Training and re-training liquid crystal elastomer metamaterials for pluripotent functionality
Savannah D. Gowen, Elina Ghimire, Charlie A. Lindberg, Ingrid S. Appen, Stuart J. Rowan, Sidney R. Nagel · 2025
Read Training and re-training liquid crystal elastomer metamaterials for pluripotent functionality by Savannah D. Gowen, Elina Ghimire, Charlie A. Lindberg, Ingrid S. Appen, Stuart J. Rowan, Sidney R. Nagel (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-17238.txt.
Introduction
Gowen and co-authors show, in a 2025 paper, that a soft metamaterial can be trained, wiped clean, and trained again for a new job. Their arrays are built from liquid crystal elastomers, rubbery sheets that bend in set ways. Training means aging the material under a chosen load until it holds the response.
First the team tuned the array's Poisson ratio until squeezing it one way made it bulge the other, the auxetic trick. Then they erased that training and taught the same sheet allostery, where a push at one spot moves a far spot. One material thus carries pluripotent function: no cutting, no new chemistry, just retraining. The work hints at hardware that adapts on demand.
Catalog record
| CREATOR | Savannah D. Gowen, Elina Ghimire, Charlie A. Lindberg, Ingrid S. Appen, Stuart J. Rowan, Sidney R. Nagel |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 23 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-17238
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Training and re-training liquid crystal elastomer metamaterials for pluripotent functionality Preprint: APS/123-QED Savannah D. Gowen, Elina Ghimire, Charlie A. Lindberg, Ingrid S. Appen, Stuart J. Rowan, Sidney R. Nagel Abstract Training has emerged as a promising materials design technique in which function can be achieved through repeated physical modification of an existing material rather than by direct chemical functionalization, cutting or reprocessing. This work investigates both the ability to train for function and then to erase that function on-demand in macroscopic metamaterials made from liquid crystal elastomers (LCEs). We first show that the Poisson’s ratio of these disordered arrays can be tuned via directed aging to induce an auxetic response. We then show that the arrays can be reset and re-trained for another local mechanical function, allostery, thus demonstrating pluripotent functionality. I Introduction Focused training often leads to stored memories Sprenger 1999 . Learning in the brain can occur by repeated exposure to input stimuli; likewise, “muscle memory” is the result of a training regimen that targets specific physical performance Snijders et al. 2020 . Over time, training can improve the performance of specific mental or physical tasks. However, forgetting previous training is also often advantageous for acquiring new, perhaps incompatible, skills Gravitz 2019 . The abilities to remember, forget, and train are not properties exclusively concerned with biological systems Jaeger et al. 2024 ; Stern and Murugan 2023 ; Dillavou et al. 2023 ; even ordinary physical materials exhibit various forms of memory as a result of training Keim et al. 2019 ; Nagel et al. 2023 . Training is unlike material design techniques where the microscopic or structural features responsible for function are incorporated at the time of fabrication. Instead, function is achieved through the modification of existing material properties in response to external stimuli; for example, via extended or repetitive exposure to strains involved in a desired mechanical function. However, some materials can be trained more easily than others, and indeed, some materials may be incapable of being trained at all. As such, more work is necessary to determine what materials are trainable and why. Less consideration has been given to materials that can both train and forget. Admittedly, forgetfulness has often been presented as an obstacle rather than a feature Keim and Nagel 2011 . However, the ability to forget introduces the possibility of creating materials that can first be trained for one function and then, when there is a need, retrained for another. In this work, we examine the ability of macroscopic disordered geometric arrays made from liquid crystal elastomers (LCEs) to train in one function and then to forget it on-demand. Disordered macroscopic arrays, composed of nodes connected by struts, are model materials for many elastic systems Jaeger et al. 2024 . Simulations and experiments have demonstrated that such structures can be trained to exhibit novel bulk and