Provision of Desalinated Irrigation Water by the Desalination of Groundwater Abstracted from a Saline Aquifer

: Globally, about 54 million ha of cropland are irrigated with saline water. Globally, the soils associated with about 1 billion ha are affected by salinization. A small decrease in irrigation water salinity (and soil salinity) can result in a disproportionally large increase in crop yield. This study uses a zero-valent iron desalination reactor to effect surface processing of ground water, obtained from an aquifer, to partially desalinate the water. The product water can be used for irrigation, or it can be reinjected into a saline aquifer, to dilute the aquifer water salinity (as part of an aquifer water quality management program), or it can be injected as low-salinity water into an aquifer to provide a recharge barrier to protect against seawater intrusion. The saline water used in this study is processed in a batch ﬂow, bubble column, static bed, diffusion reactor train (0.24 m 3 ), with a processing capacity of 1.7–1.9 m 3 d − 1 and a processing duration of 3 h. The reactor contained 0.4 kg Fe 0 . A total of 70 batches of saline water (average 6.9 g NaCl L − 1 ; range: 2.66 to 30.5 g NaCl L − 1 ) were processed sequentially using a single Fe 0 charge, without loss of activity. The average desalination was 24.5%. The reactor used a catalytic pressure swing adsorption–desorption process. The trial results were analysed with respect to Na + ion removal, Cl − ion removal, and the impact of adding trains. The reactor train was then repurposed, using n-Fe 0 and emulsiﬁed m-Fe 0 , to establish the impact of reducing particle size on the amount of desalination, and the amount of n-Fe 0 required to achieve a speciﬁc desalination level.


Introduction
Over the last two decades, irrigation has accounted for about 70% of global anthropogenic water usage from freshwater [1][2][3][4]. It accounts for 90% of anthropogenic consumptive water usage [1,3]. About 40% of freshwater withdrawals are from groundwater [1]. About 66% of the global population lives under water stress conditions [4]. The global agricultural land area is about 5 billion ha, of which about 1.24 billion ha is arable cropland [5]. Since 2000, the global cropland area has expanded by about 10% [5]. The global distribution of cropland and associated soil types are discussed elsewhere [5,6].
There is a clear link between global food availability and global poverty [7]. It is estimated that a 1% increase in global food production could result in a 0.48% decline in global poverty [7]. Since 1850, global cropland has increased from about 5% of global land area to about 10% of global land area [8]. Over the same time period, the proportion of global land used to rear livestock has increased from about 10% to about 25% of the global land area [8].
Extrapolation of this trend to 2050 indicates that the global irrigated area could tentially expand from 270 million ha in 2014 to >700 million ha by 2050. Irrigated crop currently accounts for around 40% of global food production [9,10]. More than 1 bi ha of land is negatively affected by salinization, including about 20% of the global gated cropland (54 million ha) [11]. About 34 million ha of salinized land has become inized as a direct result of irrigation [11]. Globally, about 424 million ha of topsoil ( cm) and 833 million ha of subsoil (30-100 cm) are salinized [11]. The term salinized is u to refer to soil or water with an electrical conductivity of >4 dSm −1 (mScm −1 ) [11]. The tential future global cropland land resource in areas where the soil is salinized is estim at >200 million ha [11].
Global food demand is expected to rise by between 60% and 100% by 2050 [9,10]. rise in food demand results from the expected increase in global population from 7.7 lion in 2017 [11] to 9.1 billion people in 2050 [9,10], and 11.2 billion by 2100 [11]. The jority of the future increase in food production required to meet this demand will be fined to areas that are salinized or are currently irrigated with saline water or bec newly irrigated with saline water.
Desalination, partial desalination, freshwater dilution, additional processing treatment of this saline irrigation water has the potential to substantially increase yields, increase crop varieties, and increase crop quality and value, while improving quality, groundwater quality, and riparian water quality. The process flow adopted in study to assist in raising the yields of crops currently irrigated with saline water is s marized in Figure 1. Saline irrigation, when compared with freshwater irrigation, results in reduced yields (t ha −1 ) and reduced crop value (USD t −1 ). It reduces the number of alternative c that can be grown on an agricultural unit and results in increased soil salinization. gated crops have a low sale price (USD t −1 , USD ha −1 ), a high fixed operating cost (U • A single ZVI charge to be able to process multiple batches (or volumes) of water without loss of activity. Commercial ZVI water remediation plants, removing dissolved organic matter (DOM), organic solutes, and metals have demonstrated: >3.8 million m 3 water processed t −1 Fe 0 to remove DOM in a 22,500 m 3 d −1 reactor [17]. >24,000 m 3 water processed t −1 Fe 0 to remove organic matter in a 60,000 m 3 d −1 reactor [17]. >466 m 3 water processed t −1 Fe 0 to remove arsenic in a 9.6 m 3 d −1 reactor [17]. >17,356 m 3 water processed t −1 Fe 0 to remove arsenic in a 5.2 m 3 d −1 reactor [17]. >2500 m 3 water processed t −1 Fe 0 to remove gold in a 30 m 3 d −1 reactor [17].
To date, a single ZVI charge in an experimental ZVI desalination reactor has not been able to process >14,000 m 3 water t −1 Fe 0 . One of the objectives of this study is to determine if the amount of water that can be processed by ZVI, can be increased to >42,000 m 3 water-processed t −1 Fe 0 without a loss of desalination activity.

•
A method or process to be established and incorporated in a plant, which allows the highly variable product outcome from each batch to be compensated for. Another of the objectives of this study is to assess the product outcome variability and determine how the product quality problems associated with this variability may be overcome in a future commercial unit. • Consideration as to whether replacement of Fe 0 [17] with emulsified n/m-Fe 0 or with n-Fe 0 will give a higher level of desalination and will allow the amount of Fe used to be decreased. Another of the objectives of this study is to assess whether Fe 0 could be replaced by n-Fe 0 in a future commercial unit.
This study contains two appendices, Appendix A, and Appendix B. Appendix A provides a summary of the different desalination approaches. Appendix B, Tables A1 and A2, provide the systematic dataset used in this study.
Hydrology 2022, 9, x FOR PEER REVIEW 4 of 43  A single ZVI charge to be able to process multiple batches (or volumes) of water without loss of activity. Commercial ZVI water remediation plants, removing dissolved organic matter (DOM), organic solutes, and metals have demonstrated: o >3.8 million m 3 water processed t −1 Fe 0 to remove DOM in a 22,500 m 3 d −1 reactor [17]. o >24,000 m 3 water processed t −1 Fe 0 to remove organic matter in a 60,000 m 3 d −1 reactor [17]. o >466 m 3 water processed t −1 Fe 0 to remove arsenic in a 9.6 m 3 d −1 reactor [17]. o >17,356 m 3 water processed t −1 Fe 0 to remove arsenic in a 5.2 m 3 d −1 reactor [17]. o >2500 m 3 water processed t −1 Fe 0 to remove gold in a 30 m 3 d −1 reactor [17].
To date, a single ZVI charge in an experimental ZVI desalination reactor has not been able to process >14,000 m 3 water t −1 Fe 0 . One of the objectives of this study is to determine if the amount of water that can be processed by ZVI, can be increased to >42,000 m 3 waterprocessed t −1 Fe 0 without a loss of desalination activity.  A method or process to be established and incorporated in a plant, which allows the highly variable product outcome from each batch to be compensated for. Another of the objectives of this study is to assess the product outcome variability and determine how the product quality problems associated with this variability may be overcome in a future commercial unit.  Consideration as to whether replacement of Fe 0 [17] with emulsified n/m-Fe 0 or with n-Fe 0 will give a higher level of desalination and will allow the amount of Fe used to be decreased. Another of the objectives of this study is to assess whether Fe 0 could be replaced by n-Fe 0 in a future commercial unit.
This study contains two appendices, Appendix A, and Appendix B. Appendix A provides a summary of the different desalination approaches. Appendix B, Tables A1 and A2, provide the systematic dataset used in this study.

Reactor
The reactor process flow used in this study is summarized in Figure 2. The reactor is filled from a saline water feed tank. Product water is discharged to a product water tank. A cartridge containing ZVI is placed in the reactor. An air compressor is used to pressurize and fill (20 L) a pressured (0.8 MPa) gas tank with air. Air contained in the gas tank is discharged through the reactor (Figure 2). The product air is vented into the atmosphere.
The batch reactor ( Figure 2) was constructed using a tank with a u-PVC shell. It had a 0.45 m diameter and contained 240 L of saline water. The water column depth exceeded 1.5 m. An 8 mm O.D. Cu 0 conduit connected the gas tank ( Figure 2) to a brass gas distributer. The brass gas distributer was placed at the base of the batch reactor. A 4 cm diameter, 0.7 m long, open-ended ABS cartridge containing 0.4 kg of steel wool (Fe 0 ) was placed in the reactor ( Figure 3). The cartridge was placed within a cartridge holder (Figure 4). An air compressor was used to supply 60 L air h −1 to the saline water.

Reactor
The reactor process flow used in this study is summarized in Figure 2. The filled from a saline water feed tank. Product water is discharged to a product w A cartridge containing ZVI is placed in the reactor. An air compressor is used to and fill (20 L) a pressured (0.8 MPa) gas tank with air. Air contained in the g discharged through the reactor (Figure 2). The product air is vented into the at The batch reactor ( Figure 2) was constructed using a tank with a u-PVC sh a 0. 45

Reactor
The reactor process flow used in this study is summarized in Figure 2. The reactor is filled from a saline water feed tank. Product water is discharged to a product water tank. A cartridge containing ZVI is placed in the reactor. An air compressor is used to pressurize and fill (20 L) a pressured (0.8 MPa) gas tank with air. Air contained in the gas tank is discharged through the reactor (Figure 2). The product air is vented into the atmosphere.
The batch reactor ( Figure 2) was constructed using a tank with a u-PVC shell. It had a 0.45 m diameter and contained 240 L of saline water. The water column depth exceeded 1.5 m. An 8 mm O.D. Cu 0 conduit connected the gas tank ( Figure 2) to a brass gas distributer. The brass gas distributer was placed at the base of the batch reactor. A 4 cm diameter, 0.7 m long, open-ended ABS cartridge containing 0.4 kg of steel wool (Fe 0 ) was placed in the reactor ( Figure 3). The cartridge was placed within a cartridge holder ( Figure 4). An air compressor was used to supply 60 L air h −1 to the saline water.

