Are Ionic Liquids Suitable as New Components in Working Mixtures for Absorption Heat Transformers? Are Ionic Liquids Suitable as New Components in Working Mixtures for Absorption Heat Transformers?

The working mixture almost exclusively used to operate absorption heat transformers (AHT) is {H 2 O + LiBr} ({H 2 O + NH 3 } can also be used). Unfortunately, both working pairs present some drawbacks: corrosivity, toxicity, crystallization or high working pressure. Ionic liquids (ILs) possess very interesting properties (thermal stability, possible miscibility with water, negligible vapor pressure) that make them good candidates to be used as absorbents in AHT. This paper aims at providing an overview of available thermodynamic data concerning {H 2 O + IL} mixtures that could be used to operate an AHT.


Introduction
Most of the industrial and domestic activities require large amounts of thermal energy to generate steam or heat by burning fossil fuel. After being used and degraded, low temperature heat is released to the environment as low grade waste heat. Large quantities of low temperature thermal waste heat streams from many industrial facilities such as power plants are discharged as thermal pollutants to the air and to the water at temperatures ranging from 60 to 100°C on a daily basis [1].
Among heat-driven devices are the absorption cycles. They can be divided into three classes: absorption heat pump (AHP), absorption chiller (AC), and absorption heat transformer (AHT). Absorption cycles become of great interest since electrical energy is replaced with low grade or waste heat allowing both primary energy savings and energetic efficiency improvements [2]. Consequently, absorption cycles enhance the atmospheric conditions by reducing the emissions of greenhouse gases. Environmental impacts of absorption cycles can even be reduced by the adoption of environmental friendly working mixtures [3,4].
One of the key points to the performance of an absorption cycle is the working fluid used. Nowadays, the most used binary systems in the absorption heat cycle are {water + lithium bromide (H 2 O + LiBr)} and {ammonia + water (NH 3 + H 2 O)}. The aqueous solution of LiBr is the most successful working mixture in absorption cycles and widely spread all over the world [2,3]. Nevertheless, LiBr aqueous solution has some main drawbacks as follows: • Absorption heat pumps cannot operate at an evaporation temperature below 0°C because of the use of water as a refrigerant, which makes it unusable for subfreezing refrigeration or heating/domestic hot water (DHW) supplementation in cold regions. Crystallization of {H 2 O + LiBr} at high concentrations is a common problem. High vacuum conditions should be preserved in the system for suitable operation of the {H 2 O + LiBr} system; otherwise, the performance of the absorption cycle would be greatly reduced [5]. {H 2 O + LiBr} is corrosive to metals [2][3][4][5][6].
• {NH 3 + H 2 O} requires high working pressure and ammonia is toxic.
Due to these disadvantages, which have not been solved properly, the absorption technology has known a very limited expansion [7,8]. That is why heat pump and absorption chiller technologies suffer from lack of suitable working pairs. Hence, searching for new beneficial and reliable binary systems (to overcome these technical limitations) has become of great importance lately.
Limited numbers of critical reviews have been published in the literature on the subject of absorption technologies. In 2001, Srikhirin et al. [6] reviewed different configurations and types of absorption refrigeration cycles and working pairs. Performance development and enhancement of absorption cycles were evaluated. They concluded that double-stage absorption refrigeration cycle based on {H 2 O + LiBr} has the highest coefficient of performance (COP) if compared to other systems in the market. In addition, they stated that multistage absorption cycles have a promising future.
In 2012, Sun et al. [3] have shown that {H 2 O + LiBr} and {NH 3 + H 2 O} mixtures can be improved by the use of additives. They also stated that working pairs dedicated to specific applications such as solar or geothermal energy should use hydrofluorocarbons (HFCs) as a refrigerant.
Ionic liquids (ILs) are environmentally friendly solvents, which have attracted considerable attention recently. Ionic liquids are salts in liquid state having melting point below some arbitrary temperature, such as 100°C (373 K). These solvents consist of ions (an asymmetric, large organic cation, and organic or inorganic anion). A great advantage of ILs is that their physical properties such as melting points, density, and hydrophobicity can be adjusted to design different types of ILs that can be used for various applications.
ILs could be used as alternative working mixtures in absorption heat pump cycles. Hence, the possibility to have ionic liquids with a low melting point (lower than the temperature of the cold heat source of absorption heat pumps) allows overcoming the crystallization problem of the {water + LiBr} solution that can occur under some conditions [2]. Moreover, aqueous solution of ionic liquids seems to be less corrosive than the {water + LiBr} solutions. Finally, many ionic liquids show a high miscibility with water, which is a recommended refrigerant for absorption cycles (high latent heat, low viscosity, nontoxic, etc.). Consequently, the analysis of binary systems composed of {ILs + water} for this application has to be explored [10].
Few papers were published concerning working fluids containing ILs and a refrigerant such as NH 3 , water, ethanol, or halogenated hydrocarbon series. Although many working fluids are proposed in the literature, there is not a complete review with comparison of their properties and performances. Studies mainly focus on the evaluation of {H 2 O + IL} systems for their potential use in absorption heat cycles [11,12]. Khamooshi

