We apply a Transformer architecture, specifically BERT, to learn flexible and high quality molecular representations for drug discovery problems. We study the impact of using different combinations of self-supervised tasks for pre-training, and present our results for the established Virtual Screening and QSAR benchmarks. We show that: i) The selection of appropriate self-supervised task(s) for pre-training has a significant impact on performance in subsequent downstream tasks such as Virtual Screening. ii) Using auxiliary tasks with more domain relevance for Chemistry, such as learning to predict calculated molecular properties, increases the fidelity of our learnt representations. iii) Finally, we show that molecular representations learnt by our model `MolBert' improve upon the current state of the art on the benchmark datasets.

Machine olfaction is usually crystallized as electronic noses (e-noses) which consist of an array of gas sensors mimicking biological noses to'smell' and'sense' odors [1]. Gas sensors in the array should be carefully selected based on several specifications (sensitivity, selectivity, response time, recovery time, etc.) for specific detecting purposes. On the other side, some general-purpose e-noses may have an array of gas sensors that are sensitive to a variety of odorous materials so that such e-noses can be applied to many fields. An increasing number of researches and applications utilized machine olfaction in recent years. In the early 20th century, some studies applied e-noses to the analysis of products along with gas chromatography-mass spectrometers (GC-MS) [2]. Some linear methods such as principal component analysis (PCA), linear discriminant analysis (LDA), support vector machines (SVM), etc. were used in the analysis [3].

Researchers at MIT's Center for Bits and Atoms have created tiny building blocks that exhibit a variety of unique mechanical properties, such as the ability to produce a twisting motion when squeezed. These subunits could potentially be assembled by tiny robots into a nearly limitless variety of objects with built-in functionality, including vehicles, large industrial parts, or specialized robots that can be repeatedly reassembled in different forms. The researchers created four different types of these subunits, called voxels (a 3D variation on the pixels of a 2D image). Each voxel type exhibits special properties not found in typical natural materials, and in combination they can be used to make devices that respond to environmental stimuli in predictable ways. Examples might include airplane wings or turbine blades that respond to changes in air pressure or wind speed by changing their overall shape. The findings, which detail the creation of a family of discrete "mechanical metamaterials," are described in a paper published in the journal Science Advances, authored by recent MIT doctoral graduate Benjamin Jenett PhD '20, Professor Neil Gershenfeld, and four others.

NIR Calibration-Model Services Spectroscopy and Chemometrics News Weekly 46, 2020 NIRS NIR Spectroscopy MachineLearning Spectrometer Spectrometric Analytical Chemistry foodindustry Analysis Lab Laboratories Laboratory Software IoT Sensors QA QC Testing Quality LINK Get the Spectroscopy and Chemometrics News Weekly in real time on Twitter @ CalibModel and follow us. Near-Infrared Spectroscopy (NIRS) "Near infrared absorption spectroscopy for the quantification of unsulfated alcohol in sodium lauryl ether sulfate" LINK "Estimation of Organic Carbon in Anthropogenic Soil by VIS-NIR Spectroscopy: Effect of Variable Selection" LINK "Near infrared spectroscopy (NIRS) based high-throughput online assay for key cell wall features that determine sugarcane bagasse digestibility") LINK "Authentication of barley-finished beef using visible and near infrared spectroscopy (Vis-NIRS) and different discrimination approaches" LINK "Energetic Distribution of States in Irradiated Low-Density ...

Graph neural networks have recently become a standard method for analysing chemical compounds. In the field of molecular property prediction, the emphasis is now put on designing new model architectures, and the importance of atom featurisation is oftentimes belittled. When contrasting two graph neural networks, the use of different atom features possibly leads to the incorrect attribution of the results to the network architecture. To provide a better understanding of this issue, we compare multiple atom representations for graph models and evaluate them on the prediction of free energy, solubility, and metabolic stability. To the best of our knowledge, this is the first methodological study that focuses on the relevance of atom representation to the predictive performance of graph neural networks.