local properties Pashine et al. 2019 ; Hexner et al. 2020 ; Gowen 2025 ; Altman et al. 2024 ; Lee et al. 2022 . For example, Pashine et al. Pashine et al. 2019 show that by leaving quasi-two-dimensional disordered foam arrays to age under uniform compression, they can be trained to exhibit a negative Poisson’s ratio, in which compression along one axis produces contraction (rather than expansion) along the transverse direction. This training method referred to as directed aging, benefits from a material’s ability to adapt over time to externally induced stress by lowering the energy cost of the imposed deformation Pashine et al. 2019 . Negative Poisson’s ratio materials are rarely found in nature. Thus, the ability to tune such a material response without intensive engineering demonstrates the power of training as a design technique. Our metamaterials are disordered arrays fabricated out of thin sheets of liquid crystal elastomer. Liquid crystal elastomers are elastomeric polymer networks that contain covalently bound liquid crystals (mesogens) De Gennes 1975 ; Finkelmann et al. 1981 ; Warner and Terentjev 2007 . In a polydomain LCE, the liquid crystalline phases form locally ordered microscopic domains within the bulk material. The manipulation of the domain alignment leads to a host of exceptional material properties including stimuli-responsive mechanical and optical transformations, enhanced stress dissipation, and “soft elasticity” so that the material endures large strains with little increase in the stress as the mesogens rotate in the direction of the applied strain Warner and Terentjev 2007 ; Jin et al. 2022 ; de Castro et al. 2024 ; Luo et al. 2020 ; Wu et al. 2024 . The mesogen alignment can be controlled by the imposition of an external electrical, magnetic, or mechanical field Herbert et al. 2022 ; Jiang et al. 2024 . Altering that alignment with mechanical strain produces a large number of configurations in these materials that can be accessed by simple inputs. This is advantageous in developing materials that can be trained. Additionally, liquid crystal elastomers exhibit shape memory behavior Herbert et al. 2022 and by simply heating the material above its nematic to isotropic transition temperature, the original network configuration can be recovered. This enables the erasure of the trained functionality as desired. We demonstrate that disordered arrays of liquid crystal elastomers can be trained via directed aging to exhibit a negative Poisson’s ratio; we then examine the material components that enable this adaptive behavior. Finally, we demonstrate the ability to reset these materials as a means of forgetting to retrain for a different function, in this case an allosteric-inspired mechanical behavior Hexner et al. 2020 ; Gowen 2025 . These findings demonstrate a new class of trainable pluripotent metamaterials. II Results Training for Negative Poisson’s Ratio The polydomain liquid crystal elastomers (Fig. 1 a) in this study are synthesized using Aza-Michael chemistry, aiming for a molecular weight of 2000 g/mol between crosslinks Ghimire et al. 2024 . After synthesis, the LCE gels are washed, dried, and pressed into sheets approximately 2 mm thick with a lateral dimension of 45 mm. Using a laser cutter, disordered array geometries are cut from these sheets (Fig. S1). To examine the trainability of these metamaterials, we first train them to induce a negative Poisson’s ratio (auxetic) response. Figure 1: Training disordered arrays of liquid crystal elastomers (LCEs) for auxetic behavior. (a) A schematic of microstructure in a polydomain LCE. (b) A sample LCE array is shown before (top row) and after aging under 40% compression (bottom row). Before aging, the sample has \nu>0 so that it expands horizontally under uniaxial vertical compression. After aging the sample contracts along the horizontal when uniaxial strain is applied, demonstrating an auxetic response, \nu<0 . Blue and red dots are used for tracking strain at the boundary. (c) The Poisson’s ratio, \nu , versus aging compression for six sample array geometries indicated by different symbols and colors. (Images of the arrays are shown in the SI.) \nu drops significantly for aging compressions of 30-40%. The array imaged in (b) is shown by duplicate experiments in dark blue (o and + signs). As initially fabricated, the arrays have a positive Poisson’s ratio so that when they are compressed uniaxially, they expand in the transverse directions. This is shown by the first row of images in Fig. 1 b. This response is typical of nearly