WO2014001419A1

WO2014001419A1 - Microfiltration, method, device and use

Info

Publication number: WO2014001419A1
Application number: PCT/EP2013/063447
Authority: WO
Inventors: Jeanette Lindau; Fredrik Innings; Kristina PETERSSON
Original assignee: Tetra Laval Holdings & Finance S.A.
Priority date: 2012-06-28
Filing date: 2013-06-27
Publication date: 2014-01-03

Abstract

The present invention relates to reduction or removal of microorganisms from different kinds of milk. The milk is filtered over a membrane filter having pores with a uniform minimum pore size between 0.35 μm and 0.8 μm, preferably with an elongated shape. A method, device and use is provided.

Description

MICROFILTRATION, METHOD, DEVICE AND USE

Technical Field of the Invention

The present invention relates to reduction or removal of microorganisms from different kinds of milk. More specifically, the invention relates to a microsieve with a very specific pore size.

Background

In production of milk, such as milk with extended shelf life (ESL milk), or milk for manufacture of cheese (Cheese milk), there are different processes available, of which one is membrane filtration. This type of filtration, where a feed is allowed to pass through a filter, gives a concentrate (retentate) comprising matter which do not pass the filter; and a permeate (filtrate) comprising matter which do pass the filter. The driving force behind the membrane filtration process is a pressure difference across the membrane, a trans membrane pressure (TMP).

TMP is calculated with the formula:

P1 + P2

TMP = - P₃

2 ³

where Pi is inlet pressure feed, P₂ is outlet pressure concentrate and P3 is outlet pressure permeate.

Thus, when designing an industrial membrane filtration process, it is important to consider the pore size of the filter, but also the pressure characteristics of the filter. Currently, industrial filtration processes for reducing microorganisms commonly utilize ceramic membranes with two different nominal pore sizes, 0.8 μιη and 1.4 μιη respectively.

However, since ceramic membranes have a rugged surface, it is hard to control the exact size of the pores. Thus, currently used membranes have pores with a distributed pore size, which leads to a risk that microorganisms pass through the filter (pores too large), while simultaneously require a relatively high TMP because some of the pores are very small. Summary of the Invention

Consequently, the present invention seeks to overcome the above-identified deficiencies in the art and disadvantages singly or in any combination by providing a filter with substantially no pore size distribution, i.e. a uniform pore size.

The general solution according to the invention is to provide a microsieve with a very specific pore size and shape, which simultaneously provide sufficient removal of microorganisms and acceptable pressurizing properties.

According to a first aspect of the invention, a method for removing bacteria and bacterial spores from milk is provided. The milk is filtered over a membrane filter having pores with a uniform minimum pore size between 0.35 μιη and 0.8 μιη.

This is advantageous, since a controlled and sufficient removal is obtained.

In an embodiment, the pores have an elongated shape.

This is advantageous, since the pressure characteristics are improved.

In an embodiment, the milk is milk treated to obtain extra shelf life (ESL milk) and the minimum pore size is between 0.35 μιη and 0.45 μιη, preferably 0.45 μιη.

In an embodiment, the milk is Cheese milk and the minimum pore size is between 0.7 μιη and 0.8 μιη, preferably 0.8 μιη.

According to a second aspect of the invention, a method of producing cheese from Cheese milk is provided, wherein the Cheese milk is filtered according to the first aspect of the invention.

According to a third aspect of the invention, a device for removing bacteria and bacterial spores from milk is provided. The device comprises a pump and a membrane filter having pores with a uniform minimum pore size between 0.35 μιη and 0.8 μιη, preferably 0.45 μιη or 0.8 μιη.

This is advantageous, since a controlled and sufficient removal is obtained.

In an embodiment, the pores have an elongated shape.

This is advantageous, since the pressure characteristics are improved.

According to a fourth aspect of the invention, a use of a device according to the third aspect is provided, for removing bacteria and bacterial spores from milk.

In an embodiment, the milk is Cheese milk.

In an embodiment, the milk is milk treated to obtain extra shelf life (ESL milk). According to a fifth aspect of the invention, a filter for removing bacteria and bacterial spores from milk, having pores with a uniform minimum pore size between 0.35 μιη and 0.8 μιη is provided.

This is advantageous, since a controlled and sufficient removal of bacteria and bacterial spores from milk is enabled.

In an embodiment, the pore size is between 0.35 μιη and 0.45 μιη.

This is advantageous, because it specifically enables a controlled and sufficient removal of bacteria and bacterial spores from milk with extra shelf life (ESL milk).

In an embodiment, the pore size is between 0.7 μιη and 0.8 μιη.

This is advantageous, because it specifically enables a controlled and sufficient removal of bacteria and bacterial spores from milk for manufacture of cheese (Cheese milk).

Further advantageous features of the invention are elaborated in embodiments disclosed herein.