Discrete response of structures is often a key probabilistic quantity of interest. For example, one may need to identify the probability of a binary event, such as, whether a structure has buckled or not. In this study, an adaptive domain-based decomposition and classification method, combined with sparse grid sampling, is used to develop an efficient classification surrogate modeling algorithm for such discrete outputs. An assumption of monotonic behaviour of the output with respect to all model parameters, based on the physics of the problem, helps to reduce the number of model evaluations and makes the algorithm more efficient. As an application problem, this paper deals with the development of a computational framework for generation of probabilistic penetration response of S-2 glass/SC-15 epoxy composite plates under ballistic impact. This enables the computationally feasible generation of the probabilistic velocity response (PVR) curve or the $V_0-V_{100}$ curve as a function of the impact velocity, and the ballistic limit velocity prediction as a function of the model parameters. The PVR curve incorporates the variability of the model input parameters and describes the probability of penetration of the plate as a function of impact velocity.

Developing machine learning enabled smart manufacturing is promising for composite structures assembly process. To improve production quality and efficiency of the assembly process, accurate predictive analysis on dimensional deviations and residual stress of the composite structures is required. The novel composite structures assembly involves two challenges: (i) the highly nonlinear and anisotropic properties of composite materials; and (ii) inevitable uncertainty in the assembly process. To overcome those problems, we propose a neural network Gaussian process model considering input uncertainty for composite structures assembly. Deep architecture of our model allows us to approximate a complex process better, and consideration of input uncertainty enables robust modeling with complete incorporation of the process uncertainty. Based on simulation and case study, the NNGPIU can outperform other benchmark methods when the response function is nonsmooth and nonlinear. Although we use composite structure assembly as an example, the proposed methodology can be applicable to other engineering systems with intrinsic uncertainties.

A free software tool that can find the best conditions for organic synthesis reactions often does as well as expert chemists – somewhat to the surprise of the researchers. The software, called LabMate.ML, suggests a random set of initial conditions – such as the temperature, the amount of solvent and the reaction time – for a specific reaction, with the aim of optimising its yield. After those initial reactions are carried out by a human chemist, their resulting yields are read with nuclear magnetic resonance and infrared spectroscopy, digitised into binary code and then fed back into the software. LabMate.ML then uses a machine-learning algorithm to make decisions about the yields, and then recommends further sets of conditions to try. Researcher Tiago Rodrigues of the University of Lisbon says LabMate.ML usually takes between 10 and 20 iterations to find the greatest yield, while the number of initial reactions varies between five and 10, depending on how many conditions are being optimised.

This week, the U.S. Environmental Protection Agency (EPA) celebrates America Recycles Week by hosting two days of free, virtual events that focus on creating a more robust and sustainable recycling system in the U.S. and abroad. Today, the America Recycles: Innovation Fair will feature more than 40 innovators from across the recycling system via virtual exhibit halls demonstrating their state-of-the-art products, services, outreach, and technologies. They are advancing the recycling system through strategies such as: deploying artificial intelligence robots to enhance operations at recycling facilities; using hard-to-recycle plastics in 3D printing materials; installing small system sorting units in stadiums and small communities; creating new construction materials from hard-to-recycle plastics; and using automated technology and recycled glass bottles to create new glassware. "EPA is proud to showcase top recycling innovators at the virtual Innovation Fair today," said EPA Administrator Andrew Wheeler. "Tomorrow's America Recycles Summit will include EPA's announcement of the first National Recycling Goal, which will prompt a whole new level of dialogue among stakeholders on how to improve our domestic recycling infrastructure."

