Low-k nano-dielectrics facilitate electric-field induced phase transition in high-k ferroelectric polymers for sustainable electrocaloric refrigeration (2024)

The overall performances

ND was selected owing to its high thermal conductivity, as we aimed to improve the heat transfer performance of polymeric nanocomposites. Besides the thermal conductivity, we anticipated that the introduction of low-k ND (ε = 5 ~ 6)^50,51 into P(VDF-TrFE-CFE) would also enhance the interfacial polarization, thereby improving the ECE. As schematically shown in Fig.1a, the permittivity of the ND was merely 1/8 of the polymeric matrix (ε ~ 45) at room temperature (RT) and hence would also cause the electric field distortion at the heterogeneous interface of composites as what their high-k counterparts do⁴⁵. Therefore, we hypothesized that the low-k fillers should enhance the ECE as well. Electron microscopic characterizations, e.g., transmission electron microscopy (TEM), and scanning electron microscope (SEM), indicated that the typical size of the filler is 5~10 nm as is shown in Fig.1b. When the NDs are dispersed into the polymer, some of the NDs will inevitably agglomerate together due to size effect (Fig.1c). The TEM results for T-ND-2.6% can be found in Supplementary Fig.19.

The laser flash thermal conductivity testing instrument was utilized to measure the thermal diffusivity of the base terpolymer and nanocomposites. The thermal conductivity of different EC materials can be easily obtained from the thermal diffusivity. As depicted in Fig.1d, the thermal conductivity of the EC materials exhibited an almost linear increase with the content of ND. The thermal conductivity of T-ND-1.2% and T-ND-2.6% was 35% and 75% higher than that of the base terpolymer, respectively. This trend is attributed to the large thermal conductivity of the ND particles to reduce the thermal resistance within the polymeric matrix, as schematically shown in Supplementary Fig.2. The incorporation of 2.6 vol% of the ND resulted in an increase of thermal conductivity Δλ of 0.15 W·m⁻¹·K⁻¹. The ratio Δλ/vol% is comparable to the best-reported values in the literature⁴⁴.

As naturally a great electrical insulator, ND-incorporation was observed to improve the electrical breakdown field of the EC nanocomposites, as evidenced by Fig.1e. T-ND-2.6% exhibited a 25% enhancement in the electrical breakdown strength compared to that of the base terpolymer. We developed a phase-field model to evaluate the breakdown process of the base terpolymer and T-ND-2.6%. The results of the numerical modeling corroborated well with our experimental results (see Supplementary Section 2.2). Owing to the lower permittivity of the ND than that of the base terpolymer, the induced large localized electric field would be shared on the side of the ND rather than the polymer, which is normally the case when one utilizes the high-k filler^52,53. As the ND exhibits a higher dielectric strength (over 400 MV·m⁻¹) than the relaxor ferroelectric polymer, the ND-incorporation improves the electrical stability of the nanocomposites (Supplementary Fig.4) as many other low-k fillers have shown^35,54.

Different from boron nitride nanosheet (BNNS), the other filler with high thermal conductivity, low-k and high dielectric strength that has been widely applied in EC polymer as fillers, the ND-incorporation observably enhanced the polarization of the base terpolymer rather than deteriorate the polarization as the BNNS does³⁵ (This is likely to indicate that the doping of BNNS reduces the ECE of the terpolymer⁴⁵, which can also be proved by our experiment in Supplementary Section 2.13). The EC-induced entropy changes (ΔS) and temperature changes (ΔT) of the base terpolymer and nanocomposites are presented in Fig.1f, g, respectively. Detailed methods of measuring the EC-induced heat flux and temperature changes were elaborated in Supplementary Section 2.4. The cooling performance of the nanocomposites was greatly improved compared to that of the base terpolymer, i.e., at 100 MV·m⁻¹, a ΔS ∼50.8 J·kg⁻¹·K⁻¹, corresponding to a ΔT ∼ 10.7 K, was induced in T-ND-2.6%, which is about 60% higher than that of the base terpolymer (ΔS ∼33.6 J·kg⁻¹·K⁻¹, corresponding to a ΔT ∼ 6.7 K).

To validate the directly measured results of the EC-induced ΔS, we further directly measured the ΔT via an infrared (IR) camera (see Supplementary Section 2.5). Owing to the heat loss that inevitably exists in the test condition, the IR measurement can only provide a result that is approaching the adiabatic temperature change (Supplementary Fig.10). It is observed that the results of the temperature measurement are in agreement with those of the heat flux measurement. The IR images under 50 MV·m^-1 for nanocomposite films (T-ND-2.6%) with the electrode pattern (SJTU) are clearly shown in Fig.1h, which is slightly lower than the result of the ΕC-induced ΔT deduced from the directly measured ΔS (Supplementary Fig.11).

