Application Note

A Hands-On Guide to Ultrafiltration/

Diafiltration Optimization using
Pellicon® Cassettes
In ultrafiltration (UF) tangential flow filtration (TFF) The purpose of a process simulation is to duplicate the
systems, operating parameter selection will have a far entire manufacturing process in a scale-down format,
reaching impact as the process is scaled to full-scale to confirm sizing, and to assess preliminary product
manufacturing levels. While there are many factors quality and yield. The intent is to develop an optimized
that contribute to final system design, several key process (Figure 1), on the bench, that will efficiently
parameters should be optimized early in the process scale-up to meet full-scale manufacturing expectations.
development phase. The goal is to develop a robust
process with the following success criteria: superior Sequence Purpose
product quality, consistent and high product yield, 1. TMP Excursion at Initial • Determine TMP for UF/DF
reproducible process flux and time, and a cleaning Concentration (Cb initial) • Determine Feed Flow (QF) for UF/DF
regime that allows extended membrane reuse. • Demonstrate Flux Stability
• Confirm Retention of Product
The following basic experiments should be considered
2. C
 oncentration/Volume • Determine Flux as Function of
during development of processing methodology: Reduction (Cb initial Cb final) Concentration
• Determine Placement of Diafiltration Step
• Optimization
• Determine Flux as Function of Buffer
– Impact of transmembrane pressure (TMP) and Conditions
feed flow on process flux and retention 3. T
 MP Excursion at Final • Determine TMP for UF/DF
Concentration (Cb final) • Determine Feed Flow (QF) for UF/DF
– Impact of product concentration and buffer • Confirm Retention of Product
conditions on process flux and retention
4. D
 iafiltration/Buffer • Determine Diavolume Requirement
– Impact of diavolumes on buffer exchange and Exchange • Confirm Retention of Product during DF
contaminant removal
• Paper design and full process simulation with chosen 5. Product Recovery • Crude Assessment of Step Yield
processing parameters • Product Quality Evaluation

Typically, the first three experiments are performed Figure 1. Basic Optimization Experiments
sequentially to bracket process performance and
obtain data for analysis. This information is then
combined with actual manufacturing considerations
Use this step-by-step guide to
(batch volume, process time, etc.) to design a
develop a robust UF/DF process
process simulation.
with Pellicon® cassettes (cutoffs
of 100 kD and lower) that will
deliver superior product quality,
reproducible results, and high yields.

MilliporeSigma is the U.S. and Canada

Life Science business of Merck KGaA,
Darmstadt, Germany.
The following are step-by-step protocols for basic optimization experiments.

Set-up and Installation Procedure Equilibration Procedure

Refer to the Pellicon® 2 Cassettes Installation, User, 1. Add 3 L/m2 of the appropriate buffer to the feed tank.
and Maintenance Guide (UG2996EN) or the Pellicon® 3 Example: 0.1 m2 membrane area x 3 L/m2 = 0.3 L buffer
Cassettes Installation and User Guide (AN1065EN00)
2. Direct the retentate and permeate to a waste
when performing actual set-up and installation of
container.
Pellicon® cassettes.
3. Start the feed pump and achieve the following
1. Assemble the TFF system as shown in Figure 2.
conditions by partially closing the retentate valve
2. Install the Pellicon® cassette(s) (Pellicon® 2 Mini with and adjusting the pump speed:
0.1 m2 membrane area, Pellicon® 3 with 0.11 m2
– Feed flow of 5 L/min/m2
membrane area) in the appropriate Pellicon® holder.
– Retentate pressure of 2–15 psi (0.14–1.03 bar) to
3. Flush the system with water, clean with the
achieve approximately 30% conversion
appropriate cleaning agent (per appropriate
maintenance guide), and flush again. 4. When half the buffer has been flushed, put the
system in total recycle mode 1 and recirculate for
Diafiltration
Buffer QR Retentate 10 minutes; verify that the pH and conductivity in
the system have been equilibrated to the level of
Feed Retentate PR the starting buffer.
Tank Valve
5. Direct the retentate and permeate to a waste
PP
container.
PF
QF 6. When the feed tank level reaches the minimum level,
Permeate open the retentate valve fully and stop the feed pump
Feed QP to prevent the introduction of air into the system.
Filtration
Feed Module
Pump
Figure 2. Schematic of a TFF System

