Guide: How to measure forward osmosis membrane performance

There’s a need for standardized methods of measuring FO membrane performance

Forward osmosis membrane performance results from a non-trivial combination of external test parameters such as (but not excluded to):

  • Osmotic strength of feed and draw solutions
  • Membrane orientation
  • Cross flow velocity
  • Flow conditions in the test chamber/module
  • Temperature
  • Molecular nature of osmolytes (e.g. NaCl, MgCl, etc)
  • Design of the FO membrane test system
  • Experimental design (e.g. run-in times, equilibration times, measuring times, etc.)

Given the parameters above and their non-linear interdependence it is virtually impossible (without doing the actual physical experiment) to project how a given FO membrane will perform under test conditions differing from the ones the membrane is being physically tested under. Hence, comparing performance of FO membranes from different suppliers requires said membranes to be tested under identical standard test conditions. Conversely, without widely accepted standard test conditions, FO membrane performance cannot be compared between different suppliers. We – at ForwardOsmosisTech believe – that transparent comparison of FO membrane performance contributes positively to breaking down commercialization barriers within the FO field.

So it all boils down to defining (and adopting) standard test conditions for determining FO membrane performance

Several good suggestions of standard forward osmosis membrane test conditions have already been proposed by leading experts in the field (see for example “Standard Methodology for Evaluating Membrane Performance in Osmotically Driven Membrane Processes” by Cath et. al.) but – to our knowledge – not yet widely adopted.

Here we present ForwardOsmosisTech’s take on the subject, which is inspired in part by scientific literature and in part by our many years of practical experience assessing forward osmosis membrane performance.

Membrane form factors and suggested sample sizes for performance evaluation

Form factor Suggested sample size (surface area) Notes
Flat sheet 50-100 cm2 Traditionally, flat sheet coupons are cut to fit Sterlitech’s CF042 chamber (5.5 x 11cm = 60.5 cm2). Coupons of this size will transport roughly 60 g of water per hour given a typical FO membrane performance of 10 L/m2h (abbreviated LMH)
Hollow Fiber 50-100 cm2 Hollow fibers are a new FO membrane form factor and – as of yet – there’s no consensus on fiber diameter, fiber length, and number of fibers in the hollow fiber test module. ForwardOsmosisTech suggests a total fiber inner surface area of 50-100 cm2 as this results in a water transport of roughly  50-100g per hour, which fits well with the system component specifications summarized below
Tubular 50-100 cm2  Same as above for the hollow fiber form factor

Membrane conditioning and storage

Stage Hydration level Conditioning Storage Notes
Upon delivery Wet No conditioning needed 4°C in ultra pure water Exchange water on a weekly basis to avoid bacterial growth. Use within 1 month of delivery unless otherwise specified by the supplier
Upon delivery Dry No conditioning needed Room temperature in a dry environment Use within 6 months of delivery unless otherwise specified by the supplier
Before testing Wet Rinse thoroughly in ultra pure water  N.A. Take care not to damage the rejection layer / active layer during the washing process
Before testing Dry Hydrate by rinsing thoroughly in ultra pure water followed by 15 minutes of low-pressure driven permeation of ultra pure water through the membrane

Alternatively, hydration can be done by soaking for 5 minutes in a 50% mixture of ultra pure water and ethanol or isopropyl alcohol (IPA)

 N.A. Care must be taken in the hydration step to ensure complete wetting of the FO membrane’s support structure as well as to remove any protective coatings (such as glycerol) added by the supplier

Take care not to damage the rejection layer / active layer during the washing process

After testing Wet Rinse thoroughly with ultra pure water to remove any buildup of salt  4°C in ultra pure water ForwardOsmosisTech recommends no more than 3 independent performance measurements on individual membrane samples

 

Overview of system components needed for determining FO membrane performance

The table below summarizes the main system components needed in a bench-top system for determining FO membrane performances of membrane samples ranging in membrane area from 50cm2 to 100cm2.