Structural characterization

To understand the EC enhancement of the ND-incorporated EC nanocomposites, we explored the crystalline structural variation under the varying electric field. Small-Angle X-ray scattering (SAXS) and in-situ Wide-angle X-ray diffraction (WAXD) characterizations were employed to monitor the structural evolution of the base terpolymer and nanocomposites. As shown in Fig.2a, after the ND-incorporation, the peaks for the non-polar phase (close to α phase in PVDF, see Supplementary Section 2.7) of the nanocomposites shifted to a lower diffraction angle, which is corresponding to a 0.02 Å expansion of the spacing of the crystal plane compared to that of the base terpolymer^55,56. Our focus is on the distinctive Bragg peaks resulting from the two diffraction lines associated with the crystallographic planes of (200) and (110). This phenomenon can be correlated with the interplanar distance ratio, denoted as d200/d110, characterizing the orthorhombic pseudohexagonal phase, which is commonly identified as the relaxor ferroelectric feature^57,58,59. In addition, the average crystal size in T-ND-2.6% was reduced to 29 nm from 35 nm of the base terpolymer, as shown in Fig.2b. The decrease in the averaged size of crystalline structures in T-ND-2.6% indicates a significant rise in the number of crystallites and the presence of regions characterized by crystalline–amorphous interfaces. In addition, the reduced size of the crystalline structures could also be verified by differential scanning calorimetry (DSC) (Supplementary Fig.12a, b). Although the melting enthalpies of the base terpolymer and nanocomposites showed no clear distinctions, the melting temperatures of the nanocomposites were observably lower than that of the base terpolymer (133.5 °C vs. 135.6 °C), which corroborates the results from the WAXD tests.

The extrinsic, small ND fillers introduce a dramatically increased area of the heterogeneous interface, which could strongly modulate the crystallization of the base terpolymer⁶⁰. The expanded spacing of the crystalline plane and the reduction of the crystalline size would contribute to a more randomized polar structure that could enhance the ECE in the ND-incorporated nanocomposites. The interface between the organic and inorganic components may accommodate numerous defects, acting as nucleation sites to promote the creation of diminutive crystallites. These crystallites tend to be isolated and spatially localized, thereby impeding the long-range ordering within the polymeric matrix. SAXS data in Fig.2c (also see Supplementary Section 2.7 and Supplementary Fig.14) indicates crystal thinning along the chain direction, evidenced by a correlation peak representing the periodic organization of crystalline lamellae. The long period (L_P) can provide insights into the arrangement of crystalline lamellae and amorphous regions. The reduction of the peak signal corresponding to the long cycle indicates that NDs may suppress the long-range ordering in the polymeric matrix.

Achieving a substantial ECE at low fields demands an efficient transition from a high-polar-entropy state to a low-polar-entropy state¹¹. We conducted the in-situ WAXD to study the field-induced structural change in the base terpolymer and nanocomposites (Fig.2d, e), respectively. T-ND-2.6% presented a much stronger field-induced variation of the crystalline structure. Quantitatively, T-ND-2.6% showed that about 20 vol% of the non-polar phase transformed into the polar phase when the electric field was increased to 50 MV·m⁻¹, whereas for the base terpolymer only 8 vol% was converted (Fig.2f and Supplementary Table1). Hence, the low-k ND could reduce the energy barrier of phase transitions in the terpolymer, which would work with the enhanced polar randomness to further enhance the ECE.

Dielectric analysis

To look into the correlation between the modified crystalline structure in the ND-incorporated nanocomposites and the enhanced ECE, we further studied the dielectric properties of the nanocomposites. According to the Landau–Devonshire phenomenological theory, the relationship between the change in entropy of EC materials and the change in polarization is described below^61,62,63,

where E_l is the low electric field and E_h is the high electric field. Temperature-dependent permittivity and the polarization-electric field (P-E) loops were characterized, respectively. Therefore, the β coefficient is essentially describing how efficient the maximum polarization is in generating the ECE. Owing to the low-k nature and the low content of ND, we observed no significant enhancement in the permittivity of the nanocomposites near RT (Supplementary Fig.15) and a much-enhanced polarization under high electric fields, albeit the reduced crystalline size. The phenomenon indicates the ND-incorporation does not change much of the polar correlation. The above analysis suggests that the origin of the consequently enhanced ECE is different from the polar high entropy polymers¹¹. Therefore, the enhanced ECE may be attributed to the increased polarization that was induced at the internal interfaces.