Part 1. TMP Excursion at Initial Concentration

1. Add sufficient volume of product to the feed 4. Recirculate the product for 10–15 minutes and
reservoir such that final volume or concentration ensure that stable process flux is achieved2.
target can be reached or slightly exceeded
5. Record temperatures, pressures, and flows; sample
(approximately 1–1.5 L of final product at final
feed and permeate for product retention3.
concentration per m2). Example: if Cinitial = 10 g/L
and Cretentate = 80 g/L, then the concentration factor 6. Increase TMP by 5–10 psid (0.34–0.69 bar) by
is 8X. If the minimum achievable final volume manipulating the retentate valve while keeping
for 0.1 m2 is 0.1 L, calculate the required initial the feed flow constant. For more open membranes
volume: increase by 2–5 psid (0.14–0.34 bar). Repeat steps
4 and 5.
Vinitial = Vminimum x VCF = 0.1 L x 8X = 0.8 L
7. Repeat step 6 until flux begins to level off 4; typically
2. Open the retentate valve fully and configure system 4–6 TMP values are evaluated in total.
in total recycle mode.
8. Open the retentate valve fully and allow system to
3. Start the feed pump and achieve the following continue in a total recycle.
conditions by partially closing the retentate valve
and adjusting the pump speed: 9. Increase or decrease the feed flow by 2–3 L/min/m2
and repeat steps 4 through 8. If desired, a third feed
– Recommended feed flow (QF) rate for the membrane flow rate can be investigated.
device, typically 5 L/min/m2 for Pellicon® 2 and
3 cassettes 10. Plot the data as shown in Figure 3.

– Minimal TMP, typically 2–5 psi (0.14–0.34 bar) for

more open membranes and 10 psi (0.69 bar) for
tighter membranes.

Tight membranes (1 kD, 5 kD, etc.) Can use large TMP increases since optimum is typically >30 psi
Open membranes (50 kD, 100 kD, etc.) Can use small TMP increases since optimum is typically <10 psi

2
 200 Feed Flux = 5 L/min/m2
Optimum Point Feed ∆P = 20 psid
TMP = 30 psid
160 Jf = 150 LMH

Flux (LMH)
2
Feed Flux = 3 L/min/m
120 Feed ∆P = 10 psid

80
Optimum Point
TMP = 25 psid
40 Jf = 86 LMH

0
0 10 20 30 40 50
TMP (psid)
Figure 3. TMP Excursion at Two Feed Flows

Q1 = 5 L/min/m2 Q2 = 3 L/min/m2 Q1/Q2

A [m ]
2
Volume/Time/150 LMH Volume/Time/86 LMH 0.57
QF [L/min] (5 L/min/m2) x Volume/Time/150 LMH (3 L/min/m2) x Volume/Time/86 LMH 0.95

Table 1. Membrane Area vs. Pump Feed Rate (Figure 3)

Calculations

The appropriate combination of feed flow rate and Note:

TMP will maximize flux while minimizing the impact of
• Anticipated final volume of over-concentrated product
pumping and shear on the product. The appropriate
must exceed minimum working volume of membrane
combination of these two parameters will also minimize
system at selected feed flow rate (QF); avoid
processing time and/or membrane area. To calculate the
introduction of air and maintain uniform mixing at
optimum feed flow, compare the required membrane area
end of volume reduction.
with the required pump rate at each of the two feed flow
conditions, as shown in Table 1. • Move from least to greatest fouling conditions:
Membrane Area [m ] = 2
– Do not test into pressure-independent regime (past
the knee of the flux vs. TMP curve)4
    Process Volume [L]/(Flux [LMH] x Process Time [h])
– Avoid exceeding 30–40% conversion ratios
In Figure 3:
• Check hysteresis if possible by returning the system
AreaQ1 = 0.57 x AreaQ2
to the initial conditions and taking a final flux
Pump feed rate [L/min] = measurement; compare initial flux performance to
final flux performance at initial conditions.
   Feed flux [L/min/m2] x Area [m2]
• Ensure that choice of TMP and feed flow have
In Figure 3:
corresponding retention values that are acceptable
Pump feed rateQ1 = 0.95 x Pump feed rateQ2 (> 0.998) at both initial and final product concentration
and in each buffer5.
In this example, it is advantageous to run at the higher
feed flow, Q1, since it only requires 57% of the membrane • There is often very little performance difference versus
area used at the lower feed flow rate at almost the feed flow rate at low product concentration. However,
identical pump feed rate. at the higher concentrations that will be investigated in
Parts 2 and 3, the benefits of different feed flow rates
should become more pronounced.

