Introduction

Current commercial and residential cooling devices, such as air conditioners, refrigerators, etc., are mostly based on vapor-compression technology1. The releases of gas refrigerants used in this mature technology have generated and exacerbated the global climate change, e.g., global warming2,3,4. Solid-state cooling technology based on caloric effects (CEs) has the potential to revolutionize vapor-compression refrigeration technology owing to its high energy efficiency and zero-greenhouse gas emissions1,2,3. CEs, as the core principle of solid-state cooling technology, are typically characterized by an adiabatic temperature change (ΔTadi) and an isothermal entropy change (ΔSiso) when the working material undergoes a reversible first-order phase transition on the application or removal of an external field5,6,7,8. CEs were first investigated by Joule with natural rubber (NR) as the research object9. At present, CEs can be classified into magnetocaloric, electrocaloric, elastocaloric, barocaloric and twistocaloric effects depending on the nature and mode of action of the external field10,11,12,13,14,15. In particular, due to the enormous elastocaloric effects (E-CEs) found in the martensitic transition of shape-memory alloys (SMAs) around room temperature16,17,18,19,20, solid-state cooling technology based on E-CEs is considered to be the best promising alternative to conventional refrigeration devices21,22,23,24. However, the corresponding tensile stress (σ) values of SMAs can reach several hundreds of megapascal, which remains a challenging issue for practical application (Supplementary Table 1).

The σ of a polymer working medium is one or two orders of magnitudes smaller than those applied to SMAs, but the commercialization of elastocaloric refrigeration based on polymer-tension technology is obstructed by their low E-CEs. Enhanced E-CEs via strain-induced crystallization (SIC) in NR at room temperature were reported recently25,26,27. Polyvinylidene difluoride based on the α-β phase transition also shows promising E-CEs28,29. Nevertheless, the regular arrangement of polymer molecular chains into a lattice is constrained due to the relatively high molecular weight, resulting in restricted crystalline rates with limited latent heat contributions. In contrast, E-CEs caused by reversible conformational changes of polymer molecular chains are independent of the amorphous-crystalline transition. When a single polymer molecular chain transitions from curled to straight, large conformational adjustments are required for this process. The conformational changes of the integrated network chains in a polymer working medium from curled to straight would be astronomical, which would result in enormous E-CEs. However, the conformational changes during the uniaxial tensile processes of elastomers are constrained by the uniformity of the molecular chain-length. Therefore, we speculate that an effective path to enhancing the E-CEs in polymer elastomers is to improve the uniformity of the polymer molecular chain-length.

In this study, a series of commercial triblock poly(styrene-b-ethylene-co-butylene-b-styrene) (SEBS) thermoplastic elastomers (TPEs), with polystyrene (PS) blocks (≈30 wt%) surrounding a central block of poly(ethylene-co-butylene) (PE/PB) segments (≈70 wt%), were selected to demonstrate the dependence of the E-CEs on the chain-length uniformity in SEBS (“Methods”). Furthermore, a rotary-motion cooling device was tailored to high-strains characteristics of rubbers, which effectively discharged the cooling energy of polymer elastomers.

Results

E-CEs in SEBS films

The molecular properties and sample labeling of these SEBS samples are summarized in Supplementary Table 2 and Supplementary Fig. 1. The ΔTadi associated with the E-CEs in the TPEs was induced under a uniaxial-strain-controlled system in an open indoor environment (Methods, Supplementary Fig. 2). Thus, the simplification of the ideal elastocaloric cooling cycle could be analogically derived from the reverse Brayton cycle as shown in Fig. 1a30. To quantify the E-CEs, ΔTadi can be directly measured by an infrared thermometer when TPEs undergo a strain rate (15 s−1) higher than the adiabatic strain rate (Supplementary Discussion 1, Supplementary Fig. 4). Infrared measurements provide the resulting temperature variations of TPEs with the narrowest molecular weight distribution (TPEs-1) during a single elastocaloric cooling cycle (Fig. 1b, c). During the adiabatic stretching process, the surface temperature of the TPEs-1 was found to increase significantly from the ambient temperature Ta = 299.0 K to the temperature T2 = 313.5 K when the strain abruptly increased from 0% to 600%. During the exothermic process, the surface temperature decreased exponentially to room temperature (Supplementary Fig. 5). When the TPEs-1 sample returned to the initial length, the surface temperature dropped from Ta = 299.0 K to T4 = 283.7 K, which corresponding to a ΔTadi = −15.3 K. Finally, TPEs-1 returned to room temperature through an isothermal endothermic process.

