Dynamic gating of infrared radiation in a textile

See allHide authors and affiliations

Science  08 Feb 2019:
Vol. 363, Issue 6427, pp. 619-623
DOI: 10.1126/science.aau1217

A cloth that adapts to the heat

Textiles trap infrared radiation, which helps keep us warm in cold weather. Of course, in hot weather, this is less desirable. Zhang et al. constructed an infrared-adaptive textile composed of polymer fibers coated with carbon nanotubes. The yarn itself expanded and collapsed based on heat and humidity, which changed the spacing of the fibers. Wider fiber spacing allowed the textile to breathe but also altered the infrared emissivity of the textile. This allowed for better heat exchange under hot and wet conditions. The self-adjusting emissivity of the textile could help toward wearable thermal-management attire.

Science, this issue p. 619


The human body absorbs and loses heat largely through infrared radiation centering around a wavelength of 10 micrometers. However, neither our skin nor the textiles that make up clothing are capable of dynamically controlling this optical channel for thermal management. By coating triacetate-cellulose bimorph fibers with a thin layer of carbon nanotubes, we effectively modulated the infrared radiation by more than 35% as the relative humidity of the underlying skin changed. Both experiments and modeling suggest that this dynamic infrared gating effect mainly arises from distance-dependent electromagnetic coupling between neighboring coated fibers in the textile yarns. This effect opens a pathway for developing wearable localized thermal management systems that are autonomous and self-powered, as well as expanding our ability to adapt to demanding environments.

Many species in nature have evolved elegant strategies to manipulate infrared (IR) radiation for heating and cooling purposes. For instance, Saharan silver ants possess triangular-shaped hairs that can reflect near-IR rays according to the position of the Sun to efficiently dissipate heat (1). For the human body, IR radiation makes up more than 40% of heat exchange with the environment; however, integrating an adaptive mechanism to control IR radiation in practical textiles is challenging (24). Although a few artificial structures, including photonic crystals (5), composite architectures (6), and nanoporous polyethylene films (7), have achieved radiative cooling, these technologies are nonresponsive to environmental changes, lacking an effective mechanism to bidirectionally regulate heating and cooling (810). Adaptive tuning of optical channels in response to thermal discomfort would fundamentally improve the functionality of clothing systems (8, 1114).

We can engineer the electromagnetic spectrum and wave propagation of thermal radiation by controlling distance-dependent electromagnetic interactions between conductive elements at length scales that are smaller than or comparable to the desired wavelength (1518). Incorporating tunable electromagnetic interactions that we can control at the fiber level of a fabric should allow us to produce a metatextile with smart and dynamically adaptive IR optical properties. In this study, we designed an IR-adaptive textile to directly regulate thermal radiation for personal thermal management using a few conceptual principles (Fig. 1). Each textile yarn is composed of a bundle of metafibers that function via (i) a meta-element (typically a conductive material) added in controlled quantities to the polymer textile fibers and (ii) an actuation mechanism that responds directly to changes in temperature and/or the relative humidity of skin. When hot and/or wet, the yarn collapses, bringing the meta-elements on neighboring fibers closer together to induce resonant electromagnetic coupling. The coupling shifts the emissivity of the textile to better match the human body’s thermal radiation, which effectively enhances heat exchange. When cold and/or dry, the yarn responds in an opposite manner to reduce heat dissipation. This makes it possible to effectively gate (i.e., “open” and “close”) the IR radiation through the textile in response to environmental change (Fig. 1). Accompanying this electromagnetic coupling–induced IR gating mechanism is the decrease of the interfiber spacing within the yarn. This process concomitantly enlarges pore sizes of the textile, which facilitates conventional heat exchange mechanisms (convective, conductive, and evaporative), as well as direct IR transmission (24), to synergistically enhance IR gating effects.

Fig. 1 Design principles of an IR gating textile.

Each yarn knitted into the textile is composed of multiple metafibers that contain IR-active nanostructures. The yarn is fluffy with large distances between the fibers. When hot and wet, the yarn collapses into a tight bundle, bringing the neighboring metafibers into resonant electromagnetic coupling that shifts the IR emissivity to spectrally overlap with that of the skin. This effectively “opens” the cloth to promote radiative cooling of the human body. When cold and dry, the reverse effect occurs. In this way, the coupled temperature–sweating response of humans to thermal discomfort is directly used to allow adaptive gating of radiation through the textile. The IR “open” state here corresponds at the yarn level to a collapsed (i.e., smaller interfiber spacing) structure but larger pores between yarns at the fabric level. This design allows this new adaptive IR gating mechanism to synergistically occur with conventional heat exchange channels, including convection and evaporation (not shown), to maximize personal cooling (when hot and sweating) and heating (when cold and dry).