Steel Wool
The steel wool was prepared prior to use by soaking the steel wool in 10 L of water containing FeSO 4 and urea (1:2.5 by weight (7 g L −1 )) for 24 h. The steel wool has a structural porosity of >80%. The 4 cm diameter, 0.7 m long ABS cartridges were structured with a high-permeability-filter wool plug at each end ( Figure 3). Each cartridge was filled with 0.4 kg of drained pretreated steel wool.
The pretreatment resulted in the iron fibres developing an irregular surface and a thin organic Fe(OH) x coating.

Zero-Valent Iron Characterization
All catalytic reactants have an inherent reactivity that can be defined in terms of an intrinsic forward rate constant, k ins . This intrinsic rate constant (change per unit time) is normally standardized in batch ZVI studies to a particle surface area (a s ) of 1 m 2 g −1 and a particle concentration (P w ) of 1 g L −1 . The observed forward rate constant (change s −1 ) in a batch flow (or continuous flow) reactor is then calculated [17,18] as Over the last 30 years, the principal focus of catalytic and adsorption ZVI studies, in improving k obs has been to concentrate on reducing particle size and increasing particle shape irregularity to increase a s . This has the effect of increasing the number of available catalytic and adsorption sites for a specific particle size.
The underlying assumptions are as follows: • At any specific time, a proportion of the unoccupied sites will be available for either adsorption or catalytic activity. • All catalytic and adsorption activity occurs at sites on the n-Fe 0 .
Laboratory catalyst and adsorbent studies, which have focused on this aspect of ZVI remediation, and are trying to develop a new form of n-Fe 0 , commonly use TEM, SEM-EDX, BET, XRD, XRF, microprobe, or similar analytical tools to define their product. They assume that k obs will increase if a s increases.
Another group of laboratory studies have focused specifically on increasing the ability of a site to operate as a catalyst or adsorbent. These studies tend to focus on increasing the porosity of particles [17], mixing metal combinations [17,18], and controlling the dispersion of promoters on and within the particles [17]. Their specific objective is to increase k ins [17,18].
Fe 0 , m-Fe 0 , and n-Fe 0 are highly reactive in the presence of oxygen and water [17]. n-Fe 0 is pyrophoric. Placement of Fe 0 , m-Fe 0 , and n-Fe 0 in acidic water (pH ≤ 7) will result in the dissolution of the particles to release Fe 2+ ions. In alkali water (pH = 7), the Fe 0 , m-Fe 0 , and n-Fe 0 will initially be coated in white rust (Fe(OH) 2 ) [17]. Further oxidation will result in the formation of green rust and various other iron oxyhydroxides [17].
Remediation studies, which seek to use Fe 0 , m-Fe 0 , and n-Fe, as an adsorbent or catalyst, will commonly pretreat the Fe 0 to provide a surface coating to retard the formation of Fe 2+ [17]. These coatings can also remove the pyrophoric characteristic of n-Fe 0 and improve safety when handling this material [17].
Traditionally, steel wool is used to treat water by altering its Eh and pH, and thereby affects remediation by altering redox equilibria of pollutants. The release of Fe 2+ ions has a side effect which results in the steel wool being an effective remediator of pollutants by coagulation. This results in both biota and pollutants being removed by the Fe x O y H z coagulants [19,20]. A number of pretreatment and characterization tools have been developed for steel wool [21]. Pretreatment does not enhance the reactivity of steel wool [21] but can reduce the rate of Fe 2+ dissolution [22].
If, in this study, the steel wool was to be used to produce Fe 2+ ions, then this could be characterized by measuring the rate of Fe 2+ release. ZVI desalination is considered to be a catalytic process [17]. Therefore, the pretreatment, and catalyst placement in the reactor is designed to decrease the probability that Fe 2+ ions produced by the steel wool are released into the main water body. The cartridge design used in this study is intended to ensure that the majority of any Fe 2+ ions formed are retained in the catalyst cartridge.
There is no consensus on how ZVI should be characterized for specific water remediation studies. Examples of the different approaches taken by different research groups are provided in references [23,24].
Steel wool has a low a s , and a very low catalytic k ins for Na + ion and Cl − ion removal. These two factors would appear (Equation (2)) to render steel wool as being unsuitable for use as a desalination catalyst. However, steel wool is widely available and has a low cost, typically in the range USD 200-1000 t −1 . This compares with USD 10,000-100,000 t −1 for n-Fe 0 .
A typical characterization study of Fe using one or more of TEM, SEM-EDX, BET, XRD, or XRF may cost in the range USD 10,000-100,000. While this may serve a purpose in defining a s in n-Fe 0 , it is not cost-effective for steel wool. If you purchase your steel wool from a supplier [A] at USD 200 t −1 , there is no point in spending >USD 10,000 characterizing its morphology and chemistry to determine if it might be, say, 10% more effective than if the steel wool was purchased from a supplier [B] at USD 300 t −1 when placed in a specific reactor process.

ZVI Desalination Characteristics
Equation (2) indicates that increasing the residence time of the water in the reaction environment will increase k obs . In ZVI desalination reactors, this is achieved through the use of either a water recycle loop, where water is continually circulated through a fluidized n-Fe 0 bed (e.g., US Patent US10919784B2), or a diffusion reaction environment, where water is continually circulated between a storage tank or reservoir and a reaction area containing a static ZVI bed (e.g., UK Patent GB 2520775A), or the use of long reaction times (e.g., US Patent US10919784B2; UK Patent GB 2520775A).
In a fixed bed or fluidized bed reactor, increasing P w will increase the water residence time within the reaction environment for a specific water flow rate (Equation (2)). In a ZVI desalination diffusion reactor, k obs increases until P w exceeds a critical level. Thereafter, further increases in P w have little or no impact on P w (UK Patent GB 2520775A).
In a ZVI desalination reaction environment, changing the particle size (i.e., a s ) markedly affects the rate and amount of desalination. Comparisons, as a function of particle size, are provided for a fluidized bed environment in US Patent US10919784B2 and for a diffusion bed environment in UK Patent GB 2520775A.
A reactor operating with a Fe 0 particle size of around 50 nm (a s of around 20 m 2 g −1 ), and with a similar P w , will produce markedly different desalination outcomes depending on the reactor type and reactor process used.
For example, a static water, static bed diffusion reactor may take 100 days to achieve around an 80% desalination (e.g., UK Patent GB 2520775A). The same particle characteristics may take around 16 h to achieve >60% desalination in a fluidized bed reactor incorporating fluid recycle (US Patent US10919784B2); around 10 min to obtain effective seawater desalination in a continuous flow contact flow reactor (Spanish patent ES2598032); or around 30-60 min to obtain 50-100% removal in an emulsified bubble column reactor (US Patents, US9624113B2; US9828258B2; US2018/0009678A). In each example, the intrinsic characteristics of the n-Fe 0 used are similar.
The patent literature demonstrates that the major control on k obs in the desalination reaction is the reactor process used. This change can reduce the time required to achieve Hydrology 2022, 9, 128 8 of 41 80% desalination from >100 days to less than 1 h. This process change results in an increase in k obs by a factor of around 4.8 × 10 3 .
A common feature of the ZVI desalination processes used in US Patents US9624113B2; US9828258B2; US2018/0009678A; US10919784B2, and Spanish Patent ES2598032, is that they create a turbulent environment, and k obs , increases as a function of turbulence.

Steel Wool Catalytic Design Assumptions
Most n-Fe 0 studies (e.g., [18]) assume that Fe 0 operates by chemically adsorbing the pollutant, i.e., In 1905, Professor P. Sabatier (French Patent 355,900) demonstrated that Fe 0 can remove and modify chemicals by a physical adsorption process, as opposed to a chemical bonding process.
A change in the reaction environment resulted in physical desorption, leaving the Fe 0 unchanged.
[Fe a @D b ] +cF = aFe 0 + D b F c , In this reaction series, D to F are chemicals, and the subscripts are their molar concentrations. This type of Fe catalysis is known as [Fe@D] catalysis, where [D] is the physically adsorbed species. In this type of adsorption/catalysis, the catalytic sites will release the adsorbed species as a product in response to minor flow and pressure wave changes in the environment. In this type of catalysis, k obs will increase as a function of increased turbulence.
Since this is a chemophysical process, where each adsorbed species rests on the catalytic material without necessarily blocking the adsorption site, each mole of catalytic site [a] has the potential to adsorb more than one mole (D b ) of the adsorbed species.
For example, if the steel wool contained one site per m 2 and was left in quiescent water for 365 days, then no effective ion removal would be observed. If a pressure wave was applied to the steel wool, with an amplitude of x and a frequency of y, then the proportion of sites (z) which adsorb and release [D] will be a function of [y], where 0 < z < 1. If the wave frequency is 10 per second, then for a steel wool containing one site m −2 , the number of exchanges in a year becomes [[z] x 315,360,000]. This represents an increase in catalytic activity by a factor of 10 9 .
This study establishes, through the use of a control experiment, that when [z] = 0, no desalination occurs. However, when [z] is greater than 0, desalination occurs. The steel wool is considered to operate as an [Fe a @D b ] catalyst.
UK Patent GB2470764A established that if Fe 0 is placed in an oscillating pressure environment, then the adsorption rate, k ods , equation was rewritten as k ods = k ins a s P w P u , (6) and the desorption rate, k des , equation was rewritten as k des = k ins a s P w P d0 , P u is the magnitude of the positive pressure change from the wave trough to the wave crest. P d is the magnitude of the negative pressure change from the wave crest to the wave trough. GB2470764A established that for a constant value of k ins , a s, and P w , this process could increase the magnitude of the forward rate constant by several orders of magnitude.
The forward rate constant is essentially dependent on the pressure wave frequency, shape, and amplitude. All of these items can be controlled by design in a reactor, and create the potential for the operator to adjust the forward rate constant by adjusting the pressure wave form. Under this paradigm, changes in the design pressure oscillation become the controlling reaction parameter.
This study uses a pressure wave generated by the gas distributor to produce significant desalination over a 3 h period and makes recommendations as to how the pressure wave design may be improved to increase the forward desalination rate constant.
Elsewhere, in other processes, use of catalytic pressure swing adsorption desorption in catalytic reactors has demonstrated substantial increases in k obs [25,26].

