betaFeedback

WorksPhysics and Astronomy

Giant photocaloric effects across a vast temperature range in ferroelectric perovskites

Abstract

Solid-state cooling presents an energy-efficient and environmentally friendly alternative to traditional refrigeration technologies that rely on thermodynamic cycles involving greenhouse gases. However, conventional caloric effects face several challenges that impede their practical application in refrigeration devices. Firstly, operational temperature conditions must align closely with zero-field phase transition points; otherwise, the required driving fields become excessively large. But phase transitions occur infrequently near room temperature. Additionally, caloric effects typically exhibit strong temperature dependence and are sizeable only within relatively narrow temperature ranges. In this study, we employ first-principles simulation methods to demonstrate that light-driven phase transitions in polar oxide perovskites have the potential to overcome such limitations. Specifically, for the prototypical ferroelectric KNbO3_{3} we illustrate the existence of giant photocaloric effects induced by light absorption (ΔSPC∼100ΔS_{\rm PC} \sim 100~J~K−1^{-1}~kg−1^{-1} and ΔTPC∼10ΔT_{\rm PC} \sim 10~K) across a vast temperature range of several hundred Kelvin, encompassing room temperature. These findings are expected to be generalizable to other materials exhibiting similar polar behavior.

Solid-state cooling is an environmentally friendly and highly energy-efficient technology that harnesses caloric effects in materials for refrigeration purposes [1, 2]. Through the application of external fields to caloric materials, large reversible entropy and temperature changes (ΔS∼10\Delta S \sim10–100100 J K−1^{-1} kg−1^{-1} and ΔT∼1\Delta T \sim1–1010 K) can be achieved, seamlessly integrated into cooling cycles without reliance on greenhouse gases. Particularly, polar materials undergoing ferroelectric to paraelectric phase transitions under small electric fields are well-suited for solid-state cooling applications based on the electrocaloric (EC) effect [3, 4, 5, 6, 7], which can be miniaturized and coherently integrated into electronic circuits.

For caloric phenomena to be integrated in practical applications, the involved zero-field phase transition points must be close to room temperature. Otherwise, the necessary driving fields may reach impractically high levels, creating an energetically unfavorable scenario. Furthermore, the materials performance may suffer due to factors such as leakage/eddy currents, dielectric/magnetic losses, and mechanical fatigue. Unfortunately, only a limited number of compounds exhibit phase transitions near room temperature, which limits the range of solid-state refrigeration applications. Additionally, caloric effects typically exhibit strong temperature dependence and are substantial only within relatively narrow temperature ranges.

In this study, we propose a solution to the described caloric materials challenges by focusing on polar to non-polar phase transitions triggered by light absorption [8]. Specifically, we theoretically demonstrate the existence of giant photocaloric (PC) effects (ΔSPC∼100\Delta S_{\mathrm{PC}} \sim100 J K−1^{-1} kg−1^{-1} and ΔTPC∼10\Delta T_{\mathrm{PC}} \sim10 K) in the archetypal ferroelectric KNbO3_3 (KNO). These PC effects persist across a vast temperature range spanning several hundred Kelvin, including room temperature, and are comparable to state-of-the-art EC effects [3, 4, 5, 6, 7]. The substantial anharmonicity of the paraelectric phase, naturally occurring at high temperatures (≳700\gtrsim700 K [9, 10]), coupled with its stabilization through light irradiation, are the primary factors responsible for the unravelled giant PC effects.