Brief Description of the Drawings

The invention will below be described more in detail, having reference to a preferred embodiment thereof shown on the accomplishing drawings, in which:

Fig. 1 is a temperature graph according to an embodiment of the invention; Figs 2-4 are pressure graphs according to embodiments of the invention;

Fig. 5 is a temperature graph according to an embodiment of the invention Figs 6 are pressure graphs according to embodiments of the invention;

Fig. 7 is a temperature graph according to an embodiment of the invention Figs 8-14 are pressure graphs according to embodiments of the invention; Figs 15-17 are SEM pictures according to embodiments of the invention.

Detailed Description of Preferred Embodiments

The present invention relates to reduction or removal of microorganisms from different kinds of milk. More specifically, the invention relates to a microsieve with a very specific pore size. Microsieve Production

A microsieve was produced according to the low pressure chemical vapor deposition (LP-CVD) process, well known to a person skilled in the art. Wafers were loaded in a quartz tube and the tube was evacuated to high vacuum. The tube was kept at a temperature of approximately 700-750°C. After loading, the tube is heated to a temperature of approximately 750-850°C. After heating, the tube is filled with a flow of process gasses, dichlorosilane (S1H₂CI₂) and ammonia (NH₃), at a total pressure which is below atmospheric pressure. Due to the temperature, a surface reaction at the wafer surface takes place forming S1₃N₄.

Depending on temperature and partial gas pressures a stoiciometric form of

S1₃N₄ is formed, giving the best properties for application in the micro sieve wafers, which is well known to a person skilled in the art. After processing, the tube is purged by N₂ gas and vented to atmospheric pressure.

The following material was used in the process.

Photo-resist: SPR 3612D or SPR6112D

Primer: HMDS (Hexa Methyl Di-Silazane)

Developer: AD 10

Specific Etchants: for S1₃N₄, Reactive Ion Etch : gas - mix of Ar, CF₄, CHF₃ Specific Etchants: for Si: Isotropic etchant : solution of HF, HNO₃ and H₂O

Specific Etchants: for Si: Anisotropic etchant : solution of KOH and H₂O Cleaning agents: Solution of HC1 and H20

Solution of HN03 and H20

All etching and cleaning steps are followed by rinsing in Reverse Osmosis De-Ionized water. The process gasses used are very pure, having contamination levels below ppm level (-0.0001%).

Specific Embodiment 1

According to an embodiment, retention tests were performed with Fluxxlab 10, from Fluxxion B.V., Netherlands, and a 0.45 μιη slit sieve produced according to the part Microsieve production above. Milk, which has been treated to extend shelf life was used, so called ESL milk. Before the retention tests the Fluxxlab was cleaned with a 1 vol% Divos 109 solution at 50 °C for 1 hour and sterilized with a 1 vol% Divosan Forte solution for 10 min at room temperature, prior to building in the microsieve. After the sieve was incorporated into the Fluxxlab, it was activated with a combined treatment with Divos 109 and Booster (both from Johnson Diversey) at 75 °C for 30 min and sterilized also with Divosan Forte for 10 min at room temperature. Just before the first retention test, a second activation of the sieve was executed according to above, but with an additional hydrogen peroxide rinsing step as the last step in front of the milk filtration. The milk (4 L ESL skim milk, obtained from Albert Heijn) was preheated to 50 °C for at least 30 min. After the milk was fed into the feed vessel, a start-up procedure according to the manufacturers instruction was applied and then the transmembrane pressure (TMP) settings were adjusted to gain a permeate flow of 100 ml/min.

Suspended spores of Bacillus pumilus were added after reaching the required permeate flow and re-circulated for 5 min to ensure that the spores are well mixed throughout the system. Bacillus pumilus was chosen as challenge test organism. The Bacillus pumilus test strain 020315 has been cultured according to methods well known to a person skilled in the art. The size of the spores has been measured by scanning electron microscopy at Campden & Chorleywood UK. The result is based on average of 50 measurements and the width is 0.50 μιη ±0.043 μιη and the length is 1.05 μιη ±0.065 μιη.

Then permeate samples of three 500-600 ml samples were taken. Two samples for the actual spore determination and one a control for the temperature during the following up pasteurization of the permeate samples. Also a few milliliters of the feed were taken as a sample. In total three measurements with the same 0.45 μιη slit sieve were executed. For the first run a spore load of 10⁶ cfu/ml^"1 was used and for the second and third samples, a load of 10⁷ ml^"1 based on a skim milk volume of 4 L for each run. The tests were performed at a temperature of 46 - 48 °C. After each run the system was cleaned using a 1 vol-% solution of Divos 109 re-circulated for 30 min at 50 °C. After the last flushing step the system was sterilized for 10 to 20 min with a 1 vol-% solution of Divosan Forte. This procedure was followed by the activation procedure using Divos 109 and Booster at 75 °C for at least 30 min.

