We use Monte Carlo simulations to investigate the effect of incorporating calcium chloride salt into nanoporous carbon on the performance of an ammonia–carbon adsorption refrigeration system. Simulations of ideal carbon slit-pores with pore sizes of 1, 2, and 3 nm, each containing calcium chloride with ion densities of 0.0, 0.25, and 0.5 nm−3, were carried out at temperatures between 0 and 30 °C and ammonia pressures up to 15.0 bar. The results reveal that ideal 1 nm pores are able to achieve a good refrigeration performance using waste heat below 100 °C to drive the process, but adding salt to these pores increases the waste heat temperature required beyond 100 °C. However, ideal 2 nm pores require the addition of 0.25 nm−3 salt to achieve a similar performance, while the 3 nm pores were unable to achieve a satisfactory refrigeration performance. Considering that real nanoporous carbons usually feature a variety of specific adsorption sites and non-ideal geometries that should have a similar impact to adding salt, these results indicate that nanoporous carbons with pores in the range of 1–2 nm are likely to hold the most promise for adsorption refrigeration applications and that the addition of salt may not always be helpful.

To help reduce carbon emissions and attain net-zero emission targets, there is an increasing demand for more sustainable cooling technologies as global temperatures continue to rise.1–3 Currently, refrigeration systems rely on synthetic fluorinated gases (F-gases), which are potent greenhouse gases with significant global warming effects.1,2 Recent UK regulations require the replacement of these synthetic gases with natural refrigerants like ammonia.1,4 There are three primary types of cooling technologies available: vapor compression, adsorption, and hybrid refrigeration systems.5–9 Adsorption refrigeration systems offer a significant advantage as they can be powered by low-grade thermal waste energy.5–9 Moreover, they utilize natural refrigerants such as ammonia, water, ethanol, and methanol, which have a minimal impact on global warming and ozone depletion compared to F-gases.10 In previous work,1 we provided a comprehensive review of the working principles of a basic adsorption refrigeration system as well as detailed guidelines on designing such a system using Monte Carlo simulation results.1,11

Selection of a suitable adsorbent–adsorbate working pair is crucial in adsorption refrigeration systems. Ammonia, as a natural refrigerant, offers numerous advantages over water, methanol, and ethanol. For example, ammonia is stable as a fluid over a wide range of driving temperatures,1,12 while water, of course, freezes at 0 °C, which limits its application in refrigeration systems. In addition, ammonia has a much higher latent heat than ethanol or methanol, which allows for more efficient refrigeration processes. Moreover, due to its much higher saturation pressure, it can be used at higher pressures compared to water, methanol, and ethanol, resulting in an improved mass transfer performance.

The choice of sorbent is equally important. Physical adsorbents, such as activated carbon, alumina, silica,6,13 and zeolites,9,14,15 are known for their rapid mass transfer and high ammonia uptake. However, they often suffer from low thermal conductivity.6 Meanwhile, solid absorbents such as CaCl2 and MgCl2 salts offer high sorption capacities but are prone to swelling and agglomeration.16 To address these limitations, researchers have proposed composite or hybrid adsorbents. For instance, CaCl2 has been impregnated in porous alumina,17 activated carbon,18 and graphite.19 These composite adsorbents have demonstrated an improved system performance and enhanced heat and mass transfer compared to the salt by itself.6 

Kang et al. reviewed the adsorption of ammonia in a range of porous materials for gas removal applications.9 Unfortunately, the temperatures specified in their review were higher than 20 °C, which is not sufficiently low for most refrigeration purposes. However, in a previous report,1 we employed Monte Carlo simulations to study the performance of several ammonia–zeolite working pairs in adsorption refrigeration applications. Our results indicated that zeolites with significant mesoporosity could be highly effective in such applications.1 This is because they offer substantial refrigeration capacities and favorable operating conditions, along with good coefficients of performance (COP).

Furthermore, in recent work,11 we studied adsorption refrigeration using Monte Carlo simulations of ammonia adsorption in ideal carbon slit-pores. Our aim was to isolate and elucidate the effect of pore width on system performance. Our findings revealed that using nanoporous carbon with very small pore sizes, <1 nm, results in reduced refrigeration performance compared to larger pores. In particular, high waste heat temperatures >100 °C are required combined with reduced thermal efficiency. As for zeolites, the reason for this is related to the higher isosteric heat of adsorption in smaller pores. Nevertheless, given the interest mentioned earlier in using composite adsorbents consisting of a porous material impregnated with salt for adsorption refrigeration processes, in the present study, we reprise our earlier work by using Monte Carlo simulations to assess the influence of impregnating nanoporous carbon with small quantities of CaCl2. Our main aim is to estimate the combined influence of pore width and salt loading on adsorption refrigeration performance indicators such as the refrigeration capacity, the coefficient of performance, and, crucially, the evaporation and desorption temperatures. The latter quantity is very important in relation to the use of waste heat to drive the process. As waste heat is often more readily available at temperatures below 100 °C, we consider this an important criterion.

