The basic principle of nanofiltration membrane technologies
In principle, membrane filtration is a simple filtration process with physical retention of substances on a filter surface with small pore sizes . Hence, membrane filtration is a separation process where unwanted substances (e.g. bacteria, virus, heavy metals, and other soluble & suspended contaminants) are physically removed without conversion.
A membrane unit can be configured in different ways effecting both the hydraulics and overall filtration process, but – in general – all membranes function as illustrated below in figure 1.
The feed stream is pressurized and pumped into the membrane unit where it is split into two streams. The permeate stream consists of the filtered water without unwanted substances that passes across the membrane. The concentrate stream contains concentrated unwanted substances, which are discarded. Often the concentrate stream is partially recirculated and mixed with the incoming feed stream to achieve a higher degree of utilization (i.e. higher permeate production per unit volume of incoming feed stream).
Membrane filtration techniques are categorized according to pore size and, thus, the particle size of retained substances. With decreasing particle size, there are four main categories, each with their own typical operating pressure range (table 1, ):
- Microfiltration (MF)
- Ultrafiltration (UF)
- Nanofiltration (NF)
- Forward osmosis (FO)
- Reverse osmosis (RO)
|Membrane Process||Typical Operating Pressure Range|
|Microfiltration||1.5 – 45 psi (0.1 – 3.1 bar)|
|Ultrafiltration||30 – 100 psi (2.1 -6.9 bar)|
|Nanofiltration||50 – 200 psi (3.4 – 13.8 bar)|
|Forward osmosis||Typically, below 30 psi (2.1 bar)|
|Reverse Osmosis (brackish water)||145 – 800 psi (10 – 55.2 bar)|
|Reverse Osmosis (seawater)||800 – 1015 psi (55.2 – 70 bar)|
Nanofiltration is the newest membrane process variant and bridges the gap between reverse osmosis and ultrafiltration . Nanofiltration (NF) membranes have a pore size ranging from 1 nanometer (0.001 µm) to 10 nanometer (0.01 µm) and a molecular weight cutoff (MWCO) of 1000 to 100,000 Dalton. Briefly, the MWCO of a given membrane technology refers to the lowest molecular weight solute that is retained by 90% or above by the membrane. The differential operating pressure across a nanofiltration membrane typically ranges from 50 psi (3.4 bar) to 200 psi (13.8 bar). Nanofiltration processes are limited by diffusion and not by seeping through the filtration cake formed by retained substances as it is the case in microfiltration (MF) and ultrafiltration (UF) processes.
As illustrated in figure 2 below, nanofiltration membranes remove bacteria, viruses, and most organic contaminants. Nanofiltration membranes also remove smaller molecules with charge, such as calcium and sulfate, which constitute the majority of hardness and alkalinity in water. Nanofiltration membranes were – in fact – developed for water softening applications. It is often of advantage to pretreat hard water, where small ions are to be removed, before passing the water through a nanofiltration membrane unit. This is to avoid excessive precipitation of scalants on the membrane surface. Pretreatment of the feed stream usually also involves removal of suspended solids. Due to the removal of alkalinity, permeate streams from nanofiltration processes can induce corrosion without appropriate posttreatment.
Cross-flow filtration vs dead-end filtration
There are two main operational modes for membrane filtration in water treatment applications, namely cross-flow filtration and dead-end filtration as illustrated in the figure below.
Cross-flow filtration is a filtration mode where the feed stream passes in parallel along the surface of the membrane at a high velocity. A fraction of the feed stream passes across the membrane as filtered permeate in a perpendicular direction to the feed stream (hence the term “cross-flow”). Due to the turbulent flow across the membrane surface, which continuously prevents accumulation of matter (the so-called filtration cake) on the membrane surface, cross-flow filtration is the preferred method of filtering water with high content of contaminants and foulants. For the same reason, cross-flow filtration is extensively used in both microfiltration (MF), ultrafiltration (UF), nanofiltration (NF), and reverse osmosis (RO) processes [1,2,3].