Absorption heat cycles
Absorption cycles perform heat exchange between several heat sources or sinks. In the simplest case, there are three heat reservoirs characterized by their relative temperature level (high, medium, and low) as shown in Figure 1.
Absorption systems can operate according to different modes differing by the nature of the driving heat and by the desired useful effect. These modes are as follows: • Refrigerator: The driving heat is provided to the absorption cycle by a high temperature heat source. The low temperature source also provides heat to the cycle producing the cooling effect (useful effect). Heat is released to the medium temperature heat sink (generally the environment).
• Heat pump: As well as in a refrigerator, the cycle is driven by the heat provided by the high temperature source. The cycle also receives heat from the low temperature source. The useful heat is released to the medium temperature sink (generally a building or process that requires to be heated).
• Absorption heat transformer: Compared to both previous modes, sink and sources are reversed. The driving heat is a medium temperature heat (generally a waste heat). The upgraded useful heat is rejected to the high temperature sink and degraded heat is rejected to the low temperature sink (generally the environment).
Absorption cycles are composed of five main components: evaporator, condenser, generator, absorber, and solution heat exchanger (economizer) ( Figure 2). They generally use a binary working mixture composed of a low boiling component called the refrigerant and a high boiling component called the absorbent.
In an absorption heat transformer (Figure 3), the driving heat (medium temperature waste heat) is provided to the mixture of an absorbent and a refrigerant (weak solution) in the generator at a low pressure producing two streams: a pure refrigerant vapor stream and a liquid mixture stream (strong solution).
This vapor is condensed in the condenser releasing heat to the low temperature heat sink. The condensate is increased to high pressure through a pump and vaporized in the evaporator, thanks to medium temperature heat. The strong solution passes through a pump to high pressure and is sent to the absorber where it absorbs the vapor produced in the evaporator. This operation releases high temperature useful heat. The resulting weak solution is throttled through a valve and sent back to the generator. The solution heat exchanger allows preheating the strong solution entering the absorber by exchange with the weak solution leaving the absorber [15]. In refrigeration or heat pump absorption cycles (Figure 2), pressure levels are reversed (high pressure in the generator and in the condenser and low in the evaporator and the absorber), high temperature heat is provided to the generator, low temperature heat is provided to the evaporator and medium temperature heat is rejected at the condenser and absorber.
The working mixture properties will directly affect the absorption cycle performance. The following criteria can be followed to properly choose a working mixture [3]: i.
The presence of absorbent in the refrigerant must increase as high as possible the boiling point of the solution. ii. In order to reduce refrigerant flow rate, its vaporization latent heat has to be as high as possible. i.
In order to reduce exchange areas, pressure decreases and more generally the size and cost of equipment, viscosity of the solutions have to be the lowest as possible whereas thermal conductivity and diffusion coefficient have to be the highest as possible.
ii. Components of the working mixture should not be too expensive.
iii. Components of the mixtures should be noncorrosive and nontoxic.
iv. The environmental impact of the working mixture should be the lowest as possible, especially in terms of GWP and ODP.
To assess the energetic performance level of an absorption cycle, a criterion was defined: the coefficient of performance. Its expression depends on the kind of absorption cycle (refrigeration, heat pump, or AHT). Nevertheless, it is possible to define it as the ratio of the useful heat flow exchanged to the costly heat flow consumed [1]. For example, in a refrigeration cycle the expression of COP becomes [7][8][9][10][11][12][13][14][15][16]: COP ¼ Low temperature heat flow exchanged at the evaporator Hig htemperature heat flow provided at the generator þ Mechanical pumping power For an absorption heat transformer the COP expression is [8]