ZVI Reuse Assumptions
The process design used in US Patents US9624113B2; US9828258B2; US2018/0009678A; and Spanish Patent ES2598032 results in the ZVI being removed after a single batch operation. The process design used in US patent US10919784B2 allows for the possibility that a single batch of ZVI could be used to desalinate multiple batches of water. The process design used in UK Patent GB 2520775A allowed for the reuse of a batch of ZVI with multiple batches of water. The development of the design issues and associated characterizations required to allow the use of ZVI to process multiple batches of water in a diffusion reactor are documented as follows: (i) initial trials; (ii) expansion of process types trialled and associated mass balance analyses; (iii) analysis of diffusion options for desalinating a saline aquifer; (iv) development of an initial desalination model that would allow a batch of ZVI to process multiple water volumes in a diffusion environment [27].
This study extends and redefines the pre-existing hypotheses associated with ZVI reuse during desalination.

Synthetic Saline Water
The sequential reactor trials processed synthetic saline water which was manufactured by dissolving halite in natural spring water. The water was sourced from a natural freshwater spring at National Grid Reference: NO 02817 14399; 56 • 18 43" N, 003 • 34 21" W. The water was abstracted from fractured andesitic rocks and was placed in a 5 m 3 storage tank (

Measurements
The following parameters were measured for each feed water and product water: pH, Eh, EC, Cl − ion concentration, Na + ion concentration, and temperature.

Instruments
The instruments used were: additional (third) corroboration of desalination through the use of photomicrographs of evaporated feed and product water and the residual solids concentration of the evaporate. These photomicrographs (Appendix B, Figures A5-A7) provide a qualitative corroboration of the ISE results, and indicate that a full sample (e.g., 1 to 10 L) evaporation of the feed and product waters, followed by an evaporated weight measurement, of the amount of residual NaCl would probably confirm the ISE results.
In the case of the 70 sequential reactor trials, for each trial, the following salinity measurements are provided (in Appendix B, Table A1) for both the feed water and product water. They are ISE measurements of Cl − ion concentration and Na + ion concentration; EC measurements are provided for the feed and product waters as a corroboration measurement.

Reaction Time
Each water batch was placed in the reactor for 3 h before being replaced. A total of 70 batches of water were processed with the single ZVI charge. The product water was moved to a holding tank ( Figure 2) before being used to irrigate plants.

Control Reactor
A primary control reactor was placed in operation, containing no gas distributor and five 4 cm diameter, 0.7 m long ABS cartridges ( Figure 3); each cartridge contained 0.4 kg of steel wool (Fe 0 ). The product water salinity was sampled after 3 h, 24 h, 30 days, 120 days, and 365 days. Each product water sample had the same salinity as the feed water sample (8 g NaCl L −1 ). Therefore, any changes in salinity in the product water in the replication experiments is due to the presence of air and the partial pressure changes, resulting from bubbling air, through the water.
A second control experiment was undertaken, where the reactor was filled with water without the presence of a ZVI cartridge. Air was bubbled, at a rate of 60 L m −1 , through water containing 10 g NaCl L −1 . No desalination was observed over a 6 h period. Therefore, any changes in salinity in the product water in the replication experiments are due to the presence of zero-valent iron. In this study, desalination is primarily measured using Approach 1. Approaches 2, 3, and 5, are used to provide a confirmatory check. In agricultural holdings (and government regulatory authorities), Approach 3 is the most widely used tool to check the salinity of irrigation water. ISEs provide accurate, rapid measurements of actual specific ion concentrations in the water and are therefore the preferred measurement tool.
EC meters are commonly marketed as salinity measurement tools. If pure salt (NaCl) is added to pure water, the salinity (g NaCl L −1 ) will approximate to between 0.5 EC (measured as mScm −1 ) and 0.54 EC. When Fe 0 is added to pure water, the EC values commonly rise due to the dissolution of Fe 0 to form Fe n+ ions. In the original 2010 ZVI desalination discovery studies, EC actually rose, while the NaCl content of the water reduced, due to the presence of Fe n+ ions [15]. In this study, the pretreatment of ZVI is used to reduce the rate of dissolution of Fe 0 to form Fe n+ ions.
In most natural groundwater, the relationship between salinity (g NaCl L −1 ) and EC will approximate to between 0.7 EC (measured as mScm −1 ) and 1.5 EC. This relationship is set by the water composition. A further complication is that the EC for a specific ion-rich water may change with temperature.
In water containing organic species, a further complication arises. Organic species, particularly carboxylic acids, show a pattern whereby initially the EC rises as the acid concentration increases. The EC then declines, with increasing concentration, to a low equilibrium level. ZVI will, in addition to removing NaCl, remove organic species. This creates a mismatch between the ISE results and the EC results.
Examples of this type of mismatch are provided in the results for Trials RT1 to RT7. In these examples, the ISE results indicate a more substantive desalination than the EC results (Appendix B, Table A2). In the case of Trials RT1 to RT3, this study provides an additional (third) corroboration of desalination through the use of photomicrographs of evaporated feed and product water and the residual solids concentration of the evaporate. These photomicrographs (Appendix B, Figures A5-A7) provide a qualitative corroboration of the ISE results, and indicate that a full sample (e.g., 1 to 10 L) evaporation of the feed and product waters, followed by an evaporated weight measurement, of the amount of residual NaCl would probably confirm the ISE results.
In the case of the 70 sequential reactor trials, for each trial, the following salinity measurements are provided (in Appendix B, Table A1) for both the feed water and product water. They are ISE measurements of Cl − ion concentration and Na + ion concentration; EC measurements are provided for the feed and product waters as a corroboration measurement.

Reaction Time
Each water batch was placed in the reactor for 3 h before being replaced. A total of 70 batches of water were processed with the single ZVI charge. The product water was moved to a holding tank ( Figure 2) before being used to irrigate plants.

Control Reactor
A primary control reactor was placed in operation, containing no gas distributor and five 4 cm diameter, 0.7 m long ABS cartridges ( Figure 3); each cartridge contained 0.4 kg of steel wool (Fe 0 ). The product water salinity was sampled after 3 h, 24 h, 30 days, 120 days, and 365 days. Each product water sample had the same salinity as the feed water sample (8 g NaCl L −1 ). Therefore, any changes in salinity in the product water in the replication experiments is due to the presence of air and the partial pressure changes, resulting from bubbling air, through the water.
A second control experiment was undertaken, where the reactor was filled with water without the presence of a ZVI cartridge. Air was bubbled, at a rate of 60 L m −1 , through water containing 10 g NaCl L −1 . No desalination was observed over a 6 h period. Therefore, any changes in salinity in the product water in the replication experiments are due to the presence of zero-valent iron.

Data and Data Handling
The primary data generated in this study are placed in Appendix B, Tables A1 and A2. Graphical cross plots from this dataset were constructed using MS Excel 2019. All trendlines, regression equations, and coefficients of correlation (R 2 ) were calculated by MS Excel's trendline function. Where there was a choice between different regression trend fits (e.g., linear or exponential), the trend fit, which gave the largest coefficient of correlation, was selected. The purpose of including the regression correlation equations is to allow use in modelling and help identify trends. Probability distributions associated with catalytic pressure swing adsorption desorption reactions have a wide outcome range. With this type of graph, the regression best fits can require a polynomial equation with five or six elements. These probability equations for desalination can be directly programmed into a Monte Carlo desalination analysis for reactor or plant sizing and operational analyses.

Results
The quantitative systematic results (Na + ion concentration, Cl − ion concentration, EC, pH, Eh, and temperature) from 70 sequential replication trials are provided in Appendix B ,  Table A1. The quantitative systematic results (Na+ ion concentration, Cl − ion concentration, EC, pH, Eh, and temperature) from 7 n-Fe 0 and m-Fe 0 trials are provided in Appendix B, Table A2. These results are examined and analysed further in this section.

Salinity
Salinity measurements based on EC and ion analysis show similar trends ( Figure 6). These trends confirm that EC changes can be used to measure desalination. The measured regression relationship for seawater between EC and salinity, g NaCl L −1 , was

Results
The quantitative systematic results (Na + ion concentration, Cl − ion concentration, EC, pH, Eh, and temperature) from 70 sequential replication trials are provided in Appendix B, Table A1. The quantitative systematic results (Na+ ion concentration, Cl − ion concentration, EC, pH, Eh, and temperature) from 7 n-Fe 0 and m-Fe 0 trials are provided in Appendix B, Table A2. These results are examined and analysed further in this section.

Salinity
Salinity measurements based on EC and ion analysis show similar trends ( Figure 6). These trends confirm that EC changes can be used to measure desalination. The measured regression relationship for seawater between EC and salinity, g NaCl L −1 , was EC = electrical conductivity, mScm −1 ; R 2 = coefficient of correlation. The measured regression relationship for the partially desalinated product water between EC and salinity, g NaCl L −1 , was

Redox Relationships
The redox parameters for the feed water and the product water were similar ( Figure  7). The product water pH was increased slightly (by about 2%), while the product water Eh was decreased by about 6%. This indicates that the dominant reaction in the water may have been EC = electrical conductivity, mScm −1 ; R 2 = coefficient of correlation. The measured regression relationship for the partially desalinated product water between EC and salinity, g NaCl L −1 , was NaCl g L −1 = 0.7591 EC, R 2 = 95.62%,

Redox Relationships
The redox parameters for the feed water and the product water were similar ( Figure 7). The product water pH was increased slightly (by about 2%), while the product water Eh was decreased by about 6%. This indicates that the dominant reaction in the water may have been Hydrology 2022, 9, x FOR PEER REVIEW 13 of 43

Ion Compositions
The concentrations of both Na + and Cl − ions changed within the reactor ( Figure 8). The statistical regression relationship between the feed and product ion concentrations followed a polynomial function. This relationship implies that a larger proportion of ions are removed from higher salinity feed water.

Ion Compositions
The concentrations of both Na + and Cl − ions changed within the reactor ( Figure 8). The statistical regression relationship between the feed and product ion concentrations followed a polynomial function. This relationship implies that a larger proportion of ions are removed from higher salinity feed water.

Desalination
The feed water contains a higher ion concentration than the product water (Table 1, Figures 6 and 8). This indicates that desalination occurred. At the end of 70 trials, there was no noticeable evidence that the amount of desalination in each successive trial decreased with time ( Figure 9). The amount of desalination that occurred was independent of the feed water salinity (when the water contained <10 g NaCl L −1 ) or showed a weak trend of increasing desalination with increasing feed water salinity (when the water contained >15 g NaCl L −1 ) (Figures 6 and 9).