Dead-end filtration is a filtration mode where the entire feed stream is pushed through the membrane and becomes permeate. As such, dead-end filtration is to be considered a batch process where no waste-water is created. In comparison to cross-flow filtration, dead-end filtration only has two streams; the feed stream and the permeate stream. Another main difference is that dead-end filtration has lower energy loss since all the water being pressurized enters the membrane. Since all contaminants and foulants present in the feed stream are pushed towards the membrane surface during dead-end filtration, material rapidly accumulates on the membrane surface resulting in the build-up of a thick filtration cake. The filtration cake reduces membrane performance causing a steadily declining water flux and – as a result – dead-end filtration processes require frequent cleaning steps and/or membrane exchange. Dead-end filtration is primarily used in microfiltration and ultrafiltration processes [1,2,3].
Nanofiltration membrane materials
Cellulosic nanofiltration membranes
Cellulose acetate and cellulose triacetate membranes were among the first polymeric membranes developed for commercial large scale water treatment processes and are still used today due to comparably low cost and high tolerance toward chlorine . Another advantage of cellulosic membranes is that fabrication is a single-step process where a precursor polymer solution undergoes phase inversion and subsequent hot water annealing . Disadvantages include poor tolerance towards feed streams above pH 6 or below pH 4 and towards feed streams with temperatures above 30° Celsius. These limitations exclude cellulosic membranes from being used in many industrial water treatment applications .
Thin film composite nanofiltration membranes with polyamide-based rejection layers
The majority of commercially available nanofiltration (NF) membranes are composite membranes formed from a wide range of different polymers. Composite membranes typically consist of 3 layers:
- A porous textile polyester (non-woven or woven) backing with a thickness of around 120 µm to give the membrane mechanical strength.
- A micro-porous support membrane composed mainly of polysulfone or polyethersulfone with an average pore size of 15 nanometres and a thickness of 40 µm
- A thin polymeric rejection layer, which broadly determines the membrane’s overall performance characteristics (i.e. permeability and rejection properties). The thickness of the polymeric rejection layer is typically only a few hundred nanometres.
Composite membrane formation is a two-step process. First, the micro-porous support membrane is formed by phase inversion of a polymeric dope solution on top of the porous textile polyester backing. Next, the polymeric rejection layer is formed on the support membrane by interfacial polymerization of carefully selected chemicals. For standard nanofiltration membranes formed by interfacial polymerization, chemicals may include piperazine (PIP) in the aqueous phase and trimesoyl chloride (TMC) in the organic phase .
Composite membranes have several advantages over cellulosic membranes made from cellulose acetate or cellulose triacetate
- Ability to operate at higher working temperatures (in excess of 60° Celsius).
- Increased tolerance towards pH extremes (a pH range of 2-11 is tolerated).
Thin film composite nanofiltration membranes with rejection layers formed by layer-by-layer deposition
Composite nanofiltration membranes with rejection layers formed by layer-by-layer deposition of oppositely charged polyelectrolytes share many similarities to the thin film composite nanofiltration membranes with polyamide-based rejection layers described in the previous paragraph. The main difference being the rejection layer’s chemistry and production process. To our knowledge there is only a single commercial supplier of commercially available nanofiltration membranes with rejection layers formed by layer-by-layer deposition, which is why we won’t go further into describing this particular nanofiltration membrane technology .
Ceramic nanofiltration membranes
Since the capital cost of ceramic membranes is considerable higher than conventional polymeric membrane, ceramic membranes are typically used in hostile environments with high levels of solvents, extreme pH ranges, high temperatures and other environmental process parameters that inhibit polymeric members from being used. The high initial capital cost is offset by a longer operational lifetime under harsh operating conditions where polymeric membranes would rapidly fail. Ceramic membrane technologies for water filtration applications, such as ultrafiltration or microfiltration, typically use a rejection layer chemistry based on alumina or zirconia .