Component Main functions Specifications Importance
1 laptop Record data from conductivity meters and the electronic scale Any Windows based laptop will suffice Need to have
2 gear pumps (one for the feed loop and one for the draw loop) Maintain cross velocity flow speeds of 20cm/s on the membrane feed and draw sides 1000-3000ml/min pumping capacity depending on the inlet cross sectional area of the membrane test cell Need to have
1 laboratory chiller and associated cooling coils Maintain stable feed and draw solution temperature of 20-22°C Any bench-top chiller with the following specifications will do:

  • Working Temperature: -20° to +40°C
  • Temperature Stability: ±0.1°C
  • Cooling Capacity: Up to 1290 watts @ 20°C
Nice to have if the ambient lab temperature is stable around 20-22°C

Need to have if the ambient lab temperature fluctuates excessively

2 conductivity meters Continuously measure the feed loop increase in conductivity due to reverse salt flux from the draw solution and the draw loop decrease in conductivity due to continuous dilution as water is transported across the membrane
  • Automatic data logging
  • Sensitivity range for feed loop conductivity probe: 0-400μS
  • Sensitivity range for draw loop conductivity probe: 50-100mS
  • Probe placement: the feed and draw probes should be placed after the bulk feed and draw loop reservoirs and before the FO test chamber feed and draw solution entrances, respectively
Need to have: feed loop probe

Nice to have: draw loop probe

1 electronic scale Continuously measure the decreasing feed reservoir weight to determine the FO membrane mediated water flux from feed to draw
  • Automatic data logging
  • 3-5kg max weight
  • 0.1 g accuracy
Need to have
1 magnetic stirrer Continuously stir the draw reservoir solution to ensure uniform bulk osmolyte concentration Any standard lab-scale magnetic stirrer capable of providing 200-1000 RPM will do the trick Need to have
1 FO test chamber (for flat sheet membrane coupons) Provide an enclosed & sealed environment with stable (non-turbulent), uniform, and identical flow conditions on both the draw side and the feed side of membrane coupons Many FO researchers opt for the CF042 test chamber from Sterlitech. Alternatively, for those researchers with access to engineering/workshop facilities, “homemade” acrylic/PMMA test chambers are equally suitable providing they secure  an enclosed & sealed environment (i.e. leak free) with stable (non-turbulent), uniform, and identical flow conditions Need to have
Tubing, feed&draw reservoir containers,  and various fittings Tubing and containers make up the feed and draw loops while the various fittings ensure that individual system components can easily be removed for cleaning and maintenance Silicone-based or other flexible (and transparent) tubing is preferable. The tubing must be chemically resistant to intended feed and draw solution components Need to have

 

Overview of system components needed for determining intrinsic FO membrane properties

The table below summarizes the main system components needed in a bench-top system for determining intrinsic FO membrane performance properties of membrane samples ranging in membrane area from 50cm2 to 100cm2.

Component Main functions Specifications
Importance
1 gear pump (identical to the ones used for measuring FO membrane performance) can be used for simultaneous measurements in two individual test chambers Maintain cross flow velocity speeds of 20cm/s on the membrane feed side as well as generate hydraulic pressure of up to 5 bar on the membrane feed side 1000-3000ml/min pumping capacity depending on the inlet cross sectional area of the membrane test cell Need to have
1 laboratory chiller and associated cooling coil Maintain stable feed solution temperature of 20°C Any bench-top chiller with the following specifications will do:

  • Working Temperature: -20° to +40°C
  • Temperature Stability: ±0.1°C
  • Cooling Capacity: Up to 1290 watts @ 20°C
Need to have
1 handheld conductivity meter Measure the conductivity of feed and permeate to determine membrane rejection properties towards various salts A handheld/portable, battery powered conductivity meter is preferable. The meter must be able to measure conductivity reliably in the range 1µS – 200mS Need to have
2 dead-end membrane filtration cells Provide an enclosed & sealed environment at a pressure up to 5 bar with stable (non-turbulent) and uniform feed flow. The cell should have a porous support on the permeate side to reduce the likelihood of membrane breakage. Finally the cell should allow for easy sampling/collection of permeate The CF042 dead-end RO filtration cell from Sterlitech works well Need to have
1 digital pressure gauge Provide real-time readings of the pressure in the feed loop Any digital pressure gauge designed for in-line pressure recordings in the interval 1-10 bar will suffice Need to have
Pressure resistant tubing, feed reservoir container,  needle valves, and various fittings Together, the tubing, container, and various fittings make up the feed loop. The needle valve restricts flow through the feed loop to build up pressure All tubing and fittings must be pressure resistant up to 10 bars. However, it is not recommended to operate the system at more than 5 bars Need to have

 

Standard conditions, equations, and protocols for determining FO membrane performance

Below we summarize experimental standard conditions to be used when determining FO membrane performance.

