news & views crosslinker, they were able to fabricate a family of synthetic ECM substrates with equivalent stiffness but varying pore size. On culturing MSCs and ASCs on these gels in the presence of differentiation factors, they found that stiffness rather than pore size is most predictive of spreading and lineage commitment (with soft gels promoting adipogenesis and stiff gels promoting osteogenesis, as expected). When they chemically varied the degree of ECM–protein anchoring (as verified by atomic force microscopy), they again found that MSC differentiation remained firmly dependent on ECM stiffness. The authors also showed that this result may be recapitulated by incorporating short adhesive peptides to the backbone of the PAAm gel rather than by conjugating fulllength ECM proteins to its surface. From their set of results, Engler and collaborators conclude that hydrogel ECM stiffness drives stem cell adhesion and differentiation in the absence of ligand-tethering effects (Fig. 1). Because stiffness is a bulk material property and cells are microscale entities with nanoscale force sensors, studies that attempt to bridge these dramatically different length scales are critical to understanding how cells sense and process stiffness cues from the ECM. In this respect, on top of the substantial effort that has been devoted to identifying molecular mechanosensors that channel stiffness cues, Mooney and co-authors’ study reveals an unexpected layer of nuance by showing that a given mechanosensor (integrin β4) can induce either normal or malignant behaviour depending on the abundance of an accessory factor (laminin). Moreover, the authors make rBM — which
is ubiquitous as a model matrix despite its lack of chemical definition and its limited capacity for tailoring — more amenable to mechanobiological studies by interweaving it with a semi-synthetic matrix system. And Engler and colleagues inject a fresh measure of confidence in studies that have used PAAm and other hydrogel substrates to probe stiffnessdependent effects. Given that for nearly two decades PAAm has been a mainstay tunable substrate system for mimicking the stiffness range of natural ECM11, this is no small contribution. Engler and colleagues’ findings also synergize with recent reports showing that effective ECM stiffness may be varied using silicone microposts of defined lengths, and that results in these systems are consistent with those obtained using two-dimensional hydrogels12. Furthermore, the studies of Engler, Mooney and collaborators expose major open questions and unmet needs. For instance, rBM is typically harvested from mouse tumour tissue and is acknowledged to suffer from batch-to-batch variability. Even if the alginate/rBM IPN system of Mooney and co-workers is significantly more controllable than rBM alone, it still requires rBM for the provision of bioactive matrix cues; hence, a fully defined epithelial-morphogenesis matrix system that may be reliably scaled up and could be applied clinically is clearly needed. And even if the stiffness of a hydrogel substrate influences cell behaviour independently of ligand tethering, this does not rule out the possibility that celladhesive tethers may play important roles in three-dimensional or other tissue-like matrix environments. ECM protein ligands
represent an important mechanical link between cell and scaffold, and it is at least conceivable that these linkages may vary over time and space in complex, dynamic microenvironments. For the same reason, the stiffening of actual tissue may be driven or accompanied by changes in ligand composition and density, making the end phenotype the net result of many interdependent inputs. Addressing these questions will again likely require tissuemimetic materials that preserve the design modularity of synthetic matrices while retaining the biological instructiveness associated with native matrices. Ample opportunity thus remains for materials scientists to make important contributions to cell biology, and the studies of Mooney, Engler and their colleagues provide valuable roadmaps for doing so. ❐ Sanjay Kumar is in the Department of Bioengineering, University of California, Berkeley, California 94720, USA. e-mail: [email protected] References
1. Pathak, A. & Kumar, S. Proc. Natl Acad. Sci. USA 109, 10334–10339 (2012). 2. Engler, A. J., Sen, S., Sweeney, H. L. & Discher, D. E. Cell 126, 677–689 (2006). 3. Keung, A. J., de Juan-Pardo, E. M., Schaffer, D. V. & Kumar, S. Stem Cells 29, 1886–1897 (2011). 4. Paszek, M. J. et al. Cancer Cell 8, 241–254 (2005). 5. Ulrich, T. A., Pardo, E. M. D. & Kumar, S. Cancer Res. 69, 4167–4174 (2009). 6. Pathak, A. & Kumar, S. Integr. Biol. 3, 267–278 (2011). 7. Trappmann, B. et al. Nature Mater. 11, 642–649 (2012). 8. Chaudhuri, O. et al. Nature Mater. 13, 970–978 (2014). 9. Wen, J. H. et al. Nature Mater. 13, 979–987 (2014). 10. Provenzano, P. P., Inman, D. R., Elicieri, K. W. & Keely, P. J. Oncogene 28, 4326–4343 (2009). 11. Pelham, R. J. & Wang, Y. L. Proc. Natl Acad. Sci. USA 94, 13661–13665 (1997). 12. Fu, J. P. et al. Nature Methods 7, 733–736 (2010).