Here, we describe the process of creating our initial CaCl2 impregnated carbon slit-pore configurations, the selection of force fields, and the Monte Carlo simulation procedure. Each carbon slit-pore consists of two graphite walls separated along the z-axis. These walls are made up of 14 × 8 × 1 (3.4 × 3.4 × 0.671 nm3) graphite unit cells in the x, y, and z directions, representing two layers of graphene in each wall. This corresponds to a total of 1792 carbon atoms in each simulation. We utilized chemical pore widths of 1, 2, and 3 nm, where the chemical pore width is defined as 0.24 nm less than the physical pore width between atom centers in the inner layers. Our previous work suggests that pores <1 nm result in reduced overall system performance and high waste heat temperature requirements.11 

CaCl2 molecules were randomly placed into each chemical pore volume using the Packmol tool.20 The amount of CaCl2 inserted corresponds to 0.0, 0.25, and 0.5 ion-pairs/nm3 of chemical pore volume. A CaCl2 concentration of 0.25 ion-pairs/nm3 corresponds to a total number of CaCl2 molecules of 3, 6, and 9 for pore widths of 1, 2, and 3 nm. In samples with a CaCl2 concentration of 0.5 ion-pairs/nm3, these numbers were doubled. We opted for these low concentrations to avoid ammonia–CaCl2 crystallization, which could otherwise complicate our analysis. Consequently, we obtained three initial configurations for each pore width. In all simulations, we applied periodic boundary conditions solely in the x and y directions. Therefore, the slit-pores under investigation were non-periodic in the z direction.

Next, we present details regarding the force fields we use to describe the interactions between ammonia, carbon, and CaCl2. For ammonia, we used the ammonia force field optimized by Eckl et al.21 Through benchmarking against various other ammonia models, our earlier work determined that this force field could accurately predict experimental saturation densities and pair distribution functions of ammonia liquids across a wide range of temperatures and pressures.1 Regarding the carbon adsorbent, we used parameters commonly employed with the graphitic Steele potential,22 which has been intensively used to study the adsorption of many gases, including ammonia, by carbon pores.23–26 Given the good performance demonstrated for Eckl et al.’s ammonia potential1 and the widespread use of the carbon Steele potential,22 we did not attempt to tune the C–N cross-interaction parameters beyond the standard use of Lorentz–Berthelot rules. Nevertheless, it should be possible to fine-tune these cross-interactions using experimental graphite–ammonia adsorption isotherms. Carbon atoms in the slit-pore walls were fixed in place during Monte Carlo simulations.

Regarding CaCl2, we employed a modified version of the force field developed by Zeron et al.27 In particular, we adjusted the Lennard-Jones sigma (size) parameter for the Cl ion to reproduce experimental and ab initio radial distribution functions and coordination numbers in dilute ammonia–CaCl2 solutions. A forthcoming publication will focus on the evaluation and optimization of this CaCl2 interaction model. We treated ammonia molecules as rigid units and excluded their intramolecular interaction energy from total energy calculations. This allowed us to isolate the ammonia–carbon intermolecular energy, which is very useful for isosteric heat analysis. Table I reports the Lennard-Jones parameters and partial atomic charges for each type of atom and ion. Standard Lorentz–Berthelot cross-interactions were employed.

TABLE I.

Lennard-Jones parameters and partial atomic charges of atomic interactions.

Atomε (K)σ (Å)q (e)
182.9 3.376 −0.9993 
0.3331 
28.0 3.4 0.0 
Ca2+ 61.0 2.6656 1.70 
Cl- 9.3 6.5 −0.85 
Atomε (K)σ (Å)q (e)
182.9 3.376 −0.9993 
0.3331 
28.0 3.4 0.0 
Ca2+ 61.0 2.6656 1.70 
Cl- 9.3 6.5 −0.85 

The Cassandra Monte Carlo code28 was used to perform our Grand canonical Monte Carlo (GCMC) simulations. All interactions were truncated at a distance of 1.2 nm. The Ewald summation method29 was employed to calculate long-range electrostatic interactions. The probabilities for molecular insertion, deletion, translation, and rotation moves were uniformly set at 25%. GCMC simulations were conducted over a temperature range from 0 to 30 °C with 10 °C intervals. We did not simulate temperatures below 0 °C to avoid unrealistically long simulation times required to equilibrate the system when CaCl2 is present.