Fig. 1: E-CE cycle and temperature change of TPEs-1.
figure 1

a Schematic T-S diagram of the E-CE cooling cycle. The insets show the variations of the sample temperature as a function of strain. During stage 1–2, the sample is stretched adiabatically by a uniaxial tension, increasing the surface temperature (Ta → T2). During stage 2–3, the strain remains constant, and the thermal entropy decreases isothermally until T = Ta. During stage 3–4, the specimen retracts adiabatically with the unloading of the tension, further decreasing the temperature of the specimen (Ta → T4). During stage 4–1, the thermal entropy of the material increases until T = Ta. b Typical strain evolution (analogized from the displacement of the slide table) and the resulting temperature variation of TPEs-1 during a single E-CE cycle. c Infrared thermal images of different processes recorded using an infrared thermal imager.

Full size image

Origin of E-CE entropy change and its influencing factors

In the initial state, PS domains with hexagonal packed cylindrical (HPC) microstructures were randomly embedded in the amorphous PE/PB matrix (Fig. 2a). PS hard segments, as physical crosslinking points, and PE/PB soft segments, as the elastic matrix, were linked by chemical bonds to transfer applied tension. As shown in Fig. 2b, a wide and bright dispersion ring of the unstretched TPEs-1 appeared near q = 1.33 Å−1. The relative intensity of the dispersion ring along the stretching direction diminished as the strain increased, while that perpendicular to stretching direction was continuously concentrated and increased without any indication of crystallization. The two-dimensional wide-angle X-ray diffraction (2D WAXD) pattern of TPEs-1 returned to the wide dispersion ring when the strain returned to the initial state (Supplementary Fig. 6). This indicated that during the stretching–recovery cycle, the amorphous PE/PB matrix only underwent reversible transitions from curled chain conformations to oriented chain conformations along the stretching direction without any phase transition such as SIC (details in Supplementary Discussion 2, Supplementary Figs. 7–10). The ΔSiso of TPEs-1 at Ta = 298 K was estimated to be 145 J kg−1 K−1 by integrating \(\frac{{c}_{{{{{{\rm{p}}}}}}}}{T}\) from T4 to Ta (Fig. 2c, Supplementary Eq. (12)), where cp is the specific heat capacity of TPEs-1. This superior ΔSiso value stemmed from the excellent uniformity of the molecular chain-length in TPEs-1.

Fig. 2: Origin of E-CE entropy change and its influencing factors.
figure 2

a TEM images of TPEs-1 at a strain of 0%. PS domains (black regions) were stained dark by RuO4, and the light regions correspond to PE/PB domains. b 2D WAXD diffractograms of TPEs-1 at different strain levels. c Heat capacity cp and cp/T versus temperature during the heating process of TPEs-1. The green shadow represents the integral region of cp/T from T4 to Ta. d Schematic diagram of PE/PB conformational evolution in the stretching process of TPEs with different MRs. The red fibers, blue fibers and orange cylinders represent straight PE/PB segments, curled PE/PB segments and PS segments, respectively. e GPC curves of TPEs with different MRs. f |ΔTadi| versus MR of TPEs within the ultimate strains (former one before fracture strain) during the cooling process.

Full size image

Upon tensile stress, shorter molecular chains in the TPEs network would be straightened first, which would inhibit the further stretching of longer molecular chains (Fig. 2d) and consequently obstruct some conformational transformations. These obstructed conformational transitions resulted in lower apparent values of the E-CE. The obstructed conformational transitions were related to the uniformity of the molecular chain-lengths. The uniformity of the molecular chain-length can be characterized by the molecular weight distribution range (MR), which is defined as the difference value between the maximum and the minimum molecular weights of the complete molecular chains in a sample. Gel permeation chromatography (GPC) analysis indicated that the seven samples possessed different MRs (Supplementary Table 2, Fig. 2e). TPEs-1 possessed the smallest MR, suggesting that its molecular chain-length was the most uniform. As illustrated in Fig. 2f and Supplementary Figs. 18, 19, the |ΔTadi| of the TPEs decreased from 15.3 to 9.0 K with the increase in MR, indicating that the higher the MR was, the more obstructed the conformational transitions in the TPEs became. As shown in Supplementary Fig. 11b, even for TPEs-1, with the most uniform chain-length, the theoretical E-CE (|ΔTadi| = 23.6 K at a strain of 600%) was also much higher than the experimental E-CE under a large strain, highlighting the importance of reducing the obstructed conformational transitions by improving the uniformity of the molecular chain-length (details in Supplementary Discussion 3).