Test Method

Spores of Bacillus pumilus were inoculated into approximately 3-4 liter of skim milk to get an initial load of log 6-log 7 cfu/ml. To determine the actual initial load the start (zero) sample was taken from the balance tank at the same time as the permeate samples were taken. Permeate from each trial was collected in 2 sterile bottles (marked A and B) approximately 0.5 liter in each. The 0-sample was heated to 75-80°C for 10 minutes to eliminate the vegetative cells. The permeate samples were heated to 70-75°C for 5 minutes and then cooled before the analyses were made. The evaluation was done by using 3M Petri- film and by dead-end filtration technique, well known to a person skilled in the art. The dead-end filtration technique was used to be able to evaluate larger volumes. The skim milk was diluted approximately 10 times with sterile water before it was filtered through a 0.45 μιη membranes. The membranes were placed on plate count agar (PCA) plates and incubated. 10 ml of the skim milk was filtered. Both the Petri-films and the PCA plates were incubated at 37°C for 4 to 5 days and also checked after 24 hours. Comparative Example 1

Skim milk was heated to about 51 °C and was held at this temperature until it was filled in to the feed vessel of the Fluxxlab. The time the milk was at a temperature of above 50 °C was 64 minutes, see Figure 1. After starting up the milk filtration, which took about 8 minutes, the TMP values were adjusted in order to create a permeate flow of 100 ml/min, which is the aimed target value for the new 0.45 μιη slit sieves. Figure 2 shows pressure values for different settings during the star-up phase. At an average TMP of 90 mbar (settings: TMP+ 160 mbar, TMP- 144 mbar, Dynamic Crossflow Pulse (DCP) frequency 10 Hz, pulse width 21 ms) the permeate flow just exceeded 100 ml/min. After the Bacillus pumilus spores were added the flow dropped slightly below the target value. During sampling the permeate flow dropped further and the average TMP had to be increased in order to keep the permeate flow above 90 ml/min. Figure 3 shows average TMP and permeate flow. The final settings were TMP+ 190 mbar and TMP- 171 mbar. The DCP frequency was kept constant at 10 Hz. In total 1.8 L of permeate was removed from the system leaving about 2.2 Liters of skim milk remaining in the system. After a run time of approximately 25 minutes with the spores the DCP was stopped in order to verify if the sieve is intact, which would be indicated by dropping of the permeate flow to zero. The permeate flow indeed dropped down to zero. Due to a software failure the DCP could not be started again. The whole Fluxxlab had to be reset by switching off the power supply. Afterwards the filtration was started again to check if it is possible to continue the filtration with a decent permeate flow after a power failure. As a result a permeate flow of 90 - 100 ml/min could be reached again Figure 4 shows average TMP and permeate flow after restarting the filtration. But to generate a constant permeate flow the DCP frequency had to be increased from 10 to 12 Hz. The time between the stopping and restarting was about 2 minutes. Comparative Example 2

Skim milk (4 L), inoculated with bacteria as described above, was heated to about 51 °C and was held at this temperature until it was filled in to the feed vessel of the Fluxxlab. The time the skim milk was at a temperature of above 50 °C was 78 minutes (see Figure 5). The second run with the same 0.45 μιη slit sieve was performed the day after the comparative example 1. Since the sieve was activated already on the day before after the first retention test only a short activation procedure of 10 min was used. After the start-up which was carried out as described above, the TMP settings were adjusted to values resulting in permeate flow of 100 ml/min. This time this value was not absolutely reached and the TMP settings as well as the average TMP had to be adjusted a bit higher compared to the previous run. As observed before the addition of the spores resulted in an immediate drop of the permeate flow. For this run the Bacillus pumilus spore load was 10⁷ ml^"1. During the removal of permeate the flow dropped as expected. In total 1.85 Liters of permeate were removed from the system leaving 2.15 Liters of skim milk inside. The flow decreased from about 90 ml/min to 80 ml/min while the average TMP was increased from 10⁵ mbar (TMP+ 180; TMP- 162) to 130 mbar (TMP+ 210; TMP- 199). Figure 6 shows average TMP and permeate flow. The DCP frequency was kept constant at 10 Hz.

Comparative Example 3

Skim milk (4 L) was heated to about 50 °C and was held at this temperature until it was filled in to the feed vessel of the Fluxxlab. The time the milk was at a temperature of above 50 °C was 68 minutes (see Figure 7). After the second test the system was cleaned, sterilized and activated again. The start-up was performed in the usual way. Before adding the Bacillus pumilus spores a permeate flow of above 100 ml/min could be reached but at higher TMP settings compared to the 1st and 2nd run. The average TMP had to be raised to 117 mbar (TMP+ 200; TMP- 180). The DCP frequency was 10 Hz. The load of the Bacillus pumilus spores was again 10⁷ ml^"1.

During the removal of permeate the flow dropped from 100 ml/min down to about 80 ml/min. The average TMP was increased from 120 to 130 mbar in order to keep the flow constant, but permeate kept on declining after each increase of the TMP indicating the build-up of a fouling layer. After stopping the removal of about 2.05 L permeate it was tried to get the permeate flow constant by increasing the DCP frequency from 10 to 12 Hz. The permeate flow was not becoming constant but the decline slowed down. Figure 8 shows average TMP and permeate flow.