We performed density functional theory (DFT) calculations [11] with the projector augmented wave method [12] and generalized-gradient PBEsol approximation [13] (energy cutoff of 650 eV). The electrons treated as valence were: K 4s13p63s24s^1 3p^6 3s^2, Nb 4d45s14p64d^4 5s^1 4p^6, and O 2p42s22p^4 2s^2. The first Brillouin zone (BZ) was sampled with a 12 ×\times 12 ×\times 12 k-points grid and the atomic positions were optimized until the atomic forces were smaller than 0.5 meV Å−1^{-1}. Electric polarizations were calculated within a linear approximation using Born effective charges [14]. Photoexcitation was mimicked by constraining the partial occupancies of the electronic orbitals through adjustment of the Fermi distribution smearing. This effective DFT approach is effectively equivalent to those employed in previous works where the concentration of electron-hole pairs was constrained via the introduction of two adjustable quasi-Fermi levels [8, 15]. The second-order interatomic force constants were computed by finite differences with Phonopy [16] using 5 ×\times 5 ×\times 5 supercells for the polar phases and 4×4×44 \times4 \times4 supercells for the nonpolar phase. Thermal expansion effects were appropriately accounted for with the quasi-harmonic approximation (QHA) [17]. The DynaPhoPy code [18] was used to calculate the anharmonic lattice dynamics of the room-temperature polar phase (TT-renormalized phonons) from ab initio molecular dynamics (AIMD) simulations. DFT Gibbs free energies, GG, and entropies, SS, were computed with the QHA method considering TT-renormalized phonons for the phases that are dynamically unstable at zero-temperature conditions (“DFT-rQHA”). AIMD simulations were carried out in the (NN, VV, TT) ensemble at T=300T = 300 K using Nosé-Hoover thermostats. A large simulation cell containing N=625N = 625 atoms was employed with periodic boundary conditions applied along the three Cartesian directions. Newton’s equations of motion were integrated using the customary Verlet’s algorithm with a time step of 1.5⋅10−31.5 \cdot10^{-3} ps. Γ\Gamma-point sampling was used for BZ integration and the total duration of the AIMD simulations was of 60 ps.

The ground state of KNO is a polar rhombohedral phase (R3mR3m, RR). At temperatures of 220≤T≤470220 \le T \le470 K, KNO stabilizes in a different polar orthorhombic phase

(Amm2Amm2, OO). [9, 10] Upon absorption of above-bandgap light, some electrons in KNO are photoexcited and promoted from the valence to the conduction band, where they gain increased mobility and become delocalized throughout the crystal. These carriers effectively screen the long-range dipole-dipole interactions that are key to the existence of polar order in the ferroelectric phase. Consequently, the nonpolar cubic phase (Pm3mPm3m, CC), which is stable at temperatures T≳700T \gtrsim700 K under “dark” conditions [9, 10], becomes the ground state under “light” conditions (Figure 1a).

Image not extracted

Figure 1. Light-driven phase transition mechanism in ferroelectric KNbO₃. (a) Upon photoexcitation, electrons are promoted to the conduction band and ferroelectric KNO transforms into a nonpolar cubic phase, CC. (b) Vibrational phonon spectrum of KNbO₃ in the nonpolar CC phase at T=0T = 0 K conditions in the absence (red lines) and presence (blue lines) of photoexcited electrons (nˉ=2.69⋅1021\bar{n} = 2.69 \cdot10^{21} cm−3^{-3}).

Our zero-temperature DFT calculations reveal that the nonpolar CC phase becomes vibrationally stable at photoexcited electronic densities exceeding nˉ=2.69⋅1021\bar{n} = 2.69 \cdot10^{21} cm−3^{-3} (Figure 1b). Importantly, our DFT simulations demonstrate that both the polar RR and OO phases spontaneously relax into the nonpolar CC phase under sufficiently high light irradiation (or equivalently, density of photoexcited electron-hole pairs). Thus, under such photoexciting conditions the polar phases RR and OO are unstable rather than metastable. A similar light-induced phase transition has been shown to produce sizable lattice thermal conductivity changes, with possible applications in energy scavenging and phonon-based logic [19].

In this study, above-bandgap light absorption is crucial since it drives the promotion of charge carriers from the valence to the conduction band. Hence, in practice the light frequency range should be in the ultraviolet region (i.e., hν≥EgKNO≈3h\nu\ge E_{\mathrm{g}}^{\mathrm{KNO}} \approx3 eV) [20]. An essential parameter for the practical relevance of the described photoinduced phase-transition mechanism is the light penetration length, denoted as Δz\Delta z. If Δz\Delta z is too small, it could limit technological applications since the light-induced phase transition may not occur homogeneously unless Δz∼L\Delta z \sim L, where LL represents a characteristic size of the sample. Based on experimental optical absorption data for KNO [20], we estimate Δz≈30\Delta z \approx30 μm, which significantly exceeds the usual thickness of oxide perovskite thin films (∼0.1\sim0.1–1.01.0 μm). Therefore, for practical implementations of the photocaloric phenomena revealed in this study, films are the most advisable systems.