Electronic skins (e-skins) are flexible electronic devices that emulate properties of human skin, such as high stretchability and toughness, perception of stimuli, and self-healing. These devices can serve as an alternative to natural human skin or as a human-machine interface ([ 1 ][1]–[ 3 ][2]). For on-skin applications, an e-skin should be multimodal (sense more than one external stimulus), have a high density of sensors, and have low interference with natural skin sensation. On pages 961 and 966 of this issue, You et al. ([ 4 ][3]) and Lee et al. ([ 5 ][4]), respectively, report advances of skin-like electronic devices. You et al. present a stretchable multimodal ionic-electronic (IE) conductor–based “IEM-skin” that can measure both strain and temperature inputs without signal interference. Lee et al. describe an ultrathin capacitive pressure sensor based on conductive and dielectric nanomesh structures that can be attached to a human fingertip for grip pressure and force measurement without affecting natural skin sensation. The human skin contains a large number of mechanoreceptors and thermoreceptors (nerve endings that sense deformation and temperature, respectively) that provide distinct perception of the spatial distributions of strain and temperature on our skin induced by touch stimulations ([ 6 ][5]). To replicate these sensory functions of the natural skin, different types of sensors that act as artificial receptors are integrated onto an e-skin for multimodal sensation ([ 7 ][6]). However, an e-skin containing a high-density array of sensory “pixels” of different types for sensing different physical quantities tends to have a complex structure and is challenging to manufacture. A preferred strategy for realizing multimodal sensation on an e-skin is to use the same sensory unit for detecting different physical quantities without signal interference, an approach called decoupled multimodal sensing. Traditional stretchable sensors are sensitive to both strain and temperature and cannot be used as artificial multimodal receptors without signal interference. Targeting interference-free strain and temperature sensing by a single sensory unit, You et al. creatively used the ion relaxation dynamics of an ion conductor (an elastomer mixed with an ionic liquid) to decouple the strain and temperature measurement and developed the IEM-skin composed of an array of artificial multimodal ionic receptors. They fabricated the IEM-skin by sandwiching a thin layer of ion conductor with two layers of orthogonally patterned stretchable electrode strips (see the figure, top ). A pixelated matrix of millimeter-sized artificial receptors formed between the top and bottom electrodes. The electrical properties of each receptor are affected by the externally applied strain and temperature stimuli and can be measured through impedance measurement. You et al. used a strain-independent intrinsic electrical parameter of the ion conductor, the charge relaxation time, which reflects the ionic charge dynamics of the ion conductor and is equal to the ratio of material's dielectric constant and ion conductivity ([ 8 ][7], [ 9 ][8]). The charge relaxation time is the signal readout for temperature and is not affected by the deformation of the IEM-skin. For strain measurement, the bulk capacitance of the ion conductor is measured. The effect of temperature on the capacitance is eliminated through normalization against a reference capacitance at the temperature measured by the receptor. Thus, an external strain input only changes geometric parameters of the ion conductor, whereas a temperature input primarily modulates the intrinsic electrical properties (dielectric constant and ion conductivity) of the ion conductor. Another enabling factor of the IEM-skin design is its emulation of the epidermis and dermis bilayer of the human skin by suspending the receptor matrix layer over a low-friction interface layer filled with talcum powder. This design allows three-dimensional wrinkle-like deformations of the IEM-skin under different contact modes (such as shear, pinch, tweak, and