At each temperature, we employed a range of pressures from 0.02 to 15.0 bar, with step-size increasing with pressure. It should be noted that the independent variable for our GCMC simulations was the chemical potential not pressure. Before conducting the reported GCMC simulations, we carried out several ammonia adsorption bulk simulations at wide ranges of temperatures and chemical potentials. From these simulations, we obtained an accurate chemical potential–pressure calibration curve at each temperature. This procedure is detailed in our previous paper.1 

To evaluate adsorption hysteresis, we simulated both adsorption and desorption isotherms at each temperature. For adsorption isotherms, simulations were started from an empty box at each pressure. For desorption isotherms, simulations were started from the output of an adsorption simulation at the same temperature and 15.0 bar. Each MC simulation comprised 75 × 106 MC cycles at each pressure and temperature, with average properties collected every 1000 cycles. The isosteric heat of adsorption, Qst, was calculated using the well-known fluctuation method,1,30 which incorporates fluctuations of the configurational energy and the adsorbed amount during GCMC simulations. Finally, adsorption refrigeration cycles were designed following the same procedure described in our previous work.1,11

Before reporting our main results, we must first evaluate the statistical quality of our MC simulations. In Fig. S1 (supplementary material), we report ammonia uptake (top) and energy (bottom) trajectories for the different chemical pore widths and CaCl2 contents. All trajectories correspond to an adsorption temperature of 0 °C and a pressure of 15.0 bar in order to focus on the most demanding simulation conditions. From this figure, it is clear that all our MC simulations are well-equilibrated. Based on the trajectories reported in Fig. S1, it is reasonable to compute final properties from the final 25 × 106 MC cycles.

In Fig. 1, we report snapshots of final configurations for carbon slit-pores with different chemical pore widths (1, 2, and 3 nm) and CaCl2 contents (0.0, 0.25 and 0.5 molecule/nm3). We selected fully saturated pores obtained at the lowest temperature of 0 °C and largest pressure of 15.00 bar. From this figure, it is evident that the presence of such small amounts of CaCl2 did not significantly modify the layer structure of ammonia adsorbed in these carbon slit-pores. Furthermore, the CaCl2 content was sufficiently low that the salt did not tend to cluster or phase-separate.

FIG. 1.

Final configurations for carbon slit-pores with chemical pore widths of 1 nm (a)–(c), 2 nm (d)–(f), and 3 nm (g)–(i), containing CaCl2 with ion densities of 0.0 (a), (d), and (g); 0.25 (b), (e), and (h); and 0.5 nm3 (c), (f), and (i). All configurations correspond to fully saturated pores at the lowest temperature of 0 °C and largest pressure of 15.0 bar. Gray, blue, and white circles represent the carbon, nitrogen, and hydrogen atoms, respectively. Ca2+ and Cl ions are represented by the large and small green circles.

FIG. 1.

Final configurations for carbon slit-pores with chemical pore widths of 1 nm (a)–(c), 2 nm (d)–(f), and 3 nm (g)–(i), containing CaCl2 with ion densities of 0.0 (a), (d), and (g); 0.25 (b), (e), and (h); and 0.5 nm3 (c), (f), and (i). All configurations correspond to fully saturated pores at the lowest temperature of 0 °C and largest pressure of 15.0 bar. Gray, blue, and white circles represent the carbon, nitrogen, and hydrogen atoms, respectively. Ca2+ and Cl ions are represented by the large and small green circles.

Close modal

In Fig. 2, we report ammonia adsorption and desorption isotherms obtained at a range of temperatures for carbon slit-pores with several different chemical pore widths (1, 2, and 3 nm) and CaCl2 contents (0.0, 0.25, and 0.5 molecule/nm3). Capillary condensation and hysteresis are evident for the 2.0 and 3.0 nm pores when salt is absent, and, as expected, increasing the pore width shifts the capillary condensation pressure to higher values. When salt is present, however, it is not clear if any capillary condensation and hysteresis remain as the differences observed between the adsorption and desorption curves in this case are very small. These observations are in good agreement with a recent experimental study,31 where a reduction in hysteresis was observed in composite adsorbents consisting of expanded natural graphite and CaCl2 or MgCl2.31 The impact of CaCl2 becomes very pronounced at low pressures. In fact, at the higher CaCl2 concentration studied, a complete desorption process is more challenging, probably due to higher isosteric heat of adsorption in the low coverage regime, which is discussed next.

FIG. 2.