3
Part 2. Concentration
1. Use the product from Part 1 in the starting buffer. 8. Repeat the TMP excursion outlined in Part 1 to
Based on desired final product concentration factor, determine optimum TMP at the final concentration
add additional feed volume as needed to ensure in the starting buffer.
sufficient volume at end of concentration6.
9. Diafilter with one diavolume to get product into
2. Sample feed to confirm product concentration. final buffer and dilute with final buffer back to initial
concentration.
3. Put the system in total recycle.
10. Repeat the TMP excursion to determine the
4. Start the feed pump and achieve the optimum
optimum TMP at the initial concentration in the
TMP and feed flow as determined in Part 1 by
final buffer.
partially closing the retentate valve and adjusting
the pump speed. 11. Repeat Part 2 steps 2–7 once in final buffer using
the optimum TMP as determined above.
5. Direct the permeate to a separate container to
concentrate product and reduce volume. 12. Plot the data as shown in Figure 4, remembering
to apply a temperature correction in the flux
6. Record temperatures, pressures, and flows
calculations8.
throughout the concentration; sample feed and
permeate for product retention7. Note, concentration should be plotted in a
logarithmic manner.
7. Concentrate slightly beyond desired final product
concentration.

Starting Buffer
Diafiltration Buffer
Flux (LMH)

0.1 1 10 100 1000

Product Concentration [g/L]
Figure 4. Flux vs. Concentration

4
Calculations

The tradeoff between flux and diafiltration buffer Plotting the DF optimization parameter as a function of
volume create an optimum bulk concentration at which product concentration yields the optimum concentrations
to perform diafiltration; this can be calculated using the for diafiltration in both the starting and final buffers, as
DF optimization parameter at each data point: shown in Figure 5.
DF Optimization Parameter =
   Concentration [g/L] x Flux [LMH]

Starting Buffer Final Buffer

1000 C opt = 29.0 g/L C opt = 55.0 g/L
DF Optimization Parameter

Starting Buffer
800
Diafiltration Buffer

600
Optimum
Cb for DF
400

200

0
0 20 40 60 80 100 120

Product Concentration [g/L]

Figure 5. DF Optimization

There is an alternative approach that may be used Note:

to calculate the optimum concentration at which to
• Ensure enough feed material and appropriate
perform diafiltration (Copt). It assumes that the product
system working volume in order to achieve the
is completely retained and that the passage of the
final concentration.
permeating species is constant.
• Based on the results of the additional TMP excursions
If the flux versus concentration data is plotted as
performed in Part 2, the TMPs used for concentration
shown in Figure 4, then the gel concentration, Cg, is the
in both the starting and final buffers should be changed
concentration at which the permeate flux reaches zero
and the concentration should be repeated to obtain
(example: ~80 g/L in the starting buffer, ~110 g/L in
more accurate data.
the final buffer). The optimum concentration at which
to perform diafiltration is then calculated as9: – If the optimum TMP for the dilute solution occurs in
the pressure-independent region (past the knee of
Copt [g/L] = Cg [g/L]/e
the curve) for the concentrated solution, then the TMP
In Figure 4: should be decreased to the lowest optimum value.
Starting buffer Copt = 80/2.71828 = 29.4 g/L – If the optimum TMP for the dilute solution occurs
within the pressure-dependent region (before the
Final buffer Copt = 110/2.71828 = 40.5 g/L
knee of the curve) for the concentrated solution,
The Cg/e method can only be used when the flux vs. then the TMP may be increased to the highest
concentration data allows for accurate extrapolation optimum value to further optimize the flux and
to zero flux. reduce the processing time.
• Optimum concentration for diafiltration will be
different for each buffer; choose an average or the
most conservative.
– Restrictions on buffer usage or minimum recirculation
volume often dictate the concentration at which
diafiltration occurs.
– If the required final concentration is significantly less
than the optimum concentration for diafiltration, over-
concentration followed by dilution is a possible option,
although rarely chosen. It should only be considered
in cases where diafiltration buffer is limited and the
product is stable at the higher concentrations.

5
Part 3. TMP Excursion at Final Concentration
1. Use the product from Part 2 at the final concentration Calculations
in the final buffer.
Reference Part 1.
2. Repeat steps 2–10 of Part 1.
Note: Reference Part 1 and Part 2 notes.