According to our first-principles DFT calculations, the polar RR phase is characterized by a lattice parameter of 4.034 Å, rhombohedral angle of 89.89° and ferroelectric polarization of P=35.05P = 35.05 μC cm−2^{-2}. These results compare remarkably well with the corresponding experimental values (T=230T = 230 K) of 4.016 Å, 89.83° and 30–40 μC cm−2^{-2}, respectively [9, 21, 22]. For the polar OO phase, we estimated the unit cell parameters a=3.996a = 3.996 Å, b=c=4.053b = c = 4.053 Å, α=90.18∘\alpha= 90.18^\circ, and β=γ=90∘\beta= \gamma= 90^\circ, and a ferroelectric polarization of P=38.53P = 38.53 μC cm−2^{-2} along the pseudocubic direction [11]. These calculated values are also in very good agreement with the corresponding experimental results (T=295T = 295 K) 3.973 Å, 4.035 Å, 90.27°, and ≈40\approx40 μC cm−2^{-2} [9, 21, 22]. Therefore, the employed DFT approach offers a robust description of KNO.

Figure 2 displays the vibrational phonon spectrum of the polar RR and OO phases at “dark” conditions. In the case of the OO phase, the phonon frequencies were renormalized to account for thermal effects at room temperature using a normal-mode decomposition method [18]. Our computational results accurately replicate the dynamical stability observed in experiments for the polar RR and OO phases at low and room temperature, respectively [9, 10].

Phonon spectra of KNbO3 in the ferroelectric R and O phases

Figure 2. Phonon spectrum of KNbO3 in the ferroelectric RR and OO phases at “dark” conditions. The RR phase is dynamically stable at zero temperature (a) whereas the OO phase is not as shown by the thin red lines in (b). The TT-renormalized phonon spectrum of the OO phase renders dynamical stability around room temperature (blue lines) in agreement with the experimental observations [9, 10].

The Gibbs free energy (GG) of the polar RR and OO phases was estimated with a numerical uncertainty of approximately 1 meV per formula unit (f.u.) as a function of temperature (i.e., based on the size of the employed supercells and other technical parameters), using the DFT-rQHA approach described earlier. These GG functions enable the prediction of the temperature at which the phase transition R→OR \to O occurs, TtT_{\mathrm{t}}, which is determined from the condition GO(Tt)=GR(Tt)G_{O}(T_{\mathrm{t}}) = G_{R}(T_{\mathrm{t}}). Figure 3 presents our DFT-rQHA Gibbs free energy results. We found that, considering the numerical uncertainties of 1 meV/f.u. and given the proximity in energy (near-degeneracy) of the two polymorphs, the Gibbs free energy of the RR and OO phases could not be distinguished in the temperature range 0≤T≲2000 \le T \lesssim200 K. However, at higher temperatures it is clearly appreciated that GO<GRG_{O} < G_{R}. Consequently, from our calculations it can be assuredly concluded that the polar OO phase becomes stable at temperatures above ≈200\approx200 K (Figure 3), which is in reasonable agreement with the experimental transition temperature of 220 K [9, 10].

DFT-rQHA prediction of the T-induced R → O phase transition in KNbO3

Figure 3. DFT-rQHA prediction of the TT-induced R→OR \to O phase transition in KNbO3. The Gibbs free energy of the RR and OO phases cannot be distinguished in the temperature range 0≤T≲2000 \le T \lesssim200 K, taking into account the numerical uncertainties of 1 meV per formula unit. The polar OO phase is predicted to become stable at temperatures ≳200\gtrsim200 K (arrow) in fair agreement with the experiments [9, 10].