Results

In comparison, start permeate flows of around 100 ml/min could be realized for each of the three comparative examples performed with the same sieve. But with time, it was more and more difficult to reach this value. In Table 1, the permeate flows just before the addition of the Bacillus pumilus spores are listed. It is clearly to see that higher average TMP's had to be applied in order to reach the 100 ml/min hence the permeability decreased from run to run.

Table 1. Permeate flow before addition of the Bacillus pumilus spores and the corresponding TMP settings. Pump speed (85%), DCP Frequency (10 Hz) were constant for all three runs. Comp. ex. TMP+ TMP- Aver. TMP Aver, Permeability (mbar) (mbar) (mbar) permeate (ml min ¹ flow mbar^"1) (ml/min)

1 160 144 91±1 100±3.5 1.09

2 170 153 99±1 97±1.6 0.99

3 200 180 117±3 101.3±5.4 0.86

If comparing the permeate flow during the runs it is interesting to notice that flow decline became more and more pronounced even though that it was tried to keep the flow constant by increasing the average TMP. This means that the sieve became more and more fouled from run to run either by the spores themselves, by an ingredient of the used milk or by a system dependent foulant. From the flow resistance of the sieve after activation listed in Table 2, which is determined from the water permeability during flushing the system after the treatment with Divos 109 and Booster, it can be seen that the sieve became more and more fouled indicated by an increase of the resistances.

Table 2. Permeability and flow resistance as a result of the activation procedure prior to the retention tests.

Comp. Activation Activation Permeability Flow Flow Permeability ex. time last 10 min resistance resistance last 2 min of

(min) of activation last 10 after H202

(ml min-1 min of activation rinsing (ml mbar-1) activation (m-3) min-1 mbar- (m-3) 1)

1 2 31 4.74±0.31 328±12 265±7 3.45±0.08

2 3 34 4.10±0.36 389±29 269±6

4 10.5 4.15±0.28 393±18 279±3 3.34±0.12

3 5 45 4.03±0.26 398±18 344±9 2.56±0.08 The flow resistances of water just after the first and second comparative examples are equal. A slight increase occurred after the third comparative example, see Table 3. However there seems to be a significant difference in the flow resistance of water before and after the cleaning step.

Table 3. Flow resistance during the different cleaning steps performed after the retention tests.

The addition of the spores caused a slight drop of the permeate flow. During the retention tests the flow dropped further due to the removal of permeate. In all three cases a VCF of 1.8 - 2 was obtained. From data collected during the activation and cleaning procedures in can be concluded that the sieve was increasingly fouled after each run and the cleaning and activation could not remove the foulant. It might be attributed to something from the Fluxxlab itself.

The results from Comparative examples 1-3 are shown in Table 4. The initial load of Bacillus pumilus in test 1 was log 6.66 cfu/ml and it was decided to increase the load slightly in the next two tests. The initial load in test 2 and 3 was log 7.14 cfu/ml and log 7.13 cfu/ml respectively. The logarithmic reduction has been calculated by using multiple MPN calculation (Assessment of Logarithmic Cycle Reduction by MPN with any Number of Inoculum Levels by G.Moruzzi, Tetra Pak). The confidence level has been set to 95% in the calculations. When using all the data from Table 4 in the MPN calculation there are a few results that deviate from the others. Test 1B2 seems a bit odd compared to the other result of comparative example 1. The most likely explanation to this could be handling errors as during this test there were serious problems with the dead end filtration. Due to the filtration problems it was decided to do an extra set of 3M Petri film samples. The test 1B2 3M was done after the filtration. The usual test sequence was starting with 3M followed by the filtration. Test 1B2 indicates a LCR of 7 (and lowest 6.7) this result seems to be the worst of all the result possible to calculate. Also comparative example 3 deviates from the other two tests and the LCR (log cycle reduction) indicates 8.1 (with lowest confidence limit of 7.9). A possible explanation to the increase in retention during this test could be that the fouling of the membranes has increased by time most likely due to improper cleaning. If test 1B2 and all of comparative example 3 is eliminated from the MPN calculation the LCR is 7.6 with a lower confidence level of 7.4. The conclusion from these tests is that it is possible to reach the target of 7 log reductions.

Table 4. Summary test with Bacillus pumilus Fluxxion Lab.

Date Test no Analytical Initial Sample Log Number Number method load volume load per of of

(log (ml) sample samples positive cfu/ml) samples

20080409 1A 3M 6.66 ¹ 6.66 25 5

20080409 1A2 3M 6.66 ¹ 6.66 25 1

20080409 IB 3M 6.66 ¹ 6.66 25 3

20080409 1B2 3M 6.66 ¹ 6.66 25 10

20080409 1A Filtration 6.66 10 7.66 3 3

20080409 IB Filtration 6.66 7.36 4 4

20080410 2A 3M 7.14 ¹ 7.14 25 5

20080410 2B 3M 7.14 ¹ 7.14 25 7

20080410 2A Filtration 7.14 10 8.14 6 6

20080410 2B Filtration 7.14 10 8.14 6 5

20080410 3A 3M 7.13 ¹ 7.13 25 1

20080410 3B 3M 7.13 ¹ 7.13 25 0 20080410 3A Filtration 7.13 1 8.13 6 5

20080410 3B Filtration 7.13 1 8.13 6 5

At the start it was possible to reach the target value of 100 ml/min (equal to 10 m3 h-1 m-2) in all runs. However as permeate was removed the permeate flow decreased towards 80 ml/min. It was possible to reach the 7 log reduction in the Fluxxlab 10 using a 0.45 μιη slit sieve which is the target set for the ESL milk. However the result needs to be verified in a larger system such as the pilot plant at the target capacity of 10 m³/h and a volume concentration factor (VCF) of 20.