Working fluids containing {water + ILs} for absorption cycles
Water can be considered as a green refrigerant, nontoxic, having high latent heat and excellent thermal characteristics. ILs used in the working fluids {H 2 O + ILs} have to be hygroscopic and stable in aqueous solution. Numerous articles have studied the behavior of ILs with water, but there is still a lack of thermodynamic data for such mixtures.
The binary systems {H 2 O + dialkylimidazolium alkylphosphate}, 1-dimethylimidazolium dimethylphosphate, and 1-ethyl-3-methylimidazolium dimethylphosphate were extensively studied in numerous papers, where not only the data of vapor-liquid equilibria (VLE) but also density, viscosity, heat capacity, and excess enthalpy are available [7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][27][28][29]. These data make it possible to simulate the performance of these binary mixtures as working fluids in the absorption refrigeration cycle [7-25, 27, 30]. The simulation results show that the cycle performance of both systems is lower but close to the value obtained with the conventional working pair {H 2 O + LiBr}. Yokozeki and Shiflett [31] also examined the feasibility of different binary systems {H 2 O + IL} in an absorption cooling cycle and they found out that the best system is {H 2 [35]. Even if these systems have some interesting properties such as low density and heat capacity or strong negative deviations from Raoult's law, no simulation of the performance of these working fluids in an AHT was presented.

Thermodynamic properties of {H 2 O + IL}
The knowledge of thermodynamic properties, phase behavior, and safety/environmental hazards of {H 2 O + IL} is required for the evaluation of this system in an AHT. The following section presents the behavior of ILs in the presence of water and the influence of their structure on thermodynamic properties. and completely water soluble. Therefore, good working pairs are those presenting a highly negative deviation from Raoult's law [47].
Most binary systems {water + ILs} present activity coefficients lower than unity. The deviation from Raoult's law of {H 2 O + IL} is proportional to the IL content [17-19, 23-33, 36]. With respect to the anion, Ficke [16] has shown that the γ values decrease according to: It is important to note that dialkylimidazolium [(CF 3 SO 2 ) 2 N] with water present a miscibility gap [28]. The ability of IL to increase the water boiling temperature can be estimated using a simple relationship based on solvation model's parameters such the hydrogen-bond basicity [48].  39]. This work shows clearly that the observed trend is related to the anion and to its capacity to interact with water. Other research groups confirmed that anion has an essential rule that aiming to lower the vapor pressure of water H 2 O [25,27,30,32,33]. Seiler et al. [50] have shown that some ILs such as acetate and chloride-based ionic liquids are not suitable for absorptions cycles due to their insufficient stability and/or too high corrosion rates.
All VLE of binary systems {H 2 O + IL} found in the literature have been correlated using the NRTL model. The average relative deviations on activity coefficient and pressure obtained using the NRTL model range between 0.01 and 3.5%. Deviations of the 35 investigated systems are within ±13%.