Subsequently, we computed the entropy, S=−∂G/∂TS = -\partial G/\partial T, of the RR, OO, and CC phases as a function of temperature and light irradiation conditions, denoted as S(T,x)S(T, x) (where x=d,lx = d, l stand for “dark” and “light” conditions, respectively). The resulting S(T,x)S(T, x) curves exhibit smooth behavior, as explicitly shown in Figure 4. Analogous to quasi-direct calorimetry experiments [23, 24, 25], the determination of the photocaloric isothermal entropy, ΔSPC\Delta S_{\mathrm{PC}}, and adiabatic temperature changes, ΔTPC\Delta T_{\mathrm{PC}}, can be readily inferred from these entropy curves [26]. In particular, we have that ΔSPC=S(T,l)−S(T,d)\Delta S_{\mathrm{PC}} = S(T, l) - S(T, d) and ΔTPC=T0(S,l)−T(S,d)\Delta T_{\mathrm{PC}} = T_{0}(S, l) - T(S, d), where T0T_{0} fulfills the condition S(T0,l)=S(T,d)S(T_{0}, l) = S(T, d) (black solid lines in Figure 4). We recall that, under sufficiently high light irradiation (or equivalently, for photoexcited electronic densities ≥nˉ=2.69⋅1021\ge\bar{n} = 2.69 \cdot10^{21} cm−3^{-3}) the polar RR and OO phases spontaneously relax into the nonpolar CC phase.

Image not extracted

Figure 4. Evaluation of photocaloric effects in KNbO3_3. (a) At low temperatures, for the polar RR (dark) and nonpolar CC (light) phases. (b) Around room-temperature, for the polar OO (dark) and nonpolar CC (light) phases. ΔSPC\Delta S_{\mathrm{PC}} and ΔTPC\Delta T_{\mathrm{PC}} can be straightforwardly determined from the shown entropy curves (black solid lines) as done in quasi-direct calorimetry experiments [23, 24, 25]. “Light” conditions correspond to a photoexcited electronic density of nˉ=2.69⋅1021 cm−3\bar{n} = 2.69 \cdot10^{21}\ \mathrm{cm}^{-3}.

Figure 4 encloses examples of our evaluations for ΔSPC\Delta S_{\mathrm{PC}} and ΔTPC\Delta T_{\mathrm{PC}}, considering different temperature intervals in which the polar RR and OO phases are observed to be thermodynamically stable. For instance, at a temperature of ≈220 K\approx220\ \mathrm{K} and considering the RR phase under dark conditions, we estimated a giant isothermal entropy change of 89.5 J K−1 kg−189.5\ \mathrm{J\ K^{-1}\ kg^{-1}} and an adiabatic temperature change of −8.0 K-8.0\ \mathrm{K} (ΔTPC<0\Delta T_{\mathrm{PC}} < 0 since the state stabilized with light has higher entropy). Similarly, at room temperature and considering the OO phase under dark conditions, we computed a substantial isothermal entropy change of 89.2 J K−1 kg−189.2\ \mathrm{J\ K^{-1}\ kg^{-1}} and an adiabatic temperature change of −9.2 K-9.2\ \mathrm{K}. These photocaloric results are highly promising for technological applications as they are comparable in magnitude to those obtained from state-of-the-art electrocaloric (EC) effects [3, 4, 5, 6, 7]. Furthermore, the magnitude of the ∣ΔTPC∣|\Delta T_{\mathrm{PC}}| predicted for KNO is approximately one order of magnitude larger than the EC adiabatic temperature changes measured for similar polar materials such as BaTiO3_3 (BTO) [27]. The relatively large ∣ΔTPC∣|\Delta T_{\mathrm{PC}}| as compared to ∣ΔTEC∣|\Delta T_{\mathrm{EC}}| (or equivalently, ∣ΔSPC∣>∣ΔSEC∣|\Delta S_{\mathrm{PC}}| > |\Delta S_{\mathrm{EC}}|) can be attributed to the unique nature of the light-induced phase transition: the nonpolar CC state is directly stabilized, bypassing the intermediate polar states that are typically encountered during conventional TT-induced transformations [9, 10]. Consequently, polar materials similar to KNO and presenting high Curie temperatures are probably most suitable for achieving substantial photocaloric responses.

