in two sections: the first section discusses the consistency of the results with the established formula, whereas the second section focuses on microscopic inspection of the microchannels to explain any inconsistencies in the data.
I. Threshold Pressure and Filling the Microchannels with Liquids
Figure 2 shows the threshold pressure required to initiate flow of two different liquids, water and TSB, through a cylindrical microchannel of various diameters. However, the real shape of the laser-drilled microchannels was conical instead of cylindrical, and the inlet of these cones had a larger diameter than that of the outlet. Because the difference between the surface tension of water and TSB is insignificant, the threshold pressure required to initiate flow did not differ considerably between the two liquids either.
The theoretical threshold pressure required to initiate the flow of water (gray dashed line) and TSB (solid black line) through cylindrical microchannels of various diameters. The experimental data points for PE film (circles) and for EVA film (triangles) show the leak size at which liquid leakage begins for pressures of 70, 150, and 300 mbar. TSB, tryptic soy broth; PE, polyethylene; EVA, ethylene vinyl acetate.
At test pressures of 70, 150, and 300 mbar, the data points showed the minimum sizes of the leaks at which liquid leakage began. In most cases, the experimental and theoretical results showed a good data fit. However, an outlier was obtained for 70 mbar, which will be explained in the next section. Once the pressure was high enough to overcome the surface tension, the microchannel was filled with liquid. However, because of the smaller diameter that the channel may have at the outlet, the pressure needed to overcome the surface tension may not be sufficient and liquid leakage might not occur. There were also some other factors related to the geometry and material of the channel that could cause delayed formation of the first drop of liquid.
The experimental leakage times confirmed that for small leak sizes, it takes longer for leakage to occur, whereas for large leak sizes, it takes less time for this to happen. Because the experiments ran for 30 days, this time in minutes is considered the leakage time for the leak sizes that did not show any leakage at all.
Figure 3 depicts the volumetric flow rate calculated from the experimental liquid leakage time for different leak sizes. The experimental flow rates matched the theoretical flow rates derived from the Hagen–Poiseuille equation (solid line). A correction factor of 0.39 was used, which was based on the hydrophobic nature of the film material and was the same factor employed by Keller et al. (28) and Gibney (30). The negative values for the flow rate at lower pressures showed that the effect of surface tension was greater, and we should not expect that liquid flow will occur.
Experimental volumetric flow rates as a function of leak size are shown at applied pressures of 300 mbar (gray circles), 150 mbar (white triangles), and 70 mbar (black triangles). The theoretical flow rates derived from the Hagen–Poiseuille equation are shown with dashed, solid, and point-dashed lines for 300 mbar, 150 mbar, and 70 mbar, respectively.
For a leak size <5 µm, apparently other factors like capillary effects or surface tension were more important; therefore, the Hagen–Poiseuille equation did not yield a valid flow rate. The experimental leakage times for such a small leak size also resulted in a very low flow or no flow.
II. Microscopic Inspection of Leaks
As mentioned previously, laser drilling was used to introduce an artificial leak into the film patches.
This method is best known to make microscale leaks in a controlled manner. Each leak produced was calibrated based on the flow rate in the microchannel at an applied pressure of 1000 mbar. The flow rate was calibrated at 15 psi and the accuracy varied depending on the nominal leak size.
Laser drilling is expected to result in clean-cut microchannels without any tears or flaps at the outlet (36); however, microscopic inspection reveals that the shapes of the target laser-drilled holes were not always round and that the microchannels formed through the film thickness were not cylindrical but rather more conical.
Because the size of each leak was calculated based on the flow rate, it was possible to obtain the same flow for two different geometries and different leak areas (Figure 4).
Light microscopy images of polyethylene (PE) film patches with nominal leak size of 2 µm and measured area of (A) 28.1 µm2 and (B) 19.7 µm2.
Figure 4 shows light microscopy examination of the laser-drilled hole for a nominal size of 2 µm. For both Figure 4A and B, the flow rate at 15 psi was 0.09 SCCM, but the differences in shape were considerable.
Figure 5 shows the microscopic images of both sides of one leak with a nominal size of 2 µm.
Light microscopy images of polyethylene (PE) film patches with a nominal leak size of 2 µm; there is a considerable difference between the inlet area (A) and the outlet area (B).
As is obvious in the images, the microchannel looks more conical than cylindrical; image processing showed that the ratio between the inner and the outer diameter was around 12.