THERMAL EMISSION
Ultrafast dynamic control
Control of thermal emission with microsecond switching times has been achieved by using sub-band transitions in composite quantum-well and photonic-crystal structures.
Ognjen Ilic and Marin Soljačić
T
he most familiar example of a thermal emitter is the Sun. Because of the thermal motion of charged particles, its 5,800-K hot surface most strongly emits radiation in the green (at a wavelength around 500 nm). More generally, the relationship between temperature of a black body and wavelength of emission is captured by Wien’s displacement law: the peak wavelength 920
of emitted light is inversely proportional to a black body’s absolute temperature. Hence, peaks of emission of hotter stars are shifted towards the blue, and those of colder stars towards the red. Thermal emission can also be controlled in the absence of a temperature change by incorporating periodic features with length scales comparable to the wavelength of thermally emitted light 1–3. Writing in
Nature Materials, Susumu Noda and colleagues have now pushed this concept a step further. They developed a composite structure that combines photonic crystals and quantum wells, allowing for dynamic control of thermal emission at speeds that are more than four orders of magnitude faster than conventional means of temperature modulation4. Photonic crystals exhibit a photonic
NATURE MATERIALS | VOL 13 | OCTOBER 2014 | www.nature.com/naturematerials
© 2014 Macmillan Publishers Limited. All rights reserved
news & views
a
b
Bulk Photonic crystal Emissivity
No bias With bias
Energy
Vbias(t)
Emissivity
bandgap that prevents light of certain frequencies from propagating through the crystal. Also, because of the unique way crystal periodicity affects the flow of light, photonic crystals can alter emission in drastic ways5. For example, the introduction of a patterned array of holes can enhance significantly the emission of a block of metal3 (Fig. 1a). The properties of the periodic structure, such as the shape and size of the individual periodic elements, and the lattice type and constant, directly determine the position and the size of the photonic bandgap as well as the location of the photonic crystal resonances. These resonances, in turn, influence optical properties of the structure, such as reflectance, transmittance and emittance. Once the desired pattern is implemented, however, this approach doesn’t allow for dynamic control of the resulting emission spectrum. Noda and co-authors have solved this issue by adding a multiple quantum well structure to an otherwise typical photonic crystal lattice. The composite structure is formed of thin gallium arsenide layers sandwiched between a wider-bandgap material, aluminum gallium arsenide, that confines the electrons to move in essentially two dimensions. Such confinement results in the formation of discrete energy subbands6, and it is the transition between these sub-bands that the authors exploit to modify thermal absorption. At the same time, a triangular lattice of air holes is introduced into the quantum-well layer, turning the structure into a photonic crystal (Fig. 1b). By carefully tuning the parameters of the quantum-well structure — such as its width, depth and number of layers — the authors match the absorption associated with a sub-band transition to one of the emission peaks of the photonic crystal. These emission peaks are manifestations of photonic crystal resonances, which have the benefit to both sharpen and enhance the spectral features of absorption. Traditional means of varying thermal emission of radiation by temperature modulation are limited to approximately hundred cycles per second, and are only possible in objects with a small thermal mass7. Noda and co-authors’ design dramatically outperforms, both in speed and magnitude, any previous methods of controlling thermal emission. The authors achieved such fast variation in emission by directly controlling the absorption rate of the quantum wells through electrical modulation with an external bias voltage. They report modulation rates that are as high as several hundred kHz, and argue that further optimization of the structure could increase this value to more than 10 MHz.
λtrans Distance
λtrans
Wavelength
Wavelength
Figure 1 | Structures for the control of thermal emission and emissivity spectra. a, The emission (solid line, bottom) from a metallic photonic crystal (top) has strong spectral selectivity relative to the emission of a bulk metal (dashed line, bottom). The shape of the periodic structure determines the positions of the emission peaks, and emission can be controlled by modulating the temperature T(t) of the entire device over time t. b, The composite structure of Noda and colleagues, which incorporates a multiple-quantumwell structure (green and yellow layers, top), allows for dynamic control of emission (bottom) by means of electrical tuning (through a time-dependent voltage bias, Vbias(t)). The wavelength λtrans of the inter-subband transition of the multiple-quantum-well structure (inset) is matched to one of the resonances of the photonic crystal, thereby enhancing the emission of light at that wavelength.
Moreover, the magnitude of the variation of the emittance is Δε = 0.5 (a perfect reflector has emittance of 1 whereas that of a perfect absorber is 0), an order of magnitude larger than that of previously reported methods8. The result is of fundamental interest, as it allows basic quantities in thermal physics, such as the amount of emitted heat by a body at a given temperature, to be dynamically controlled without changing the temperature of the emitting body itself. Furthermore, this can be achieved at rates that are much faster than the time required to achieve thermal equilibrium, opening avenues for the study of non-equilibrium heat processes, in particular for applications in spectroscopy as well as chemical and biological sensing. Still, the wavelength tunability that Noda and colleagues’ structures allow is constrained by the fact that sub-band transitions are generally restricted to infrared wavelengths and to transverse magnetic polarization of light, and by the structures’ limited potential for operation at high temperatures. Furthermore, the wavelength selectivity in these structures is still static: both the photonic
NATURE MATERIALS | VOL 13 | OCTOBER 2014 | www.nature.com/naturematerials
© 2014 Macmillan Publishers Limited. All rights reserved
crystal and the quantum-well layers are specifically — and irreversibly — designed to switch thermal emission at a specific wavelength. Although photonic crystals can be designed with almost arbitrary resonances, dynamic control of the wavelength of these resonances remains a challenge. Nevertheless, the work of Noda and co-workers is a marked progress towards a more complete control of thermal emission. ❐ Ognjen Ilic and Marin Soljačić are at the Research Laboratory of Electronics, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA. e-mail: [email protected]; [email protected] References
1. Arpin, K. A. et al. Nature Commun. 4, 2630 (2013). 2. Fan, S., Raman, A. & Yu, Z. Proc. Natl Acad. Sci. USA 107, 17491–17496 (2010). 3. Chan, D. L. C., Soljačić, M. & Joannopoulos, J. D. Opt. Express 14, 8785–8796 (2006). 4. Inoue, T., De Zoysa, M., Asano, T. & Noda, S. Nature Mater. 13, 928–931 (2014). 5. Joannopoulos, J. D., Meade, R. D. & Winn. J. N. Photonic Crystals: Molding the Flow of Light (Princeton Univ. Press, 1995). 6. West, L. C. & Eglash, S. J. Appl. Phys. Lett. 46, 1156–1158 (1985). 7. Hildenbrand, J. et al. IEEE Sensors J. 10, 353–362 (2010). 8. Vassant, S. et al. Appl. Phys. Lett. 102, 081125 (2013).
921
is ubiquitous as a model matrix despite its lack of chemical definition and its limited capacity for tailoring — more amenable to mechanobiological studies by interweaving it with a semi-synthetic matrix system. And Engler and colleagues inject a fresh measure of confidence in studies that have used PAAm and other hydrogel substrates to probe stiffnessdependent effects. Given that for nearly two decades PAAm has been a mainstay tunable substrate system for mimicking the stiffness range of natural ECM11, this is no small contribution. Engler and colleagues’ findings also synergize with recent reports showing that effective ECM stiffness may be varied using silicone microposts of defined lengths, and that results in these systems are consistent with those obtained using two-dimensional hydrogels12. Furthermore, the studies of Engler, Mooney and collaborators expose major open questions and unmet needs. For instance, rBM is typically harvested from mouse tumour tissue and is acknowledged to suffer from batch-to-batch variability. Even if the alginate/rBM IPN system of Mooney and co-workers is significantly more controllable than rBM alone, it still requires rBM for the provision of bioactive matrix cues; hence, a fully defined epithelial-morphogenesis matrix system that may be reliably scaled up and could be applied clinically is clearly needed. And even if the stiffness of a hydrogel substrate influences cell behaviour independently of ligand tethering, this does not rule out the possibility that celladhesive tethers may play important roles in three-dimensional or other tissue-like matrix environments. ECM protein ligands
represent an important mechanical link between cell and scaffold, and it is at least conceivable that these linkages may vary over time and space in complex, dynamic microenvironments. For the same reason, the stiffening of actual tissue may be driven or accompanied by changes in ligand composition and density, making the end phenotype the net result of many interdependent inputs. Addressing these questions will again likely require tissuemimetic materials that preserve the design modularity of synthetic matrices while retaining the biological instructiveness associated with native matrices. Ample opportunity thus remains for materials scientists to make important contributions to cell biology, and the studies of Mooney, Engler and their colleagues provide valuable roadmaps for doing so. ❐ Sanjay Kumar is in the Department of Bioengineering, University of California, Berkeley, California 94720, USA. e-mail: [email protected] References
1. Pathak, A. & Kumar, S. Proc. Natl Acad. Sci. USA 109, 10334–10339 (2012). 2. Engler, A. J., Sen, S., Sweeney, H. L. & Discher, D. E. Cell 126, 677–689 (2006). 3. Keung, A. J., de Juan-Pardo, E. M., Schaffer, D. V. & Kumar, S. Stem Cells 29, 1886–1897 (2011). 4. Paszek, M. J. et al. Cancer Cell 8, 241–254 (2005). 5. Ulrich, T. A., Pardo, E. M. D. & Kumar, S. Cancer Res. 69, 4167–4174 (2009). 6. Pathak, A. & Kumar, S. Integr. Biol. 3, 267–278 (2011). 7. Trappmann, B. et al. Nature Mater. 11, 642–649 (2012). 8. Chaudhuri, O. et al. Nature Mater. 13, 970–978 (2014). 9. Wen, J. H. et al. Nature Mater. 13, 979–987 (2014). 10. Provenzano, P. P., Inman, D. R., Elicieri, K. W. & Keely, P. J. Oncogene 28, 4326–4343 (2009). 11. Pelham, R. J. & Wang, Y. L. Proc. Natl Acad. Sci. USA 94, 13661–13665 (1997). 12. Fu, J. P. et al. Nature Methods 7, 733–736 (2010).
THERMAL EMISSION
Ultrafast dynamic control
Control of thermal emission with microsecond switching times has been achieved by using sub-band transitions in composite quantum-well and photonic-crystal structures.
Ognjen Ilic and Marin Soljačić
T
he most familiar example of a thermal emitter is the Sun. Because of the thermal motion of charged particles, its 5,800-K hot surface most strongly emits radiation in the green (at a wavelength around 500 nm). More generally, the relationship between temperature of a black body and wavelength of emission is captured by Wien’s displacement law: the peak wavelength 920
of emitted light is inversely proportional to a black body’s absolute temperature. Hence, peaks of emission of hotter stars are shifted towards the blue, and those of colder stars towards the red. Thermal emission can also be controlled in the absence of a temperature change by incorporating periodic features with length scales comparable to the wavelength of thermally emitted light 1–3. Writing in
Nature Materials, Susumu Noda and colleagues have now pushed this concept a step further. They developed a composite structure that combines photonic crystals and quantum wells, allowing for dynamic control of thermal emission at speeds that are more than four orders of magnitude faster than conventional means of temperature modulation4. Photonic crystals exhibit a photonic
NATURE MATERIALS | VOL 13 | OCTOBER 2014 | www.nature.com/naturematerials
© 2014 Macmillan Publishers Limited. All rights reserved
news & views
a
b
Bulk Photonic crystal Emissivity
No bias With bias
Energy
Vbias(t)
Emissivity
bandgap that prevents light of certain frequencies from propagating through the crystal. Also, because of the unique way crystal periodicity affects the flow of light, photonic crystals can alter emission in drastic ways5. For example, the introduction of a patterned array of holes can enhance significantly the emission of a block of metal3 (Fig. 1a). The properties of the periodic structure, such as the shape and size of the individual periodic elements, and the lattice type and constant, directly determine the position and the size of the photonic bandgap as well as the location of the photonic crystal resonances. These resonances, in turn, influence optical properties of the structure, such as reflectance, transmittance and emittance. Once the desired pattern is implemented, however, this approach doesn’t allow for dynamic control of the resulting emission spectrum. Noda and co-authors have solved this issue by adding a multiple quantum well structure to an otherwise typical photonic crystal lattice. The composite structure is formed of thin gallium arsenide layers sandwiched between a wider-bandgap material, aluminum gallium arsenide, that confines the electrons to move in essentially two dimensions. Such confinement results in the formation of discrete energy subbands6, and it is the transition between these sub-bands that the authors exploit to modify thermal absorption. At the same time, a triangular lattice of air holes is introduced into the quantum-well layer, turning the structure into a photonic crystal (Fig. 1b). By carefully tuning the parameters of the quantum-well structure — such as its width, depth and number of layers — the authors match the absorption associated with a sub-band transition to one of the emission peaks of the photonic crystal. These emission peaks are manifestations of photonic crystal resonances, which have the benefit to both sharpen and enhance the spectral features of absorption. Traditional means of varying thermal emission of radiation by temperature modulation are limited to approximately hundred cycles per second, and are only possible in objects with a small thermal mass7. Noda and co-authors’ design dramatically outperforms, both in speed and magnitude, any previous methods of controlling thermal emission. The authors achieved such fast variation in emission by directly controlling the absorption rate of the quantum wells through electrical modulation with an external bias voltage. They report modulation rates that are as high as several hundred kHz, and argue that further optimization of the structure could increase this value to more than 10 MHz.