Desalination
The feed water contains a higher ion concentration than the product water (Table 1, Figures 6 and 8). This indicates that desalination occurred. At the end of 70 trials, there was no noticeable evidence that the amount of desalination in each successive trial decreased with time ( Figure 9). The amount of desalination that occurred was independent of the feed water salinity (when the water contained <10 g NaCl L −1 ) or showed a weak trend of increasing desalination with increasing feed water salinity (when the water contained >15 g NaCl L −1 ) (Figures 6 and 9). Table 1. Statistical parameters. F w = feed water; P w = product water. T = temperature; SD = standard deviation; Min = minimum; Max = maximum; 1st = first quartile; 3rd Q = third quartile. Dataset: Appendix B Table A1.

Catalytic Removal of NaCl
The catalytic removal of NaCl results in the concentration of NaCl, NaOH, NaHCO3, and NaClO in the porosity, associated with Fe 0 corrosion products such as layered double hydroxides (LDH, green rust) and FeOOH.

Catalytic Removal of NaCl
The catalytic removal of NaCl results in the concentration of NaCl, NaOH, NaHCO 3 , and NaClO in the porosity, associated with Fe 0 corrosion products such as layered double hydroxides (LDH, green rust) and FeOOH.
The associated reactions are complex, and the nature of the complexity was described in reference [27]. This complexity is summarized in Figure 10. The surface of the ZVI corrodes to be covered in LDH (Fe(OH) x ) and FeOOH. The LDH adjacent to the ZVI has a high permeability, which allows ion migration within the LDH. The LDH adjacent to the water develops a high proportion of dead-end pores. These dead-end pores act as a depository for removed NaCl [17]. OH − + Na + + Cl − ions enter the LDH from the water.

Catalytic Removal of NaCl
The catalytic removal of NaCl results in the concentration of NaCl, NaOH, NaHCO3, and NaClO in the porosity, associated with Fe 0 corrosion products such as layered double hydroxides (LDH, green rust) and FeOOH.
The associated reactions are complex, and the nature of the complexity was described in reference [27]. This complexity is summarized in Figure 10. The surface of the ZVI corrodes to be covered in LDH (Fe(OH)x) and FeOOH. The LDH adjacent to the ZVI has a high permeability, which allows ion migration within the LDH. The LDH adjacent to the water develops a high proportion of dead-end pores. These dead-end pores act as a depository for removed NaCl [17]. OH − + Na + + Cl − ions enter the LDH from the water.
This model was developed [27] to explain the initial ZVI desalination observations, where large amounts of NaCl were found to be concentrated in the LDH. It is unable to account for the substantially larger volumes of NaCl removed during the course of this study (Appendix B Table A1).
Therefore, an alternative explanation for the removal of the NaCl is required.  This model was developed [27] to explain the initial ZVI desalination observations, where large amounts of NaCl were found to be concentrated in the LDH. It is unable to account for the substantially larger volumes of NaCl removed during the course of this study (Appendix B Table A1).
Therefore, an alternative explanation for the removal of the NaCl is required.

Role of Radicals
In an environment where oxygen is bubbled through water, a portion of the oxygen will interact with the components in the water to form radicals. Oxygenation may create a reaction of the form: This interpretation indicates that ZVI desalination could be a variant of the Fenton reaction, where The Eh:pH relationship for the product water is consistent with a ratio for [ P [O 2 ]:moles H 2 O 2 ] of between 1 and 5000 ( Figure 11), where the ratio [ P [O 2 ]:moles H 2 O 2 ] decreases as the Eh decreases [28]: Eh, volts = 0.

Sodium and Chloride Ion Removal
The sodium and chloride ions showed high rates of removal until Trial 40 ( Figure  12). In a few trials, individual ions showed no removal, or a slight increase in concentration, within the product water (Appendix B Table A1). This may indicate that adsorption, followed by desorption, from the ZVI can occur.
There is a linear relationship between Cl − removal and Na + removal ( Figure 13). Sodium ion removal is favoured by increasing Eh (Figure 13). Chloride ion removal (and total desalination) is favoured by increasing Eh above 300 mV and by decreasing Eh below 300 mV ( Figure 13). These relationships are interpreted as indicating that Cl − may be removed as follows: This relationship may imply that Cl − ions are removed as HClO or ClO • .

Sodium and Chloride Ion Removal
The sodium and chloride ions showed high rates of removal until Trial 40 ( Figure 12). In a few trials, individual ions showed no removal, or a slight increase in concentration, within the product water (Appendix B Table A1). This may indicate that adsorption, followed by desorption, from the ZVI can occur.  There is a linear relationship between Cl − removal and Na + removal ( Figure 13). Sodium ion removal is favoured by increasing Eh (Figure 13). Chloride ion removal (and total desalination) is favoured by increasing Eh above 300 mV and by decreasing Eh below 300 mV ( Figure 13). These relationships are interpreted as indicating that Cl − may be removed as follows:

Predicting NaCl Removal
The control trial indicated that no desalination occurred when no air was bubbled through the reactor. Therefore, the desalination is either a direct result of a chemical reaction involving O2 and Na + + Cl − ions, or a function of the pressure wave created by the rising air within the bubble column ( Figure 14).

Predicting NaCl Removal
The control trial indicated that no desalination occurred when no air was bubbled through the reactor. Therefore, the desalination is either a direct result of a chemical reaction involving O 2 and Na + + Cl − ions, or a function of the pressure wave created by the rising air within the bubble column ( Figure 14).

Pressure Swing Adsorption/Desorption
The results of each trial are variable ( Figure 12). Sequential trials give different outcomes ( Figure 12). This is a characteristic of a pressure swing adsorption (PSA) process, where a loose adsorption onto catalytic material occurs as the pressure rises. This is followed by a rapid desorption of a product from the catalytic material as the pressure drops. Pressure fluctuations and pressure waves are created by the rising gas bubbles within the reactor ( Figure 14).
Increases in pressure are associated with adsorption. Decreases in pressure are associated with desorption. Bubbling air through water creates a vertically rising pressure wave ( Figure 14) where the pressure is maximized above the bubble and minimized in its wake. The water displaced by the rising air (60 kg h −1 ) creates a horizontal pressure wave within the reactor ( Figure 14). The pressure fluctuations associated with this lateral pressure wave interact with the ZVI cartridge. The amplitude of these waves in the reactor varied between 1 and 4 cm. This would have applied an oscillating driving force (∆P) of between +/− [1000 to 4000 Pa] to the ZVI within the cartridge.
The fluid flow rate [QJ] between the ZVI and the water body can be defined using the general flux equation: where kJ = permeability, m 3 m −2 s −1 Pa −1 . In the reactor, the ΔP wave form is pulse initiated and is associated with a multitude of pulse sources. It can, therefore, take a multiplicity of forms, and will include interference patterns. Figure 14 demonstrates that higher perio-

Pressure Swing Adsorption/Desorption
The results of each trial are variable ( Figure 12). Sequential trials give different outcomes ( Figure 12). This is a characteristic of a pressure swing adsorption (PSA) process, where a loose adsorption onto catalytic material occurs as the pressure rises. This is followed by a rapid desorption of a product from the catalytic material as the pressure drops. Pressure fluctuations and pressure waves are created by the rising gas bubbles within the reactor ( Figure 14).
Increases in pressure are associated with adsorption. Decreases in pressure are associated with desorption. Bubbling air through water creates a vertically rising pressure wave ( Figure 14) where the pressure is maximized above the bubble and minimized in its wake. The water displaced by the rising air (60 kg h −1 ) creates a horizontal pressure wave within the reactor (Figure 14). The pressure fluctuations associated with this lateral pressure wave interact with the ZVI cartridge. The amplitude of these waves in the reactor varied between 1 and 4 cm. This would have applied an oscillating driving force (∆P) of between +/− [1000 to 4000 Pa] to the ZVI within the cartridge.
The fluid flow rate [Q J ] between the ZVI and the water body can be defined using the general flux equation: where k J = permeability, m 3 m −2 s −1 Pa −1 . In the reactor, the ∆P wave form is pulse initiated and is associated with a multitude of pulse sources. It can, therefore, take a multiplicity of forms, and will include interference patterns. Figure 14 demonstrates that higher periodicity waves were supplemented with shorter periodicity waves. The wave structures are composite. If adsorption is a function of ∆P, then for a simple model, the adsorption rate [J A ] can be defined as where k A = adsorption constant, sites m −2 s −1 Pa −1 . Similarly, the desorption rate [J D ] can be defined as J D = k D ∆P; when ∆P < 0, where k D = desorption constant, sites m −2 s −1 Pa −1 . As a first approximation, a simple model would provide ∆P with a constant wavelength (e.g., 1 s) and a constant amplitude (e.g., 4000 Pa). This would create a situation where ∆P associated with adsorption is +4000 Pa and ∆P associated with desorption is −4000 Pa. If k A = k D , then adsorption will equal desorption.
If more than one species can be adsorbed by the ZVI when the pressure increases, then only a proportion of the total number of sites will be open for occupation by a specific species. Each species will have a different value of k A and will be adsorbed by the ZVI at different rates. Each species has a different value of k D and will desorb from the ZVI at different rates in response to a change in ∆P. This will create a situation if a desorbed species results in desalination, where the forward rate constant for desalination decreases with increasing reaction time.
This A batch flow, bubble column, static bed, recirculating diffusion reactor was operated with a N 2 gas feed. It displayed long periods of no desalination, followed by short abrupt periods of substantial desalination. This observation is consistent with a slow site adsorption of N 2 followed by rapid desorption of the product.

Role of O 2 and CO 2 in Pressure Swing Adsorption/Desorption
In a pressure swing environment, the adsorption of O 2 by Fe 0 can be >4.5 times greater than the adsorption of N 2 [29].
Desorption of O 2 creates fast adsorption sites [S 1 ] which result from O 2 desorption. These fast adsorption sites [S 1 ] remove CO 2 from the air. The adsorbed CO 2 then reacts with modified reaction sites [S 2 ] or unmodified adsorption sites [S 3 ] to form bicarbonates [30]. A possible reaction sequence of adsorption and desorption, in response to pressure changes, is provided by Equations (17)-(21): These relationships imply that active CO 2 removal will occur from the feed gas. The reactor structure used in this study does not allow this concept to be tested ( Figure 2). The process identified in Equations (19)-(21) is a CO 2 -LDH carbon capture approach (Appendix A). This concept can be tested in a batch flow, bubble column, static bed, recirculating diffusion reactor.

Role of Na + and Cl − Ions in Pressure Swing Adsorption/Desorption
The positive correlation between Na + ion removal and Cl − ion removal ( Figure 13) indicates that if they directly interact with the Fe 0 , then their rates of adsorption and desorption are similar.

Desalination Probability Distribution
The statistical feed and product water composition data are provided in Table 1. The nonparametric NaCl concentration and desalination probability distributions are provided in Figure 15. The mean and standard deviation provide an assessment of the aggregated desalination, associated with a series of batch trials (in accordance with the central limit theory).
Hydrology 2022, 9, x FOR PEER REVIEW 19 of 43 provided in Figure 15. The mean and standard deviation provide an assessment of the aggregated desalination, associated with a series of batch trials (in accordance with the central limit theory). Figure 15d provides a regression equation that can be used in a Monte Carlo analysis to predict the desalination outcome of any specific water batch, where [x] is a random number between 0 and 1.
Each water batch will have a different value of [x]. Aggregating the water batches will produce an aggregated water product with an average desalination value similar to that provided in Table 1 and within the range predicted by Figure 15e,f.
Every saline water has a different water composition and will react slightly differently. Consequently, the results shown in Figure 15 are only directly applicable to the source water used in this study. Other water compositions from other sources may give a higher or lower desalination value.   Table A1. Mean desalination = 24.5%; standard deviation = 22.8%; minimum = 0%; 1st quartile = 9.2%; median = 14.3%; 3rd quartile = 33.9%; maximum = 91%. The methodology for use of the expected normal order statistic, u, (which is sometimes termed Z p ) is provided in reference [31]. Figure 15d provides a regression equation that can be used in a Monte Carlo analysis to predict the desalination outcome of any specific water batch, where [x] is a random number between 0 and 1.
Each water batch will have a different value of [x]. Aggregating the water batches will produce an aggregated water product with an average desalination value similar to that provided in Table 1 and within the range predicted by Figure 15e,f. Every saline water has a different water composition and will react slightly differently. Consequently, the results shown in Figure 15 are only directly applicable to the source water used in this study. Other water compositions from other sources may give a higher or lower desalination value.

Forward Rate Constant
Bubbling a gas through water creates a series of cyclic pressurization and depressurization waves (Figure 14). In a catalytic PSA system, the ions are adsorbed during pressurization. The products are removed and released during the associated counter depressurization.
The forward rate constant, k f , takes the form [17] where t = reaction time, s; M p = moles product (NaCl removed); M ZVI = moles ZVI. The rate constant is highly variable (Figure 16). It increases with increasing feed water salinity ( Figure 16).

Forward Rate Constant
Bubbling a gas through water creates a series of cyclic pressurization and depressurization waves (Figure 14). In a catalytic PSA system, the ions are adsorbed during pressurization. The products are removed and released during the associated counter depressurization.
The forward rate constant, kf, takes the form [17] where t = reaction time, s; Mp = moles product (NaCl removed); MZVI = moles ZVI. The rate constant is highly variable (Figure 16). It increases with increasing feed water salinity ( Figure 16).

Predicting the Impact of ZVI Desalination on Crop Yield
All crops and crop varieties are adversely affected by increasing water salinity. The results (Figure 15) indicate that a water product constructed from 70 batches of water will show an expected mean desalination of 24.5%.
For most crops, there is a linear relationship between soil water salinity and crop yield. Soil water salinity is typically 10-30% greater than the irrigation water salinity.
If a crop has the potential, when the soil salinity is <3 g L −1 , to produce 8 t ha −1 and ceases to produce a measurable yield when the soil salinity is >10 g L −1 and the current soil salinity is 7 g L −1 , then the expected current crop yield is 3.4 t ha −1 .
If the current irrigation water salinity is 6.3 g L −1 , then the desalination distribution in Figure 15f indicates that there is a 50% probability of increasing the crop yield to >6 t ha −1 following irrigation with partially desalinated water ( Figure 17).

Predicting the Impact of ZVI Desalination on Crop Yield
All crops and crop varieties are adversely affected by increasing water salinity. The results (Figure 15) indicate that a water product constructed from 70 batches of water will show an expected mean desalination of 24.5%.
For most crops, there is a linear relationship between soil water salinity and crop yield. Soil water salinity is typically 10-30% greater than the irrigation water salinity.
If a crop has the potential, when the soil salinity is <3 g L −1 , to produce 8 t ha −1 and ceases to produce a measurable yield when the soil salinity is >10 g L −1 and the current soil salinity is 7 g L −1 , then the expected current crop yield is 3.4 t ha −1 .
If the current irrigation water salinity is 6.3 g L −1 , then the desalination distribution in Figure 15f indicates that there is a 50% probability of increasing the crop yield to >6 t ha −1 following irrigation with partially desalinated water ( Figure 17). yield. Soil water salinity is typically 10-30% greater than the irrigation water salinity.
If a crop has the potential, when the soil salinity is <3 g L −1 , to produce 8 t ha −1 and ceases to produce a measurable yield when the soil salinity is >10 g L −1 and the current soil salinity is 7 g L −1 , then the expected current crop yield is 3.4 t ha −1 .
If the current irrigation water salinity is 6.3 g L −1 , then the desalination distribution in Figure 15f indicates that there is a 50% probability of increasing the crop yield to >6 t ha −1 following irrigation with partially desalinated water ( Figure 17).   Partial desalination of water to produce a product water with a predictable salinity can only be effectively undertaken using a multi-train reactor.

Livestock
In many marginal areas with saline water, the principal agricultural product is livestock. Saline irrigation is used to grow animal feed crops to supplement grazing. Different animals have different degrees of tolerance to different water salinities. Their tolerance to water salinity can be split into three categories: (i) safe, where they are tolerant to the salinity and gain weight; (ii) suitable, where they can tolerate the salinity but will only gain weight at a slow rate; (iii) poor, where the high salinity will result in weight loss. The salinity ranges associated with these categories varies with animal type and variety. Figure 19 indicates the generic salinity tolerance ranges for a number of different types of animals.
The results of this study indicate that the partial desalination of livestock feed water, which had been unsuitable for pigs, horses, and cattle, may allow pigs and horses to be Partial desalination of water to produce a product water with a predictable salinity can only be effectively undertaken using a multi-train reactor.

Livestock
In many marginal areas with saline water, the principal agricultural product is livestock. Saline irrigation is used to grow animal feed crops to supplement grazing. Different animals have different degrees of tolerance to different water salinities. Their tolerance to water salinity can be split into three categories: (i) safe, where they are tolerant to the salinity and gain weight; (ii) suitable, where they can tolerate the salinity but will only gain weight at a slow rate; (iii) poor, where the high salinity will result in weight loss. The salinity ranges associated with these categories varies with animal type and variety.  (Table 1); orange arrow indicates the salinity removed by partial desalination (Table 1).

Figure 20.
Illustrative salinity tolerance of four livestock grass crops-these vary with agricultural practice, climate, soil type, soil moisture content, soil permeability, and plant variety. Solid black line = expected soil salinity if it is irrigated with water containing 6.9 g NaCl L −1 (Table 1); dotted black line = expected soil salinity if it is irrigated with partially desalinated water containing 4.77 g NaCl L −1 (Table 1). Red square indicates that a 33% increase in tall wheat grass yields is expected if irrigation involves partially desalinated water; green square indicates that a 75% increase in wheat grass and Bermuda grass yields is expected if irrigation involves partially desalinated water; blue square indicates that a 175% increase in barley (hay) yields is expected if irrigation involves partially desalinated water.

Effect of ZVI Particle Size on Desalination
Steel wool is a rigid material that will stay in the reactor when the reactor is emptied. It is therefore a suitable material for the processing of multiple batches of water.
It is well established, as demonstrated in references and patent [US10,919,784B2], that the forward rate constant for desalination increases as the ZVI particle size decreases. The reactor train was retested using m-ZVI and n-ZVI, where the ZVI was only used to process a single batch of water. Five of the seven trials used a seawater feed water and a nano- Figure 19. Tolerance of livestock to salinity-illustrative relationships. These vary with variety and region. Green = safe; blue = suitable; green = poor. Black line = average saline water salinity of 6.9 g L −1 ; dashed black line = average partially desalinated water salinity of 4.77 g L −1 ( Table 1); orange arrow indicates the salinity removed by partial desalination ( Table 1).
The results of this study indicate that the partial desalination of livestock feed water, which had been unsuitable for pigs, horses, and cattle, may allow pigs and horses to be grown and could supply emergency relief water for cattle ( Figure 19).
The yield of livestock feedstock, such as hay and silage, is a function of irrigation water salinity. Figure 20 illustrates some typical crop yield ranges. Integrating the desalination data in Table 1 with the yield ranges in Figure 20 indicates that the desalination demonstrated using the reactor train could, depending on the crop, result in an increased yield of between 33% and 175% ( Figure 20).  (Table 1); orange arrow indicates the salinity removed by partial desalination (Table 1).

Figure 20.
Illustrative salinity tolerance of four livestock grass crops-these vary with agricultural practice, climate, soil type, soil moisture content, soil permeability, and plant variety. Solid black line = expected soil salinity if it is irrigated with water containing 6.9 g NaCl L −1 (Table 1); dotted black line = expected soil salinity if it is irrigated with partially desalinated water containing 4.77 g NaCl L −1 (Table 1). Red square indicates that a 33% increase in tall wheat grass yields is expected if irrigation involves partially desalinated water; green square indicates that a 75% increase in wheat grass and Bermuda grass yields is expected if irrigation involves partially desalinated water; blue square indicates that a 175% increase in barley (hay) yields is expected if irrigation involves partially desalinated water.

Effect of ZVI Particle Size on Desalination
Steel wool is a rigid material that will stay in the reactor when the reactor is emptied. It is therefore a suitable material for the processing of multiple batches of water.
It is well established, as demonstrated in references and patent [US10,919,784B2], that the forward rate constant for desalination increases as the ZVI particle size decreases. The Figure 20. Illustrative salinity tolerance of four livestock grass crops-these vary with agricultural practice, climate, soil type, soil moisture content, soil permeability, and plant variety. Solid black line = expected soil salinity if it is irrigated with water containing 6.9 g NaCl L −1 (Table 1); dotted black line = expected soil salinity if it is irrigated with partially desalinated water containing 4.77 g NaCl L −1 (Table 1). Red square indicates that a 33% increase in tall wheat grass yields is expected if irrigation involves partially desalinated water; green square indicates that a 75% increase in wheat grass and Bermuda grass yields is expected if irrigation involves partially desalinated water; blue square indicates that a 175% increase in barley (hay) yields is expected if irrigation involves partially desalinated water.

Effect of ZVI Particle Size on Desalination
Steel wool is a rigid material that will stay in the reactor when the reactor is emptied. It is therefore a suitable material for the processing of multiple batches of water.
It is well established, as demonstrated in references and patent [US10,919,784B2], that the forward rate constant for desalination increases as the ZVI particle size decreases. The reactor train was retested using m-ZVI and n-ZVI, where the ZVI was only used to process a single batch of water. Five of the seven trials used a seawater feed water and a nano-iron. These trials are labelled RT1, RT2, RT3, RT6, and RT7. Trials RT4 and RT5 used saline water constructed by dissolving halite in water.

n-Fe 0
The nano-iron was manufactured in accordance with the instructions in GB Patent GB2,520,775A. The ingredients used were (a) iron (II) sulphate heptahydrate [CAS 7782-63-0] (molecular weight = 278.02 g/mol) as a source of Fe n+ ions; (b) black tea (Tanay Ceylon Turkish tea leaves Yaprak Cayi) soaked in water for 24 h, as a source for polyphenols.
The ZVI composition used in each trial is provided in Appendix B, Table A2.

Trial Results
Trials RT1, RT2, and RT4 were designed to determine if an emulsified Fe 0 provided a more effective desalination solution. Trials RT3, RT5, RT6, and RT7 were designed to determine if low concentrations of n-Fe 0 could provide an effective desalination solution.
The quantitative trial results are provided in Appendix B, Table A2. Each trial displayed a trend of increasing desalination with increasing reaction time. With the exception of Trial RT1, the trials showed a trend of decreasing forward rate constant, k f , with increasing trial duration.
Trial RT5 indicated that the process was less effective in acidic water than pH-neutral or alkaline water.

Trial RT1
The ISE trial data (Appendix B, Table A2) indicated that 79% desalination was achieved after 96 days. This was accompanied by a 31.6% decline in EC. The EC discrepancy is interpreted as being associated with the presence of organics in the ZVI mixture. Microscopic analysis of evaporated feed and product water qualitatively confirmed this quantitative observation that significant NaCl removal had occurred (Appendix B, Figure A5). These observations confirm the view proposed in US Patents [US9624113B2; US9828258B2; US2018/0009678A] that emulsified ZVI may be able to remove Cl − and Na + ions from water.

Trial RT2
The ISE trial data (Appendix B, Table A2) indicated that 96.5% desalination was achieved after 48 days. The product water contained 0.19 g Cl − L −1 + 1.15 g Na + L −1 . This was accompanied by a 24.0% decline in EC. The EC discrepancy is interpreted as being associated with the presence of organics in the ZVI mixture. Microscopic analysis of evaporated feed and product water confirmed this observation (Appendix B, Figure A6) and indicated the presence of some NaCl in the product water.

Trial RT3
The ISE trial data (Appendix B, Table A2) indicated that 74.9% desalination was achieved after 3 h, and 86% desalination after 71 h. This was accompanied by a 16.0% decline in EC. The EC discrepancy is interpreted as being associated with the presence of gallic acid and other polyphenols which were present in the n-Fe 0 slurry. Microscopic analysis of evaporated feed and product water confirmed this observation (Appendix B, Figure A7) and indicated the presence of some NaCl in the product water.

Amount of NaCl Removed by Sequential Trials
The removal of NaCl by Fe 0 is a relatively new research field (Appendix A). The use of a single charge of Fe 0 to process multiple water volumes has only previously been demonstrated in three experimental studies. These studies (Appendix A) used a batch flow, bubble column, static bed diffusion reactor and a batch flow, bubble column, recirculating diffusion reactor.
The amount of NaCl that can be removed by a gram of Fe 0 in sequential trials was extended in this study from 14.79 g NaCl g −1 Fe 0 to 89.3 g NaCl g −1 Fe 0 . This represents a six-fold increase in the operational efficiency for partial desalination of Fe 0 , relative to the earlier study.

Catalytic Implications
The RT series of trials (Appendix B, Table A2) established that reducing the particle size will allow 52 L water to be processed by 1 g n-Fe 0 and will allow 950 g NaCl (Trial RT3) to be removed by 1 g n-Fe 0 .
These observations confirm that the desalination process includes a catalytic element. The exact nature of the catalytic products has not been identified, but it is known that:

•
The products are not retained in the reactor when the water is removed from the reactor (i.e., are not settled precipitates).

•
When the product water is evaporated, the amount of NaCl present in the resultant precipitate is substantially reduced when compared with the feed water (Appendix B, Figures A5-A7).

•
The products are either miscible and are contained in the product water, or they were removed from the water with the expelled air.
Five groups of hypotheses have been proposed [18] to address the observed removal of pollutants (e.g., NaCl) by ZVI from water. They are:

•
Removal by adsorption and reaction, entirely within ZVI coagulants and flocculates. • Removal, using the electrons released by the oxidation of Fe 0 in water, to facilitate a reduction reaction. This concept may apply to the reduction of HClO or NaClO but it is unclear what the remediation products from this approach would be. • Pollutant removal by hydrogenation, where the hydrogen is produced by reaction with water, and by the catalytic decomposition of water. This could allow concentration and removal of Na + ions as NaOH.

•
Redox remediation, where the ZVI changes the Eh and pH of the water to force a change in the stable equilibrium species associated with the pollutant. The Eh:pH relationships are consistent with the formation of O 2 radicals, and the operation of a Fenton type of reaction, to remove Na + and Cl − ions. • Pollutant removal by Fe 0 catalysis to one or more products. The removal of 89-950 g NaCl g −1 Fe is consistent with this hypothesis.
Adsorption and reaction models are relatively easy to reconcile with the presence of an adsorbent when the amount removed (moles) is a fraction of the amount of ZVI present (moles). The reconciliation is more complex when the number of moles of NaCl removed is significantly greater than the number of moles of ZVI present in the water.
If the products constructed from NaCl are retained in the water, then these products have properties that result in them not being recognized by a chloride ISE, a sodium ISE, or an EC meter. Figure 21 identifies the principal product groupings that have been identified by ZVI desalination studies.

Removed Products-Evidence from the ZVI Cartridge
Following the completion of the 70 trials, the ZVI cartridge was drained, and the r sidual FexOyHz product examined under the microscope. The original high-porosity iro structure had been removed, and the fibres had been replaced by a hydrated network o iron oxyhydroxides (Figure 22). These hydroxides, when drying, discharged a saline w ter (Figure 22), confirming the view [17] that these hydroxides can actively remove NaC from the water.

NaCl Concentration Mechanism
The ZVI bed in a diffusion reactor commences operation with a network of inte connected pores. As the ZVI corrodes (oxidizes), some of these pores become dead-en pores which do not allow viscous flow through them.
The initial corrosion product is a layered double hydroxide (LDH). LDHs are ion compounds. Two types of defect processes occur within the LDH, which allow them t capture ions from the adjacent water. They are the Schottky defect and the Frenkel defec The LDH is a 3-dimensional structure where each M n+ ion (Fe n+ ) and each A m− (OH − ) io occupy a separate lattice site, i.e., [

Removed Products-Evidence from the ZVI Cartridge
Following the completion of the 70 trials, the ZVI cartridge was drained, and the residual Fe x O y H z product examined under the microscope. The original high-porosity iron structure had been removed, and the fibres had been replaced by a hydrated network of iron oxyhydroxides (Figure 22). These hydroxides, when drying, discharged a saline water (Figure 22), confirming the view [17] that these hydroxides can actively remove NaCl from the water.

Removed Products-Evidence from the ZVI Cartridge
Following the completion of the 70 trials, the ZVI cartridge was drained, and the residual FexOyHz product examined under the microscope. The original high-porosity iron structure had been removed, and the fibres had been replaced by a hydrated network of iron oxyhydroxides (Figure 22). These hydroxides, when drying, discharged a saline water (Figure 22), confirming the view [17] that these hydroxides can actively remove NaCl from the water.

NaCl Concentration Mechanism
The ZVI bed in a diffusion reactor commences operation with a network of interconnected pores. As the ZVI corrodes (oxidizes), some of these pores become dead-end pores which do not allow viscous flow through them.
The initial corrosion product is a layered double hydroxide (LDH). LDHs are ionic compounds. Two types of defect processes occur within the LDH, which allow them to capture ions from the adjacent water. They are the Schottky defect and the Frenkel defect.

NaCl Concentration Mechanism
The ZVI bed in a diffusion reactor commences operation with a network of interconnected pores. As the ZVI corrodes (oxidizes), some of these pores become dead-end pores which do not allow viscous flow through them.
The initial corrosion product is a layered double hydroxide (LDH). LDHs are ionic compounds. Two types of defect processes occur within the LDH, which allow them to capture ions from the adjacent water. They are the Schottky defect and the Frenkel defect. The LDH is a 3-dimensional structure where each M n+ ion (Fe n+ ) and each A m− (OH − ) ion occupy a separate lattice site, i.e., [M] n+ ] lattice site and [A m− ] lattice site . At any given time, the concentration of an ion on the LDH surface will be higher (or lower) than the concentration of the same ion within the LDH. The chemo-potential flow is from zones of high concentration to zones of low concentration. In a solid 3-dimensional lattice structure, this flow movement results in vacated lattice sites, termed vacancies. The residual vacant site, [V], has a charge which is opposite and equal to the charge of the removed ion.
The Schottky defect allows the removed ions to migrate to the surface of the lattice, thereby creating positive and negative charged lattice surface sites, where: The Frenkel defect allows the removed ions to migrate within the lattice towards the surface of the lattice or away from the surface of the lattice, thereby increasing the effective positive charge and negative charge of peripheral lattice sites: The thermodynamic equilibrium between interstitial cations, cation vacancies, and cations in normal lattice positions also occurs in the hydrated layers, where: H 3 O + are protons on interstitial sites and OH − are proton vacancies. These vacancies create a mechanism which allows Na + and Cl − ions to be transported, from inter-particle porosity, through intra-particle porosity and concentrated in dead-end pores (patent specification [US10919784B2]).

