news & views
MATERIAL WITNESS
JOINT ENTERPRISE Tissue engineering is as much a mechanical as a biological problem. Although the field’s focus has been largely on getting cells to grow in vitro in the right morphology — a question increasingly now concerned with eliciting the requisite modes of stem-cell differentiation through appropriate environmental cues — there are many applications in which an engineered tissue’s dynamic properties are at least as important as the static ones. That’s why tissues associated with moving parts of the body, such as the trachea, remain among the hardest to ‘grow’ in a medically viable form. It is precisely because the mechanical performance of some biological tissues is so admirable that they present such a challenge. Of these, cartilage — the connective tissue at the ends of long bones — is among the most impressive. It can provide lubrication and joint articulation without compromising wear or damage for seven or more decades of constant use. Of course, the problem is that it does degenerate eventually, and the resulting osteoarthritis is a source of pain for millions of people. Although the static mechanical properties of native cartilage have been reasonably well matched in artificial tissues1, it is harder to replicate the lubricating function2. An alternative to growing real cartilage with the right mechanics is to mimic its properties in a
biocompatible material. Indeed, there is plenty worth emulating here in a synthetic material that might find tribological applications beyond medicine. It has been long recognized that cartilage, like so many biological tissues, has a hierarchical structure3; the question is how much of, or how closely, this structure need be copied to achieve acceptable mimicry of function. Previous efforts to make wholly synthetic cartilage have tended to fixate on just one aspect of its behaviour 4 — not because that is all that’s needed, but because it was all that’s feasible. But the tribological superiority of natural cartilage relies on a synergy that will only be attainable by approaching it as an integrated system. Greene et al. have now taken a step towards a more sophisticated cartilage mimic that combines two of the key features responsible for its lubrication5. One is that the tissue is a fluid-filled porous matrix in which most of the compressive load is carried by the fluid itself, as it is trapped and pressurized. This reduces the stress on the matrix, but it doesn’t alone account for lubrication and wear resistance. Furthermore, highly charged macromolecules called proteoglycans bound to the surface both resist collapse of the pores through electrosteric repulsion and provide detachable ‘boundary lubricants’ under shear.
PHILIP BALL In the analogue material developed by Greene et al., cellulose with submicrometre porosity mimics the collagen fibril network of cartilage, while the boundary lubrication is supplied by copolymers of the polyelectrolyte carboxymethylcellulose with polyethylene glycol. The resulting material shows not only low friction (albeit higher than real cartilage) but also the time-dependent static frictional response diagnostic of the fluid-pressurization mechanism of lubrication. This composite isn’t yet robust enough for medical applications, but it shows that the strategy is sound. ❐ References 1. Natoli, R. M., Revell, C. M. & Athanasiou, K. A. Tissue Eng. A 15, 3119–3128 (2009). 2. McNary, S. M., Athanasiou, K. A. & Reddi, A. H. Tissue Eng. B 18, 88–100 (2012). 3. Mow, V. C., Ratcliffe, A. & Poole, A. R. Biomaterials 13, 67–97 (1992). 4. Chen, M., Briscoe, W. H., Armes, S. P. & Klein, J. Science 323, 1698–1701 (2009). 5. Greene, G. W., Olszewska, A., Osterberg, M., Zhu, H. & Horn, R. Soft Matter http://dx.doi.org/10.1039/ c3sm52106k (2013).
SHAPE-MEMORY MATERIALS
Nanoscale oxides shape up
Reversible strains up to 14% driven by changes in temperature or electric field have been realized in a thin film of bismuth ferrite oxide.