The production of ceramic nanofiltration membranes follows a sol-gel process where a colloidal solution gradually evolves towards a gel-like structure as solvent is removed. A final thermal treatment (sintering) is applied to enhance the structural stability and overall mechanical properties of the ceramic nanofiltration membrane. Depending on the desired molecular cut-off, the top rejection layer chemistry is based on titanium dioxide (TiO2), or a mixture of titanium dioxide (TiO2) or zirconium dioxide (ZrO2) [2,3].
Nanofiltration membrane form factors
The membrane geometry, or form factor, in combination with the choice of membrane material are determining factors for the overall separation process and hydraulic conditions within individual membrane elements. Importantly, the geometrical design of a membrane element directly impacts the hydraulic conditions within the element and thus significantly influences the fouling propensity or lack of the same. In addition, membrane geometry also determines the physical dimensions of the resulting membrane system as well as maintainability during operating cycles (e.g. back washing, chemical cleaning, etc.).
For water filtration applications there are five overriding membrane geometries:
- Spiral wound membrane elements
- Plate-and-frame membrane elements
- Disc tube membrane elements
- Hollow fiber membrane elements
- Tubular membrane elements
Refer to the table below for an overview of advantages & disadvantages between different membrane geometries.
|Membrane geometry||Maximum packing density||Comparative capital cost|
Maximum operating pressure
Comparative fouling propensity
1200 psi (83 bar)
|100 m2/m3||High||870 psi (60 bar)||Low|
|200 m2/m3||High||2320 psi (160 bar)||Low|
|1600 m2/m3||Low||870 psi (60 bar)||High|
|500 m2/m3||High||145 psi (10 bar)||Low|
|500 m2/m3||High||550 psi (38 bar)||Low|
Spiral wound membrane elements
Spiral wound membrane elements are the most common water treatment membranes for production of drinking water and are extensively used for desalination (i.e. removal of sodium chloride and other contaminants from seawater and brackish water to produce drinking water) using reverse osmosis (RO) and nanofiltration (NF) membrane technologies.
Spiral wound membrane elements are constructed from membrane envelopes with an internal permeate spacer and external feed spacer glued to a perforated central permeate collection tube . Once the membrane element is pressurized, water from the feed stream is forced across the membrane’s rejection later and filtered water enters the inside of the membrane envelope and exits the membrane element as a permeate stream through the permeate collection tube. The resulting retentate or concentrate stream consists of the remaining water that does not enter the permeate stream alongside rejected contaminants within the original feed stream. Said retentate or concentrate stream exits the membrane element through a designated outlet, which is physically separated from the permeate collection tube to avoid re-contamination. The overall filtration and separation properties of a spiral wound membrane element depend on the choice of membrane material and – especially – the chosen membrane’s rejection and permeability properties.
One of the main advantages of spiral wound membrane elements is their high packing density (i.e. ratio of membrane surface area to the volume of the membrane element itself), which results in a better value per membrane area, smaller footprint, and relatively low capital and operating costs [1,6]. The spiral wound configuration also allows for easy cleaning in place (CIP). The high packing density requires reduction of total suspended solids (TSS) to a minimum (<5mg/L) in the feed stream through various means of pre-treatment to prevent plugging of the membrane [6,7]. As an example, spiral wound elements for desalination of seawater can reach packing densities as high as 1200 m2/m3 . In this example the distance between membrane layers becomes less than 1mm, which demands stringent pre-treatment to avoid membrane fouling and plugging .
The design & functionality of spiral wound membrane elements is illustrated in the following YouTube video:
Plate-and-frame membrane elements
Plate-and-frame membrane elements – also known as stacked membrane elements – are yet another membrane configuration where flat membrane sheets are sandwiched between plates or frames creating small flow channels. The plates or frames are then mounted in a cassette system, which can be connected in parallel or series according to the nature of the application at hand. In plate-and-frame membrane elements, the distance between neighboring membrane sheets can be tailored according to the nature of the feed stream. As a result, plate-and-frame elements are used in many water treatment applications where the waste streams to be treated have high viscosities and/or contain high amounts of fouling agents.
