A metal–organic framework for efficient water-based ultra-low-temperature-driven cooling

Efficient use of energy for cooling applications is a very important and challenging field in science. Ultra-low temperature actuated (Tdriving < 80 °C) adsorption-driven chillers (ADCs) with water as the cooling agent are one environmentally benign option. The nanoscale metal-organic framework [Al(OH)(C6H2O4S)] denoted CAU-23 was discovered that possess favorable properties, including water adsorption capacity of 0.37 gH2O/gsorbent around p/p0 = 0.3 and cycling stability of at least 5000 cycles. Most importantly the material has a driving temperature down to 60 °C, which allows for the exploitation of yet mostly unused temperature sources and a more efficient use of energy. These exceptional properties are due to its unique crystal structure, which was unequivocally elucidated by single crystal electron diffraction. Monte Carlo simulations were performed to reveal the water adsorption mechanism at the atomic level. With its green synthesis, CAU-23 is an ideal material to realize ultra-low temperature driven ADC devices.

MIP-200 recently published data was used. 13 Solid lines represent the model fit.   Table 2).

IR spectroscopy
The infrared spectra of the reaction products were measured on a Bruker Alpha-P IR spectrometer. For the drying of the compound it was heated to 100 °C for 16 h (Supplementary Figure 2).

Nitrogen sorption measurement
The nitrogen sorption experiment was carried out at a BEL Japan Inc. Belsorpmax at 77 K.
The sample was activated at 150 °C for 16 h under reduced pressure (< 0.1 mbar).
The nitrogen sorption measurement of CAU-23 shows a type I isotherm (Supplementary

Thermogravimetric measurements
The thermogravimetric curve was recorded in air on a Linseis STA PT 1600 (heating rate = 4 K min -1 , gas flow = 20 mL/min). For the evaluation the formula [Al(OH)(C6H2O4S)] • 3.8 H2O was used (water content is varying with the external rh and temperature). Since the decomposition product at 1000 °C is not crystalline the composition Al2O2S was used (Supplementary Figure 4).

Temperature dependent PXRD measurements
Temperature dependent powder X-ray diffraction patterns were recorded on a Stoe Stadi P Combi diffractometer in transmission geometry equipped with Mo-Kα1 radiation, a curved germanium monochromator and a linear MYTHEN detector with an aperture angle of 17° using a furnace (Supplementary Figure 5).

Structure determination details from electron diffraction data with cRED of CAU-23
The structure of CAU-23 was solved by direct methods using SHELXT, and all the atom positions were found directly (Supplementary Figure 6). The structure was refined against the cRED data using SHELXL. 3 Atomic scattering factors for electrons and isotropic atomic displacement parameters were used for the refinement without adding any restraints. The structure converged rapidly with chemically reasonable bond lengths and angles. The selected crystallographic data are given in Supplementary Table 3, and the structure details are provided in the supporting cif file.

Structure refinement details of CAU-23 using PXRD data
Due to the extremely high intensities typical for synchrotron based PXRD measurements or in-house measurements with long expose time (38 h), Rexp is very low, resulting in a GoF that is higher than the typical value for in-house data (Supplementary Figures 7-8, Supplementary   Table 3).

Stability test of coated CAU-23
The Le Bail fits were performed using TOPAS Academic 6 (Supplementary Table 8). 10 The water uptake capacity of coated CAU-23 was investigated gravimetrically. The sample was exposed to different temperatures, 100 °C for 2 h, 16 h with 71.1 % relative humidity at 20.1 °C realized by saturated aqueous sodium chloride solution (150 g NaCl in 100 mL water) and 100 °C for 2 h again. The mass of every sample was measured and the difference was calculated and used to determine the uptake (Supplementaey Table 9).