To visually verify the existence of the interfacial polarization, we employed EFM to evaluate the interfacial coupling effect of the ND-incorporated nanocomposites (See Supplementary Section 2.10 for details). The surface morphology and potential distribution of the nanocomposites are shown in Fig.3a, b, respectively. Owing to the low-k, electric potential at the locations of ND was in average lower than that at the polymeric matrix. However, a significant build-up of potential can be observed at the interface between the polymeric matrix and inorganic fillers (Fig.3b). In principle, the dipolar polarization can largely affect the surface potential of ferroelectric polymers⁴⁶. Therefore, the microscopic images suggest the origin of the enhanced polarization after the incorporation of the low-k fillers.

On the other hand, the dielectric losses of the nanocomposites were greatly enlarged at temperatures slightly above the RT, which is consistent with the P-E loops at RT (Supplementary Fig.17). The increased dielectric loss may be attributed to the random defects induced by the ND-polymer interface, which facilitates the hopping of the charged carriers. We observed that the conduction loss was increasing with the increasing content of ND, which may cause errors when we study the contribution of the orientation polarization that directly contributes to the ECE.

To single out the orientation polarization, we subtracted the conduction loss to correctly reveal the polarization-entropy correlation⁶⁴. The conduction loss from leakage current contributes to the nominal polarization that was measured, so we subtracted the contribution from the measured nominal polarization to obtain the real polarization and the correction method can be found in Supplementary Section 2.8. The corrected P-E loops of the base terpolymer and nanocomposites under 100 MV·m^-1 at 10 Hz and RT are shown in Fig.3c. The maximum polarization was significantly increased in the nanocomposites. T-ND-2.6% exhibited an induced polarization of 0.066 C·m⁻² under 100 MV·m⁻¹, which was about 37.5% higher than the base terpolymer (0.048 C·m⁻²). Combing the reduction of the energy barrier and the enhanced interfacial polarization⁴⁷, the ND-incorporated nanocomposite exhibited the modified field-induced entropy changes compared to the base terpolymer.

In ferroelectrics and dielectrics, not all polarization could contribute to the ECE⁶⁰. The effectiveness of polarization-induced entropy changes can vary over a great range¹¹. To determine how effective the interfacial polarization induces ECE, we studied the β coefficient in the Landau–Devonshire phenomenological theory (Eq. (2)). According to Fig.3d, the β coefficient of the ND-incorporated nanocomposites was slightly smaller than that of the base terpolymer. Although the overall ECE was enhanced, the reduced β coefficient indicated that the interfacial polarization is not as effective as the base terpolymer in terms of generating polar entropy changes. The polar inactive carbon material may restrain the polymer chain mobility under the electric field⁶⁵. It is important to note that significantly enhancing the local electric field could overcome the reduced β coefficient and improve the overall ECE. In addition, we compared the volumetric enhancement of ECE (the ratio between the EC enhancement in percentage and the volume percentage of the filler) at 100 MV·m⁻¹ and RT, for different nano-fillers. The ND introduced the highest volumetric EC enhancement (~23%/vol%) when compared with other fillers in the previous study^45,47,48 (see Supplementary Fig.8).

In addition, we note that the low-k ND incorporation could efficiently enhance the temperature independency of the ECE. For conveniently designing an EC refrigeration device, its working body is expected to feature a temperature-invariant cooling capability across a broad temperature range. The EC-induced temperature changes of the base terpolymer and T-ND-2.6% were tested from 0 to 70 °C, as shown in Fig.4a, b. Under 100 MV·m⁻¹, T-ND-2.6% can reach an EC-induced temperature change of more than 6 K in a temperature range of over 70 K, while the corresponding temperature range for the base terpolymer was approximately 40 K.

The improved ECE, EC strength, and temperature independency in the T-ND-2.6% could lower the overall carbon emission of a potential EC device. To quantify the emission reduction, we studied the material-specific total equivalent warming impact (TEWI_mat) of the working body under a fixed device working condition. The TEWI_mat is a measure of the greenhouse gas emissions associated with a particular product, including its activity, or process, expressed in terms of its impact on global warming²⁰. It takes into account both direct emissions (if a refrigerant with a finite GWP was used), as well as indirect emissions (electricity consumption) associated with the energy used to operate the product, activity, or process.

TEWI is a commonly-used metric for evaluating energy-consuming technologies with CO_2-associated origins. In conventional vapor compression refrigeration systems, TEWI comprises two key components: (1) direct emissions of high GWP refrigerant vapors and (2) electricity consumption, quantified as an equivalent CO₂ emission, contributing indirectly to TEWI.

where CO_{2, direct}, CO_{2, indirect} are the direct and indirect equivalent CO₂ emission.

where x denotes the weight of vapor emissions, Q represents the cooling capacity of the refrigeration system, \({T}_{{{{{{{\rm{CO}}}}}}}_{2}}\) signifies the average mass of CO₂ released per kWh of electrical energy in power plants, and L stands for the lifetime (in years) of the refrigeration system.