Material properties

TPEs can produce enormous E-CEs over a wide Ta range due to their broad rubber elastic plateau region (Fig. 3a, b). For a temperature range of 238–353 K, |ΔTadi| exhibited a strong dependence on Ta. In the glass transition region of the PE/PB segment (Ta = 263–293 K, region II in Fig. 3b), the storage modulus (E’) of TPEs-1 increased with the decrease in temperature, which indicated that the PE/PB soft segments began to stiffen and the conformational transition became more difficult. Thus, the experimental value of |ΔTadi| in this region was lower than the theoretical value. This can also be evidenced by comparing Young’s modulus (E) values at different stretching temperatures (Supplementary Discussion 4, Supplementary Fig. 20). When Ta was below 263 K, the sample temperature was further reduced (ΔTc in Fig. 3a) owing to the recovery cooling, which caused the specimen temperature to reach a value closer to the glass transition temperature of the soft segments (S-Tg ≈ 247 K) and resulted in the freezing of elastic soft segments. It is reasonable that the recovery strain rate of the freezing segments dropped below the adiabatic strain rate, so that the effect of heat transfer caused by convection and radiation between TPEs-1 and the ambient environment was enhanced, which led to a significant drop in |ΔTadi|. This drop in |ΔTadi| was also observed when Ta increased above 328 K, which was related to the destruction of the crosslinking points at the glass transition temperature of hard segments (H-Tg). When Ta increased to 328 K, the sample temperature further increased (ΔTh in Fig. 3a) due to the tensile heating, which caused the specimen temperature to enter the region IV in Fig. 3b. In this region, the destruction of the PS crosslinking points led to the slip of the whole molecular chains accompanied by the irreversible deformation under tensile stress. Nonetheless, in the wide ambient temperature range of 263–328 K, TPEs-1 could yield a |ΔTadi| of about 11.5–16.1 K.

Fig. 3: Ambient temperature range and COPmat of TPEs-1.
figure 3

a Experimental and theoretical |ΔT| of TPEs-1 as a function of the ambient temperature under the ultimate strains during the cooling process. The theoretical change of |ΔTadi| is proportional to Ta when the elongation ratio (λ) and cp are constant, and the slope is estimated as the ratio of |ΔTadi| = 15.3 K to Ta = 298.15 K according to Supplementary Eq. (4) in Supplementary Discussion 3. b E’ and tan δ versus temperature curves of TPEs-1. Region I is the glass state region of TPEs-1. Region II is the glass transition region of the soft segments. Region III is the rubber elastic plateau region. Region IV is the viscous flow region. c Surface ΔTadi evolution of TPEs-1 during 20 times of elastocaloric cycles with 600% strain. d COPmat and input work of unit mass in TPEs-1 during elastocaloric cycles.

Full size image

The material coefficient of performance (COPmat) for TPEs near room temperature is described by the ratio of the cooling energy per unit mass (Q/m) to the input work per unit mass (ΔW/m) (details in Supplementary Discussion 5, Supplementary Eq. (13)). As illustrated in Supplementary Discussion 6, all TPEs generally exhibit stable cyclic behavior after their first 20 cycles. The cyclic behaviors of the first 20 times of E-CE cycles for TPEs-1 are shown in Fig. 3c. It is encouraging that the ΔTadi and the Q/m of TPEs-1 remained constant during 20 cycles, explaining its outstanding and steady cycling performance. Nevertheless, TPEs-1 showed an extremely low COPmat value in the 1st cycle (Fig. 3d), which was mainly due to the large ΔW/m. As is well known, there are high internal stresses in polymer elastomer products, which are generated during the molding process, and it takes considerable work to eliminate these internal stresses. Gratifyingly, TPEs-1 exhibited an exceedingly stable COPmat value after the first cycle, suggesting that a single stretch could basically eliminate the internal stress of TPEs-1. The average COPmat of TPEs-1 during the stable cycles (2nd to 20th) was estimated to be 16.2 by Supplementary Eq. (13) (details in Supplementary Discussion 5, Supplementary Fig. 21).