A highly promising aspect of the photocaloric effect theoretically unveiled in this study, which sets it apart from other known caloric effects, is its operability over vast temperature intervals defined by the ranges of thermodynamic stability of the polar phases of the material. This exceptional characteristic is explicitly demonstrated in Figure 5a, where we represent the values of the ∣ΔSPC∣|\Delta S_{\mathrm{PC}}| and ∣ΔTPC∣|\Delta T_{\mathrm{PC}}| estimated for KNO over an unprecedented wide temperature range spanning more than a hundred Kelvin, encompassing room temperature. Interestingly, both the isothermal entropy and adiabatic temperature changes slightly but steadily increase with increasing temperature. For example, at T=380T = 380 K, these quantities amount to 102.2 J K−1^{-1} kg−1^{-1} and 12.50 K, respectively. This behavior arises from the fact that the variation of the entropy upon increasing temperature, or equivalently the heat capacity Cv=1T(∂S∂T)VC_v = \frac{1}{T}\left(\frac{\partial S}{\partial T}\right)_V, is somewhat higher for the nonpolar CC phase than for the polar OO phase (Figure 4). The likely persistence of PC effects at very low temperatures makes the proposed scheme an appealing alternative for achieving cryogenic temperatures, which can be particularly useful in applications such as quantum computing. This possibility warrants future detailed theoretical exploration due to the inevitable consideration of quantum nuclear effects [17].

Image not extracted

Figure 5. Extremely ample photocaloric operation conditions and envisaged solid-state refrigeration cycle. (a) The magnitude of the two photocaloric descriptors ∣ΔSPC∣|\Delta S_{\mathrm{PC}}| and ∣ΔTPC∣|\Delta T_{\mathrm{PC}}| evaluated for KNbO3_3 as a function of temperature. (b) A possible four-step solid-state refrigeration cycle based on the photocaloric effect unveiled in this study.

A possible four-step refrigeration cycle based on the photocaloric effect observed in KNO is envisioned (Figure 5b). (0) → (1) Adiabatic light exposure: Initially, starting from the polar low-entropy OO phase at room temperature, T0T_0, the KNO system is adiabatically irradiated until stabilizing the nonpolar high-entropy CC phase. As a result, the temperature of the KNO sample decreases, T1<T0T_1 < T_0 (ΔTPC<0\Delta T_{\mathrm{PC}} < 0). (1) → (2) Heat transfer: With the light conditions maintained, the KNO sample is placed in contact with the targeted body to be refrigerated, facilitating the transfer of heat towards KNO, ∣Q∣=T⋅∣ΔSPC∣|Q| = T \cdot|\Delta S_{\mathrm{PC}}|, until restoring its initial temperature conditions, T2=T0T_2 = T_0. (2) → (3) Adiabatic light removal: The light is adiabatically switched off, stabilizing the polar low-entropy OO phase again. Consequently, the temperature of the KNO sample increases, T3>T0T_3 > T_0. (3) → (0) Heat removal: Finally, the KNO sample is

placed in contact with a heat sink, allowing heat to flow from KNO until restoring the initial temperature conditions, thereby completing a cooling cycle.

In conclusion, we have presented compelling theoretical evidence for the existence of giant photocaloric effects driven by light absorption in the archetypal ferroelectric perovskite KNbO3_3. The magnitude of the unveiled photocaloric effects is remarkably large, that is, comparable to the most promising electrocaloric effects measured in ferroelectrics. An unparalleled and distinctive feature of these photocaloric effects is that they remain very large over vast temperature intervals spanning several hundred Kelvin, only limited by the ranges of thermodynamic stability of the polar phases of the material. Therefore, this study provides motivation for exploring new concepts and strategies to develop environmentally friendly and highly miniaturizable solid-state cooling technologies.