We fabricated a series of regular arrays of carbon-coated polymer pillars with tunable interpillar spacing by three-dimensional (3D) laser direct-write lithography printing and investigated their IR response to illustrate the distance-dependent electromagnetic coupling (fig. S1). The total emissivity of each carbon pillar array evolves with the pitch of the pillars (Fig. 2A), which we determined by numerical integration of the Fourier-transform infrared (FTIR) absorption spectrum based on Kirchhoff’s law of thermal radiation. We also experimentally evaluated Au as an alternative meta-element and compared its performance with that of carbon. Although the Au-coated pillar arrays also manifest similar tunable emissivity, the resonant emissivity and tuning range are much smaller than those of the carbon-coated arrays. We visualized this interpillar spacing dependence of the emissivity using a calibrated microbolometer IR camera (Fig. 2B). The thermal radiation from the carbon-coated pillar arrays at a sample temperature of 150°C showed a strong, nonlinear dependence on the pitch between pillars. The radiance was low for small-pitch arrays, reaching a maximum at a pitch of ~6 μm and then dropping at increased spacing. We reproduced this dependence in all three replicate rows and for different sample temperatures. The absolute intensity changed with temperature, as we expected for blackbody radiators (fig. S1). Hence, we found a strong, nonlinear optical coupling effect among the carbon pillars within the array with thermal imaging and FTIR spectroscopy measurements.

Fig. 2 The IR emissivity strongly depends on the electromagnetic coupling distance between the nanostructures.

(A) Distance dependence of the emissivity of carbon (blue solid circles) and gold-coated (red circles) pillar arrays. Inset: SEM image from a small area of a pillar array. Scale bar: 5 μm. (B) IR images of the arrays featuring different spacing. The interpillar distance d [defined as center-to-center distance between two neighboring pillars as depicted in the inset of (A)] within each array (fabricated in triplicate) is 3.5, 4, 4.5, 5, 6, 7, and 8 μm, from left to right. The substrate temperature was set at 150°C.

To confirm that the tunable emissivity we observed was indeed due to electromagnetic wave interactions among the carbon-coated pillars (as prototypical meta-elements), we performed finite-difference time-domain (FDTD) simulations by solving Maxwell’s equations under the boundary conditions defined from the fabricated carbon pillar arrays. Our FDTD simulation results confirmed that the IR absorption spectra follow a strong dependence on interpillar spacing, in excellent agreement with the experimental measurements (fig. S2).

The metacoupling mechanism is not sensitive to the randomness and nonuniformity that typically exist in a textile. By incorporating random variations in spacing and orientation in the FDTD simulations, we found that in-plane spatial randomness reduced emissivity by only ~10%, even for large disorder of up to 30%. The coupling effect is also insensitive to axial randomness, causing only 2% variation in emissivity for 30% axial randomness (fig. S3). We conclude from our experiments and our modeling that electromagnetic metacoupling offers a facile and robust mechanism for directly and dynamically modulating the IR optical properties of textiles.

The simplified arrays and simulations motivated us to develop a method for fabricating metafibers that meet the requirements for self-adjustable emissivity in response to environmental changes (Fig. 3). We fabricated metafibers with few-walled carbon nanotubes (CNTs) as meta-elements to coat the fibers. Carbon-based materials have widely tunable emissivity (Fig. 2). CNTs are a special class of carbon-based metals that feature excellent thermal conductivity and are chemically stable, mechanically flexible, and lightweight (~1.3 g/cm3) (1921). In addition to these properties that make CNTs excellent candidates for meta-elements, CNTs are also polymers that allow the formation of strongly adhered and conformal coatings on commodity textile fibers (21, 22). To enable a robust actuating mechanism to control the electromagnetic coupling between metafibers, we used bimorph fibers composed of hydrophobic triacetate and hydrophilic cellulose as the base polymer materials (Fig. 3A). This choice offers seamless integration with our metafiber technology, because the bimorph fibers are already commercially available in the textile industry and their humidity response has been used to enhance ventilation in sportswear.

Fig. 3 Bimorph metafibers.

(A) A metafiber design based on CNT-coated triacetate-cellulose side-by-side bimorph fibers. (B) TEM image of the microtomed cross-section of a metafiber. Scale bar: 5 μm. (C) Photograph of a fabric knitted from the bimorph fibers. (D and E) Confocal fluorescent microscopy images showing the knitted fabric in the closed state (D) (low humidity) and the open state (E) (high humidity). (D) and (E) were acquired from the same region of the sample. To illustrate the side-by-side bimorph structure of the fibers, the hydrophilic cellulose component was dyed with an aqueous solution of rhodamine B (red), and the hydrophobic triacetate component was dyed using coumarin 6 (green) from a mixed organic solvent. Absorption of water at higher humidity causes the hydrophilic cellulose to swell more than the hydrophobic triacetate side, effectively actuating the spacing between the metafibers in response to the increased local humidity. Scale bars: 200 μm.