Supplementary Notes 1 Benefits for low adsorption temperature on a adsorption driven chiller system
There are three important factors why a low desorption temperature has multiple benefits compared to finding MOFs with high adsorption capacities (ideally high capacities and low desorption would be the ultimate goal). 1 First, by lowering the driving temperature other applications may come into range for adsorption cooling that yet cannot be used efficiently (district heating, geothermal heating, data centers. …). For example our main effort in the past was to achieve High Performance Cooling using the waste heat from CPUs or GPUs in order to drive the process to cool the boundaries.
Although new CPUs may withstand temperatures up to 90°C, throttling is activated for temperatures above 70°C.
Second, whereas the cooling energy per cycle is directly correlated to the water uptake, the regeneration energy is directly correlated to the heat of adsorption and the temperature needed to remove the water molecules. In a real device, all components like the heat exchanger, binder and pipes have to be heated up additionally to the necessary driving temperature to remove the adsorbed water. This can be summed up as sensible heat Qsens of these parts and is consumed in every cycle (QIH = isosteric heating of the MOF). This additional term is added to the COPC (coefficient of performance for cooling) calculation, thus lowering the COP.
The higher the necessary driving temperature the more energy is lost by the needed sensible heat in every single cycle. Taking this point into account, we have carried out an estimation on the importance of a reduced desorption temperature on the COP for a coated heat exchanger (Supplementary Table 1), keeping the other parameters constant.
The reduction of the desorption temperature from 90 to 60 °C leads to a significant enhancement of the COP from 0.67 to 0.78, an improvement of 15 %. This calculation demonstrates the change for only one cycle, but the consumed sensible heat has to be multiplied by the amount of cycles performed during lifetime of the device (>100 000). Thus it leads to considerable overall amounts, showing the immense influence of the driving temperature.
Taking the higher necessary desorption temperature for Al-fum of 90 °C into account, whereas its adsorption enthalpy is comparable to CAU-23 (ΔHevap,Al-fum ~ -50 kJ/mol, 2 ΔHevap,CAU-23 = -48.2 KJ/mol, this work), the sensible heat Qsens will have a much larger negative impact on the COP of Al-fum than it is for CAU-23 (Tdesorption = 60°C).
Third, under working conditions a low driving temperature (here < 60 °C) is also advantageous even if higher driving temperatures are available (e.g. waster heat). For a real system containing a MOF-coated heat exchanger (device) thermodynamics comes into play and the heat flow from the source to the fluid to the heat exchanger to the MOF (heat resistance chain) must take place to desorb the water molecules. In order to have a relevant heat transfer a temperature difference is needed. This is described in Fourier´s law in its integral form as with ̇= heat flow rate, λ = thermal conductivity, A = cross sectional area, T1, T2 = temperatures on the different sides and d = thickness of the wall. This equation can also be given in the differential form.
with q = heat flux density, k = conductivity, T = temperature gradient Reaching the minimum desorption temperature, the MOF starts to desorb efficiently while consuming heat (desorption is endothermic). Thus a larger heat flux into the material leads to a faster desorption. A higher temperature difference (T1-T2 or T) will lead to a higher heat flux and consequently a shorter desorption cycle can be achieved. Thus more cycles per time are possible, which directly leads to higher cooling power output of the system.

Comparison of the water uptake of CAU-2and Al-fum
To get an insight into the water sorption kinetics of CAU-23 and to be able to compare it to another state of the art material (Al-fum) 2 , cycling thermogravimetric measurements were carried employing the same experimental set up and parameters (Tmin, Tmax, relative humidity, gas flow rate, similar sample weight). The results are shown in Supplementary Figure 11.
Both samples, Al-fum and CAU-23 show a very similar adsorption/desorption behavior under the cycling conditions. The desorption behavior of CAU-23, following the equilibrium step after 20 cycles (t ~ 4 d), was compared to those available for Al-fum (Supplementary Figure 12).
Both materials show a similar desorption speed, small differences are due to a slightly larger coating thickness, as well as the mass of the coatings, were measured and thus the composite density was determined to be 0.46 g/cm³ (each measurement was repeated five times and the values were averaged). Taking the binder mass and the gravimetric water uptake capacity for CAU-23 into account, a volumetric uptake capacity of 0.14 gH2O/cm³ composite is obtained.
These values are highly depending on the way the coating is manufactured, as changing the binder content or the binder itself will change the density of the coating. To provide an estimate value for a full scale heat exchanger, a heat exchanger with 1 L volume and 0.9 m² surface is considered with a coating thickness of 300 µm. This results in a volumetric density of 37.8 gH2O/Lheat exchanger.