Ammonia adsorption and desorption isotherms for carbon slit-pores with chemical pore widths of 1 nm (a)–(c), 2 nm (d)–(f), and 3 nm (g)–(i), containing CaCl2 with number densities of 0.0 (a), (d), and (g); 0.25 (b), (e), and (h); and 0.5 nm3 (c), (f), and (i). Adsorption data are shown by full symbols and solid lines, while desorption data are shown by empty circles and dashed lines.

FIG. 2.

Ammonia adsorption and desorption isotherms for carbon slit-pores with chemical pore widths of 1 nm (a)–(c), 2 nm (d)–(f), and 3 nm (g)–(i), containing CaCl2 with number densities of 0.0 (a), (d), and (g); 0.25 (b), (e), and (h); and 0.5 nm3 (c), (f), and (i). Adsorption data are shown by full symbols and solid lines, while desorption data are shown by empty circles and dashed lines.

Close modal

The isosteric heat of adsorption plays a crucial role in the development of adsorption-refrigeration systems. This parameter represents the heat released during the binding of an adsorbate to the surface of an adsorbent, indicating the strength of the interaction between the two.30,32,33 Moreover, the value of the isosteric heat of adsorption influences the temperatures and pressures needed for the desorption of specific concentrations of adsorbate. In Fig. 3, we report the dependence of the calculated isosteric heat of adsorption on ammonia loading, CaCl2 content, and chemical pore width.

FIG. 3.

Dependence of the isosteric heat of adsorption on ammonia loading and CaCl2 content in carbon pores with widths of 1 nm (a), 2 nm (b), and 3 nm (c).

FIG. 3.

Dependence of the isosteric heat of adsorption on ammonia loading and CaCl2 content in carbon pores with widths of 1 nm (a), 2 nm (b), and 3 nm (c).

Close modal

In agreement with our previous work11 on nanoporous carbon, when CaCl2 is not present in carbon pores, the isosteric heat generally increases with ammonia uptake in all studied pore widths. This is due to increasing ammonia–ammonia interactions as the pore fills. The effect of CaCl2 on the isosteric heat of adsorption is very clear in Fig. 3. We find that the strongest adsorbent–adsorbate interactions occur with the lowest ammonia uptakes. These strong interactions weaken as ammonia uptake increases initially, but then strengthen again at higher ammonia loadings. This is expected, since at low ammonia loadings, the most energetic ammonia–salt and ammonia–carbon interaction sites are available. Although these strong interactions saturate as ammonia uptake increases, they are compensated by increased ammonia–ammonia interactions as pressure increases. In fact, when the pores are saturated with ammonia, the isosteric heat in the presence and absence of CaCl2 nearly converges. From these results, we therefore anticipate that pores with large CaCl2 concentrations will be less useful for adsorption refrigeration applications since they would require higher desorption temperatures. This conclusion appears to be robust against fine-tuning the C–N interaction using experimental adsorption isotherm data.

The most important part of the present Monte Carlo study is the design of ammonia–carbon–CaCl2 adsorption refrigeration cycles, with a focus on studying the effect of CaCl2 addition on the overall performance of these thermally driven processes. As explained in detail in our previous study on ammonia–zeolite adsorption refrigeration systems,1 in order to design an adsorption refrigeration cycle, we need first to construct Clausius–Clapeyron PTx diagrams. Moreover, to construct these diagrams, we should first choose maximum (xmax) and minimum (xmin) ammonia uptake levels and then plot ln(P) vs T−1 for a range of fixed loadings between these extrema. Following common practice, the obtained P–T–x diagrams were fitted to straight lines.

The extreme loadings, xmax and xmin, determine the maximum amount of ammonia in the adsorption bed upon completion of the adsorption and desorption parts of the refrigeration cycle, respectively. For a given adsorbent–adsorbate working pair, the difference between these two concentrations will determine the cooling effect. Since the choice of these extreme loadings is somewhat arbitrary, and since our underlying data are noisy, their choice will slightly influence our results. Therefore, to improve the statistical quality of our final results, we select three values for both xmin (0.05, 0.06, and 0.07 kg/kg) and xmax (0.16, 0.17, and 0.18 kg/kg) and calculate the system performance averaged over all combinations of these extreme loadings. P–T–xmax and P–T–xmin diagrams were constructed from the adsorption and desorption isotherms reported in Fig. 2, respectively. See the supplementary material for a schematic of a complete adsorption-refrigeration cycle and our earlier work for a comprehensive description of this process.1 

In our refrigeration cycle process design, we select a condenser operating pressure (Pc) corresponding to an ambient condenser temperature of 25 °C: that is, the pressure corresponding to a temperature of 25 °C on the experimental ammonia saturation curve. For each slit pore, we also select an evaporator pressure (Pe) corresponding to a temperature of 25 °C on the P–T–xmax curve. The combination of xmax, xmin, Pc, and Pe defines a complete ideal adsorption-refrigeration cycle. Thus, for each xmax–xmin pair, we complete the entire thermodynamic cycle design to obtain values for four key performance criteria: the evaporator (refrigeration) temperature, the working (waste heat) temperature, the specific cooling effect (SCE), and the coefficient of performance (COP). The SCE and COP performance criteria are defined in our earlier work1,11 and also in the supplementary material for convenience. Briefly, SCE measures the heat extracted by the refrigerator, while COP measures the efficiency of this process.