The design & functionality of plate-and-frame elements is illustrated in the following YouTube video
With a typical packing density of around 100 m2/m3 , plate-and-frame membranes elements have the lowest packing density of the five membrane geometries considered here .
Disc tube membrane elements
Disc tube membrane elements are, to a certain extent, a hybrid between spiral-wound and plate-and-frame configurations, which allow flat sheet membrane sheets to treat highly concentrated wastewater at hydraulic pressures up to 160 bar (2320 psi) [11,12]. Both reverse osmosis (RO), nanofiltration (NF), and ultrafiltration (UF) flat sheet membranes can be mounted in a disc tube configuration. Essentially, disc tube membrane elements consist of alternatively stacked flat sheet membrane cushions and hydraulic discs with a patterned surface providing improved feed stream permeability and enhanced cleaning. The pressurized feed stream flows into the disc tube membrane element from below and moves to the top of element after which it flows in an s-shaped flow path down through the individual stacks and – ultimately – exits the element as a concentrate stream at the bottom of the element. In a similar way to spiral wound elements, water from the feed stream is forced across the membrane’s rejection later and filtered water enters the inside of individual membrane envelopes, moves to the permeate collector and exits the membrane element as a permeate stream through the permeate outlet at the bottom of the element [8,11,12].
Importantly, disc tube membrane elements can be opened allowing for inspection, cleaning, and replacement of individual membrane cushions and hydraulic disks.
With a packing density of up to 200 m2/m3 , disc tube membrane elements have the second lowest packing density of the five membrane geometries considered here . Typically, disc tube membrane elements have an overall length of 1200 mm or 1400 mm, an external diameter of 214 mm – 234 mm, about 200 individual membrane discs, and a total surface area of 7 m2 – 10 m2 [8,11,12,13].
The design & functionality of disc-tube membrane elements is illustrated in the following YouTube video:
Hollow fiber membrane elements
Hollow fiber membranes are small tubes with inner diameters typically ranging from 200 µm to 1 mm. The rejection layer pore size of hollow fiber membranes can be tailored to accommodate all types of membrane processes, namely microfiltration, ultrafiltration, nanofiltration, and reverse osmosis.
Hollow fiber membrane elements are constructed from bundles of up to thousands of individual hollow fiber membranes potted in both ends into a cylindrical housing to allow for pressurization. The packing density of a hollow fiber membrane element can reach up to 1600 m2/m3 . The most densely packed elements are prone to fouling and care must be taken in pre-treating the feed stream before it enters the element.
In a similar way to tubular membrane elements, hollow fiber membrane elements are either operated in outside-in mode (i.e. the permeate stream flows to the inside of individual tubes) or inside-out mode (i.e. the permeate stream flows to the outside of individual tubes) . As the driving force for water filtration can be either a positive (hydraulic) or negative (vacuum) pressure there are essentially 4 filtration modes for hollow fiber membrane elements:
- Inside-out with positive pressure
- Outside-in with positive pressure
- Inside-out with negative pressure
- Outside-in with negative pressure
The design & functionality of hollow fiber elements is illustrated in the following YouTube video:
Tubular membrane elements
Tubular membrane elements are predominantly used for challenging ultrafiltration applications where the feed streams have either high fouling propensity, high viscosity, or a combination of both, which inhibits the use of other membrane geometries. In the case of ultrafiltration membranes, porous tubes with inner diameters ranging from 5mm to 25mm and lengths of 150cm to 610cm are typically coated with thin micro-porous rejection layers of polyvinylidene fluoride (PVDF) or polyethersulfone (PES) on either the inside or outside walls of the porous tubes.
To our knowledge, commercially available tubular nanofiltration membranes are exclusively of the ceramic type with a reject layer chemistry based on titanium dioxide (TiO2), or a mixture of titanium dioxide (TiO2) or zirconium dioxide (ZrO2) [9,10].
Tubular membrane elements are constructed from bundles of individual tubular membranes potted in both ends into a cylindrical housing. The packing density of a tubular membrane element can reach up to 500 m2/m3. Depending on the orientation of the rejection layer, tubular membrane elements are either operated in outside-in mode (i.e. the permeate stream flows to the inside of individual tubes) or inside-out mode (i.e. the permeate stream flows to the outside of individual tubes) .