Antoni Planes and Lluís Mañosa
B
end a paper clip, and its shape will persist even if the clip is heated up. This is because the deformed clip has undergone irreversible plastic deformation through the flow of dislocations. This type of crystallographic defect — which
6
can be viewed as misfit lines between crystallographic planes — confers ductility to metals because they can easily move under an applied load. Yet, in contrast to most metals, shape-memory alloys respond very differently to applied loads. In this
case, the response of the material is not controlled by dislocation flow, but by a transition between two crystallographic phases, a low-symmetry martensitic phase and a high-symmetry parent phase1. At the transition, and in the absence of external
NATURE MATERIALS | VOL 13 | JANUARY 2014 | www.nature.com/naturematerials
© 2014 Macmillan Publishers Limited. All rights reserved
news & views
a
b
Shape-memory alloy
Shape-memory effect
Nanoscale BFO
Tetragonal phase
Parent phase
ε
Single-variant martensite Applied load
Twinned martensite
Temperature-induced transition
forces, the low-symmetry phase consists of a number of (symmetry dependent) degenerate martensitic variants that have specific orientations with respect to the high-symmetry phase. The variants selforganize in such a way that no macroscopic change in shape occurs. However, when an applied load deforms the alloy, variant degeneracy breaks, which makes the new shape persist when the load is removed. The original shape can then be recovered by heating up the material across the phasetransition boundary. In some cases, the low-temperature deformed shape can also be memorized, thus allowing for reversible shape switching by temperature cycling through the martensitic transition (Fig. 1a). For this behaviour to occur it is necessary to create internal stresses in the material that mimic the effect of the externally applied load and promote the appearance of variants that give rise to the particular low-temperature shape. In practice, this is achieved by means of a ‘training’ thermomechanical procedure that induces oriented dislocation or precipitate patterns. Most materials displaying a shapememory effect are metallic alloys. Among them, Ni–Ti-based alloys are technologically relevant because of their excellent thermomechanical properties; in particular, shape-memory strains up to ~10% are available2. Although a shape-memory effect associated with a martensitic phase transition has also been shown for some ferroelectric and antiferroelectric oxides3, in these materials recoverable strains are, in general, less than 1%. Now, writing in Nature Communications, Jinxing Zhang and co-workers show that strains larger than 10% can reversibly be induced and removed in a thin film of bismuth ferrite oxide (BFO) — a material with unique multiferroic properties4 — by means of temperature and/or electric-field cycling across a phase transformation between rhombohedral (R) and tetragonal (T) lattices that resembles the typical martensitic transformation seen in shape-memory alloys5. Although it is known that BFO films can show coexistence of R and T phases and that their relative amount can be tuned by an applied electric field, only strains of about 1% have been achieved so far 6. Zhang and colleagues demonstrate that the effective distribution of stresses along the surface of the BFO film can be tuned by reducing the clamping of the film to the LaAlO3 substrate on which it has been grown. They achieved this by reducing the lateral size of the film from 10 μm to 1 μm (which corresponds to the unclamped situation) by means of focused ion-beam milling. When the degree of clamping is
ε
Shape-memory effect
Rhombohedral phase Clamping reduction
Clamped thin film with two-phase coexistence
Figure 1 | Shape-memory effect in a conventional shape-memory alloy and in a nanoscale BFO. a, Cooling a shape-memory alloy from the parent phase (top) across the martensitic transition (from top to bottom) leads to the formation of martensite variants (bottom). Twin variants self-organize (for example, in the form of stripes; bottom) in such a way that no macroscopic shape change occurs. A deformed single-variant martensite (middle) can be stabilized by cycling the alloy across the transition under an applied load (thermomechanical training). Subsequent temperature cycling between the parent phase and the single-variant martensite enables a reversible shape-memory effect with a typical strain ε in the range 5% to 10%. b, Temperature cycling between tetragonal (T; top) and rhombohedral (R; middle) phases enables a reversible shape-memory effect with a strain up to 14%5. The R phase is stabilized by using focused ion-beam milling to induce a lateral reduction of surface clamping in an as-grown thin BFO film composed of coexisting T and R phases (bottom).