Catalytic Pressure Swing Adsorption-Desorption
ZVI is a "smart material" that can undertake a multitude of functions simultaneously [18]. Its properties change in response to changes in external stimuli, such as Eh, pH, temperature, pressure, light, and salinity [18].
The removal of Na + and Cl − ions is related to the oscillating pressure waves in the water. The resultant pressure swing adsorption/desorption (PSAD) can enhance catalytic reactions [25,32].
This study establishes that a PSAD process can potentially be used to remove >950 g NaCl g −1 Fe 0 in an aqueous oxygenated environment (Appendix B, Table A2, Trial RT3).
Catalytic yields of >900 g reactant removed g −1 Fe 0 have only previously been recorded in a reducing gas environment.
It had previously been established [33] that Fe 0 , operated using the Fischer Tropsch process, in a reducing environment (nH 2 + CO) could remove >130 g CO g −1 Fe 0 h −1 (with an operating life of >250 h and a potential removal of >32,000 g CO g −1 Fe 0 ). These results, from 2022, represent a substantial improvement on the position when this catalysis was first examined in the 1920s [18]. In the 1920s, the observed removal was around 0.15 g CO g −1 Fe 0 [18].
These observations indicate that if the future development of oxygenated Fe 0 catalysts follows a similar yield development curve to Fe 0 catalysts operating in a reducing environment [34], then it may be possible to develop a reusable n-Fe 0 catalyst that has a lifetime yield expectancy of 10-40 kg NaCl removed g −1 Fe 0 .
The next stage in the development of this process is to repeat the sequential reactor trials, using: • A batch flow, bubble column, static bed, recirculating diffusion reactor. This allows the ∆P to be increased into the range 10,000 to 60,000 Pa, and allows: The wavelength of the pressure wave to be reduced, while increasing its amplitude. The shape of the pressure wave to be modified from a complex variable pressure wave form to a simple pressure wave, to improve outcome predictability.
• A larger batch volume (860 L). This will allow a more compact construction for a commercial reactor unit. • A n-Fe 0 catalyst charge of around 10 g-100 g, with the intent of building on the results of Trial RT3 (Appendix B, Table A2). • A minimum of 50 sequential batches with a single catalyst charge.
The intent of this next stage of the process development is to obtain a NaCl removal rate that is substantially higher than the 89.3 g NaCl removed g −1 Fe 0 demonstrated in Table 1 and Appendix B, Table A1. The target removal of this next phase of the process development is 5000-10,000 g NaCl removed g −1 Fe 0 .

Ground Water Processing
The provision of desalinated water for irrigation traditionally requires a large (1000-100,000 m 3 d −1 ) plant with disposal facilities for 4000-400,000 m 3 of waste brine (reject brine). This study demonstrates, using a reactor train capable of processing up to 1.7 m 3 d −1 (and up to 600 m 3 a −1 ), that it is possible to produce partially desalinated water in a compact facility that could be located on a small agricultural unit (e.g., 1 to 100 ha). The process flow diagram for a facility capable of processing 15.3 m 3 d −1 (5585 m 3 a −1 ) is provided in Figure 23.
In this example, saline groundwater is abstracted from a saline aquifer ( Figure 23). The water is then pumped to a feedwater holding tank. This tank supplies nine reactor trains operating in parallel. Product water is pumped from these reactors to a product water tank. The product water can then either be pumped and used for irrigation or, alternatively, it can be pumped into the saline aquifer to dilute its salinity. Each reactor contains ZVI. The air supply to the reactors is controlled by a pressure meter and a flow meter.
A reactor of this size would be suitable for the supply of partially desalinated water for irrigation to an agricultural holding of 1 to 5 ha. Increasing the reactor size to 5 m 3 (1.9 m diameter × 2.1 m high) would increase the plant's potential processing capacity to 270 m 3 d −1 .

Partially Desalinating a Saline Aquifer
The plant configuration in Figure 23 will allow a slow moving or static saline aquifer to be partially desalinated. This requires construction of a single abstraction well, which is surrounded, radially, by n-injection wells located a distance x metres away from the abstraction well. The well spacing will be aquifer-and site-specific. The product water is injected into the aquifer while abstracting saline water from the aquifer. This dilution process will gradually allow the aquifer to become partially desalinated.
This aquifer management strategy concept has previously been proposed [27] as a potential future application of ZVI technology.
The concept of injecting large volumes of surplus desalinated water into a freshwater aquifer for subsequent abstraction and use is being explored in the Middle East [35].
The focus of this study is on providing partially desalinated water for irrigation. In coastal areas, seawater intrusion into an aquifer can be a major cause of saline groundwater. Potential protection solutions that have been adopted include the installation of subsurface cut-off walls; the installation of subsurface dams; and the injection of low-salinity wastewater to provide a recharge barrier that will protect against seawater intrusion [34,36].
The aquifer modification approach illustrated in Figure 23 can be used to inject lowsalinity water to provide a recharge barrier that will protect against seawater intrusion.

Adjusting the Process Flow to Allow Aquifer Reconstruction
This study demonstrates the following: mean desalination = 24.5%; standard deviation = 22.8%; minimum = 0%; 1st quartile = 9.2%; median = 14.3%; 3rd quartile = 33.9%; maximum = 91%. Based on this mean desalination value, a process of continually abstracting water from a saline aquifer followed by processing in the reactor and subsequent reinjection would be expected to result in partial desalination of the aquifer.
The abstraction borehole will receive a fraction [a] of new saline water coming into the aquifer (J) and a fraction [b] coming from recycled partially desalinated water within the aquifer, JPD. Equilibrium recycle modelling indicates that the achievable equilibrium

Partially Desalinating a Saline Aquifer
The plant configuration in Figure 23 will allow a slow moving or static saline aquifer to be partially desalinated. This requires construction of a single abstraction well, which is surrounded, radially, by n-injection wells located a distance x metres away from the abstraction well. The well spacing will be aquifer-and site-specific. The product water is injected into the aquifer while abstracting saline water from the aquifer. This dilution process will gradually allow the aquifer to become partially desalinated.
This aquifer management strategy concept has previously been proposed [27] as a potential future application of ZVI technology.
The concept of injecting large volumes of surplus desalinated water into a freshwater aquifer for subsequent abstraction and use is being explored in the Middle East [35].
The focus of this study is on providing partially desalinated water for irrigation. In coastal areas, seawater intrusion into an aquifer can be a major cause of saline groundwater. Potential protection solutions that have been adopted include the installation of subsurface cut-off walls; the installation of subsurface dams; and the injection of low-salinity wastewater to provide a recharge barrier that will protect against seawater intrusion [34,36].
The aquifer modification approach illustrated in Figure 23 can be used to inject lowsalinity water to provide a recharge barrier that will protect against seawater intrusion.

Adjusting the Process Flow to Allow Aquifer Reconstruction
This study demonstrates the following: mean desalination = 24.5%; standard deviation = 22.8%; minimum = 0%; 1st quartile = 9.2%; median = 14.3%; 3rd quartile = 33.9%; maximum = 91%. Based on this mean desalination value, a process of continually abstracting water from a saline aquifer followed by processing in the reactor and subsequent reinjection would be expected to result in partial desalination of the aquifer.
The abstraction borehole will receive a fraction [a] of new saline water coming into the aquifer (J) and a fraction [b] coming from recycled partially desalinated water within the aquifer, J PD . Equilibrium recycle modelling indicates that the achievable equilibrium salinity is a function of [a] and [b] (Figure 24). The modelling indicates that if it is possible to reduce [a] to <0.1 [b] then it may be possible to reduce the aquifer salinity by >50%.

gation. 
A partially desalinated aquifer will always require continual water injection to prevent saline incursions from outside the abstraction and injection area.  ZVI can remove other contaminants from the water [17]. The abstraction of water, passage of water through a ZVI reactor, followed by reinjection of the water into the aquifer can be used to decontaminate an aquifer ( Figure 24). Figure 24. Partial desalination of an aquifer. Modelled equilibrium salinities in the aquifer, where the initial aquifer salinity is assumed to be 5 g L −1 .

Costs
The viability of the application of this process is a function of cost. Table 2 provides the reactor manufacturing cost for the reactor used in this study ( Figure 2). This cost excludes manufacturers' profit, associated financing costs, transport costs, installation costs, feed water tank costs, product water tank costs, valve costs, pump costs, and the cost of associated conduits.
The ZVI cartridge is a consumable. The life expectancy of the cartridge is for more than 70 batches of water. If the cartridge only processes 70 batches, then the ZVI processing cost will be around USD 0.54 m −3 (Table 3). This observation has implications for regional aquifer salinity management and for the management of seawater incursion into an aquifer.
Specific points to note are as follows: • If no water is abstracted for irrigation, then it is possible to design the abstraction such that [a] decreases relative to [b] over time. The rate of decrease of [a] increases as the rate of injection increases. • If no water is abstracted for irrigation, then potentiometric surface for the aquifer may be altered to prevent and reverse seawater incursion while reducing the aquifer salinity.

•
The time taken to desalinate the aquifer is a function of the aquifer size, aquifer recharge rate, the number of abstraction wells, the number of injection wells, the abstraction rate, the injection rate, and the leakage rate provided by water used for irrigation. • A partially desalinated aquifer will always require continual water injection to prevent saline incursions from outside the abstraction and injection area. • ZVI can remove other contaminants from the water [17]. The abstraction of water, passage of water through a ZVI reactor, followed by reinjection of the water into the aquifer can be used to decontaminate an aquifer ( Figure 24).