The design & functionality of tubular membrane elements is illustrated in the following YouTube video. Please note that although the YouTube video in question illustrates the functionality of hollow fiber membranes, the same general principles apply to tubular membranes, which are essentially a large-diameter version of hollow fiber membranes.
The basic principle of nanofiltration systems
Membrane systems are designed and constructed in a modular way for ease of scalability, and optimization of capacity, utilization, and the quality of the permeate and concentrate. Modularity comes in the form of treatment stages each with a well-defined array of membrane elements (figure 4).
With reference to figure 3, the feed stream for stage 2 is the concentrate stream from stage 1 resulting in successively lower water recovery rate (i.e. utilization) in later stages. In each stage, the concentrate stream is often recirculated and mixed with the incoming feed stream to achieve a higher overall recovery. The overall plant design will depend on feed water quality, desired degree of utilization, and space restraints (if any).
Operating cycles for water treatment plants vary from plant to plant depending on the design of the plant itself the design of the membrane stages and the feed water quality. Usually, periodic back washing / chemical cleaning of membranes is required. Such clean in place processes are typically fully automated in a similar way to what is know from large-scale water utilities. The period cleaning regime is supplemented on an ad hoc basis with more thorough cleaning routines using stronger chemicals.
Acid & alkaline solutions as well as oxidizing agent are commonly used for periodic cleaning. Thorough cleaning may involve chlorinated solutions. The choice of chemical type and concentration depends on the feed water quality and the membrane type as some membrane types do not tolerate exposure to certain chemicals. When evaluating which membranes to use for any given plant, special care must be given to the range of chemicals recommended by the membrane supplier. And the suppliers standard operating procedures for cleaning should always be followed.
How to choose the right membrane for a given water treatment application?
Membrane systems can be constructed as complete water treatment plants where membrane-based processes comprise the primary water treatment. Alternatively, membrane systems can supplement existing water treatment processes. When choosing the right membrane system for a water treatment plant, several factors must be considered:
- The membrane’s application field
- Effect on existing water quality
- Installation costs
- Operational and environmental aspects
When selecting a membrane system for existing plants, the footprint is often the most important factor to consider. Membrane systems will have different footprint requirements depending on how they are designed & constructed. I general, however, membrane systems are considered compact in comparison to alternative processes. As an example, submerged membranes impose demands on the capacity of the container (i.e. the area of the container) while membrane elements designed for pressure vessels require space in a dry room.
Selecting the right membrane for a given feed water quality tends to be a complicated process. Most suppliers will offer a range of membranes tested on different types of water qualities and will therefore be able to suggest a suitable membrane and membrane system for the feed water in question based on a water quality analysis. In most cases, the supplier will install an onsite pilot system to ensure that the selected membrane fulfills all demands and performance criteria.
Membrane systems are usually sensitive to fouling and it is therefore beneficial to operate the systems under anaerobic conditions to maintain solubility of iron and manganese and – thus – prevent fouling. By doing so, any existing water treatment processes can be fully utilized with the added functionality of the membrane system to remove organics or other hard-to-treat contaminants.
Since nanofiltration and reverse osmosis membrane systems remove ions from the feed stream, the quality of the permeate in terms of pH, hardness, and alkalinity will be influenced. It is therefore recommended to evaluate if additional chemical treatment is needed to regulate pH, hardness, and alkalinity to ensure a satisfactory water quality.
It is traditionally CAPEX intensive to establish membrane filtration processes. Besides the membrane system itself with its pumps, piping, and cleaning stations, investments are also needed for pretreatment, posttreatment, handling of the reject stream, control & monitoring, civil works, storage & mixing tanks, and assistance from external consultants. As a rule of thumb, total capital expenditure amounts to USD 8000 – 16000 per m3/hour installed treatment capacity. For smaller systems, the normalized capital expenditure can be higher.