TEWI_mat is closely related to the material COP of the EC material, as detailed by Qian et al.¹⁴. Specifically for caloric cooling, COP_mat can be represented as:

where W_net represents the net input work per unit mass, T_C and T_H denote the temperatures at the cold and hot ends, respectively, and Hy stands for the hysteresis in caloric materials.

Naturally, as a solid-state refrigerant, the TEWI of the EC material has no direct emission part (Fig.4c)⁶⁶. By fixing the working condition of a potential EC cooling device, we could use the TEWI to evaluate any EC material that can fit into the working condition¹⁴. The major input to the TEWI of the EC materials is the hysteresis and conduction loss of the materials during the electrical cycling. By utilizing the directly measured P-E loops, we can conclude that the TEWI of the ND-incorporated nanocomposites was significantly lower compared to the base terpolymer (from 30.1 kgCO₂-eq to 27.4 kgCO₂-eq). The concurrently enhanced ECE, EC strength, and reduced dielectric loss demonstrate the great potential of the ND-incorporated EC nanocomposites for a sustainable future of refrigeration.

System cooling performance evaluation

The performance of EC materials can be assessed through their application as core components in refrigeration devices. By analyzing the cooling power and efficiency of EC devices utilizing various materials, one can identify the most suitable EC material for a given refrigeration system. In this study, we proposed a design concept for a rotary EC device that incorporated fluid-solid conjugated heat transfer to meet the demand for highly efficient and compact cooling where the size and weight had high priority. The structure of the designed EC device is shown in Fig.5a and numerical simulation was conducted to further evaluate its performance.

The simulation parameters were obtained based on the experimental results mentioned previously. Referring to the previous study⁶⁷, deionized water was selected as the heat transfer fluid. The detailed working principles, system structures, initial and boundary conditions, and other information are presented in Supplementary Section 3. The primary objective of improving the thermal conductivity of EC materials is to enhance the heat exchange between EC materials and heat transfer fluid.

The hot fluid outlet is designated as “face 1,” while the cold fluid outlet is termed “face 2” (fluid flow occurs only during the fluid flushing stage, as observed in Supplementary Figs.29 and31). Once the system achieves stability, the deionized water temperature within the EC part and at the outlet cyclically varies with operating time. Figure5b, c present the temporal resolutions of average fluid temperatures on face 2 and the total cooling capacity, based on different EC materials.

As shown in Fig.5c, the optimal performance in terms of total cooling power and cooling power density (CPD) for the EC device is achieved when T-ND-2.6% is utilized as core components. This can be attributed to the exceptional cooling and heat transfer capabilities exhibited by T-ND-2.6% compared to the other two types of EC materials. At a temperature span of 10 K, the system can achieve a CPD of 28.24 W·cm⁻³ and total cooling power of 241.2 W with the utilization of T-ND-2.6%. Compared to the refrigeration system based on neat terpolymer, it offered more than a 10-fold cooling power improvement.

The comparative cooling efficiency of the EC device based on various materials is clearly shown in Fig.5d–f. Consistent with the trend observed for CPD, the system utilizing T-ND-2.6% as core components exhibited the highest coefficient of performance (COP), followed by T-ND-1.2%. These findings further underscore the notable benefits of incorporating the nanocomposites as core components in the designed EC device. When operating at a temperature span of 10 K, the refrigeration system utilizing T-ND-2.6% as core components exhibited a COP of 5.3, which is more than 6 times higher than that of the EC device based on the neat terpolymer. These results suggest that the incorporation of nanocomposites with superior heat transfer and cooling performance can significantly advance the research and application of high-power EC devices.

We can also determine the best working temperature span of certain EC materials based on the variation trend of 2^nd law efficiency. As depicted in Fig.5g, the EC device utilizing T-ND-2.6% as core components can achieve the maximum 2^nd law efficiency (17.6%) when operating at a temperature span of 10 K. In contrast, the efficiency of the refrigeration system based on the neat terpolymer is merely 2.6% under the same conditions. In other words, in terms of cooling efficiency, terpolymer-based cooling systems are more suitable for operation at a 5 K temperature span. This is also a good example of the advantages of the ND-incorporated nanocomposites, i.e., systems based on the nanocomposites can achieve better cooling efficiency and operate over a wider temperature span.

In addition, we developed a film-like EC oscillating refrigerator that is in a similar configuration and operating strategy as the previously reported EC device²⁷ (see Supplementary Section 4). The experimental results of this platform also illustrated the superior cooling capacity (the same CPD can be obtained at lower electric fields) and the better heat transfer performance (shorter heat dissipation time, see Supplementary Fig34) of the ND-incorporated nanocomposites compared to the base terpolymer.