Experimental condition Value Units Notes
Feed and draw solution cross-flow velocity across the membrane surface  20  cm/s Use the cross-sectional area of the chamber’s feed and draw inlets to evaluate respective cross-flow velocities from the equation Q(flow rate) = A (cross-sectional area) * V (cross-flow velocity)
Draw solution concentration 1  M NaCl
Feed solution concentration 0  M De-ionized (MilliQ) water as standard. Pollutants/contaminants may be added to the feed solution to assess application-specific forward rejection properties
Hydraulic pressure ≈ 0 bar Care should be taken to minimize the hydraulic pressure difference across the membrane’s active layer (the dense rejection layer). Aim for less than 0.2 bar hydraulic pressure difference
Feed and draw solution temperature 20-22 °C No need for temperature control if the ambient lab temperature is stable within the required interval
Feed and draw solution pH ≈ 7 N.A Use feed and draw solutions as close to neutral pH as possible. However, there’s no need for pH adjustment
Membrane orientation FO mode and PRO mode N.A FO membrane performance (water flux (Jw) and reverse salt flux (Js))  should be determined in both FO and PRO mode. Forward rejection to specific feed solution contaminants (Rxxx) should be determined in FO mode only.

Below we summarize our take on standard protocols for determining FO membrane performance.

Performance parameter
Equation
Measurement protocol
Notes
 Water flux (Jw) Jw = AΔ∏e
  • Start the experiment by pumping feed and draw solutions around the fully assembled system and make sure to remove air-bubbles  in the test chamber
  • Set the electronic scale to record readings in 5 minute intervals. Start recording (t=0) immediately after all air-bubbles have been removed
  • Run the experiment for 45 minutes in total: 15 minutes initial run-in (not included in the data analysis) followed by 30 minutes membrane operation (included in the data analysis)
  • Calculate the average water flux for the 30 minute operation interval from the total feed reservoir weight reduction during said 30 minute interval. For a more detailed water flux analysis, the average water flux can be calculated and plotted for each 5 minute interval
  • Jw for any given FO membrane type is reported as the average of at least 3 measurements (in both PRO and FO mode) – as described above – of individual randomly selected membrane coupons
  • A is the “pure water permeability coefficient” – an intrinsic membrane property
  • Δ∏e is the effective osmotic pressure difference (assuming the absence of any hydraulic pressure difference) across the membrane’s active layer (the dense rejection layer)
  • Jw is predominantly reported in L/m2h or LMH in short
Reverse salt flux (Js)  Js = BΔC
  • Prepare a calibration curve relating conductivity to NaCl concentration in the feed probe’s sensitivity range (0-400μS)
  • Calibrate the feed probe according to the manual
  • Set the conductivity meter and the electronic scale to record readings in 5 minute intervals. Start recording (t=0) immediately after all air-bubbles have been removed
  • Run the experiment for 45 minutes in total: 15 minutes initial run-in (not included in the data analysis) followed by 30 minutes membrane operation (included in the data analysis)
  • Calculate the average reverse salt flux for the 30 minute operation interval from the increase in feed loop conductivity during said 30 minute interval. For a more detailed reverse salt flux analysis, the average reverse salt flux can be calculated and plotted for each 5 minute interval
  • Js for any given FO membrane type is reported as the average of at least 3 measurements (in both PRO and FO mode) – as described above – of individual randomly selected membrane coupons
  • B is the “salt permeability coefficient” – an intrinsic membrane property depending on the solutes used in the draw solution
  • ΔC is the difference in solute concentration across the active layer
  • Js is predominantly reported in g/m2h or GMH in short
Forward rejection to specific feed solution contaminants (Rxxx) Rxxx =

(1-CP/CF)*100%

  • A separate set of protocols must be developed to quantify trace amounts of feed solution contaminants of interest within a 1M NaCl solution (the draw). Alternatively, samples may be sent for external chemical analysis by external vendors
  • Start recording (t=0) immediately after all air-bubbles have been removed
  • Run the experiment for 45 minutes in total: 15 minutes initial run-in (not included in the data analysis) followed by 30 minutes membrane operation (included in the data analysis)
  • Calculate the average forward rejection for the 30 minute operation interval from the increase in contaminant concentration in the draw loop during said 30 minute interval. For a more detailed forward rejection analysis, the average forward rejection can be calculated and plotted for each 5 minute interval
  • Rxxx for any given FO membrane type is reported as the average of at least 3 measurements (in FO mode only) – as described above – of individual randomly selected membrane coupons
  • CP is the average concentration of contaminant in the permeate over the sampling time interval where forward rejection is assessed
  • CF is the average concentration of the contaminant in the feed over the sampling time interval where forward rejection is assessed