Heat capacity
Heat capacity evaluates the heat storage capacity of a fluid [51]. Only one theoretical model based on an artificial neural network is proposed in the literature to predict the heat capacity of binary systems containing ILs [52]. This approach gives good estimate of C p of mixtures containing ILs with an average absolute relative deviation of about 1.60%.
In general, heat capacity is expressed using a temperature-and composition-dependent polynomial equation [35]: where C p is the mass heat capacity in kJ kg −1 K −1 , A i and B i are adjustable parameters, T is the absolute temperature in K, and xm 2 is the mass fraction of ILs. We have correlated all heat capacity data of {H 2 O + ILs} published in the literature using Eq. (3). The mass excess heat capacity, C p E can be calculated from the heat capacities of the mixture and that of the pure compounds: where C p is the mass heat capacity in kJ kg −1 K −1 of the mixture, C p,i is the mass heat capacity of the pure compound and xmi is its mass fraction. They found that the conductor-like screening model for real system (COSMO-RS) is a successful estimating method to predict the behavior of the interaction between water and ILs. It is obvious from the literature [10][11][12][13][14][15][16] that the positive (endothermic) H E of the binary system mainly depends on the hydrogen bonding, water molecules, and hydrophobicity. Weak interaction between the water-IL binary system causes water to use the energy of the system to rearrange their molecules and the process turn to be endothermic and the reverse occurs in the case of the exothermic process. Ficke [16] stated that with the increase of the alkyl chain length the hydrophobicity increases hence decreasing the negativity of H E .
H E data found in the literature are regressed using Redlich-Kister polynomials [25]: where H E is the excess enthalpy in kJ kg −1 , A i is the adjustable parameter, and x m i is the mass fraction of species i (i = 1, 2).
We used Eq. (5) to regress all H E data found in the literature.

Density
Density is an important property because its knowledge is necessary to evaluate the pumping cost in a process. The density of pure ILs roughly ranges between 1.1 and 1.6 g cm −3 . The density of an IL depends on the type of anion and cation, but the key parameter is the anion. Hydrophobicity of ILs has also an important effect on the density of binary mixtures {H 2 O + IL}. The hydrophobicity of a dialkylimidazolium-based IL increases with an increase of the alkyl chain length [26, 31, 33-35, 37-45, 47-57]. Consequently, the density of such ILs decreases with the increase of the alkyl chain length.
An increase in water content or temperature causes a decrease in the density in most of the binary systems studied. Hence, physical properties of ILs can be adjusted to fulfill the needs of applications for hydrophilic ILs by adding water or changing the temperature [57].
The density data for the 19 investigated binary systems were fitted [35] using Eq. (6).
where ρ is the density of the solution in g cm −3 , T is the absolute temperature in K, x 2 i is the molar fraction of the ILs, and a i and b i are adjustable parameters.
Excess molar volume (V E ) is an important parameter for the process design while it gives information on the nonideality of the working fluid. In the case of binary mixtures {H 2 O + IL}, the sign of excess molar volumes is related to the structure of the IL (anion, cation, and alkyl chain length) [

Viscosity
It is well known that pure ILs have higher viscosity than other solvents such as water, methanol, and ethanol [26]. This may enlarge the AHT size (exchange area) and increase the power required for the pumping process [63]. Nevertheless, various publications [26,31,[33][34][35] stated that viscosity of ILs sharply decreases when temperature increases and/or ILs are mixed with water. Taking into account that AHT has a high generator and absorber temperature (between 80 and 150°C), the viscosity of the {H 2 O + ILs} should not be a limitation for their use as absorbents in AHT [36]. The viscosity of {H 2 O + ILs} binary systems decreases because of the weak interaction between the IL anion and cation so the mobility of ions increases and the viscosity decreases [67]. It was noticed that fluorinated anions have lower viscosity than other anions such as alkylsulfates [64].

Thermal decomposition
Thermal decomposition could possibly be one of the most important properties to measure during the initial screening of an IL, especially for the operating temperatures of the processes related to this work. Most of the decomposition temperatures of ILs are measured using weight loss thermogravimetric (TGA) experiments and selected data are given in Table 2.
Nevertheless, it must be kept in mind that data taken from TGA will not serve to determine the maximum temperature limit for working without decomposition of the IL because this technique overestimates the decomposition temperature [68][69][70]. The experimental procedure proposed by Seiler et al. [50] based on a long-time thermal decomposition analysis seems to be more appropriate.