Part 4. Diafiltration
1. Use the product from Part 3 at the optimum – Use 3–5 diavolumes as an initial estimate for
concentration for diafiltration; dilute as needed upstream UF/DF steps, or
using the final buffer.
– Use 7–12 diavolumes as an initial estimate for final
2. Configure the system for constant volume formulation UF/DF steps
diafiltration.
5. Record temperatures, pressures, and flows at every
3. Start the feed pump and achieve the optimum TMP diavolume; sample feed and permeate for both
and feed flow as determined in Part 1 and Part 3. product retention, and retention and concentration
of the contaminant of interest.
4. Diafilter the product with the chosen number of
diavolumes: 6. Plot the data as shown in Figure 6.
– Choose the number of diavolumes based on the
product purity specifications (if known, see calculation
below) and add a safety factor of 2 diavolumes, or

100
Contaminant Retained

10
[% of original]

R = 0.4
0.1

0.01 R = 0.2
R=0

0.001
0 5 10 15
Diavolumes

Figure 6. Contaminant removal vs. Diavolumes

Calculations

The percentage of the original contaminant in the For upstream steps, add 1–2 diavolumes. If the goal
retentate at each diavolume can be calculated from the of diafiltration is not to wash out a contaminant but
retention values using the following: rather to reach a target pH or conductivity, then the
measurement of that quality can be plotted against the
Remaining Contaminant [%] = 100 x e (Retention – 1) x N
number of diavolumes instead.
where N is the number of diavolumes.
Note:
However, since contaminant concentration is being
• If it appears necessary to diafilter past ~14 diavolumes,
directly measured in each feed sample throughout
any dead-legs or poor mixing areas in the system will
diafiltration, plot these concentrations as a percentage of
increase the apparent retention of the contaminant and
the original and use the above equation to plot several
make further removal difficult.
lines of theoretical retention, as shown in Figure 6. This
plot will help demonstrate the contaminant removal at • Ensure that choice of TMP and feed flow have
various retentions. corresponding product retention values that are
acceptable (>0.998) throughout diafiltration.
Select the whole number of diavolumes based on the
acceptable contaminant levels for the product; always
add 2–3 diavolumes as a 10-fold safety factor for
critical diafiltration steps, such as final formulation.

6
Part 5. Product Recovery
There are various methods for product recovery at 2. Recirculate at the selected feed flow rate with the
large-scale10. However, at small-scale, sufficient product retentate valve fully open for 10 minutes.
recovery can be achieved by manually tilting the system
3. Recover the buffer in a separate container using
and breaking the piping at low-points to drain the product.
the same methods that were used to recover the
Samples of the final retentate should then be analyzed for
product. Samples of this buffer rinse should be
product concentration and quality.
analyzed for product concentration.
1. After the product has been drained from the
4. After the product is recovered, the system should
system, add one system volume of diafiltration
be cleaned with the appropriate solutions8.
buffer to the feed tank.

Calculations

Ideally, the total product mass recovered in the retentate, The theoretical yield can also be calculated based on the
permeate, and buffer flush as well as unrecoverable membrane retention and compared to the actual yield.
holdup volume should equal the total mass of product in
Theoretical Yield [%] = 100 x e(Retention – 1)(N + lnX)
the feed. If the total product mass recovered is less than
the initial product mass, it is typically due to adsorption where N = number of diavolumes and X =
and/or solubility losses during processing11. However, it is concentration factor.
important to perform a mass balance and calculate total
yield to ensure optimum process parameters.
Actual Yield [%] =
100 x (Vretentate [L] x Cretentate [g/L]) / (Vinitial [L] x Cinitial [g/L])
Mass Balance [%] =
100 x {(Vretentate [L] x Cretentate [g/L]) + (Vpermeate [L] x Cpermeate [g/L]) + (Vrinse [L] x Crinse [g/L])}/(Vinitial [L] x Cinitial [g/L])

Note:
• All calculations are estimates; during these optimization
steps, the product has undergone more processing than
normal. Product degradation and yield may be slightly
affected. For a true indication of processing on product
quality, perform the entire optimized process using
fresh feed and new membranes.
• Product can be very viscous when recovered and may
affect assays; perform serial dilutions for more accurate
assay results.
• Actual yield and mass balance percentages should be
close to 100% and/or theoretical yield. If significant
losses occur, process parameters (including membrane
type) may have to be changed and then re-optimized.
• In a robust process, adsorption and solubility losses
should be very low.