Costs
The viability of the application of this process is a function of cost. Table 2 provides the reactor manufacturing cost for the reactor used in this study (Figure 2). This cost excludes manufacturers' profit, associated financing costs, transport costs, installation costs, feed water tank costs, product water tank costs, valve costs, pump costs, and the cost of associated conduits. The ZVI cartridge is a consumable. The life expectancy of the cartridge is for more than 70 batches of water. If the cartridge only processes 70 batches, then the ZVI processing cost will be around USD 0.54 m −3 (Table 3). Trial RTL 3 indicated that it may be possible to reduce the ZVI cartridge cost by both reducing the amount of ZVI present and increasing the amount of NaCl that can be removed per unit weight of Fe 0 . Table 4 provides a scoping cost associated with replacing the ZVI used in this study with n-ZVI. This analysis indicates that it may be possible. to reduce the ZVI cost to around USD 0.18 m −3 . The air compressor represents a major capital cost associated with the desalination process ( Table 5). The measured electricity usage associated with this study was about 0.3 kW batch −1 (about 2 kW m −3 ). In the UK, the unit cost of electricity is currently about USD 0.4 kW from a mains grid supplier. This unit cost reduces using off-grid Viridian Solar PV16 solar panels (Cambridge, UK), incorporating 8 kW battery storage, to a 20-year amortized cost of around USD 0.05 kW to USD 0.1 kW. Other off-grid power solutions may provide a more cost-effective option. Without specific cost management, the desalination cost associated with the single reactor train (Figure 2) used in this study can be summarized as follows:
This analysis indicates that ZVI desalination using a grid power supply and a batch flow, bubble column, static bed diffusion reactor will produce partially desalinated water for a cost of USD 1.37-1.58 m 3 . Switching to an off-grid power supply, combined with using n-Fe 0 , has the potential to reduce these costs by about USD 0.96 m −3 , to give a finished water product cost of around USD 0.4 m −3 to USD 0.5 m −3 . The majority of the cost reduction is associated with the cost of energy. Changing the energy source can, in some global areas, create significant cost savings.
The major cost innovations provided by this study may allow small agricultural holdings (0.5 to 10 ha) to consider using standalone partial desalination as a potential alternative to irrigation with saline water.
The next stage in the development of partial desalination ZVI diffusion reactors will involve the TRL7 testing of a batch flow, bubble column, static bed, recirculating, diffusion reactor train.

Conclusions
This study confirms that ZVI can partially desalinate saline water to allow for a number of potential applications, which include irrigation, livestock feed water, aquifer desalination, and to provide recharge barriers against seawater intrusion into aquifers.
This study used a batch flow, bubble column, static bed, diffusion reactor train to demonstrate the following: • >70 batches of batches of water can be processed sequentially using a single ZVI charge. • >42,000 m 3 water can be processed by 1 t Fe 0 . • >89.3 g NaCl g −1 Fe 0 can be removed in a multiple volume use application, processing 16.8 m 3 . • >950 g NaCl g −1 n-Fe 0 can be removed in a single volume use application, processing 240 L.
The outcome variability associated with individual water batches indicates that commercial units will be constructed from multi-train reactors. These multi-train reactors will produce an aggregated product water. The expected desalination, associated with the reactor trains and operating conditions, used in this study is 24.5%.
The ZVI desalination utilizes a catalytic pressure swing adsorption/desorption process. Analysis of the results of this process, combined with the sensitivity studies using n-Fe 0 , identified that it may be possible to use the results of this study to design a batch flow, bubble column, static bed, recirculating, diffusion reactor ( Figure 4) that would allow ∆P to be increased into the range of 10,000 to 60,000 Pa. This would create the following improvements in the reaction environment: The wavelength of the pressure wave will be reduced, and its amplitude increased. The shape of the pressure wave will be modified from a complex variable pressure wave form to a simple pressure wave.
These changes are expected to improve the desalination outcome predictability and boost the total amount of desalination that occurs. This study also indicates that the economics of partial desalination will be further enhanced if:

•
The batch volume processed is increased; • The steel wool Fe 0 is replaced with n-Fe 0 .
Funding: This research received no external funding.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data used in this study are placed in the tables and figures contained within the paper, and in Appendix B.

Acknowledgments:
The three reviewers and the Academic Editor are thanked for their helpful and constructive comments.

Conflicts of Interest:
The author declares no conflict of interest.

Appendix A. Desalination
Figure A1 provides a summary of the different desalination routes that are available.

Appendix A.1. Passive Desalination
The two passive partial desalination approaches ( Figure A1) that could be used to partially desalinate saline water for irrigation are the carbon dioxide desalination approach ( Figure A2) and the zero-valent iron desalination approach ( Figure A3).
The carbon dioxide approach ( Figure A2) converts Na + ions to NaHCO 3 or Na 2 CO 3 . The Cl − ions are either removed as NH 4 Cl or CaCl 2 or are left in the water. This approach is most effective in supersaturated brines that are in the presence of a flue gas containing CO 2 . The process can be operated at higher pressures to use CO 2 obtained from the atmosphere [37]. The principal advantage of this approach is its ability to recover Na + ions and Cl − ions in marketable products. The principal disadvantage, of this approach is that it is expensive and requires major economies of scale in order to create an economic product.
The alternative passive desalination route is ZVI desalination ( Figure A3). It was first discovered about 20 years after carbon dioxide desalination and is, therefore, not at the same stage of technology development. Figure A1. Approaches used to produce desalinated water and partially desalinated wat study uses zero-valent iron desalination to partially desalinate water.  The number of references in this paper to the papers and chapters dealing with the diffusion reactor process was reduced to two at the reviewers' request. Academic References: (i) CSR Reactor: Batch flow reactor; [15,16,45]; (ii) Fluidised Bed Reactor, Continuous flow, contact flow reactor: [46]; (iii) Diffusion reactors: 12 publications including [17] and [27].

Types of ZVI Diffusion Reactor
A diffusion reactor comprises a body of water and a body of ZVI. The ZVI is placed either in the base of the reactor to form a static bed or is placed in a cartridge with at least one open end. The cartridge (Figure 4) is either placed within the reactor, as in this study ( Figure 5), or is attached to the body of the reactor. Water within the reactor is allowed to enter the cartridge.
In a batch flow static bed reactor, during the period of operation, both the water column and the ZVI bed are static, and fluid exchange is by diffusion, or by replacement of fluids expelled from the ZVI (e.g., [18]).
In a batch flow, circulating water, static bed reactor, the ZVI body is static, but the water body is split between two or more tanks interconnected thanks, and either thermal convection or pumped flow is used to continually circulate water between the tanks. The

Types of ZVI Diffusion Reactor
A diffusion reactor comprises a body of water and a body of ZVI. The ZVI is placed either in the base of the reactor to form a static bed or is placed in a cartridge with at least one open end. The cartridge (Figure 4) is either placed within the reactor, as in this study ( Figure 5), or is attached to the body of the reactor. Water within the reactor is allowed to enter the cartridge.
In a batch flow static bed reactor, during the period of operation, both the water column and the ZVI bed are static, and fluid exchange is by diffusion, or by replacement of fluids expelled from the ZVI (e.g., [18]).
In a batch flow, circulating water, static bed reactor, the ZVI body is static, but the water body is split between two or more tanks interconnected thanks, and either thermal convection or pumped flow is used to continually circulate water between the tanks. The ZVI bed is static, and fluid exchange is by diffusion or by replacement of fluids expelled from the ZVI.
A batch flow, bubble column, static bed reactor has a static ZVI bed that is in contact with the water body. A gas is bubbled through the water body but not through the ZVI. Fluid exchange is by diffusion or by replacement of fluids expelled from the ZVI. This is the reactor type used in this study.
A batch flow, bubble column, static bed, recirculating, reactor has a static ZVI bed that is in contact with the water body. The water body is held in two or more tanks. A gas is bubbled through the water body but not through the ZVI. The bubbling gas continually recirculates water between the tanks. Fluid exchange is by diffusion or by replacement of fluids expelled from the ZVI.

Appendix A.2. Technology Readiness
A scale of nine technology readiness levels, termed TRL1 to TRL9 ( Figure A4), was developed [47] to allow any technology to be assessed on its route from discovery (TRL1) to commercial production (TRL9).
Hydrology 2022, 9, x FOR PEER REVIEW 37 of 43 Figure A4. Technology readiness levels associated with zero-valent iron desalination research in diffusion reactors. This study is an initial TRL7 demonstration of a batch flow, bubble column, static bed, diffusion reactor operation ( Figure A3). Patents [US9624113B2; US9828258B2; US2018/0009678A] have demonstrated a TRL7 operation for a batch flow, emulsified bubble column, fluidised bed reactor ( Figure A3). It is possible that unpublished TRL7 demonstrations have been undertaken for a continuous flow, contact flow, fluidised bed reactor ( Figure A3) and a continuous flow, fluidised bed, recirculating reactor ( Figure A3).

Appendix B. Systematic Dataset
The measurements from the 70 sequential trials are placed in Table A1. The measurements from the 7 RT series of trials are placed in Table A2. This represents the quantitative data set for the trials.
The following measurement procedure was adopted:  Figure A4. Technology readiness levels associated with zero-valent iron desalination research in diffusion reactors. This study is an initial TRL7 demonstration of a batch flow, bubble column, static bed, diffusion reactor operation ( Figure A3). Patents [US9624113B2; US9828258B2; US2018/0009678A] have demonstrated a TRL7 operation for a batch flow, emulsified bubble column, fluidised bed reactor ( Figure A3). It is possible that unpublished TRL7 demonstrations have been undertaken for a continuous flow, contact flow, fluidised bed reactor ( Figure A3) and a continuous flow, fluidised bed, recirculating reactor ( Figure A3).
This study provides a TRL7 [47] demonstration of ZVI desalination ( Figure A3) with a reactor train that has the potential to desalinate saline irrigation water ( Figure A4).
The information on carbon dioxide desalination in Figure A2 is provided here for completeness. This is because analysis of the trial results incorporated in this study indicate that some of the NaCl could potentially have been removed by CO 2 -LDH desalination ( Figure A2).

Appendix B. Systematic Dataset
The measurements from the 70 sequential trials are placed in Table A1. The measurements from the 7 RT series of trials are placed in Table A2. This represents the quantitative data set for the trials.
The following measurement procedure was adopted: • 48 h before a measurement was due to be made, the measurement instruments were placed in a fresh saline solution. This is required to condition the ISE electrodes.

•
The feed and product water samples (2.3 L/sample) were collected.

•
The ISE, pH, ORP, and EC tools were washed in deionized water and then they were calibrated using reference fluids.

•
The ISE, pH, ORP, and EC tool instrument readings were rechecked against the calibration fluids.

•
The ORP, ISE, pH, and EC measurements were taken for the feed water.

•
The tools were then washed in deionised water.

•
The ORP, ISE, pH, and EC measurements were taken for the product water.

•
The tools were then washed in deionised water.

•
The feed and product water measurements were then repeated using this procedure until three identical measurements had been recorded.

•
The measurement instruments were then placed in saline solution to allow them to be ready for the next set of measurements.