low enough, the film’s structure consists of pure R phase (the stable phase at room temperature); when heated past 400 °C, the film undergoes a martensitic-like transition into a T phase, which involves strains reaching up to 14%. The film can then be reversibly cycled from a low-temperature to a high-temperature phase in a manner similar to conventional, thermomechanically trained shape-memory alloys (Fig. 1b). However, there are significant differences between shape-memory alloys and shapememory nanoscale oxides. In the latter, the low-temperature shape is established by stabilizing a pure phase at low temperature rather than by breaking the degeneracy between crystallographically equivalent variants. Also, the large recoverable deformation is associated with a transformation strain that is largely dominated by a large volume change (c-axis expansion), which is quite unusual in martensitic transitions (which are mainly shear-driven). A salient feature of Zhang and collaborators’ shape-memory nanoscale
NATURE MATERIALS | VOL 13 | JANUARY 2014 | www.nature.com/naturematerials
© 2014 Macmillan Publishers Limited. All rights reserved
films is the possibility of using an electric field to induce the martensitic-like transformation. This property is a consequence of the fact that the ferroelectric properties of BFO strongly depend on its crystallographic structure. From a technological point of view, the possibility of inducing and reversing huge deformations by combining temperature and electric-field changes accompanied by a considerably large volumetric work density opens up new perspectives for applications of BFO thin films in miniaturized devices, such as actuators and energy harvesters that take advantage of electromechanical energy conversion. Still, limitations associated with hysteresis and microcrack formation during shape-memory cycling — which may originate from the large change in volume involved in the transformation — could seriously compromise the practical use of the films in devices. Zhang and co-authors’ findings should nevertheless inspire the further development of oxides with improved shapememory performance. For conventional 7
news & views shape-memory alloys, electric-field control of shape memory is unfeasible, and they are restricted to operating at low frequency owing to slow responses to temperature changes (even at the nanoscale). The possibility of using electric fields to actuate ferroelectric nanoscale oxides should then enable easily controllable actuation
at higher frequencies in nanoscale shape-memory devices.
References
❐
Antoni Planes and Lluís Mañosa are at the Departament d’Estructura i Constituents de la Matèria, University of Barcelona, Diagonal 647, 08028 Barcelona, Catalonia, Spain. e-mail: [email protected]
1. Otsuka, K. & Wayman, C. M. in Shape Memory Materials 27–48 (Cambridge Univ. Press, 1998). 2. Otsuka, K. & Ren, X. Prog. Mater. Sci. 50, 511–678 (2005). 3. Uchino, K. in Shape Memory Materials 184–202 (Cambridge Univ. Press, 1998). 4. Spaldin, N. A., Cheong, S.‑W. & Ramesh, R. Phys. Today 63, 38–43 (October, 2010). 5. Zhang, J. et al. Nature Commun. 4, 2768 (2013). 6. Zeches, R. J. et al. Science 326, 977–980 (2009).
DIRECTED COLLOIDAL ASSEMBLY
Printing with magnets
Planar patterns of colloidal microparticles have been manufactured with high yield over square centimetre areas by using magnetic-field microgradients in a paramagnetic fluid. This approach could evolve into technology capable of printing three-dimensional objects through programmable and reconfigurable ‘magnetic pixels’.
Changqian Yu, Jie Zhang and Steve Granick
A
s the manufacturing needs of modern society evolve, so must do materials research. This is exemplified by printing, an ancient but evergreen idea. Printing technologies have inspired new manufacturing paradigms, from the woodblocks invented almost two millennia ago to its advanced forms that underpin the digital age (such as lithography, chemical stamping and inkjet a
Magnetic particles
Ni
PDMS
Non-magnetic particles
Ni PDMS Magnet
Ni
printing) to the increasingly important three-dimensional (3D) printing — already widely employed for rapid prototyping. In fact, ‘ink’ has become a metaphor for depositing, with precision and on demand, countless tiny objects such as nanoparticles, viruses or transistors. Even living tissues and organs can be printed. Because most printing methods work in air, it has been a challenge to ‘print’ c
PDMS N S
AA
b
A
mb yco one
uld
mo
H
AB
A
d
Figure 1 | Magnetic moulding could form the basis of 3D printing technology for colloidal materials. a, Magnetic-field microgradients produced by Ni grids embedded in a poly(dimethyl siloxane) (PDMS) layer and placed on top of a permanent magnet can act as magnetic moulds for both magnetic and non-magnetic colloidal particles2. b, As exemplified by the confocal image of a monolayer of colloidal tetramers, magnetic moulds can be used to assemble colloidal objects over large areas2. Scale bar, 10 μm. c, Taking inspiration from the rapid reprogrammability of electronic displays, magnetic pixels might be designed for layer-by-layer 3D printing, here exemplified conceptually by the layer-by-layer deposition of colloidal spheres to produce two families of 3D lattices (AAA and ABA packings) whose pore size, pore chemistry and colloidal arrangement are precisely controlled at each deposition step. d, The concept of reconfigurable magnetic pixels remains an engineering dream. Panels a and b reproduced from ref. 2, © 2013 NPG. 8