Each fiber is elliptically shaped, with triacetate (~14.5 μm by 9.4 μm) and cellulose (~13.6 μm by 9.9 μm) components fused side by side (Fig. 3A), as characterized by transmission electron microscopy (TEM) of the microtomed cross-sections (Fig. 3B). We incorporated the meta-element (carbon) into the fiber structure by simply coating the knitted fabric with few-walled CNTs in a process similar to solution dyeing (23). We knitted a piece of fabric 0.5 m2 in size to demonstrate the scalability of the technology (Fig. 3C). We further performed 3D FDTD simulations of the metafiber structures (fig. S4), which showed resonant emissivity and tunability similar to those of the pillar arrays in Fig. 2. As expected, the modeled performance was insensitive to the random arrangement of bimorph fibers within the yarn, thus validating the application of metafiber technology for the textile industry.

Owing to the competing hydrophobic and hydrophilic effects, the bimorph fibers can actuate to change interfiber spacing as a function of relative humidity, manifested as perspiration on human skin in response to the environment (14). Because the cellulose component of the fiber can absorb water molecules whereas the triacetate side cannot, this hydrophilic-hydrophobic side-by-side structure leads to differential expansion in different relative humidities, causing subsequent mechanical actuation of the fibers within the knitted fabric to enable dynamically tunable IR gating in clothing. We captured the fiber actuation in real time by imaging the fluorescently dyed fabric under a confocal fluorescent microscope. The bimorph fibers responded instantaneously to humidity changes (movie S1). At increasing humidity level, the interfiber spacing within the same yarn decreased by ~100 μm on average. In addition to the variation of individual yarns with reduced interfiber spacing, a constructive side effect was a corresponding increase in spacing between yarns knitted in the textile, thus resulting in larger pores in the fabric compared to the same textile at lower humidity levels (Fig. 3, D and E), which promotes synergetic convective, conductive, and evaporative cooling to maximize the efficiency of thermal management. This humidity-induced actuation is driven by changes in the relative humidity of skin in response to thermal discomfort (i.e., being hot and/or sweating), and once the skin humidity returns to a normal level, the textile regains its original state. This entire process does not require any external power to modulate the electromagnetic coupling in the yarn, and thus its IR properties function in a self-powered manner. Also, these structural changes occur locally, as revealed by real-time imaging (movie S1), preventing global distortion of the fabric, which would be undesirable for apparel. Such a dynamic system makes it possible to modulate IR radiation through textiles in response to stimuli by the human body.

We knitted the cellulose-triacetate bimorph fibers into fabric, coated them with a solution of few-walled CNTs, and characterized their IR response at controlled humidity levels using an FTIR spectrometer equipped with an environmental chamber (Fig. 4A). Using a humidity controller, we were able to program the humidity inside the environmental chamber (23). We visually captured the modulation of the IR emissivity of the metatextile using a calibrated high-resolution IR camera (Fig. 4, B and C). As the relative humidity increased from 10 to 75% at room temperature, the metayarns brightened owing to increased thermal emissivity compared to the interknitted polyethylene terephthalate (PET) yarns that served as a side-by-side control. The apparent dynamically responsive IR properties were effective over a broad relative humidity (5 to 90%) window, effectively covering the range of relative humidity found on skin under various environments (24, 25).

Fig. 4 Dynamic gating of IR radiation in response to changes in humidity.

(A) Schematic setup for experimental measurements of the IR response of the knitted fabric in specular transmission mode. (B and C) IR images from the same region of a fabric piece containing metayarns knitted into a PET matrix at 10% (B) and 75% (C) relative humidity. Scale bar: 2 mm. (D) IR gating of the metatextile (black line) at different relative humidity profiles. The fabric was coated with few-walled CNTs from a 1.8 μg/ml solution. The IR transmittance was averaged over the atmospheric transmission window (8 to 14 μm) and plotted against time. (E) IR response of the control fabric for (D) without any CNT coating (black line). The only difference between (D) and (E) is the CNT coating. (F) IR response of a conventional fabric made of PET fibers (black line). Overlaid in each of (D), (E), and (F) is the relative humidity profile (orange curve) during measurements.