The ratio between the inner and outer diameter formed during laser drilling was high for very small leak sizes and decreased as the leak size increased.
Because this study mainly focused on leak sizes <10 µm, the effect of noncylindrical microchannels needs to be considered.
Two main differences between cylindrical and conical microchannels were considered: first, in a conical microchannel, the funnel effect may cause a higher flow rate than in a cylindrical channel with the same average diameter (37). Second, the pressure applied may be sufficient to overcome the threshold pressure so that liquid can enter the microchannel. However, because of the microchannel’s small outlet size, this liquid may not be able to exit.
On the other hand, because of instabilities in the drilling process or local deviations in the properties of the film material, laser drilling does not always result in a consistently shaped microchannel. Figure 6A shows the computed tomography scan of a microchannel designed to be 5 µm at the outlet, although the diameter of the pathway actually varies: as the center is approached, the diameter becomes larger, then decreases below the center. This may lead to a lower flow rate compared with that of a similar 5 µm microchannel (Figure 6B).
(A) Computed tomography image of a leak channel in a PE film with a thickness of 400 µm and a nominal leak size of 5 µm. (B) Channel diameter as a function of film thickness.
It is also possible for laser drilling to leave residues inside the channel. The blocking point results in a lower flow rate or may even impede flow entirely, which may result in an incorrect calculation of the leak size. Figure 7A shows the computed tomography scan of a PE film with a thickness of 400 µm and a nominal outlet leak size of 5 µm. The diameter of the microchannel was larger at the top than at the bottom; however, a film residue located in the length of around 280 µm can reduce the flow rate or block the channel so that liquid flow cannot occur (Figure 7B).
(A) Computed tomography image of a leak channel in a polyethylene (PE) film with a thickness of 400 µm. (B) Channel diameter as a function of film thickness.
III. Liquid Leak Testing versus Microbial Ingress Testing
In the previous study performed by the authors (29), a total of 30 patches per leak size were tested by the microbial ingress test method. The testing units and procedure for determining the MALL were reported in this earlier paper.
The experimental data for both microbial ingress testing and liquid leak testing together with the logarithmic regression analysis of the same data are presented here.
Based on the probabilistic nature of microorganisms, uneven geometry of the microchannels, and the pressure applied, different categories are distinguishable (Figures 8 and 9).
The presence or absence of bacterial growth (circles) together with liquid leakage or no liquid leakage (triangles) are shown as a function of leak size for an applied pressure of 70 mbar. Black and gray solid line curves express the probability of bacterial ingress and liquid leakage, respectively, as a function of different leak sizes with a 95% confidence limit for microbial ingress testing and liquid leak testing expressed as black and gray dashed line curves, respectively. Different combinations of the possible behavior of the two tests are distinguished within regions (1) to (5).
The presence or absence of bacterial growth (circles) together with liquid leakage or no liquid leakage (triangles) are shown as a function of leak size for an applied pressure of 300 mbar. Black and gray solid line curves express the probability of bacterial ingress and liquid leakage, respectively, as a function of different leak sizes with a 95% confidence limit for microbial ingress testing and liquid leak testing expressed as black and gray dashed line curves, respectively. Different combinations of the possible behavior of the two tests are distinguished within regions (1) and (2).
Whether the experimental data from both microbial ingress testing and liquid leak testing in each category were consistent with one other and with the results predicted by the logarithmic regression analysis is a debate that we will address here.
Figure 8 shows the probability of microbial ingress and liquid leak as a function of defect size based on a statistical analysis of the experimental data at an applied pressure of 70 mbar for PE film.
As mentioned in the previous section, the microchannels have conical shapes; therefore, if the pressure is sufficient for liquid to enter the channel and the outlet is large enough, liquid leakage can occur, as shown in Figure 8 (1) to (3). In such a case, microbial ingress will occur with a high probability, as depicted in Figure 8 (1), or may remain in a transitional area with a different probability of occurrence, as depicted in Figure 8 (2).
Figure 8 (3) is an example of the situation in which liquid leakage occurs, but experimental data shows that a very long time is required for the first liquid drop to form and to leave the microchannel, which constitutes leakage. This leakage time is much longer than the aerosolization exposure time of film patches during microbial ingress testing (29). Therefore, it is possible that bacteria did not enter the channel while a liquid path did not yet exist within this aerosolization exposure time.