Rotary-motion cooling device

The E-CE results (ΔTadi ≈ −15.3 K, ΔSiso ≈ 145 J kg−1 K−1) of TPEs-1 at room temperature exceeded those for all previously reported elastocaloric polymers (Supplementary Fig. 37), and thus, it is a feasible material for cooling devices. However, large deformation spaces are inherent for polymer elastomers (Supplementary Table 1). A rotary-motion cooling device based on the E-CE of TPEs-1 allowed a single cooling cycle to be designed for heat separation and output under high-strain with continuously flowing water (Fig. 4a, details in Supplementary Discussion 7 and Supplementary Fig. 38). The water current with room temperature flowed into the cavity from the top inlet over TPEs-1, and removing 600% of the strain cooled the outlet water by −1.1 K (Fig. 4b). According to Supplementary Eq. (14), The COPsys for the cooling process of the full system was estimated as 9.3. On the other hand, as the strain amplitude increased from 100% to 600% during the cooling processes, the temperature change of the outlet water (|ΔTout|) increased from 0.1 K to 1.1 K (Fig. 4b). In this process (no temperature span), the cooling power (SCPsys) of the system was determined by the temperature change of the cold water. As shown in Fig. 4c, the corresponding SCPsys of the rotary motion cooling device increased almost linearly with the increase of |ΔTout| (SCPsys values were calculated by Supplementary Eq. (15)). And the maximum value of SCPsys was about 1.9 W g−1. The comparison among the obtained SCPsys and that reported in the literature is summarized in Supplementary Table 4. Furthermore, the cold heat exchanger (E1) was replaced by a heat storage tank (details in Supplementary Discussion 7). After 10th cooling cycles with a constant strain amplitude of 600%, the thermal loss was stable and the system achieved a maximum temperature span of about 5.2 K.

Fig. 4: Principle and performance of rotary motion cooling device.
figure 4

a Schematic diagram of the cooling device. Stage 1: one end of sample is fixed on the body case (U1), and the other is fixed to the revolving disc (M). Stage 2: M rotates counterclockwise causing the elongation of sample. Water flows toward the hot heat exchanger (E2) to release heat to the environment. Stage 3: The sample maintains elongation until the sample and water temperature recover to Ta. Stage 4: M rotates clockwise causing the contraction of sample. Water flows toward the cold heat exchanger (E1) to absorb heat from the environment. The symbols V1 and V2 represent precision pumps. The symbols T represent thermocouples. U2 is a cavity that allows water flow and sample dilation. The mass of water in U2 is constant. The black dotted arrow points to the flow direction of the water. The white arrow points to the rotation direction of M. b Temperature changes of outlet water versus strain after a single cycle. The strain of sample was adjusted by controlling the rotation angle of M. c The SCPsys vs. ΔTout of the rotary motion elasto-based cooling device.

Full size image

Discussion

In summary, the enormous E-CEs found in the TPEs were a consequence of the reversible conformational changes at orientation structure transition, which could occur in a Ta range higher than the Tg of the elastic segments. In particular, a uniform chain-length significantly improved the E-CEs of the TPEs during deformation. We anticipate that the enhanced E-CEs will also be encountered in other polymers with uniform chain-lengths. Furthermore, the generated heat was separated and conducted of the rotary-motion cooling device, which was appropriate to the requirement of high strains. Our work would inspire the study of E-CEs in polymer elastomers, as well as the development of solid-state cooling devices that could help combat climate change4.

Methods

Materials

Triblock poly(styrene-b-ethylene-co-butylene-b-styrene) (SEBS) thermoplastic elastomers (TPEs) used in this study was provided by KRATON Polymers Co., Ltd. (U.S.), Sinopec Group Co., Ltd. (China) and TSRC Industries Co., Ltd. (Nantong, China). The weight fraction of polystyrene (PS) of SEBS elastomer was evaluated as ~30 wt%. The molecular properties of these SEBS samples were assessed by GPC. The weight average molecular weights from sample TPEs-1 to sample TPEs-7 were 76,000 g mol−1, 156,000 g mol−1, 208,000 g mol−1, 308,000 g mol−1, 318,000 g mol−1, 350,000 g mol−1 and 414,000 g mol−1 respectively. Details of the molecular properties and sample labeling of these samples are illustrated in Supplementary Table 2, Supplementary Fig. 1 and Fig. 2e.