order into microscopic objects that remain suspended in liquids. Perhaps because methods to do so are so poorly developed many scientists have taken the self-assembly route, which is the preeminent approach to organize objects experiencing Brownian motion into the ‘self-organized’ state (that is, the state of lowest thermodynamic free energy). Passive self-assembly has the appeal of working well in fluids, yet typically suffers from low yield and unwanted by-products, and its successes can be difficult to generalize and scale up. Hence, the appreciation for methods that direct colloidal assembly by external control is growing. In this spirit, by building on earlier proof-of-concept work that showed that magnetic fields can place diamagnetic and paramagnetic particles into programmed positions1, Bartosz Grzybowski and collaborators report in Nature how to print a spectrum of colloidal objects — polymeric particles, silica particles, even live bacteria — suspended in a paramagnetic fluid2. By using magnetic-field microgradients produced by metal grids embedded in a rubber layer a few hundreds of nanometres thick that is placed on a permanent magnet (Fig. 1a), the authors show that beautiful arrays of complex microstructures can be assembled with high yield over square centimetre areas (Fig. 1b). Grzybowski and co-authors refer to this assembly approach as ‘magnetic moulding’ because solutions of paramagnetic salts regulate the response of the colloids to the magnetic fields; the solution becomes in fact part of the mould. Interestingly, the controlling magnetic fields extend into the bulk fluid and therefore the method is able to assemble
NATURE MATERIALS | VOL 13 | JANUARY 2014 | www.nature.com/naturematerials
© 2014 Macmillan Publishers Limited. All rights reserved
MATERIAL WITNESS
JOINT ENTERPRISE Tissue engineering is as much a mechanical as a biological problem. Although the field’s focus has been largely on getting cells to grow in vitro in the right morphology — a question increasingly now concerned with eliciting the requisite modes of stem-cell differentiation through appropriate environmental cues — there are many applications in which an engineered tissue’s dynamic properties are at least as important as the static ones. That’s why tissues associated with moving parts of the body, such as the trachea, remain among the hardest to ‘grow’ in a medically viable form. It is precisely because the mechanical performance of some biological tissues is so admirable that they present such a challenge. Of these, cartilage — the connective tissue at the ends of long bones — is among the most impressive. It can provide lubrication and joint articulation without compromising wear or damage for seven or more decades of constant use. Of course, the problem is that it does degenerate eventually, and the resulting osteoarthritis is a source of pain for millions of people. Although the static mechanical properties of native cartilage have been reasonably well matched in artificial tissues1, it is harder to replicate the lubricating function2. An alternative to growing real cartilage with the right mechanics is to mimic its properties in a
biocompatible material. Indeed, there is plenty worth emulating here in a synthetic material that might find tribological applications beyond medicine. It has been long recognized that cartilage, like so many biological tissues, has a hierarchical structure3; the question is how much of, or how closely, this structure need be copied to achieve acceptable mimicry of function. Previous efforts to make wholly synthetic cartilage have tended to fixate on just one aspect of its behaviour 4 — not because that is all that’s needed, but because it was all that’s feasible. But the tribological superiority of natural cartilage relies on a synergy that will only be attainable by approaching it as an integrated system. Greene et al. have now taken a step towards a more sophisticated cartilage mimic that combines two of the key features responsible for its lubrication5. One is that the tissue is a fluid-filled porous matrix in which most of the compressive load is carried by the fluid itself, as it is trapped and pressurized. This reduces the stress on the matrix, but it doesn’t alone account for lubrication and wear resistance. Furthermore, highly charged macromolecules called proteoglycans bound to the surface both resist collapse of the pores through electrosteric repulsion and provide detachable ‘boundary lubricants’ under shear.
PHILIP BALL In the analogue material developed by Greene et al., cellulose with submicrometre porosity mimics the collagen fibril network of cartilage, while the boundary lubrication is supplied by copolymers of the polyelectrolyte carboxymethylcellulose with polyethylene glycol. The resulting material shows not only low friction (albeit higher than real cartilage) but also the time-dependent static frictional response diagnostic of the fluid-pressurization mechanism of lubrication. This composite isn’t yet robust enough for medical applications, but it shows that the strategy is sound. ❐ References 1. Natoli, R. M., Revell, C. M. & Athanasiou, K. A. Tissue Eng. A 15, 3119–3128 (2009). 2. McNary, S. M., Athanasiou, K. A. & Reddi, A. H. Tissue Eng. B 18, 88–100 (2012). 3. Mow, V. C., Ratcliffe, A. & Poole, A. R. Biomaterials 13, 67–97 (1992). 4. Chen, M., Briscoe, W. H., Armes, S. P. & Klein, J. Science 323, 1698–1701 (2009). 5. Greene, G. W., Olszewska, A., Osterberg, M., Zhu, H. & Horn, R. Soft Matter http://dx.doi.org/10.1039/ c3sm52106k (2013).
SHAPE-MEMORY MATERIALS
Nanoscale oxides shape up
Reversible strains up to 14% driven by changes in temperature or electric field have been realized in a thin film of bismuth ferrite oxide.
Antoni Planes and Lluís Mañosa
B
end a paper clip, and its shape will persist even if the clip is heated up. This is because the deformed clip has undergone irreversible plastic deformation through the flow of dislocations. This type of crystallographic defect — which
6
can be viewed as misfit lines between crystallographic planes — confers ductility to metals because they can easily move under an applied load. Yet, in contrast to most metals, shape-memory alloys respond very differently to applied loads. In this
case, the response of the material is not controlled by dislocation flow, but by a transition between two crystallographic phases, a low-symmetry martensitic phase and a high-symmetry parent phase1. At the transition, and in the absence of external
NATURE MATERIALS | VOL 13 | JANUARY 2014 | www.nature.com/naturematerials
© 2014 Macmillan Publishers Limited. All rights reserved
news & views
a
b
Shape-memory alloy
Shape-memory effect
Nanoscale BFO
Tetragonal phase
Parent phase
ε
Single-variant martensite Applied load
Twinned martensite
Temperature-induced transition
forces, the low-symmetry phase consists of a number of (symmetry dependent) degenerate martensitic variants that have specific orientations with respect to the high-symmetry phase. The variants selforganize in such a way that no macroscopic change in shape occurs. However, when an applied load deforms the alloy, variant degeneracy breaks, which makes the new shape persist when the load is removed. The original shape can then be recovered by heating up the material across the phasetransition boundary. In some cases, the low-temperature deformed shape can also be memorized, thus allowing for reversible shape switching by temperature cycling through the martensitic transition (Fig. 1a). For this behaviour to occur it is necessary to create internal stresses in the material that mimic the effect of the externally applied load and promote the appearance of variants that give rise to the particular low-temperature shape. In practice, this is achieved by means of a ‘training’ thermomechanical procedure that induces oriented dislocation or precipitate patterns. Most materials displaying a shapememory effect are metallic alloys. Among them, Ni–Ti-based alloys are technologically relevant because of their excellent thermomechanical properties; in particular, shape-memory strains up to ~10% are available2. Although a shape-memory effect associated with a martensitic phase transition has also been shown for some ferroelectric and antiferroelectric oxides3, in these materials recoverable strains are, in general, less than 1%. Now, writing in Nature Communications, Jinxing Zhang and co-workers show that strains larger than 10% can reversibly be induced and removed in a thin film of bismuth ferrite oxide (BFO) — a material with unique multiferroic properties4 — by means of temperature and/or electric-field cycling across a phase transformation between rhombohedral (R) and tetragonal (T) lattices that resembles the typical martensitic transformation seen in shape-memory alloys5. Although it is known that BFO films can show coexistence of R and T phases and that their relative amount can be tuned by an applied electric field, only strains of about 1% have been achieved so far 6. Zhang and colleagues demonstrate that the effective distribution of stresses along the surface of the BFO film can be tuned by reducing the clamping of the film to the LaAlO3 substrate on which it has been grown. They achieved this by reducing the lateral size of the film from 10 μm to 1 μm (which corresponds to the unclamped situation) by means of focused ion-beam milling. When the degree of clamping is