Specific Embodiment 2

The retention tests were performed in a small scale lab unit, Fluxxlab 10, as previously described under specific embodiment 1 , equipped with membranes with different pore sizes, either in slit shape or round. The sieves were produced according to the part Microsieve production above. The pores sizes of the sieves were slits 0.33, 0.50 and 0.8 μιη in width and round pores with diameter 0.5 and 0.8 μιη. Approximately 4 liters of pasteurized skim milk was inoculated with spores of Bacillus pumilus to get an initial load of log 6-log 7 cfu/ml. Permeate was collected in sterile bottles and the bacterial analyses were made by using the 3M Petri- film technique.

The wafer with 0.5 x 2.3 μιη slits gave a log reduction of 6 log and it was also possible to reach the target capacity of 10 000 l m²h. To be able to reach the target of a reduction of 7 log the width should be slightly less than 0.5 μιη. The pore shape should be slits otherwise the target flux will not be reached. The objective with this

investigation is to find the relevant pore-size that will fulfill both the required log reduction on spores and the required capacity for the ESL milk application.

This evaluation was done only on the retention of Bacillus pumilus spores. Three different pore-sizes, 0.33 μιη, 0.50 μιη and 0.8 μιη and two different shapes, slits and round, were evaluated.

Experiments were done on the Fluxxlab 10, as previously described under specific embodiment 1. The Fluxxlab system is a small experimental filtration set-up with capacities of a few liters of permeate per hour. It uses 20*20 mm membranes, and has a Dynamic Crossflow Pulse (DCP) system to prevent fouling.

The Fluxxlab has an open tank container for the feed. A lob rotor pump, well known to a person skilled in the art, (to prevent too much shear forces in the pump) creates the cross flow of approximately 2 m/sec and also creates the pressures (TMP) necessary for the filtration. The TMP is controlled by a valve on the permeate side.

Pasteurized 0% fat ESL milk from the supermarket (Albert Heijn) is used for the filtration. A heating system maintains the temperature of the milk between 40 - 45°C. The retentate flow and permeate flow are circulated into the feed tank. The pressure of the feed (PI) is set between 500 - 600 mbar. The pressure difference over the module (P1-P2) creates the cross flow and is set at 200 mbar. The back pulse frequency (DCP) was set at 15 Hertz. The TMP and the back pulse depth are set during filtration to obtain the flux. Prior to filter milk, the micro sieve needs to be activated. For all the sieves used, this was an oxygen plasma. Prior to filtering, the installation was cleaned with Divos 123 (1%) at 60°C for 15 min., flushed with RO-water, disinfected with Divosan Forte (0.25%) at room temperature for 30 min. and again flushed with RO-water.

All the filtrations are started with RO-water (preheated to 40°C) and the filtration parameters are slowly increased to prevent air bubbles in the cross flow and the filter. Once running with water, slowly milk is added. When all the milk is added and the filtration is running stable, the spores are added. After a while samples are taken to determine the retention of the filter with respect to the spore Bacillus pumilus.

For the first three experiments (comparative example 4-6) the milk was heated in the cardboard packing placed in a water bath. These experiments showed a considerably lower permeate flux (data not shown). For the last three experiments

(comparative example 7-9) it was decided to heat up the milk in a glass bottles placed in a water bath and fluxes were back to the normal levels. We assume that components from the plastic which is inside the package are dissolved in the milk at the elevated temperature and contaminate the sieve surface. From the 0.5 μιη slits microfilter in combination with the spore Bacillus pumilus, also a SEM picture was taken as shown in Figure 15. Test Organism

In order to determine the retention of bacterial spores in a 0.45 μιη membrane spores of Bacillus pumilus were chosen as challenge test organism. The Bacillus pumilus test strain, labeled 020315, has been cultured according to methods well known to a person skilled in the art. The size of the spores has been measured by scanning electron microscopy at Campden & Chorleywood UK. The result is based on average of 50 measurements and the width is 0.50 μιη ± 0.043 μιη and the length is 1.05 μιη ± 0.065 μιη.

Spores of Bacillus pumilus were inoculated into approximately 3-4 liter of ESL skim milk to get an initial load of log 6-log 7cfu/ml. Samples of the feed with added spores and of permeate were taken. The samples were heated to 77°C for 10 min in a water bath. After this 1 ml sample was put on 3M Petri-films. If necessary the samples were first diluted in sterile water. All the feed samples were diluted. The 3M Petri-films were incubated at a temperature of 32-40°C. It was very difficult to keep the

temperature in the incubator at 37°C.