torsion) and permits the IEM-skin to distinguish these contact modes through the measured temperature and strain profiles. Data confirm that the IEM-skin can perform decoupled, real-time measurement of strain and temperature with high accuracy. The IEM-skin can serve as a human-machine interface that accepts tactile inputs of different contact modes and can be integrated into prosthetic and robotic devices to provide tactile and thermal feedback with high spatial resolution. The concept of using intrinsic electrical parameters, such as conductivity and dielectric constant of sensing materials, for strain-independent temperature sensing can be generalized to developing other types of stretchable multimodal sensors for humidity, chemicals, and biomolecules. One limitation is that the method for recognizing different tactile input modes through the measured temperature and strain profiles only works for interactions with hot or cold objects at temperatures different from that of the IEM-skin. Alternative solutions may include the use of learning-based recognition models purely based on strain-distribution data or modulation of the temperature of the IEM-skin (by adding a heating layer) based on the environment. Skin-like electronic sensors also hold great potential for construction of hand-wearing devices such as instrumented gloves for quantifying tactile signals like force and pressure during finger or in-hand manipulation ([ 10 ][9]). Such data could facilitate the decoding of human hand sensation and its roles in object manipulation and enable better designs of robotic and prosthetic hands with biomimetic sensory feedback ([ 11 ][10]). Targeting imperceptible wearing and tactile sensing on fingertips, Lee et al. developed an ultrathin capacitive pressure sensor consisting of multilayers of conductive and dielectric nanomesh structures. This sensor design is derived from the design of conductive nanomesh electrodes proposed by Miyamoto et al. ([ 12 ][11]), which can be directly laminated on human skin during fabrication. The electrode is fabricated by first electrospinning a water-soluble polymer, polyvinyl alcohol (PVA) into a multilayered mesh-like network of 300- to 500-nm-wide nanofibers. A 100-nm-thick gold layer is then deposited onto the PVA nanomesh sheet, and the gold-coated nanomesh sheet is transferred onto the skin surface. The sacrificial PVA nanofibers are washed off by water, but a residual layer of the dissolved PVA greatly facilitates the attachment of the resultant gold nanomesh layer onto the textured skin surface with excellent adhesion and conformal contact. The skin-integrated nanomesh electrode is stretchable and highly breathable and has exceptionally low bending stiffness, and so it creates no mechanical constraint or dermatological irritation to the skin. To fabricate a nanomesh pressure sensor (see the figure, bottom), Lee et al. first laminated a nanomesh electrode on the skin surface and then sequentially attached a dielectric nanomesh layer made of electrospun polyurethane and parylene nanofibers and another nanomesh electrode layer to form a parallel-plate capacitor structure. Then, a nanomesh passivation layer of polyurethane nanofibers was attached to the top electrode layer with dissolved PVA nanofibers as the filler and adhesive. The total thickness of the nanomesh pressure sensor is ∼13 μm. When fingers wearing such a pressure sensor grip an object, the grip force applied to the pressure sensor deforms the middle dielectric nanomesh layer and leads to a change in the capacitance measured between the top and bottom electrodes as the sensor readout. ![Figure][12] Improved electronic skins Two goals in artificial touch sensors are to sense more than one stimulus with one receptor and to create wearable sensors that maintain natural skin sensation. GRAPHIC: C. BICKEL/ SCIENCE Through object-gripping experiments performed by human participants, Lee et al. investigated the effect of the finger-integrated pressure sensor on the natural fingertip sensation and found no decrease of the sensory feedback caused by the