Acknowledgements

We acknowledge financial support by the Spanish Ministry of Science under the grants PID2020-119777GB-I00, TED2021-130265B-C22, PID2020-112975GB-I00, RYC2018-024947-I and CEX2019-000917-S, and by the Generalitat de Catalunya under the grants 2021SGR-00343 and 2021SGR-01519. Computational support was provided by the Red Española de Supercomputación under the grants FI2022-1-0006, FI-2022-2-0003 and FI-2022-3-0014. We thank J. Carrete for useful discussions.

References

  1. [1]X. Moya and N. D. Mathur, Science 370, 797 (2020).
  2. [2]L. Mañoa, A. Planes, Adv. Mater. 29, 16030607 (2017).
  3. [3]E. Defay, R. Faye, G. Despesse et al., Nat. Commun. 9, 1827 (2018).
  4. [4]Y. Nouchkwe, P. Lheritier, Ch. Hong et al., Nat. Commun. 12, 3298 (2021).
  5. [5]S. E. Shirsath, C. Cazorla, T. Lu et al., Nano Lett. 20, 1262 (2020).
  6. [6]C. Menéndez, R. Rurali and C. Cazorla, Phys. Chem. Chem. Phys. 25, 17450 (2023).
  7. [7]J. Li, A. Torelló, V. Kovakova et al., Science 382, 801 (2023).
  8. [8]C. Paillard, E. Torun, L. Wirtz, J. Íñiguez and L. Bellaiche, Phys. Rev. Lett. 123, 087601 (2019).
  9. [9]D. Wang, G. Wang, Z. Lu et al., Front. Mater. 7, 91 (2020).
  10. [10]G. Shirane, R. Newnham and R. Pepinsky, Phys. Rev. 96, 581 (1954).
  11. [11]G. Kresse and J. Furthmüller, Phys. Rev. B 54, 11169 (1996).
  12. [12]P. E. Blöchl, Phys. Rev. B 50, 17953 (1994).
  13. [13]J. P. Perdew, A. Ruzsinszky, G. I. Csonka et al., Phys. Rev. Lett. 100, 136406 (2008).
  14. [14]C. Menéndez and C. Cazorla, Phys. Rev. Lett. 125, 17601 (2020).
  15. [15]B. Peng, Y. Hu, S. Murakami et al., Sci. Adv. 6, eabd1618 (2020).
  16. [16]A. Togo and I. Tanaka, Scr. Mater 108, 1 (2015)
  17. [17]C. Cazorla and J. Boronat, Rev. Mod. Phys. 89, 035003 (2017).
  18. [18]A. Carreras, A. Togo and I. Tanaka, Comput. Phys. Commun. 221, 221 (2017).
  19. [19]C. Cazorla, S. Bichlmaier, C. Escorihuela-Sayalero et al., Nanoscale DOI:10.1039/D4NR00100A (2024).DOI
  20. [20]Y. Kaifu, T. Komatsu, T. Hirano et al., J. Phys. Soc. Jpn. 23, 903 (1967).
  21. [21]A. I. Frenkel, F. M. Wang, S. Kelly et al., Phys. Rev. B 56, 10869 (1997).
  22. [22]M. D. Fontana, G. Metrat, J. L. Servoin et al., J. Phys. C: Solid State Phys. 17, 483 (1984).
  23. [23]P. Lloveras, A. Aznar, M. Barrio et al., Nat. Commun. 10, 1803 (2019).
  24. [24]A. Aznar, P. Lloveras, M. Barrio et al., J. Mater. Chem. A 8, 639 (2020).
  25. [25]A. Aznar, P. Negrier, A. Planes et al., Appl. Mater. Today 23, 101023 (2021).
  26. [26]C. Escorihuela-Sayalero, L. C. Pardo, M. Romanini et al., npj Comput. Mater. 10, 13 (2024).
  27. [27]F. Han, Y. Bai, L.-J. Qiao et al., J. Chem. Mater. C 4, 1842 (2016).

Paper details

vlnc:W_hfdxckr9
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents