Capacitive Deionisation (CDI) is a technology for removing salt ions from water by applying an electrical voltage difference between two porous carbon electrodes in which the ions will be temporarily stored .
The driving force of the process is the application of an electrical voltage difference ; the separation principle of salts from water is based on the capability of ions to move selectively according to their charge; in different words: positively and negatively charged salt ions in aqueous solutions migrate towards electrodes of opposite charge.
1. Technical Description
The Basics – Operating Mechanism
CDI´s preferred use is in low-salinity desalination applications (in principle below 10,000 ppm , in practice below 4,000 ppm ), where the benefit is lower energy consumption and higher recovery rates in comparison to Reverse Osmosis (RO) or distillation as alternative methods of brackish water desalination. This is mainly because CDI removes only salt ions, i.e. a tiny mass, from the saline feed water, whereas in RO or distillation a considerably larger mass of water is removed from the saline feed water .
Literature has traditionally classified CDI by its nature as an electrochemical technology, however, recent research  suggests that the most fundamental underlying principle of CDI is capacitive ion storage, a phenomenon based on the formation of a so called electrical double layer (EDL). In simple words: upon applying an electrical charge, ions are captured electrostatically and stored capacitively (analogous to electrical charge being stored in an electrical condenser/“capacitor“) in a diffuse layer formed next to the carbon surface. Ideally, pure physical accumulation of ions (sorption) dominates the process, not chemical (faradaic) reactions. The process is therefore easily reversible. Congruently, CDI may also be viewed as a process of ion sorption and desorption on electrically charged surfaces; CDI is, therefore, also named Electrosorption .
The formation of a capacitive EDL has today been identified as the actually dominant underlying principle for ion removal and is, therefore, at the heart of in-depth understanding and optimising the CDI process and its performance .
Apart from the aforementioned “capacitive ion storage“, a number of further electrochemical mechanisms are believed to accompany the “capacitive ion storage” phenomenon in a CDI cell which may influence CDI performance. Abstracting from the presence of such parallel electrochemical side effects, the CDI process – in its ideal form - is of purely physical nature and potentially enables CDI devices to have a long service life and low maintenance .
In practice, however, some electrochemical impacts are undoubtedly active, typically leading to some degree of electrochemically induced deviation from ideal performance of CDI cells over time (mainly faradaic reactions) .
Following the above explanations, CDI is a desalination technology that operates by adsorbing ions in an EDL formed in the proximity of electrodes by the application of an electrical potential difference. These electrodes are made of porous carbons optimised for high (ion) salt storage capacity and for good ion and electron transport .
In a nutshell, two fundamental technical phenomena need to be mastered and controlled in CDI systems :
- “capacitive ion storage“ (capturing ions electrostatically and storing them capacitively)
- “ion kinetics“ (the differences in mobility of the various ions on EDL formation).
While scientists may have today opposing convictions about the most dominantly underlying principle of CDI being of physical or chemical nature, categorisation also appears to depend on the breadth of definition of the term “electrochemistry”. It will be an interesting debate of whether or not it is preferable to hold on to historic practice of categorising CDI methodologically under the class of electrochemical desalination technologies or differently.
Process Description CDI
The basic mechanism underlying CDI is schematically shown in Fig. 1. It depicts a single CDI cell. In practice, multiple CDI cells are stacked on top of each other to form a stack; several stacks are packed together to form a CDI module. The process of CDI comprises three steps:
Step 1 – Purification (Ion Electrosorption) (see Fig. 1 A)
Upon applying a direct current (DC) voltage difference (typically between 1.2 and 1.5 V) between two porous carbon electrodes (typically attached to metallic current collectors) ions are attracted to the respective electrodes, i.e. cations (ions with positive charge, here, for example, Na+) to the negatively charged electrode, and anions (ions with negative charge, here, for example, Cl-) to the positively charged electrode. During this first “ion electrosorption” step, ions are stored in each of the two porous carbon electrodes. As a result, the saline feed water gets more and more deprived of the salt ions and, thus, is desalinated .
Fig. 1 A: CDI purification step Fig. 1 B: CDI regeneration step Fig. 1 C: CDI flushing step
Step 2 – Regeneration Step (Ion Desorption) (see Fig. 1 B)
As the electrodes become saturated with ions electrosorbed into the porous carbon, the electrodes require regeneration after some time. For regeneration purpose, the potential difference is reduced to zero (or reversed) and, in consequence, ions are desorbed from the electrodes and released back into the water stream  for the purpose of subsequent flushing them out with the bulk fluid as brine.
Step 3 – Flushing Step (see Fig. 1 C)
Finally, the desorbed ions are released into the water stream and flushed out of the CDI cell as brine. A further desalination cycle can restart hereafter .
To separate desalinated water from intermittent brine flushing, corresponding flow separation needs to be procured in the system´s piping arrangement.
A disadvantage of this CDI design is that – when reversing the electrical potential difference – ions may be driven to the electrode on the opposite side of the feed water channel. Once the electric potential difference is reversed, the adsorbed ions will be desorbed; at the same time, however, oppositely charged ions from the bulk fluid will be attracted and adsorbed. Therefore, ion desorption and adsorption occur simultaneously during this step. Consequently, some typical drawbacks will need to be taken into consideration, namely:
– incomplete regeneration of the electrodes leading to a deterioration in their adsorption capacity,
– longer regeneration times and
– residual ion accumulation blocking the way of other ions during the next purification step .
To avoid this drawback, a modification of CDI exists using specific membranes which is generally referred to as “Membrane Capacitive Deionisation” (MCDI).
Process Desciption MCDI
To enhance desalination performance, traditional CDI cells are modified by integration of ion selective membranes, as depicted in Fig. 2 A. This design is known as Membrane Capacitive Deionisation (MCDI) (see Fig. 2 B).
Fig. 2 A: CDI basic design Fig. 2 B: MCDI basic design
In MCDI, ion selective membranes (ion exchange membranes) are applied as thin films (typically 50 – 200 μm thickness) or coated (typically 20 μm thickness) directly to each of the two carbon electrodes, i.e. a cation exchange membrane to the cathode and an anion exchange membrane to the anode . Cations can pass cation exchange membranes freely, anions are repelled; vice versa, anions can move freely through anion exchange membranes, cations cannot.
The MCDI process works similarly to the CDI process, as shown in Fig. 3.
Fig. 3 A: MCDI purification step Fig. 3 B: MCDI regeneration step Fig. 3 C: MCDI flushing Step
As a particular advantage of MCDI, the ion selective membranes attached to the respective electrodes hinder salt ions to get electrosorbed by the opposite electrode when switching the potential difference during the regeneration step .
CDI Process Peculiarities
- Sequential production of desalinated water is a peculiarity of (M)CDI operation. In some typical CDI cell geometries (M)CDI desalination water production is intermittently interrupted by the need to flush the (M)CDI cell. Precise valve switching for the separation of brine from desalinated water flow is, therefore, essential.
- Energy requirements of (M)CDI strongly correlate with the level of feed water salinity and to the level of ions removed from the feed water .
- (M)CDI does not provide a barrier against bacteria, viruses and other microbes .
- (M)CDI requires prevention against scaling and fouling .
- Pressurised air is required when (M)CDI modules use air scouring . The latter is a method to remove sediments and to avoid biofilm build-up. In air scouring, clean (compressed) air is injected into the feed flow of the (M)CDI cell, thereby generating a series of air slugs. The slugs scour away the sediments while the water flow expels the undesired waste to an outlet to be discharged.
- (M)CDI systems are module based; (M)CDI plants are, therefore, scalable in size.
- (M)CDI modules need to be preserved against biological growth by a biocide conservation solution in case of shut-down for a prolonged period, e.g. a week or more .
Requirements pertaining to the layout of an (M)CDI plant are covered under section “Plant Design“ hereinafter.
Motivation of Use
(M)CDI has been developed as a non-polluting, energy-efficient and cost-effective alternative to desalination technologies such as RO and Electrodialysis (ED) .
- Basic CDI cell design (CDI or MCDI)
As mentioned in section 1, CDI cells are used in the traditional CDI design or in membrane enhanced MCDI design. MCDI design has operational advantages.
- Basic CDI cell geometries
CDI cells may be built in different geometries. However, the so-called “flow-by” mode in which the water channel lies in-between the electrode pair, as illustrated in Fig. 1, represents the very typical cell geometry today. The pathway of the feed water flows between the electrodes can be designed as an open channel (about 1 mm thickness) or may be constructed from spacer material, then about 100 - 300 μm in thickness. Flow may be one-dimensional from an entry point in a straight line to the exit point, as depicted in Fig. 1 hereinabove or e.g. in a 2-dimensional plane from corner to corner or from a central input with radial flow .
- Increased independence from feed water salinity fluctuation
For making the CDI process less sensitive to variations in the feed water salinity, a recycle line with reservoir to take up desalinated water being recycled may be worthwhile considering . (→ see Fig. 4).
- Electrical operation mode of CDI plant
a) Constant voltage operation
In constant voltage operation, the effluent salt concentration increases first, but then decreases over time, as electrodes will get saturated with ions over time. In other words: in the beginning, the EDLs are still uncharged and, hence, the potential difference between electrodes (the driving force of ion electrosorption) is high. Over time, with more and more ions becoming electrosorbed in the EDLs, the EDL potential (charge) increases and the remaining potential difference between the electrode pair, which drives the ion transport, decreases. Because of the decreasing ion removal rate over time, the effluent concentration increases again .
b) Constant current operation (MCDI design)
As the ionic charge transported into the porous electrodes is equal to the applied electric current, applying a constant current allows a better control of the effluent salt concentration compared to constant voltage operation. For stable effluent salt concentration, an MCDI design (i.e. incorporation of ion selective membranes into the CDI cell) is preferable 
Differences to and Similarities with Other Relevant Technologies
(M)CDI appears to compare favorably to RO on operational cost  and has come closer in recent years to get to parity with comparable RO investment amounts . CDI compares favorably to RO on energy cost and environmental burden . In particular the higher recovery rates of CDI compared to RO decrease water intake and disposal volume and add to the sustainability benefit of CDI versus RO . However, unlike RO, (M)CDI does not provide a barrier against microbes.
Compared to Ion Exchange (IX) in e.g. softening applications, (M)CDI, like RO and other electrochemical desalination technologies, is not competitive on the amount of investment, but extremely competitive on operational and environmental cost .
Compared to electrochemical desalination technologies such as Electrodialysis/Electrodialysis Reversal (EDR)  and Electrodeionisation (EDI) and Continous Electrodeionisation (CEDI) , relative differences are less clearly documented in literature and may need to be evaluated case by case. All such technologies, however, operate at high recovery rates, whereby EDI/CEDI appears to be limited even to brackish total dissolved solids (TDS) levels , but then again can produce water with extremely low salinity (e.g. boiler feed water or high purity water), which is not the domain of CDI.
2. Plant design
Besides the core desalination system (M)CDI, a full plant arrangement reguires the following components:
- feed pump (> 3 bar head pressure) for conveying the feed water through pretreatment and desalination unit
- appropriate pretreatment, particularly in applications with elevated scaling risk, to avoid scaling and fouling of (M)CDI cells, e.g. pH adjustment, anti-scalant dosing etc. 
- a protection for the (M)CDI cell against blocking by particles given the narrow spacer channels of the (M)CDI cell; typically, a 5 μm particle filter will do as the last step of pretreatment 
- further pretreatment to protect the 5 μm filtration unit (e.g. further pre-filtration step at larger mesh size) depending on application
- further pretreatment to reduce the volume of total suspended solids (TSS) and total organic carbon (TOC) depending on TSS/TOC loading of the feed water
- pressurized air (6 bar), if air scouring of (M)CDI units is desired for enhanced performance 
- a cleaning in place (CIP) unit (possibly with neutralisation) for intermittent acidic cleaning of CDI modules 
- a waste water disposal facility to allow for occasional CIP waste water discharge and intermittent flushing water discharge
- no particular post-treatment other than mandated by the application of the desalination system
- voltage, electrical current and TDS control to monitor and control system operation; in particular, precision of valve switching to separate concentrate flow from desalinated water flow for a high recovery rate
(M)CDI Plant Layout
Due to (M)CDI being module-based, (M)CDI plants are scalable in size . Fig. 4 shows a generic (M)CDI plant layout in batch mode operation, and, alternatively, in continuous mode operation (dotted line with recycle loop).
In batch mode operation the product water from a first pass through the (M)CDI desalination unit is collected and may then reinjected again and again as required for a desired level of desalination. As such batch operation is tedious and costly, a continuous mode of operation is generally preferred. This is achieved by recycling the product water via a recycle loop. In continuous mode operation the degree of desalination can be easily pre-set and variations in feed water salinity are more easily controllable.
3. Feed Water
- Different from RO and distillation technologies, energy consumption of (M)CDI is strongly related to salinity .
- (M)CDI performs well in brackish water applications, preferably at a feed water salinity below 4,000 ppm TDS .
- To avoid blockage of the spacer channel in CDI modules, particles in the feed water stream need to be controlled. Typically, a 5 μm pre-filtration will be appropriate [8/6].
- Temperatures may be slightly more elevated than in RO (< 60° C) .
- Feed water scaling potential needs to be evaluated .
- For general guidance, Table 1 characterises general (M)CDI feed water limits .
|Total Dissolved Solids||ppm||0 – 4,000|
|Total Organic Carbon||ppm||> 15|
|Chemical Oxygen Demand||ppm||> 50||> 100|
|Turbidity||NTU||> 4||> 100|
|Fats, Oils, Greases||ppm||> 0,5|
|Total Suspended Solids||ppm||> 4||> 20|
|Free Chlorine||ppm||> 1||> 25|
|pH||–||2 – 10||1 – 12|
|Iron Total||ppm||> 0,5|
|Total Hardness (as CaCo3)||ppm||> 1,000*|
|M Alkalinity (as CaCO3)||ppm||> 1,000*|
|Temperature||°C||5 – 60|
|°F||40 – 140|
* Limits depend on set TDS reduction and water recovery
Table 1: Typical feed water composition for (M)CDI 
Table 2 illustrates existing and future applicability of (M)CDI technology in relation to the water salinity in which the technology is applicable. While literature sees the brackish water salinity segment (1,000 – 10,000 ppm salinity) as the core applicational segment for (M)CDI technology today [3, 6, 7, 8], there are meanwhile also claims from (M)CDI suppliers to have supplied first (M)CDI systems to the salinity segment above 10,000 ppm, even to applications above 100,000 ppm . However, these applications have not been typical seawater desalination applications, but mainly were used to concentrate produced water in the oil & gas industry. Results on techno-commercial feasibility are not yet in the public domain.
|Feed Water Salinity||Technically feasible||Now in Use for Desalination||Potentially for techno-commercial Viability for the Future|
> 100,000 ppm
50,000 – 10,0000 ppm
10,000 – 50,000 ppm
1,000 – 10,000 ppm
< 1,000 ppm
Table 2: Stage of development of (M)CDI technology in relation to water salinity [3, 6, 7, 8]
4. Desalinated Water & Brine
The level of remaining salts in deionised water from a (M)CDI desalination process is highly dependent on the recovery rate applied and the energy input. The level of remaining ions in the desalinated water is – in limits – adjustable in (M)CDI operation .
Typically, (M)CDI produces deionised water at high recovery rates compared to RO . As a result, the volume of brine in (M)CDI systems is generally lower than in RO systems. Depending on application (e.g. antiscalant dosing, pH adjustment), the brine of (M)CDI systems may or may not be contaminated by chemical additives.
Depending on application, waste streams will come from disposing intermittent CIP (cleaning-in-place). Due to (M)CDI´s capability of deionising at comparatively high recovery rates, feed water volume and brine disposal volume will usually be lower than in RO operation .
5. Data and Information 
- Typical field of economical desalination with regard to energy consumption:
Brackish water < 4,000 ppm TDS
- Feed water temperature 5 – 60 °C
- Typically, higher recovery rate than RO (up to 95 %)
- Low pressure application, e.g. feed pressure < 3 bar 
- Compressed air requirement (6 bar) for air scouring
- Typical CDI cell voltage 1.2 – 1.5 V
- CDI specific energy consumption strongly depends on feed water salinity, as shown in Fig. 5.
- The specific energy consumption of CDI for feed water below TDS 4,000 ppm is significantly lower than that of RO, as depicted in Fig. 5.
- Exemplary module and system data 
• 1 module contains 12 CDI stacks
• 1 stack contains 23 CDI cells (flow between 2 electrodes)
• smallest system: 2 modules, total flow rate: 0.2 – 1.8 m3/h -
• system at industrial scale: 12 modules, total flow rate 1.4 – 10.8 m3/h
• typical module flow rate: 0.4 m3/h at 80 % recovery rate
- CDI is operated fully automatically (low personnel cost for operation)
- Module replacement (no reliable data available yet)
6. Preferred Use
- Brackish water desalination (energy efficient below 4,000 ppm feed water TDS) 
- Water softening
- Commercial laundry waste water deionisation (waters with temperatures too high for RO, but manageable for CDI)
- Treatment of cooling tower make-up water (side stream filtration; IX replacement)
- Waste water / RO brine recovery
- Horticulture irrigation (maintaining of tuneable ion levels in the desalinated water) 
7. Environmental Impact
With energy consumption typically below RO in comparable brackish water applications , CDI ranks on a privileged level of sustainability in brackish water desalination.
Like RO systems, also (M)CDI systems are prone to inorganic and organic fouling in practical applications, even though it has been shown that switching the potentials of electrodes, as is standard in CDI operation, may drastically reduce fouling . Also, (M)CDI is a low pressure operation compared to RO, which in principle should reduce fouling tendency.
Dosing of chemical additives in CDI operation is typically restricted to intermittent CIP (cleaning-in-place) with acids, and – as may be required – to antiscalant and/or pH adjustment in case of elevated scaling potential .
While (M)CDI systems appear to be affected by fouling in practical operation, these fouling problems may in tendency be viewed as somewhat less severe compared to RO .
Depending on application, (M)CDI brines will be or will not be contaminated by chemical additives. Depending on the application, CIP waste water requires neutralisation and disposal. Due to (M)CDI´s capability of deionising at comparatively high recovery rates, feed water utilisation and brine disposal will be lower than in RO operation
In comparison to RO, (M)CDI may in total be ranked preferable in environmental sustainability.
8. Stage of Maturity
Table 3 summarises (M)CDI technology with respect to the technology´s stage of maturity.
|Commercialization (up to 7 years)||Use/Established
(more than 7 years)
Table 3: Stage of maturity of (M)CDI technology
9. Further Developments
Challenges for improvement in CDI are:
- To develop sufficient application breadth and depth for CDI, so as to lower CDI investment from experience and volume effects (economies of scale), particularly on module assembly cost. While RO has already travelled along the experience curve for a few decades, economies of scale improvement are yet to be experienced by CDI technology .
- As highest specific surface area at low resistivity is a key contributor to increased CDI performance, improvements in the choice of suitable electrode materials are of key importance. Key requirements for CDI electrode materials and associated CDI performance are:
- large ion-accessible specific surface areas (accessibility of electrode surface to ions)
- high electrochemical stability over pH and voltage range, to avoid oxidation and to ensure system longevity
- fast ion mobility within the micro- and macro-pores of the carbon electrode (e.g. inter-particle distance, intra-particle pore sizes, pore structure and thickness)
- high electronic conductivity of electrodes (metal like electronic conductivity to ensure that the entire electrode is charged without major voltage gradients and to avoid larger energy dissipation and heating up of the electrode)
- low contact resistance between the porous electrode and the current collector
- good wetting behaviour (good hydrophilicity ensures participation of the entire pore volume in the CDI process)
- low material cost
- ease and scalability of material processing for higher volumes
- abundant availability of the material and low CO2 footprint for material purchase
- high bio-inertness to avoid biofouling of CDI cell 
- A particular challenge is the combination of high specific surface area with high ion mobility. The ultimate limit to the pore size is the bare ions size, e.g. 1.16 Angstrom for Na+ and 1.67 Angstrom for the Cl-, or, more precisely, to 3.58 Angstrom for sodium and 3.31 Angstrom for the chloride as solvated ions . Commonly the pore sizes for most common carbons will be significantly larger (in the small nm area).
For reference: 1 Angstrom = 0.1 nm
For further reference: IUPAC Classification of pore size:
1. Macropores: > 50 nm
2. Mesopores: 2 – 50 nm
3. Micropores: < 2 nm 
Carbon materials of relevance for ion adsorption in porous electrodes are amongst others:
• activated carbon (100 – 3,500 m2/g)
• ordered mesoporous carbon (OMC – typically 750 – 1,500 m2/g)
• carbon aerogels (400 – 1,100 m2/g)
• carbide-derived carbons (1,200 – 2,000 m2/g)
• carbon nanotubes and graphene (fluctuating largely from 50 – 500 m2/g)
• carbon black (< 120 m2/g) 
- Improvement in operational modes to increase yield in deionised water (e.g. move to continuous/semi-continuous operation; recycle-loop layout) 
- Improvement in cell unit design (e.g. wire type anodes) 
- Further reduction of energy cost by harvesting the energy generated in the ion adsorbing (charging) CDI module
 Porada, S.; Zhao, R.; van der Wal, A.; Biesheuvel, P.M.: Review of the science and technology of water desalination by capacitive deionization. Progress in Material Science 58, (2013), pp. 1388-1442
 Anderson, M. A.; Cudero, A. K.; Palma, J.: Capacitive deionization as an electrochemical means of saving energy and delivering clean water. Comparison to present desalination practice: Will it compete? Electrochimica Acta 55 (2010), pp. 3845-3856
 Voltea: Voltea´s Technical Bulletin – Desalination Cooling Water with CapDI, URL: http://www.aquafit4use.eu/userdata/file/Mid-term%20Conference/Session%203%20Advanced%20Treatments/Voltea%92s%20desalination%20technology%20applied%20on%20cooling%20water%20B%26J.pdf (accessed 16.01.2017)
 Al Marzooqi, F. A.; Al Ghaferi, A. A.; Saadat, I.; Hilal, N.: Application of Capacitive Deionisation in water desalination: a review. Desalination 342 (2014), pp. 3-15
 Oren, Y.: Capacitive deionization (CDI) for desalination and water treatment – past, present and future (a review). Desalination 228 (2008), pp. 10-29
 Voltea: Voltea´s Technical Bulletin – Feed Water Guidelines, URL: http://voltea.com/wp-content/uploads/2016/03/402D021_Rev02-Technical-Bulletin_Water-Quality-Criteria.pdf (last accessed on 16.01.2017)
 Voltea: Technical Specification Industrial Series – IS 2 to IS 48 Systems, URL: http://voltea.com/wp-content/uploads/2016/03/402D026EN_Rev06-IS-Technical-Specifications.pdf (last accessed on 16.01.2017)
 Voltea: Voltea´s Technical Bulletin – Preservation Handling Guidelines, URL: http://voltea.com/wp-content/uploads/2016/03/402D004_Rev01_Tech-Bulletin_Preservation-Handling-Guidelines-1.pdf (last accessed on 16.01.2017)
 Welgemoed, T. J.: Capacitive Desalination Technology: Development and Evaluation of an Industrial Prototype System, University of Pretoria; Dissertation (2005)
 Voltea: Technical Comparison CapDI vs. RO, URL: http://voltea.com/wp-content/uploads/2016/05/CapDI-vs-RO.pdf (last accessed on 16.01.2017)
 Voltea: Voltea´s Technical Bulletin – Technology Comparison, URL: http://www.voltea.com/wp-content/uploads/2015/12/402D002_Rev01_Tech-Bulletin_Technology-Comparison-1.pdf (last accessed on 16.01.2017)
 Subramani, A.; Jacangelo, J. G.: Emerging desalination technologies for water treatment: A critical review. Water Research 75 (2015), pp. 164-187
 Atlantis Technologies Website, URL: http://www.atlantis-water.com/value-proposition/ (last accessed on 29.04.2018)
 Voltea Website, URL: http://voltea.com/technology-4/ (last accessed on 16.01.2017)
 Porada, S., Sales, B. B., Hamelers, H. V. M., Biesheuvel, M.: Water Desalination with Wires. Journal of Physical Chemistry Letter 3 (2012), pp. 1613-1618