Preparation

Sample sheets (0.4–1 mm thickness) were prepared by hot pressing process at 240 °C and 15 MPa for 20 m. Test specimens for two-dimensional wide-angle X-ray diffraction (2D SAXS), two-dimensional small-angle X-ray scattering (2D SAXS), Dynamic mechanical measurement (DMA) and mechanical characterization (ASTM D4482, active area is about 25 mm in length, 4 mm in width and 1 mm in thickness) and elastocaloric effects (E-CEs) measurement (active area is about 25 mm in length, 18 mm in width and 0.4 mm in thickness) were die cut from the same molded SEBS sheet. All samples were kept at the required ambient temperature for ten minutes before testing.

Characterization

The ΔTadi associated with the E-CEs in the TPEs was induced in an open indoor environment, heat sink and heat source in Fig. 1a are ambient air environment in this paper. The E-CE produced by the SEBS test specimens was carried out using a custom instrument, including stretching device and temperature measuring device (Supplementary Fig. 2). The stretching device comprised a precision ball screw and a linear guide rail, driven by an AC servo motor and controller (ECMA, Delta Electronics Co., LTD). Two ends of test specimen were mounted respectively on the endplate and the slide table of guide rail by jaw clamps. Servo motor drive the slide table to do precision repositioning movement on the linear guide rail, which cause the test specimen to undergo a controlled uniaxial strain. During the recovery process, the sider recovers its original position and the specimen contracts spontaneously. Sample strain is derived from the slider displacement. Direct measurement of ΔTadi was performed by the examination of sample surface temperature (details in Supplementary Discussion 1). The surface temperature of deforming test specimen was measured by on-line infrared thermometer (ABSD-01A, Aobosaide Automation Technology Co., LTD; BRW600-406, Hunan Firstrate Sencer Co., Ltd) and visualized by an infrared thermal imager (T890-2, Testo SE & Co. KGaA). Temperature information detected by infrared thermometer was recorded at a frame rate of 9 s−1, and it was averaged on an area of ~0.5 cm2 at a fixed infrared emissivity of 0.91. The molecular properties of these TPEs were assessed by a gel permeation chromatography (GPC, PL-GPC220, Agilent technologies, CA, USA). The GPC measurements were carried out at 40 °C using THF as the eluent at a flow rate of 1 mL min−1. Transmission electron microscopy (TEM) images were obtained on FEI Tecnai G2 F30 with an accelerating voltage of 300 kV. The samples were ultramicrotomed at −130 °C to a section with a thickness of about 70 nm. The sections were then stained with 0.5 wt% RuO4 vapor at 50 °C for 30 min in order to stain selectively polystyrene (PS) domains. RuO4 is a highly toxic vapor, please use it carefully. The structure evolution of SEBS samples was in situ monitored by 2D WAXS and SAXS, which were performed using Genix 3D X beamline with wavelength λ = 1.54 Å at Xeuss 2.0 (Xenocs). The sample-to-detector distances for SAXS and WAXS were set to be 2497.7 and 151.7 mm, respectively. The one-dimensional SAXS profiles were acquired by the circularly averaging of the intensity with the build-in software. The value of the scattering wave vector magnitude is given by \(q=\frac{4\pi \,\sin \,\theta }{\lambda }\) where 2θ is the scattering angle. Mechanical tests were run in an electromechanical universal testing machine (E44.104, MTS systems Co., LTD) equipped with a BSA-XS-50kg force transducer and a CEC1200 temperature testing chamber. DMA measurement of SEBS samples was carried out on a dynamic mechanical analyzer (DMA, PE-DMA8000) at 1 Hz with a heating rate of 2 K min−1 in the temperature range of 173–443 K. The specific heat capacity (cp) was confirmed by differential scanning calorimetry (DSC, TA-DSC2500) measurement with a heating rate of 10 K min−1 in the temperature range of 203–443 K.