attachment of the pressure sensor. They hypothesized that the ultrathin and compliant structure of the nanomesh pressure sensor renders the device imperceptible on the fingertip. In addition, the intimate and conformal adhesion of the sensor's bottom nanomesh electrode layer to the skin surface may also contribute to the negligible interference of the finger skin sensation by the sensor attachment. This sensor also shows excellent mechanical durability under cyclic compression, shearing, and surface friction, which is attributed to the high mechanical robustness of the multilayered nanomesh structure of the pressure sensor. This work highlights another new application of the previously reported skin-integrated nanomesh electronics ([ 12 ][11]) to wearable physical sensing with unprecedented performance. Future work may involve the further examination of fundamental mechanisms for the on-skin imperceptibility of the nanomesh pressure sensor, the systematic study of the skin-integrated pressure sensor performance for grasping objects of different materials and properties (such as insulating versus conductive, hard versus soft, and smooth versus textured), and the scalable fabrication of pixelated nanomesh pressure sensors in a large area with high density. The nanomesh pressure sensor could record tactile signals of human-hand manipulation that could provide superior sensing performance and zero data artifacts over existing instrumented gloves and e-skins. Multimodal sensation and nonobstructive skin integration are two important features that are desirable in e-skin designs. The studies reported by You et al. and Lee et al. , respectively, provide new solutions to better realize these attractive features with simplified device structures and enhanced sensing performance without impeding natural sensation. These results will inspire new sensor designs and lead to applications of e-skins as wearable health care monitoring, sensory prosthetic and robotic devices, and high-performance human-machine interfaces. 1. [↵][13]1. J. C. Yang et al ., Adv. Mater. 31, 1904765 (2019). [OpenUrl][14] 2. 1. T. R. Ray et al ., Chem. Rev. 119, 5461 (2019). [OpenUrl][15] 3. [↵][16]1. T. Someya, 2. M. Amagai , Nat. Biotechnol. 37, 382 (2019). [OpenUrl][17][CrossRef][18][PubMed][19] 4. [↵][20]1. I. You et al ., Science 370, 961 (2020). [OpenUrl][21][CrossRef][22] 5. [↵][23]1. S. Lee et al ., Science 370, 966 (2020). [OpenUrl][24][CrossRef][25] 6. [↵][26]1. A. Zimmerman, 2. L. Bai, 3. D. D. Ginty , Science 346, 950 (2014). [OpenUrl][27][Abstract/FREE Full Text][28] 7. [↵][29]1. S. Jeon, 2. S.-C. Lim, 3. T. Q. Trung, 4. M. Jung, 5. N.-E. Lee , Proc. IEEE 107, 2065 (2019). [OpenUrl][30] 8. [↵][31]1. C. Gainaru et al ., J. Phys. Chem. B 120, 11074 (2016). [OpenUrl][32][CrossRef][33][PubMed][34] 9. [↵][35]1. B. A. Mei, 2. O. Munteshari, 3. J. Lau, 4. B. Dunn, 5. L. Pilon , J. Phys. Chem. C 122, 194 (2018). [OpenUrl][36] 10. [↵][37]1. S. Sundaram et al ., Nature 569, 698 (2019). [OpenUrl][38][CrossRef][39][PubMed][40] 11. [↵][41]1. E. D'Anna et al ., Sci. Robot. 4, eaau8892 (2019). [OpenUrl][42] 12. [↵][43]1. A. Miyamoto et al ., Nat. Nanotechnol. 12, 907 (2017). [OpenUrl][44][CrossRef][45][PubMed][46] Acknowledgments: X.L. acknowledges support from the Natural Sciences and Engineering Research Council of Canada (RGPIN-2017-06374). [1]: #ref-1 [2]: #ref-3 [3]: #ref-4 [4]: #ref-5 [5]: #ref-6 [6]: #ref-7 [7]: #ref-8 [8]: #ref-9 [9]: #ref-10 [10]: #ref-11 [11]: #ref-12 [12]: pending:yes [13]: #xref-ref-1-1 "View reference 1 in text" [14]: {openurl}?query=rft.jtitle%253DAdv.%2BMater%26rft.volume%253D31%26rft.spage%253D1904765%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx [15]: {openurl}?query=rft.jtitle%253DChem.%2BRev%26rft.volume%253D119%26rft.spage%253D5461%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx [16]: #xref-ref-3-1 "View reference 3 in text" [17]: {openurl}?query=rft.jtitle%253DNat.