Comparative Example 4

An experimental setup according to Specific embodiment 2 was used, with a 0.33 μιη slit wafer. The width of the slits is 0.33 μιη and the length is 1.3 μιη. The sieves were produced according to the part Microsieve production above.

The first trial did not go very well. When milk was added the flux declined too much. The system was cleaned and the test started over again. At start-up there is 600 ml of water in the tank. A sample of permeate with water was taken. The ESL skim milk, bought in the store, had been heated up in the package in a water bath, to 45 °C. The milk is slowly added to the water. The second time, the result was better but it was decided to not add more than 1500 ml of skim milk. The spore suspension, 10 ml, was added to this mixture of water and milk. After 20 minutes samples of feed and permeate were taken. The sample was taken at a flow of 9 ml/min, which corresponds to a theoretical flux of 945 l/m²h in a large scale. The flux based on the 4 cm² in the Fluxxlab was 1350 l/m²h. The water used to dilute the feed was boiled in a glass beaker. Samples of this water were put on 3M Petri- film, 2x1 ml. The results showed that it was not sterile. It was not possible to count all the colonies. Because of this it was not possible to determine the amount of spores in the feed.

The results were as follows:

Number of spores found in the permeate samples: 0, 0, 1, 0, 0, 0, 1, 0, 0, 0. Total 2 spores/ 10 ml.

Number of spores found in the water sample: 2, 2, 0, 1, 0.

Log reduction: since it is not possible to determine the correct amount of spores in the feed it is not possible to calculate a log reduction. We can estimate the amount of spores in the feed to be 1 * 10⁶/ ml, which will give a log reduction of 6,8. The two colonies in permeate might well be re-contaminations which means that we could have a log reduction of greater than 7.

Figure 9 shows different values of TMP and permeate flow.

Comparative Example 5

An experimental setup according to Specific embodiment 2 was used. Test with 0.5 μιη round pores. The sieves were produced according to the part Microsieve production above.

10 ml of the spore suspension was added when a total of 3000 ml of ESL skim milk had been added to the 600 ml of water.

Samples of permeate were taken at a flow of approximately 15 ml/min which theoretically equals to a flux of about 1700 1/m² h (15x 114) on a large scale.

Calculated on the surface area of 4 cm² it gives a flux of 2250 l/m²h. Two permeate samples were taken. Sample of the feed were taken at the same time as the permeate. Distilled water was brought from a neighbor company, according to standards well known to a person skilled in the art. The test tubes for dilution of the feed samples, with 9 ml of this water, were also heated to 78°C for 30 minutes, as an extra security.

The results were as follows:

Number of spores found in feed: -4 125, 113

-5 12, 12

-6 0, 1

Number of spores found in permeate: 1,1,0,0,0,0,1,0,1,1. Total 5 cfu/10 ml. The log load in the feed was 6.1.

Logarithmic reduction: 6.4.

Number of spores found in the water sample: 0,0,0,0.

Figure 10 shows different values of TMP and permeate flow. Comparative Example 6

An experimental setup according to Specific embodiment 2 was used, but the ESL milk was heated in glass bottles. Test with 0.5 μιη slits (0.5 μιη x 2.3 μιη). The sieves were produced according to the part Microsieve production above.

The mean pore size was 0.475 μιη, varying from 0.440 μιη to 0.490 μιη. 10 ml of the spore suspension was added when a total of 3000 ml of skim milk had been added to the 600 ml of water.

Samples of permeate were taken at a flow of 30 ml/min. This flow would equal a flux of 8 460 l/m²h in a large scale. A conversion factor is used based on the pure porosity of the membrane. This membrane was not meant to be 0.5μιη but a 0.8μιη but the layer of silica nitride became too thick. So a correct membrane would have a higher porosity, which would give a higher flux. The flux in the Fluxxlab unit during the test, based on the area 4 cm², was around 4500 l/m²h.

The results were as follows:

Number of spores found in feed:

-4 115, 105

-5 12, 4

-6 1, 2

Log load in the feed was 5.9.

Number of spores found in permeate: 0,1,4,0,0,0,1,1,2,1. Total 10 cfu/10 ml. Logarithmic reduction: 6.0

Number of spores found in the water: 0,0,0,0,0. Figure 11 shows different values of TMP and permeate flow.

Comparative Example 7

An experimental setup according to Specific embodiment 2 was used.

Test with 0.8μιη round pores. The sieves were produced according to the part

Microsieve production above.

Samples of permeate was taken at a flow of 110 ml/min which corresponds to a theoretical flux of about 13 000 l/m²h in large scale. In the Fluxxlab the flux was 16 500 l/m²h, based on the surface area 4 cm².

The results were as follows:

Number of spores found in feed:

-4 94, 109

-5 11, 7

-6

Log load in the feed was 6.0.

Number of spores found in permeate: -2 (overgrown).

Number of spores found in the water: 24, 27, 24, 25, 27.

Figure 12 shows different values of TMP and permeate flow.

Comparative Example 8

An experimental setup according to Specific embodiment 2 was used.