There is a transitional area between 3 µm and 5 µm in which it cannot be predicted in a clear-cut manner that liquid leakage will not occur (Figure 8 (4)). For a leak size <3 µm, no liquid leakage and no microbial ingress are shown (Figure 8 (5)).
The five categories for the pressure of 70 mbar decreased to two categories for a pressure of 300 mbar (Figure 9). In Figure 9 (1), the situation can be explained exactly as for Figure 8 (1); however, in Figure 9 (2), every combination of the occurrence and absence of liquid leakage and microbial ingress is possible. For instance, if the outlet is too small (<3 µm), no liquid leakage is expected to occur, but because the channel has a larger inlet, the pressure may be sufficient to overcome the surface tension at the inlet and liquid can fill the channel. This may cause microbial ingress as a result (Figure 9 (2)).
The main outcome of this comparison was the good agreement that existed between the MALL found for microbial ingress testing under different use-case conditions and the corresponding leak size in liquid leak testing in which no liquid leakage was reported. A key factor for microbial ingress was the presence of liquid in the pathway. If the liquid leakage time was longer than the microbial exposure time (aerosolization time), the risk of microbial contamination could be lower than expected. However, in reality, contamination by microorganisms can occur anytime.
The transitional area for microbial ingress testing was understandable based on the probabilistic nature of bacteria. However, the unexpected results below the transitional area observed for liquid leak testing could have different causes, which include microchannel size limitations, microchannel geometry imperfections, and overcoming of the effect of surface tension.
Finally, qualitative discussions that have shown good agreement between the results of microbial ingress testing and liquid leak testing were based on experimental results and the authors’ theoretical knowledge, and there is still room for further discussion about the relationship between the two test methods.
Conclusion
The liquid leakage mechanism through microchannels in flexible, single-use packaging systems was studied in the present article. The liquid leak test method was applied to PE and EVA multilayer film patches to establish a relationship between the MALL and the loss of package integrity for SUSs made of these materials and was performed under different use-case conditions. The experimental results were provided and discussed for film patches that had various leak sizes and were tested using water and TSB as model solutions at different applied pressures. The leak sizes showing liquid leakage and the corresponding leakage times were recorded. The leak sizes that did not show any liquid leakage after 30 days were designated as “no leak”. The rate of liquid flow through microchannels of different sizes was calculated theoretically based on the Hagen–Poiseuille equation. However, the experimental flow rate was calculated directly from the leakage time. To consider the hydrophobic effects of the film material, a correction factor was implemented based on the Hagen–Poiseuille equation (28). A very good fit between theoretical and experimental flow rates was reported. According to the Hagen–Poiseuille theory, the flow rates for “no leak” sizes were negligibly low or negative.
Microscopic inspection of leak sizes, including light microscopy and computed tomography, was performed to explain any deviation from the expected results. In order to find a relationship between microbial ingress testing and liquid leak testing, the experimental data from both tests together with the logarithmic regression analysis of the same data were plotted and compared. Based on the pressure applied, different areas appeared in the plots. There was a good fit between “no leak” sizes and the MALL obtained under different pressure conditions applied for microbial ingress testing. To summarize, it is understood that for a very small leak size, liquid leakage itself may not be that harmful if microbial ingress does not occur simultaneously. The correlation between microbial ingress and liquid leakage also showed that although the physics of the microchannel confirmed that no liquid leakage is expected to happen, the system may be contaminated because of microbial ingress caused by a liquid-filled microchannel.
Conflict of Interest Declaration
The authors declare that they have no competing interests.
Acknowledgments
The Authors wish to express their sincere gratitude to Dr. Tatjana Melnyk at Laser Zentrum Hannover for providing the computed tomography images of the leaks.
APPENDIX
Liquid Drop Size Measurement
For this study, a simple method was employed to measure the size of one drop of liquid formed at the end of a microchannel with a certain diameter. The test setup used was the same as for liquid leak testing, except that here, only one filter holder was required.
This filter holder was filled with liquid, placed on the balance pan and pressurized from the top using the standard test pressure. The weight of 10 drops falling into the filter holder base on the balance was recorded along with the time it took for the drops to fall. The average value of the drop weight was considered the actual drop size of liquid for that microchannel.
The experiment was performed for 25 µm and 30 µm channels. The results were used to extrapolate the drop size for smaller microchannel sizes.
Footnotes
↵1 For theoretical calculation, it is considered that the droplet formed at the exit point of the leak has its maximum diameter, which is equal to the microchannel diameter.
- © PDA, Inc. 2021