In forward osmosis experiments it is impractical/difficult to directly measure CP. Instead, one can measure the concentration of contaminant in the draw solution at the end of the sampling time interval (CD) and calculate CP as follows:

CP=((CD,final*VD,final)-(CD,initial*VD,initial))/VP

  • VD,final/initial is the draw volume at the end and beginning of the sampling time interval
  • CD,final/initial is the contaminant concentration at the end and beginning of the sampling time interval
  • VP is the volume of permeate passing across the membrane during the sampling time interval

The feed contaminant concentration (CF) is usually approximated as the average of the initial concentration and the final concentration at the end of the sampling time interval

Standard conditions, equations, and protocols for determining intrinsic FO membrane properties

Below we summarize experimental standard conditions to be used when determining intrinsic FO membrane properties.

Experimental condition Value Units Notes
Feed solution cross-flow velocity across the membrane surface 20 cm/s Use the cross-sectional area of the chamber’s feed and draw inlets to evaluate respective cross-flow velocities from the equation Q(flow rate) = A (cross-sectional area) * V (cross-flow velocity)
Feed solution concentration when determining the pure water permeability coefficient (A) 0 M MilliQ water as standard
Feed solution concentration when determining the salt permeability coefficient (B) 200ppm NaCl MilliQ water with 200mg/L reagent grade NaCl corresponding to an osmotic pressure of 0.15 bar
Hydraulic pressure (for both A and B value determination) 2 bar In order to avoid compaction of potential delicate FO membrane substrate materials, the applied hydraulic pressure should not exceed 2 bar
Feed solution temperature 20 °C Maintain the feed solution temperature by use of the laboratory chiller
Feed solution pH ≈ 7 N.A Use feed solutions as close to neutral pH as possible. However, there’s no need for pH adjustment
Membrane orientation N.A N.A Membrane active layer (the dense rejection layer) against the permeate

Below we summarize our take on standard protocols for determining intrinsic FO membrane properties.

Performance parameter Equation
Measurement protocol Notes
Pure water permeability coefficient (A) Jw = AΔ∏e
  • Use MilliQ water as the feed solution (zero osmotic pressure)
  • Equilibrate the system until the hydraulic pressure and permeate flow are stabilized
  • Once equilibrated set the starting time (t=0) and start collecting permeate (at least 5 g)
  • The A value for any given FO membrane type is reported as the average of at least 3 measurements – as described above – of individual randomly selected membrane coupons
  • Δ∏e is the hydraulic pressure difference across the membrane’s active layer (the dense rejection layer)
  • Jw is reported in L/m2h or LMH in short
Salt permeability coefficient (B) Js=BΔC

B = exp(-Jw/k)*Jw*(1-RNaCl)/RNaCl

  • FO membrane B values are calculated from Jw and RNaCl :
  • B = Jw*(1-RNaCl)/RNaCl
  • The B value for any given FO membrane type is reported as the average of at least 3 Jw/RNaCl  measurements of individual randomly selected membrane coupons
  • ΔC is the difference in solute concentration across the active layer
  • Js is predominantly reported in g/m2h or GMH in short
  • k is the mass transfer coefficient
  • In the case of high feed flow rate and low feed salt concentration, external concentration polarization can be neglected, meaning exp(-Jw/k) ≈ 1
Rejection to NaCl (RNaCl) RNaCl =

(1-CP/CF)*100%

  • Immediately after determining the A value, carefully empty the test chamber, tubing, permeate collection tube, and exchange the feed solution to 200ppm NaCl
  • Equilibrate the system until the hydraulic pressure and permeate flow are stabilized
  • Collect a sufficient amount of permeate (usually 5 ml) to allow for the measurement of permeate conductivity
  • Measure the bulk feed conductivity immediately after permeate collection
  • The RNaCl value for any given FO membrane type is reported as the average of at least 3 measurements – as described above – of individual randomly selected membrane coupons
  • CP (permeate concentration) and CF (feed concentration) are assumed to be proportional to the conductivities of said solutions, hence:
  • RNaCl =(1-SP/SF)*100% where SP and SF are the permeate and feed conductivities, respectively