In Table II and Fig. 4, we report the effect of carbon pore width and CaCl2 content on the operating conditions and performance criteria of an ideal ammonia–carbon adsorption refrigeration cycle, including the evaporator pressure (Pe), evaporator temperature (Te), working temperature (T3), specific cooling effect (SCE), and coefficient of performance (COP).

TABLE II.

Effects of the carbon adsorbent pore width and CaCl2 content on working temperatures, pressures, specific cooling effect (SCE), and coefficient of performance (COP) of an ideal ammonia–carbon–CaCl2 adsorption refrigeration cycle.

Pore width (nm)CaCl2 Content (molecule nm-3)Pc (bars)Pe (bars)T1 = Tc (°C)T2 (°C)T3 (°C)T4 (°C)Te (°C)xmin (kg/kg)xmax (kg/kg)SCE (kJ/kg)Ideal COP
0.0 9.71 2.26 ± 0.03 25.0 71.90 ± 1.00 79.68 ± 0.39 29.10 ± 0.37 −14.98 ± 0.35 0.06 ± 0.01 0.17 ± 0.01 131.52 ± 0.43 0.55 ± 0.01 
0.25 2.13 ± 0.08 65.54 ± 1.93 114.42 ± 2.08 53.66 ± 1.11 −16.47 ± 0.89 131.15 ± 0.36 0.49 ± 0.01 
0.50 3.30 ± 0.27 51.43 ± 2.46 215.32 ± 17.06 129.55 ± 4.24 −13.92 ± 1.48 134.15 ± 0.54 0.38 ± 0.02 
0.0 6.29 ± 0.05 40.74 ± 0.35 47.32 ± 0.90 30.14 ± 0.50 11.84 ± 0.24 138.01 ± 0.51 0.75 ± 0.01 
0.25 4.55 ± 0.06 48.40 ± 0.51 87.41 ± 3.80 54.20 ± 2.42 2.80 ± 0.37 135.82 ± 0.46 0.60 ± 0.01 
0.50 3.78 ± 0.19 51.19 ± 1.86 152.94 ± 11.30 93.72 ± 5.09 −2.37 ± 1.33 134.56 ± 0.36 0.46 ± 0.02 
0.0 7.40 ± 0.07 32.98 ± 0.28 38.34 ± 0.93 28.33 ± 0.76 16.63 ± 0.30 139.16 ± 0.51 0.83 ± 0.01 
0.25 6.26 ± 0.07 39.12 ± 0.43 74.09 ± 2.30 53.77 ± 1.66 11.69 ± 0.33 137.97 ± 0.48 0.66 ± 0.01 
0.50 5.28 ± 0.17 42.56 ± 1.21 141.73 ± 12.17 93.37 ± 5.47 6.82 ± 0.88 137.30 ± 0.58 0.50 ± 0.02 
Pore width (nm)CaCl2 Content (molecule nm-3)Pc (bars)Pe (bars)T1 = Tc (°C)T2 (°C)T3 (°C)T4 (°C)Te (°C)xmin (kg/kg)xmax (kg/kg)SCE (kJ/kg)Ideal COP
0.0 9.71 2.26 ± 0.03 25.0 71.90 ± 1.00 79.68 ± 0.39 29.10 ± 0.37 −14.98 ± 0.35 0.06 ± 0.01 0.17 ± 0.01 131.52 ± 0.43 0.55 ± 0.01 
0.25 2.13 ± 0.08 65.54 ± 1.93 114.42 ± 2.08 53.66 ± 1.11 −16.47 ± 0.89 131.15 ± 0.36 0.49 ± 0.01 
0.50 3.30 ± 0.27 51.43 ± 2.46 215.32 ± 17.06 129.55 ± 4.24 −13.92 ± 1.48 134.15 ± 0.54 0.38 ± 0.02 
0.0 6.29 ± 0.05 40.74 ± 0.35 47.32 ± 0.90 30.14 ± 0.50 11.84 ± 0.24 138.01 ± 0.51 0.75 ± 0.01 
0.25 4.55 ± 0.06 48.40 ± 0.51 87.41 ± 3.80 54.20 ± 2.42 2.80 ± 0.37 135.82 ± 0.46 0.60 ± 0.01 
0.50 3.78 ± 0.19 51.19 ± 1.86 152.94 ± 11.30 93.72 ± 5.09 −2.37 ± 1.33 134.56 ± 0.36 0.46 ± 0.02 
0.0 7.40 ± 0.07 32.98 ± 0.28 38.34 ± 0.93 28.33 ± 0.76 16.63 ± 0.30 139.16 ± 0.51 0.83 ± 0.01 
0.25 6.26 ± 0.07 39.12 ± 0.43 74.09 ± 2.30 53.77 ± 1.66 11.69 ± 0.33 137.97 ± 0.48 0.66 ± 0.01 
0.50 5.28 ± 0.17 42.56 ± 1.21 141.73 ± 12.17 93.37 ± 5.47 6.82 ± 0.88 137.30 ± 0.58 0.50 ± 0.02 
FIG. 4.