We quantified this IR gating effect against controls using FTIR transmittance measurements (Fig. 4, D and E). The orange curves show the real-time relative humidity as programmed in the environmental chamber, whereas the black curves are synchronized changes in the relative IR transmittance. Notably, the metatextile showed a relative change in IR transmittance as high as 35.4%, corresponding to an absolute ~12% change after we corrected for scattering (Fig. 4D and fig. S5). The IR response that we observed was fast (<1 min) and reversible. We attributed the IR gating effect to the environmentally adaptive electromagnetic coupling mechanism based on our control experiments. To provide further support for this hypothesis, we prepared the triacetate-cellulose–based textiles from the same humidity-actuating fibers at identical conditions except that we omitted adding CNTs in the dyeing process (Fig. 4E). Although the control also responded to the humidity profile, we found only a 4% change in absolute transmittance after correcting for scattering. This control experiment indicated that modulation of direct IR transmission through the physical pores of the textile is a minor effect. We also fabricated and measured another set of control fabric samples made from commercially available PET, which, regardless of whether a CNT coating was applied or not, showed no IR tunability in response to humidity (Fig. 4F). This result confirmed the requirement of the fiber actuation mechanism in the observed IR gating effect. Finally, we excluded the possible response from CNTs alone by directly depositing CNTs on an IR window (fig. S6). Therefore, the large difference in IR response that we observed is clearly due to the contribution of distance-dependent electromagnetic coupling of CNTs when integrated in the metatextile.

We found that this IR response depends strongly on the CNT coating on the bimorph fibers. By simply varying the solution concentration of the CNTs used in the dyeing process, we could easily control the morphology and thickness of the CNT coating. We found that the maximum gating effect occurs at an unexpectedly low CNT solution concentration of ~1.8 μg/ml, corresponding to 1.6 ± 0.1 parts per million as estimated from scanning electron microscopy (SEM) imaging (fig. S7). After this coating, the fabric remains white. Under this condition, the coated CNTs form a low-density network, featuring a conductivity of just 9.7 ± 0.7 × 10−6 S/m. The IR gating effect is much weaker for both lower and higher coating densities. At lower coating densities the effect decreases, approaching the limit of meta-element–absent controls. For higher coating densities, electrical percolation occurs, which reduces electromagnetic meta-coupling (fig. S7). This observed nonlinear dependence on CNT coating density is consistent with our proposed mechanism that the observed IR gating effect mainly arises from the electromagnetic coupling between CNTs coated on neighboring fibers.

Although more work is needed to optimize the observed gating effect and to meet cost, safety, and human subject concerns (3, 4, 13, 14), our technology is readily compatible with commercial processes as the base fibers are made of commodity materials and the dyeing process is a standard part of textile manufacturing (4). The resultant metafibers can be knitted, dyed, and washed in a similar way to that of performance fabrics (figs. S8 and S9). This IR gating textile concept is not limited to the current metafiber bimorph design. There is a wide range of conductive materials that could be selected for the meta-elements (20, 26), as well as other temperature- and/or humidity-responsive mechanisms [e.g., thermoresponsive polymers (10, 27)], which should be adaptable to achieve performance similar to that demonstrated in this study. Because IR radiation is a primary mechanism for heat exchange between the body and the environment (2), the ability to harness this optical channel may lead to breakthroughs for wearable devices featuring localized thermal management capabilities that reduce the energy costs associated with heating and cooling in buildings or expand our ability to survive demanding environments (8, 11, 13).

Supplementary Materials

Materials and Methods

Figs. S1 to S9

References (2832)

Movie S1

References and Notes

  1. See supplementary materials.
Acknowledgments: We thank the DELTA team for constructive inputs, B. Davis (North Carolina State University) for assistance with knitting, and W.-A. Chiou (Maryland NanoCenter) for performing microtome and TEM analysis. Funding: We gratefully acknowledge the Advanced Research Projects Agency–Energy (ARPA-E), U.S. Department of Energy, for the financial support of this work as a part of the DELTA program under award DE-AR0000527. Author contributions: Y.H.W. and M.O. conceived and directed the project. X.A.Z., B.X., and Z.W. performed IR measurements and analyzed the data. S.Y. fabricated and characterized the pillar arrays and performed simulations. B.X., Z.P., M.L., S.D., Y.W., and X.W. prepared samples and performed general characterization. Y.H.W., M.O., X.A.Z., and S.Y. jointly wrote the manuscript. All authors discussed the results and commented on the manuscript. Competing interests: Y.H.W., M.O., Y.W., and S.Y. are inventors on a nonprovisional U.S. patent application (no. 15/159,666) related to this work. Data and materials availability: All data are available in the manuscript or the supplementary materials.

Stay Connected to Science

Navigate This Article