Simulation of the AHT cycle performance
This work focuses on single effect absorption heat transformers (AHT). The simulations used to evaluate the performance of the AHT were performed with the following assumptions [7][8][9]: i.
Steady-state operation; ii. Negligible heat loss; iii. Pressure drops not taken into account; iv. Outlets of the generator and the absorber are liquids at their bubble point;

v.
Liquid and vapor at the outlet of the condenser and the evaporator are saturated; vi. Enthalpy of the fluid is conserved through the throttling valve; vii. Minimum temperature difference between strong and weak solutions equal 5°C in the heat exchanger; viii. Pumping mechanical power is neglected compared to heat flow exchanged.
The steady-state simulation of such a process is achieved by solving mass and energy balance equations.
The generator can be described by the overall and ionic liquid mass balance and heat balance equations: The strong solution at the outlet of the generator is a saturated liquid, where T 8 = T 1 =T G and p 8 = P 1 =p G = p c The condenser can be described by For state point 2 (saturated liquid water at the condenser outlet), we have: In the case of the evaporator: Vapor is saturated at the outlet of the evaporator, so for point 4, we have: Balance equations for the absorber give: State point 5 is described by Eq. (20) where T 5 = T A and p 5 = p A = p E Then, the heat exchanger is characterized by the minimal temperature approach between hot and cold streams: Heat balance on the solution heat exchanger can be written: with p is the total pressure, x 2 , γ 2 and p s 1 is the IL mole fraction in the liquid phase, ionic liquid activity coefficient for the liquid phase, and saturated vapor pressure of the refrigerant, respectively.
Usually, when simulating an absorption heat transformer, the temperature level of the waste heat source is known (the medium-temperature level) as well as the temperature of the environment that is used as cold heat sink (the low-temperature level). The objective temperature level of the upgraded heat is also an input in this problem (the high-temperature level). Hence, temperatures T G , T E , T C and T A of the generator, the evaporator, the condenser, and the absorber, respectively, are known and taken as independent variables in the present research.
The enthalpy of a liquid mixture is expressed as follows: with where h 1 and h 2 are the enthalpy of pure liquid H 2 O and IL, x m 1 and x m 2 are the mass fraction of H 2 O and IL, respectively, Δ mix h is mixing enthalpy of the system, which can be sometimes neglected. The performance of the AHT is evaluated through different criteria. The main one is the coefficient of performance. Its expression is given in Eq. (2) as the ratio of useful heat flow produced at the absorber to the waste heat flows provided to the generator and to the evaporator (pumping work is neglected).
Among other meaningful criteria, Δx m is the difference between ionic liquid mass fractions in the strong and weak solutions.
If COP is used to represent the quantitative aspect of heat upgrading, the gross temperature lift Δt, which is the temperature level difference between the upgraded heat and the waste heat, provides a qualitative performance criterion. It is defined as follows: Another important criterion is the solution circulation ratio f, which is defined as the ratio of the strong solution mass flow rate to the vapor mass flow rate: This criterion allows knowing if the use of one working mixture leads or not to high solution flow rate which is linked to the capital cost (cost of the required working mixture and pumps) and operating costs (pumping energy cost). Observed values of f for the {H 2 O + LiBr} and {H 2 O + NH 3 } systems are generally low (typically around 10 [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16]).
Another criterion to assess system compactness is the available heat output per unit mass of refrigerant, q (kJ kg -1 ):

COP for the absorption refrigeration cycle
The performance of {H 2 O + ILs} as working fluids was mainly evaluated for the absorption refrigeration cycle. Zhang and Hu [27]  [DMP]} presents the advantage to allow a wider temperature working range as well as avoiding corrosion and crystallization problems.