Test with 0.8 μιη slits, 0.8 μιη x 2.6 μιη. The sieves were produced according to the part Microsieve production above.

Samples of permeate were taken at a flow of around 176 ml/min which equals a theoretical flux of about 18 400 l/m²h in a large scale. Calculated on the surface area 4 cm² in the Fluxxlab it was a flux of 26 400 1/m²h.

The results were as follows: Number of spores found in permeate: -2 (overgrown).

Number of spores found in the water: 0,0,0,0,0.

Figure 13 shows different values of TMP and permeate flow. Comparative Example 9

An experimental setup according to Specific embodiment 2 was used, but the ESL milk was heated in glass bottles.

Test with 0.5 μιη x 2.6 μιη slits, i.e. a repetition of comparative example 6. 10 ml of the spore suspension was added when a total of 3000 ml of skim milk had been added to the 600 ml of water.

Two sets of permeate samples were taken. The first sample, PI, was taken at a flow of 53 ml/min which corresponds to a theoretical flux of 14 946 l/m²h in a large scale. The flux in the Fluxxlab, based on the 4 cm² area, was 7 950 l/m²h. The second sample, P2, was taken at a flow of 35 1/min, corresponding to a flux of 9 870 l/m²h in large and 5 250 l/m²h in the Fluxxlab.

The results were as follows:

Number of spores found in feed:

-4 143, 159

-5 12, 14

-6 1, 0

Log load in the feed was 6.2

Number of spores found in permeate PI : 2, 3, 2, 3, 1, 4, 4, 1, 1, 2. Total 23/ 10 ml. Log reduction was 5.8.

Number of spores found in permeate P2: 1, 2, 3, 3, 2, 4, 1, 0, 0, 1. Total 17/10 ml. Log reduction was 5.9.

Number of spores found in the water: 0, 0, 0, 0, 0.

Figure 14 shows different values of TMP and permeate flow.

Results

In Table 5, the results of the log reductions achieved and the fluxes achieved on the different membranes are summarized. It can be seen that to be able to reach the target on capacity the pore shape should be slits and to reach the target on log reduction the width of the slits shall be slightly smaller than 0.5 μιη.

Table 5. Log reductions and fluxes.

The goals set for the feasibility study are a flux of 10 000 l/m²h and a logarithmic reduction of 7 on Bacillus pumilus spores. As seen in table 5, a wafer with slits with a width of less than 0.5 μιη, such as 0.35-0.45 μιη, preferably 0.45 μιη, should be used.

It is also important to note that, to be able to reach the flux of 10 000 l/m²h, the pores have to be elongated, i.e. slits, and not round pores.

The results also show that filtration of 0% fat milk is possible with a start-up with water for pore sizes between 0.5 and 0.8 μιη. Fluxes for 0.35 μιη are lower than predicted but this size is not necessary to achieve the goals.

Fluxes of 14 m³/m²/hour have been achieved with 0.5 μιη slits.

During concentrating the milk, the permeate flux declines. Increasing the transmembrane pressure restores the flux (Comparative example 9).

SEM images of the Bacillus pumilus spores on the 0.5 μιη slits wafer used in Comparative example 6 and 9 were taken and can be seen in Figure 15 and 17. It can be seen that some spores have nearly the same size as the width of the slits.

Thus, to reach a log reduction of 7 of Bacillus pumilus in ESL milk, a minimum pore size of 0.35-0.45 μιη, preferably 0.45 μιη should be used. In order to obtain a proper pressure distribution, as described above, an elongated, slit shaped, pore is preferred. Specific Embodiment 3

A further test was performed in larger filter plant by Fluxxion. However, the principle is the same as described in previous embodiments, and the scaling is easily appreciated by a person skilled in the art. The plant was equipped with membrane sieves with elongated, slit shaped holes. The sieves were produced according to the part Microsieve production above. The pores sizes of the sieves were 0.8 μιη in width. A permeate flow of 3 1/min (flux about 5 500 l/m²h) was used. The operating temperature was about 55°C. The DCP frequency was kept constant at 22 Hz.

Test Organism

In order to determine the retention of bacterial spores in a 0.8 μιη membrane spores of Clostridium tyrobutyricum, DSM No 2637. The Clostridium tyrobutyricum test strain has been cultured according to methods well known to a person skilled in the art.

Comparative Example 10

Spores of Clostridium tyrobutyricum were inoculated into approximately 60 liters of skim milk to get an initial load of log 4-log 5 cfu/ml. The filter apparatus was operated as described under Specific Embodiment 3 above. A 0-sample to get the initial load was taken at the same time as permeate samples were taken. Permeate was collected in sterile bottles approximately 0.5 liter in each.