Dependence of (a) evaporator temperature Te, (b) desorption temperature T3, and (c) coefficient of performance (COP) on pore width and CaCl2 content.

FIG. 4.

Dependence of (a) evaporator temperature Te, (b) desorption temperature T3, and (c) coefficient of performance (COP) on pore width and CaCl2 content.

Close modal

First, notice that Table II shows that all our designed refrigeration cycles have a similar SCE because we choose the set of xmin and xmax to be the same for each pore width and salt loading. The next most important criterion is the evaporator (refrigeration) temperature, Te. Clearly, different refrigeration applications, such as air conditioning and food storage, will have different requirements, but here, we assume that to be a useful refrigerator, this temperature should not be much higher than 0 °C. Table II and Fig. 4(a) show that 1 nm pores are useful for all salt loadings and 3 nm pores are not useful for any salt loading studied here. The intermediate case of 2 nm is useful provided that salt is present.

Next, consider the desorption or working temperature, T3, which must be smaller than the waste heat temperature available if the process is to be powered in this way, which is assumed to be the case here. Since waste heat is often recovered from process steam supplies, temperatures well below 100 °C are preferred here. Table II and Fig. 4(b) show that the working temperature reduces quickly with increasing pore width (which agrees with our previous work11 on nanoporous carbon) but rises steeply with added salt because of the resulting increase in isosteric heat. Taking 100 °C as a limit, and having already excluded 3 nm pores and 2 nm pores without salt, we see that 1 nm pores without salt and 2 nm pores with low salt concentrations are preferred.

Finally, consider the COP, which is a measure of process efficiency. We see from Table II and Fig. 4(c) that COP increases with pore width but decreases steeply with salt loading. This behavior is strongly related to the isosteric heat. For the specific cases of 1 nm pores (without salt) and 2 nm pores (with low salt), their COP is quite similar with a slight preference for the 2 nm pore with low salt loading. Considering these results together, the preferred pore width representing the best overall performance is between 1 and 2 nm with low or no salt added. For these pores, it should be possible to operate a useful adsorption refrigeration process efficiently using waste heat below 100 °C.

We used Monte Carlo simulations to assess the effect of introducing CaCl2 salt into nanoporous carbon on ammonia uptake, isosteric heat of adsorption, and performance indicators of ammonia–carbon adsorption refrigeration cycles. We studied pore sizes ranging from 1 to 3 nm at temperatures between 0 and 30 °C and CaCl2 contents up to 0.5 molecule/nm3. From the designed refrigeration cycles, we found that carbon pores in the range 1–2 nm with zero or low salt loadings should be preferred for useful and efficient refrigeration. These conclusions regarding carbon pore size agree with our previous work,11 but this work also shows that adding CaCl2 salt appears to be an unnecessary complication if pore size can already be selected appropriately.

While we have focused on ideal graphitic pores with added salt in this work, we recognize that real nanoporous carbons are not formed by ideal pores. Instead, they typically incorporate a multitude of specific adsorption sites corresponding to different kinds and concentrations of surface groups along with curved and buckled pore surfaces that are far from the ideal slit pore geometry. Such non-ideal features will likely behave in a similar way to the salt added in this work by providing specific adsorption sites with a wide range of adsorption energies. We consider that this adds further weight to the view that adding salt appears to be an unnecessary complication. Instead, nanoporous carbon can be selected according to its pore size distribution and surface features. We expect that nanoporous carbons with pore sizes in the range 1–2 nm will likely display the best overall thermodynamic performance, with the optimal pore size dependent on surface features and ammonia transport performance.

The supplementary material file contains energy and particle number trajectories for each simulation as well as definitions for the SCE and COP performance criteria and a figure illustrating the adsorption refrigeration process.