7
Paper Design and Process Simulation
The optimization parameters obtained from the previous After performing the process simulation, the system
experiments can be combined to design a full process should be cleaned with the appropriate solution8.
simulation: concentration, diafiltration, (concentration,) If possible, the process should be rerun using the
and recovery. If time permits, a process simulation cleaned membranes to determine the effectiveness of
should be run immediately following the optimization the cleaning cycle and the consistency of membrane
work, and should employ the following: performance from run-to-run. If the cleaning cycle does
not prove effective, the cleaning parameters or cleaning
• New set of cassettes; same membrane type, same
solutions will need to be changed and the cleaning cycle
cassette path length
will have to be tested again.
• Fresh feedstock
• Fresh buffer(s)
• Optimized process parameters
• See detailed process simulation calculations below.

Calculations

The membrane area can be optimized to allow the Manufacturing scale volumes as determined by the user:
entire process (both concentration and diafiltration) to
• Feed volume = 5000 L
be completed in the specified timeframe (3–4 hours is
recommended). The average flux for each concentration • Retentate volume at end of 2.9X concentration =
and diafiltration step can be estimated from the 5000 L/2.9 = 1724 L
optimization data and combined with the desired volumes
to be processed. The approximate required membrane • Permeate volume removed during 2.9X concentration
area can then be calculated for both manufacturing = 5000 L – 1724 L = 3276 L
scale and scale-down runs. • 7X Diafiltration buffer volume = 7 x 1724 L = 12,068 L
Assume an example process scenario (this would have • Retentate volume at end of 3.4X Concentration =
been determined by optimization data, DF parameter, etc.): 1724 L/3.4 = 507 L
• 2.9X Concentration: • Permeate volume removed during 3.4X concentration
10 g/L to 29 g/L; flux decreases from 150 LMH to 80 LMH = 1724 L – 507 L = 1217 L
• 7X Diafiltration:
29 g/L; flux increases from 80 LMH to 85 LMH
• 3.4X Concentration:
29 g/L to 100 g/L; flux decreases from 85 LMH to 20 LMH
• Desired process time is 4 hours

8
Average process flux for concentration step12:
Javg = Jfinal + 0.33 (Jinitial – Jfinal) = Jinitial x 0.33 + Jfinal x 0.67
For 2.9X concentration:
Javg = 150 LMH x 0.33 + 80 LMH x 0.67 = 103 LMH
For 3.4X concentration:
Javg = 85 LMH x 0.33 + 20 LMH x 0.67 = 41 LMH

Average process flux for diafiltration step:

For diafiltration the average flux can be estimated as the
initial and final process flux during the diafiltration step.

Required area:
Area = [(Permeate volume/Average flux)Concentration + (Permeate volume/Average flux)Diafiltration + … ]/Time
In this example:
Area = [(3,276 L/103 LMH) + (12,068 L/83 LMH) + (1,217 L/41 LMH)] / 4 hours = 51.6 m2
Add 20% safety factor:
Area = 62 m2
To perform a scale-down process simulation, the same
volume to area ratio is used and scaled based on either
the feed volume that can be used for the simulation or
the area of the desired filtration device. For example, if
the process is to be performed on one Pellicon® 2 Mini
cassette (with an area of 0.1 m2), then the required
feed volume will be:
Scale-down feed volume = 0.1 m2 x (5000 L/62 m2) = 8 L
Instead, if there is a specific volume of feedstock to
process (example: 25 L), then the required membrane
area will be:
Scale-down membrane area = 25 L x (62 m2/5000 L) = 0.3 m2
The process parameters, including Pellicon® device type,
should be consistent between scales, allowing the process
to be completed in a similar timeframe with similar fluxes,
pressures and loadings. The concentration factors, number
of diavolumes and feed quality should be kept consistent
at all scales to ensure robust scalability. However, to
demonstrate process robustness and repeatability, the
process should be tested at pilot scale before proceeding
to manufacturing.