The 0-sample as well as the permeate samples was heated to 70 -75°C for 5-10 minutes to eliminate the vegetative cells. The evaluation was done by using dead end filtration technique, well known to a person skilled in the art. The pore size of the filters in the test set-up was 0.8 μιη. The filters were then placed on Reinforced Clostridium Agar (RCA) and incubated in anaerobic jars in 37° for 5-7 days. To determine the initial load, a volume of 1 ml from dilution step 2, 3 and 4 respectively were filtered and cultured as above. The permeate samples were evaluated by using the MPN technique and 5* 1 ml milk, 5* 10 ml milk and 5* 100 ml milk were sampled from each of the two bottles and then filtered. The permeate samples were also evaluated by adding 10 ml of milk to each of twelve bottles containing 100 ml of sterile Reinforced Clostridium Medium (RCM). The bottles were then incubated in anaerobic jars at 37°C for 8 days and after incubation 10 μΐ was streaked on RCA to confirm growth.

Results

Dead end filtration - 0 sample

The numbers of colonies were counted on each dilution the result is shown in Table 6. The average number of cfu (colony forming units)/ml was calculated to 4,5 * 10⁴ based on ISO 7218. The initial load of Clostridium tyrobutyricum was thus determined to log 4.65 cfu/ml.

Table 6 Colony count 0-sample

Dead end filtration - Permeate sample

The result from the MPN series is shown in Table 7.

Table 7 MPN result permeate sample

According to the MPN calculation tool the logarithmic reduction was app 6 when all the samples were used in the calculation. However, the outcome from the MPN calculation is inconsistent and therefore the MPN is considered meaningless. When only the result from bottle 6 was calculated the outcome was acceptable and gave a log cycle reduction of 6 with a lower limit of 5.5 based on 95 % confidence level.

Some colonies from the different MPN groups were identified as presumed not to be the test organism. This was confirmed by identification in microscope as well as plating on PCA and aerobic incubation. Both cocci as well as some different rods were found. A possible explanation could be recontamination down streams in the equipment after filtration and or a laboratory contamination as well as thermoduric micro organism passing the membrane. Incubated bottles

All the bottles with incubated milk samples showed visible changes after 8 days incubation. Microbial growth was confirmed by streaking 10 μΐ on RCA plates and after 5 days of anaerobic incubation the plates were checked. Six streaks were negative and six were positive of which three showed very weak growth. The growth was not further investigated.

With the specified pore properties 0.8 x 2.3 μιη slits, the specific logarithmic reduction of Clostridium tyrobutyricum was 5. A SEM picture was taken as shown in Figure 16.

Thus, to reach a log reduction of 5 of Clostridium tyrobutyricum in Cheese milk, a minimum pore size of 0.7-0.8 μιη, preferably 0.8 μιη should be used. In order to obtain a proper pressure distribution, as described above, an elongated, slit shaped, pore is preferred.

Without further elaboration, it is believed that one skilled in the art can, using the preceding description, utilize the present invention to its fullest extent. The preferred specific embodiments described herein are, therefore, to be construed as merely illustrative and not limitative of the remainder of the description in any way whatsoever. Further, although the present invention has been described above with reference to specific embodiments, it is not intended to be limited to the specific form set forth herein.

In the claims, the term "comprises/comprising" does not exclude the presence of other elements or steps. Additionally, although individual features may be included in different claims, these may possibly advantageously be combined, and the inclusion in different claims does not imply that a combination of features is not feasible and/or advantageous. In addition, singular references do not exclude a plurality. The terms "a", "an", "first", "second" etc do not preclude a plurality.

Claims

1. Method for removing bacteria and bacterial spores from milk, comprising the step of:

filtering the milk over a membrane filter, said filter having pores with a uniform minimum pore size between 0.35 μιη and 0.8 μιη.

2. The method according to claim 1, wherein the pores have an elongated shape.

3. The method according to any of claims 1 or 2, wherein the milk is milk treated to obtain extra shelf life (ESL milk) and the minimum pore size is between 0.35 μιη and 0.45 μιη.

4. The method according to claim 3, wherein the minimum pore size is 0.45 μιη.

5. The method according to any of claims 1 or 2, wherein the milk is Cheese milk and the minimum pore size is between 0.7 μιη and 0.8 μιη.

6. The method according to claim 5, wherein the minimum pore size is 0.8 μιη.

7. Method of producing cheese from Cheese milk, wherein the Cheese milk is filtered according to any of claims 1-2 or 5-6.

8. A device for removing bacteria and bacterial spores from milk, comprising a pump and a membrane filter having pores with a uniform minimum pore size between 0.35 μιη and 0.8 μιη.

9. The device according to claim 8, wherein the pores have an elongated shape.

10. The device according to any of claims 8 or 9, wherein the minimum pore size is 0.45 μιη or 0.8 μιη.

11. Use of a device according to any of claims 8 to 10, for removing bacteria and bacterial spores from milk.

12. The use according to claim 11, wherein the milk is Cheese milk.

13. The use according to claim 11, wherein the milk is milk treated to obtain extra shelf life (ESL milk).

14. A filter for removing bacteria and bacterial spores from milk, having pores with a uniform minimum pore size between 0.35 μιη and 0.8 μιη.

15. The filter according to claim 14, wherein the pore size is between 0.35 μιη and 0.45 μιη.

16. The filter according to claim 14, wherein the pore size is between 0.7 μιη and 0.8 μιη.