Absorption heat transformer
A VBA dedicated calculation code has been developed to evaluate the performance of {H 2 O + IL} mixture as a working fluid in an absorption heat transformer (Figures 4 and 5).
Simulation results for both {H 2 O + IL} mixtures and conventional fluids as working fluids in AHT are presented in Table 3.  In this work, evaporating temperature T E , condensing temperature T C , absorbing temperature T A , and generator temperature T G are set to 80°C, 20°C, 130°C, and 80°C, respectively (Figures 6 and 7 and Table 3).
The NRTL model, C p , H E , and density correlation parameters that were regressed by the authors and used for the simulations. The resulting calculated COP values for 12 binary systems are shown in Table 3.
The influence of the working temperature levels on the COP is shown in Figures 8-10.  [8].** Data found in the literature [60].   Figure 8 shows that an increase in the condenser temperature leads to a decrease in the COP. This behavior is due to the fact that the low pressure level evolves the same way as the condenser temperature. Hence, when the condenser temperature increases, the strong solution ionic liquid fraction will decrease and f increases. For the investigated working pairs, the COP remains unchanged for T C lower than 30°C.     Figure 9 shows that an increase of T E or T G leads to an increase of the COP. In fact, the high pressure level of AHT depends on the evaporator temperature. Increasing T E (or T G ) leads to a decrease of the weak solution concentration by decreasing the flow ratio. The lower flow ratio results in a higher heat flow released during absorption and consequently in a higher COP. Figure 9 shows that the evolution of the COP values versus the generator temperature is quite similar to {H 2 O + IL} for all binary systems studied in this work. The evolution of COP values with T G firstly increases, then stabilizes and finally decreases. When T G approaches its minimal value, f tends to reach infinity and so it requires the generation of heat. Consequently, the COP of the cycle tends toward zero. With the increase of generator temperature, f decreases, COP sharply increases and then smoothens.
It can be seen from Figure 10 that the COP of an AHT decreases at different rates depending on the working mixture when absorber temperature (T A ) increases. This behavior can be explained in Figure 11 that illustrates the ionic liquid mass fraction variation of the weak solution x m w with T A .
A decrease of x m w means that the less refrigerant has been absorbed and consequently less heat is released at the absorber, which leads to lower the COP. The same behavior was observed by Zhang and Hu [8]. Figure 10 shows that the COP of all the binary systems {H 2 O + IL} as well as {H 2 O + LiBr} is basically unchanged when the gross temperature lift is lower than 45°C. Upon increasing the gross temperature lift more than 45°C, the COP of {H 2 O + LiBr} and {H 2 O + ILs} sharply decreases.
The available heat output per unit mass of refrigerant (q) for the studied binary systems was calculated and compared with Zhang and Hu [8] (Table 4).
These values must be compared with those obtained for {H 2 O + LiBr}: 2466 kJ kg −1 and 311 kJ kg −1 for {TFE + E181}. Hence, to produce the same amount of useful heat, the refrigerant flow rate is lower when using water-based mixtures (water latent heat of vaporization is much higher than that of E181).    The concentration (mass fraction) of ILs in the strong solution is exceeding 0.9 for most of the binary systems studied, and is only 0.64 for {H 2 O + LiBr} ( Table 3). This behavior is not in favor of ionic liquid-based working mixtures and will particularly lead to increased pumping costs. In the light of these results, it would be highly recommended to further investigate these binary systems.

Conclusion
A large number of binary mixtures {H 2 O + ILs} have been identified to be used in absorption heat transformers. The resulting performances of these new working fluids were evaluated for single-effect absorption heat transformer cycles.
Ionic liquid-based working mixtures lead to slightly low COP than the classical {H 2 O + LiBr} mixture and larger circulation ratios. Nevertheless, the possibility to find ILs that are significantly less corrosive than LiBr is a condition for reliable operation and a moderate investment cost. Moreover, many ILs are totally miscible with water which avoid crystallization problems.
It must be kept in mind that thermal and chemical stability of {H 2 0 + IL} mixtures have to be assessed in order to prove their practical use for industrial applications.