λtrans Distance
λtrans
Wavelength
Wavelength
Figure 1 | Structures for the control of thermal emission and emissivity spectra. a, The emission (solid line, bottom) from a metallic photonic crystal (top) has strong spectral selectivity relative to the emission of a bulk metal (dashed line, bottom). The shape of the periodic structure determines the positions of the emission peaks, and emission can be controlled by modulating the temperature T(t) of the entire device over time t. b, The composite structure of Noda and colleagues, which incorporates a multiple-quantumwell structure (green and yellow layers, top), allows for dynamic control of emission (bottom) by means of electrical tuning (through a time-dependent voltage bias, Vbias(t)). The wavelength λtrans of the inter-subband transition of the multiple-quantum-well structure (inset) is matched to one of the resonances of the photonic crystal, thereby enhancing the emission of light at that wavelength.
Moreover, the magnitude of the variation of the emittance is Δε = 0.5 (a perfect reflector has emittance of 1 whereas that of a perfect absorber is 0), an order of magnitude larger than that of previously reported methods8. The result is of fundamental interest, as it allows basic quantities in thermal physics, such as the amount of emitted heat by a body at a given temperature, to be dynamically controlled without changing the temperature of the emitting body itself. Furthermore, this can be achieved at rates that are much faster than the time required to achieve thermal equilibrium, opening avenues for the study of non-equilibrium heat processes, in particular for applications in spectroscopy as well as chemical and biological sensing. Still, the wavelength tunability that Noda and colleagues’ structures allow is constrained by the fact that sub-band transitions are generally restricted to infrared wavelengths and to transverse magnetic polarization of light, and by the structures’ limited potential for operation at high temperatures. Furthermore, the wavelength selectivity in these structures is still static: both the photonic
NATURE MATERIALS | VOL 13 | OCTOBER 2014 | www.nature.com/naturematerials
© 2014 Macmillan Publishers Limited. All rights reserved
crystal and the quantum-well layers are specifically — and irreversibly — designed to switch thermal emission at a specific wavelength. Although photonic crystals can be designed with almost arbitrary resonances, dynamic control of the wavelength of these resonances remains a challenge. Nevertheless, the work of Noda and co-workers is a marked progress towards a more complete control of thermal emission. ❐ Ognjen Ilic and Marin Soljačić are at the Research Laboratory of Electronics, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA. e-mail: [email protected]; [email protected] References
1. Arpin, K. A. et al. Nature Commun. 4, 2630 (2013). 2. Fan, S., Raman, A. & Yu, Z. Proc. Natl Acad. Sci. USA 107, 17491–17496 (2010). 3. Chan, D. L. C., Soljačić, M. & Joannopoulos, J. D. Opt. Express 14, 8785–8796 (2006). 4. Inoue, T., De Zoysa, M., Asano, T. & Noda, S. Nature Mater. 13, 928–931 (2014). 5. Joannopoulos, J. D., Meade, R. D. & Winn. J. N. Photonic Crystals: Molding the Flow of Light (Princeton Univ. Press, 1995). 6. West, L. C. & Eglash, S. J. Appl. Phys. Lett. 46, 1156–1158 (1985). 7. Hildenbrand, J. et al. IEEE Sensors J. 10, 353–362 (2010). 8. Vassant, S. et al. Appl. Phys. Lett. 102, 081125 (2013).
921