This research is part of the ACCESS project, which was funded by EPSRC Grant Nos. EP/W027593/1 and EP/W027593/2. We thank the ACCESS project team members for their collaboration.

The authors have no conflicts to disclose.

Nasser D. Afify: Investigation (lead); Methodology (equal); Writing – original draft (lead). Martin B. Sweatman: Conceptualization (lead); Funding acquisition (lead); Methodology (equal); Project administration (lead); Supervision (lead); Writing – review & editing (lead).

The data used to support the findings of this study are included within the article. Should further data or information be required, these are available from the corresponding author upon request.

1.
N. D.
Afify
and
M. B.
Sweatman
, “
Monte Carlo simulation of ammonia adsorption in high-silica zeolites for refrigeration applications
,”
Chem. Eng. J. Adv.
18
,
100612
(
2024
).
2.
T.
Delay
,
The Emerging Cold Economy
(
Carbon Trust
,
2015
).
3.
L.
Davis
,
P.
Gertler
,
S.
Jarvis
, and
C.
Wolfram
, “
Air conditioning and global inequality
,”
Global Environ. Change
69
,
102299
(
2021
).
4.
L.
Yang
et al, “
A comprehensive review on sub-zero temperature cold thermal energy storage materials, technologies, and applications: State of the art and recent developments
,”
Appl. Energy
288
,
116555
(
2021
).
5.
R.
Nikbakhti
,
X.
Wang
,
A. K.
Hussein
, and
A.
Iranmanesh
, “
Absorption cooling systems – review of various techniques for energy performance enhancement
,”
Alexandria Eng. J.
59
(
2
),
707
738
(
2020
).
6.
T.
Zeng
,
H.
Huang
,
N.
Kobayashi
, and
J.
Li
, “
Performance of an activated carbon-ammonia adsorption refrigeration system
,”
Nat. Resour.
08
(
10
),
611
631
(
2017
).
7.
V.
Palomba
,
E.
Varvagiannis
,
S.
Karellas
, and
A.
Frazzica
, “
Hybrid adsorption-compression systems for air conditioning in efficient buildings: Design through validated dynamic models
,”
Energies
12
(
6
),
1161
(
2019
).
8.
P.
Srikhirin
,
S.
Aphornratana
, and
S.
Chungpaibulpatana
, “
A review of absorption refrigeration technologies
,”
Renewable Sustainable Energy Rev.
5
(
4
),
343
372
(
2001
).
9.
D. W.
Kang
,
S. E.
Ju
,
D. W.
Kim
,
M.
Kang
,
H.
Kim
, and
C. S.
Hong
, “
Emerging porous materials and their composites for NH3 gas removal
,”
Adv. Sci.
7
(
24
),
2002142
(
2020
).
10.
R. P.
Sah
,
B.
Choudhury
, and
R. K.
Das
, “
A review on low grade heat powered adsorption cooling systems for ice production
,”
Renewable Sustainable Energy Rev.
62
,
109
120
(
2016
).
11.
N. D.
Afify
and
M. B.
Sweatman
, “
Monte Carlo simulation of ammonia adsorption in nanoporous carbon: Optimal pore sizes for adsorption refrigeration applications
” (unpublished).
12.
E. J.
Hu
, “
A study of thermal decomposition of methanol in solar powered adsorption refrigeration systems
,”
Sol. Energy
62
(
5
),
325
329
(
1998
).
13.
D. C.
Wang
,
Y. H.
Li
,
D.
Li
,
Y. Z.
Xia
, and
J. P.
Zhang
, “
A review on adsorption refrigeration technology and adsorption deterioration in physical adsorption systems
,”
Renewable Sustainable Energy Rev.
14
(
1
),
344
353
(
2010
).
14.
C. Y.
Liu
and
K.
Aika
, “
Ammonia adsorption on ion exchanged Y-zeolites as ammonia storage material
,”
J. Jpn. Pet. Inst.
46
(
5
),
301
307
(
2003
).
15.
W.
Ouyang
et al, “
Dynamic ammonia adsorption by FAU zeolites to below 0.1 ppm for hydrogen energy applications
,”
Int. J. Hydrogen Energy
46
(
64
),
32559
32569
(
2021
).
16.
M.
van der Pal
and
R. E.
Critoph
, “
Performance of CaCl2-reactor for application in ammonia-salt based thermal transformers
,”
Appl. Therm. Eng.
126
,
518
524
(
2017
).
17.
V. E.
Sharonov
,