%2BBiotechnol%26rft.volume%253D37%26rft.spage%253D382%26rft_id%253Dinfo%253Adoi%252F10.1038%252Fs41587-019-0079-1%26rft_id%253Dinfo%253Apmid%252F30940942%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx [18]: /lookup/external-ref?access_num=10.1038/s41587-019-0079-1&link_type=DOI [19]: /lookup/external-ref?access_num=30940942&link_type=MED&atom=%2Fsci%2F370%2F6519%2F910.atom [20]: #xref-ref-4-1 "View reference 4 in text" [21]: {openurl}?query=rft.jtitle%253DScience%26rft.stitle%253DScience%26rft.aulast%253DYou%26rft.auinit1%253DI.%26rft.volume%253D370%26rft.issue%253D6519%26rft.spage%253D961%26rft.epage%253D965%26rft.atitle%253DArtificial%2Bmultimodal%2Breceptors%2Bbased%2Bon%2Bion%2Brelaxation%2Bdynamics%26rft_id%253Dinfo%253Adoi%252F10.1126%252Fscience.aba5132%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx [22]: /lookup/external-ref?access_num=10.1126/science.aba5132&link_type=DOI [23]: #xref-ref-5-1 "View reference 5 in text" [24]: {openurl}?query=rft.jtitle%253DScience%26rft.stitle%253DScience%26rft.aulast%253DLee%26rft.auinit1%253DS.%26rft.volume%253D370%26rft.issue%253D6519%26rft.spage%253D966%26rft.epage%253D970%26rft.atitle%253DNanomesh%2Bpressure%2Bsensor%2Bfor%2Bmonitoring%2Bfinger%2Bmanipulation%2Bwithout%2Bsensory%2Binterference%26rft_id%253Dinfo%253Adoi%252F10.1126%252Fscience.abc9735%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx [25]: /lookup/external-ref?access_num=10.1126/science.abc9735&link_type=DOI [26]: #xref-ref-6-1 "View reference 6 in text" [27]: {openurl}?query=rft.jtitle%253DScience%26rft.stitle%253DScience%26rft.aulast%253DZimmerman%26rft.auinit1%253DA.%26rft.volume%253D346%26rft.issue%253D6212%26rft.spage%253D950%26rft.epage%253D954%26rft.atitle%253DThe%2Bgentle%2Btouch%2Breceptors%2Bof%2Bmammalian%2Bskin%26rft_id%253Dinfo%253Adoi%252F10.1126%252Fscience.1254229%26rft_id%253Dinfo%253Apmid%252F25414303%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx [28]: /lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6Mzoic2NpIjtzOjU6InJlc2lkIjtzOjEyOiIzNDYvNjIxMi85NTAiO3M6NDoiYXRvbSI7czoyMjoiL3NjaS8zNzAvNjUxOS85MTAuYXRvbSI7fXM6ODoiZnJhZ21lbnQiO3M6MDoiIjt9 [29]: #xref-ref-7-1 "View reference 7 in text" [30]: {openurl}?query=rft.jtitle%253DProc.%2BIEEE%26rft.volume%253D107%26rft.spage%253D2065%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx [31]: #xref-ref-8-1 "View reference 8 in text" [32]: {openurl}?query=rft.jtitle%253DJ.%2BPhys.%2BChem.%2BB%26rft.volume%253D120%26rft.spage%253D11074%26rft_id%253Dinfo%253Adoi%252F10.1021%252Facs.jpcb.6b08567%26rft_id%253Dinfo%253Apmid%252F27681664%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx [33]: /lookup/external-ref?access_num=10.1021/acs.jpcb.6b08567&link_type=DOI [34]: /lookup/external-ref?access_num=27681664&link_type=MED&atom=%2Fsci%2F370%2F6519%2F910.atom [35]: #xref-ref-9-1 "View reference 9 in text" [36]: {openurl}?query=rft.jtitle%253DJ.%2BPhys.%2BChem.%2BC%26rft.volume%253D122%26rft.spage%253D194%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx [37]: #xref-ref-10-1 "View reference 10 in text" [38]: {openurl}?query=rft.jtitle%253DNature%26rft.volume%253D569%26rft.spage%253D698%26rft_id%253Dinfo%253Adoi%252F10.1038%252Fs41586-019-1234-z%26rft_id%253Dinfo%253Apmid%252F31142856%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx [39]: /lookup/external-ref?access_num=10.1038/s41586-019-1234-z&link_type=DOI [40]: /lookup/external-ref?access_num=31142856&link_type=MED&atom=%2Fsci%2F370%2F6519%2F910.atom [41]: #xref-ref-11-1 "View reference 11 in text" [42]: {openurl}?query=rft.jtitle%253DSci.%2BRobot.%26rft.volume%253D4%26rft.spage%253D8892eaau%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx [43]: #xref-ref-12-1 "View reference 12 in text" [44]: {openurl}?query=rft.jtitle%253DNat.%2BNanotechnol%26rft.volume%253D12%26rft.spage%253D907%26rft_id%253Dinfo%253Adoi%252F10.1038%252Fnnano.2017.125%26rft_id%253Dinfo%253Apmid%252F28737748%26rft.genre%253Darticle%26rft_val_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Ajournal%26ctx_ver%253DZ39.88-2004%26url_ver%253DZ39.88-2004%26url_ctx_fmt%253Dinfo%253Aofi%252Ffmt%253Akev%253Amtx%253Actx [45]: /lookup/external-ref?access_num=10.1038/nnano.2017.125&link_type=DOI [46]: /lookup/external-ref?access_num=28737748&link_type=MED&atom=%2Fsci%2F370%2F6519%2F910.atom