9
Definitions

Apparent Sieving (Sapp) Mass Flux (Jm)

The fraction of a particular protein that passes through The mass flow of protein through the membrane
the membrane to the permeate stream based on the normalized for the area of membrane (m2) through
measurable protein concentrations in the feed and which it is passing.
permeate streams. A sieving coefficient can be calculated
Jm [g m–2 h–1] = QP x CP/area
for each protein in a feedstock.
Sapp [–] = (Concentration in permeate, CP)/ Permeate Flux (Jf)
(Concentration in feed, Cb)
The permeate flow rate normalized for the area of
membrane (m2) through which it is passing.
Concentration Factor (CF)
The amount that the product has been concentrated Pressure Drop (ΔP)
in the feed stream. This depends on both the volume
concentration factor and the retention. As with the VCF, for The difference in pressure along the feed channel of
a Fed-Batch concentration process, calculate CF based only the membrane from inlet to outlet.
on the volume of feedstock added to the TFF application. ΔP [bar] = PF – PR
CF [–] = F
 inal product concentration/
initial product concentration = VCF(Rapp) Retention (R)
The fraction of a particular protein that is retained
Conversion Ratio (CR) by the membrane. It can also be calculated as either
The fraction of the feed side flow that passes through apparent or intrinsic retention. Retention is often also
the membrane to the permeate. called rejection.

CR [–] = QP / QF Rapp [–] = 1 – Sapp or Ri = 1 – Si

Diavolume (DV or N) Transmembrane Pressure (TMP)

A measure of the extent of washing that has been The average applied pressure from the feed to the
performed during a diafiltration step. It is based on permeate side of the membrane.
the volume of diafiltration buffer introduced into the TMP [bar] = [(PF + PR)/2] - PP
unit operation compared to the retentate volume. If a
constant-volume diafiltration is being performed, where Volume Concentration Factor (VCF or X)
the retentate volume is held constant and diafiltration
buffer enters at the same rate that permeate leaves, The amount that the feed stream has been reduced in
a diavolume is calculated as: volume from the initial volume. For instance, if 20 L of
feedstock are processed by ultrafiltration until 18 L have
DV or N [–] = T
 otal buffer volume introduced to passed through to the permeate and 2 L are left in the
the operation during diafiltration/ retentate, a ten-fold concentration has been performed
retentate volume so the Volume Concentration Factor is 10. In a Fed-Batch
concentration process, where the bulk feedstock is held
Intrinsic Sieving (Si) in an external tank and added to the TFF operation
The fraction of a particular protein that passes through continuously as permeate is removed, VCF should be
the membrane to the permeate stream. However, it is calculated based only on the volume that has been added
based on the protein concentration at the membrane to the TFF operation.
surface. Although it cannot be directly measured, it VCF or X [–] = T
 otal starting feed volume added to the
provides a better understanding of the membrane’s operation/current retentate volume
inherent separation characteristics.
Si [–] = ( Concentration in permeate, CP)/
(Concentration at membrane wall, Cw)

10
References/Footnotes

1. Total recycle means retentate and permeate lines 8. See Guide: Pellicon® 2 Cassettes Installation, User,
return to feed vessel and Maintenance Guide (UG2996EN) or Pellicon® 3
Cassettes Installation and User Guide (AN1065EN00)
2. If process flux is unstable, it may be necessary to allow
additional time or investigate other membrane options 9. Ng P, Lundblad J, and Mitra G, Optimization of Solute
Separation by Diafiltration, Separation Science, 11(5):
3. Retention samples are not required at every data
499-502, 1976.
point; sampling at lowest and highest TMP is typical
10. See Technical Brief: Protein Concentration and
4. The point at which the flux levels off is defined as
Diafiltration by Tangential Flow Filtration (TB032)
the point around which the slope of the flux vs. TMP
curve decreases to ≤50% of the previous slope. This 11. See Technical Note: Increase Product Yield in Your
point is also referred to as the “knee” of the curve. UF/DF Processes (AN1026EN00)
5. These other conditions are described in more detail 12. Average flux can also be calculated for each step
in Parts 2 and 3. by dividing the total process volume by the total
process time
6. Example: 10X concentration with a final volume of
300 mL requires (300 mL x 10) = 3 L of feed
7. Retention samples are not required at every data
point; initial and final concentration are typical.
Typical data recording interval is approximately
every 10–15 minutes.

11
 MilliporeSigma
400 Summit Drive
Burlington, MA 01803

For additional information, visit

EMDMillipore.com

To place an order or receive

technical assistance, visit
EMDMillipore.com/contactPS

© 2022 Merck KGaA, Darmstadt, Germany and/or its affiliates. All Rights Reserved. MilliporeSigma, the vibrant M, Millipore, MS_AN2700EN00 Ver. 3.0
and Pellicon are trademarks of Merck KGaA, Darmstadt, Germany or its affiliates. All other trademarks are the property 37140
of their respective owners. Detailed information on trademarks is available via publicly accessible resources. 06/2022