ε
Shape-memory effect
Rhombohedral phase Clamping reduction
Clamped thin film with two-phase coexistence
Figure 1 | Shape-memory effect in a conventional shape-memory alloy and in a nanoscale BFO. a, Cooling a shape-memory alloy from the parent phase (top) across the martensitic transition (from top to bottom) leads to the formation of martensite variants (bottom). Twin variants self-organize (for example, in the form of stripes; bottom) in such a way that no macroscopic shape change occurs. A deformed single-variant martensite (middle) can be stabilized by cycling the alloy across the transition under an applied load (thermomechanical training). Subsequent temperature cycling between the parent phase and the single-variant martensite enables a reversible shape-memory effect with a typical strain ε in the range 5% to 10%. b, Temperature cycling between tetragonal (T; top) and rhombohedral (R; middle) phases enables a reversible shape-memory effect with a strain up to 14%5. The R phase is stabilized by using focused ion-beam milling to induce a lateral reduction of surface clamping in an as-grown thin BFO film composed of coexisting T and R phases (bottom).
low enough, the film’s structure consists of pure R phase (the stable phase at room temperature); when heated past 400 °C, the film undergoes a martensitic-like transition into a T phase, which involves strains reaching up to 14%. The film can then be reversibly cycled from a low-temperature to a high-temperature phase in a manner similar to conventional, thermomechanically trained shape-memory alloys (Fig. 1b). However, there are significant differences between shape-memory alloys and shapememory nanoscale oxides. In the latter, the low-temperature shape is established by stabilizing a pure phase at low temperature rather than by breaking the degeneracy between crystallographically equivalent variants. Also, the large recoverable deformation is associated with a transformation strain that is largely dominated by a large volume change (c-axis expansion), which is quite unusual in martensitic transitions (which are mainly shear-driven). A salient feature of Zhang and collaborators’ shape-memory nanoscale
NATURE MATERIALS | VOL 13 | JANUARY 2014 | www.nature.com/naturematerials
© 2014 Macmillan Publishers Limited. All rights reserved
films is the possibility of using an electric field to induce the martensitic-like transformation. This property is a consequence of the fact that the ferroelectric properties of BFO strongly depend on its crystallographic structure. From a technological point of view, the possibility of inducing and reversing huge deformations by combining temperature and electric-field changes accompanied by a considerably large volumetric work density opens up new perspectives for applications of BFO thin films in miniaturized devices, such as actuators and energy harvesters that take advantage of electromechanical energy conversion. Still, limitations associated with hysteresis and microcrack formation during shape-memory cycling — which may originate from the large change in volume involved in the transformation — could seriously compromise the practical use of the films in devices. Zhang and co-authors’ findings should nevertheless inspire the further development of oxides with improved shapememory performance. For conventional 7
news & views shape-memory alloys, electric-field control of shape memory is unfeasible, and they are restricted to operating at low frequency owing to slow responses to temperature changes (even at the nanoscale). The possibility of using electric fields to actuate ferroelectric nanoscale oxides should then enable easily controllable actuation
at higher frequencies in nanoscale shape-memory devices.
References
❐
Antoni Planes and Lluís Mañosa are at the Departament d’Estructura i Constituents de la Matèria, University of Barcelona, Diagonal 647, 08028 Barcelona, Catalonia, Spain. e-mail: [email protected]
1. Otsuka, K. & Wayman, C. M. in Shape Memory Materials 27–48 (Cambridge Univ. Press, 1998). 2. Otsuka, K. & Ren, X. Prog. Mater. Sci. 50, 511–678 (2005). 3. Uchino, K. in Shape Memory Materials 184–202 (Cambridge Univ. Press, 1998). 4. Spaldin, N. A., Cheong, S.‑W. & Ramesh, R. Phys. Today 63, 38–43 (October, 2010). 5. Zhang, J. et al. Nature Commun. 4, 2768 (2013). 6. Zeches, R. J. et al. Science 326, 977–980 (2009).