J. V.
Veselovskaya
, and
Y. I.
Aristov
, “
Ammonia sorption on composites ‘CaCl2 in inorganic host matrix’: Isosteric chart and its performance
,”
Int. J. Low-Carbon Technol.
1
(
3
),
191
200
(
2006
).
18.
H. R.
Goshayeshi
,
M.
Gewad
, and
H.
Nazari
, “
Investigation on evaluation of a solar intermittent refrigeration system for ice production with ammonia/calcium chloride and activated
,”
Energy Power Eng.
07
(
10
),
433
439
(
2015
).
19.
S. L.
Li
,
Z. Z.
Xia
,
J. Y.
Wu
,
J.
Li
,
R. Z.
Wang
, and
L. W.
Wang
, “
Experimental study of a novel CaCl2/expanded graphite-NH3 adsorption refrigerator
,”
Int. J. Refrig.
33
(
1
),
61
69
(
2010
).
20.
L.
Martínez
,
R.
Andrade
,
E. G.
Birgin
, and
J. M.
Martínez
, “
Packmol: A package for building initial configurations for molecular dynamics simulations
,”
J. Comput. Chem.
30
(
13
),
2157
(
2009
).
21.
B.
Eckl
,
J.
Vrabec
, and
H.
Hasse
, “
An optimised molecular model for ammonia
,”
Mol. Phys.
106
(
8
),
1039
1046
(
2008
).
22.
W. A.
Steele
, “
The physical interaction of gases with crystalline solids
,”
Surf. Sci.
36
(
1
),
317
352
(
1973
).
23.
G. R.
Birkett
and
D. D.
Do
, “
Simulation study of ammonia adsorption on graphitized carbon black
,”
Mol. Simul.
32
(
7
),
523
537
(
2006
).
24.
Y.
Zeng
,
H.
Xu
,
D. D.
Do
, and
D.
Nicholson
, “
Adsorption of argon on graphitized carbon black preloaded with methanol, ammonia and water: The role of adsorption regions and adsorbates
,”
Chem. Eng. J.
334
,
1316
1327
(
2018
).
25.
Y.
Zeng
,
D. D.
Do
,
T.
Horikawa
,
D.
Nicholson
, and
K.
Nakai
, “
On the explanation of hysteresis in the adsorption of ammonia on graphitized thermal carbon black
,”
Phys. Chem. Chem. Phys.
18
(
2
),
1163
1171
(
2016
).
26.
L. F.
Herrera
,
D. D.
Do
, and
G. R.
Birkett
, “
Comparative simulation study of nitrogen and ammonia adsorption on graphitized and nongraphitized carbon blacks
,”
J. Colloid Interface Sci.
320
(
2
),
415
422
(
2008
).
27.
I. M.
Zeron
,
J. L. F.
Abascal
, and
C.
Vega
, “
A force field of Li+, Na+, K+, Mg2+, Ca2+, Cl, and SO2−4 in aqueous solution based on the TIP4P/2005 water model and scaled charges for the ions
,”
J. Chem. Phys.
151
(
13
),
134504
(
2019
).
28.
J. K.
Shah
,
E.
Marin-Rimoldi
,
R. G.
Mullen
,
B. P.
Keene
,
S.
Khan
,
A. S.
Paluch
,
N.
Rai
,
L.
Romanielo
,
T. W.
Rosch
,
B.
Yoo
, and
E. J.
Maginn
, “
Cassandra: An open source Monte Carlo package for molecular simulation
,”
J. Comput. Chem.
38
(
19
),
1727
1739
(
2017
).
29.
C. J.
Fennell
and
J. D.
Gezelter
, “
Is the Ewald summation still necessary? Pairwise alternatives to the accepted standard for long-range electrostatics
,”
J. Chem. Phys.
124
(
23
),
234104
(
2006
).
30.
E.
Dickinson
, “
Computer simulation and the statistical mechanics of adsorption
,”
Colloids Surf.
9
(
3
),
294
(
1984
).
31.
K. B.
Jivrakh
,
S. B.
Kaki
,
R.
Sharma
,
S. S.
Mendu
, and
A. K.
Emadabathuni
, “
Comparison of ammonia sorption properties and thermodynamic performance of adsorption‐based thermal energy storage system for MnCl2, CaCl2, and their composites
,”
Energy Storage
2
(
4
),
e138
(
2020
).
32.
A.
Nuhnen
and
C.
Janiak
, “
A practical guide to calculate the isosteric heat/enthalpy of adsorption via adsorption isotherms in metal–organic frameworks, MOFs
,”
Dalton Trans.
49
(
30
),
10295
10307
(
2020
).
33.
S.
Hlushak
, “
Heat of adsorption, adsorption stress, and optimal storage of methane in slit and cylindrical carbon pores predicted by classical density functional theory
,”
Phys. Chem. Chem. Phys.
20
(
2
),
872
888
(
2018
).
Published open access through an agreement with JISC Collections