DIRECTED COLLOIDAL ASSEMBLY
Printing with magnets
Planar patterns of colloidal microparticles have been manufactured with high yield over square centimetre areas by using magnetic-field microgradients in a paramagnetic fluid. This approach could evolve into technology capable of printing three-dimensional objects through programmable and reconfigurable ‘magnetic pixels’.
Changqian Yu, Jie Zhang and Steve Granick
A
s the manufacturing needs of modern society evolve, so must do materials research. This is exemplified by printing, an ancient but evergreen idea. Printing technologies have inspired new manufacturing paradigms, from the woodblocks invented almost two millennia ago to its advanced forms that underpin the digital age (such as lithography, chemical stamping and inkjet a
Magnetic particles
Ni
PDMS
Non-magnetic particles
Ni PDMS Magnet
Ni
printing) to the increasingly important three-dimensional (3D) printing — already widely employed for rapid prototyping. In fact, ‘ink’ has become a metaphor for depositing, with precision and on demand, countless tiny objects such as nanoparticles, viruses or transistors. Even living tissues and organs can be printed. Because most printing methods work in air, it has been a challenge to ‘print’ c
PDMS N S
AA
b
A
mb yco one
uld
mo
H
AB
A
d
Figure 1 | Magnetic moulding could form the basis of 3D printing technology for colloidal materials. a, Magnetic-field microgradients produced by Ni grids embedded in a poly(dimethyl siloxane) (PDMS) layer and placed on top of a permanent magnet can act as magnetic moulds for both magnetic and non-magnetic colloidal particles2. b, As exemplified by the confocal image of a monolayer of colloidal tetramers, magnetic moulds can be used to assemble colloidal objects over large areas2. Scale bar, 10 μm. c, Taking inspiration from the rapid reprogrammability of electronic displays, magnetic pixels might be designed for layer-by-layer 3D printing, here exemplified conceptually by the layer-by-layer deposition of colloidal spheres to produce two families of 3D lattices (AAA and ABA packings) whose pore size, pore chemistry and colloidal arrangement are precisely controlled at each deposition step. d, The concept of reconfigurable magnetic pixels remains an engineering dream. Panels a and b reproduced from ref. 2, © 2013 NPG. 8
order into microscopic objects that remain suspended in liquids. Perhaps because methods to do so are so poorly developed many scientists have taken the self-assembly route, which is the preeminent approach to organize objects experiencing Brownian motion into the ‘self-organized’ state (that is, the state of lowest thermodynamic free energy). Passive self-assembly has the appeal of working well in fluids, yet typically suffers from low yield and unwanted by-products, and its successes can be difficult to generalize and scale up. Hence, the appreciation for methods that direct colloidal assembly by external control is growing. In this spirit, by building on earlier proof-of-concept work that showed that magnetic fields can place diamagnetic and paramagnetic particles into programmed positions1, Bartosz Grzybowski and collaborators report in Nature how to print a spectrum of colloidal objects — polymeric particles, silica particles, even live bacteria — suspended in a paramagnetic fluid2. By using magnetic-field microgradients produced by metal grids embedded in a rubber layer a few hundreds of nanometres thick that is placed on a permanent magnet (Fig. 1a), the authors show that beautiful arrays of complex microstructures can be assembled with high yield over square centimetre areas (Fig. 1b). Grzybowski and co-authors refer to this assembly approach as ‘magnetic moulding’ because solutions of paramagnetic salts regulate the response of the colloids to the magnetic fields; the solution becomes in fact part of the mould. Interestingly, the controlling magnetic fields extend into the bulk fluid and therefore the method is able to assemble
NATURE MATERIALS | VOL 13 | JANUARY 2014 | www.nature.com/naturematerials
© 2014 Macmillan Publishers Limited. All rights reserved
