<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Thato Selomane on Medium]]></title>
        <description><![CDATA[Stories by Thato Selomane on Medium]]></description>
        <link>https://medium.com/@selomanethato780?source=rss-be6dbe0eba8d------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*D2JHOYy5e059iRii</url>
            <title>Stories by Thato Selomane on Medium</title>
            <link>https://medium.com/@selomanethato780?source=rss-be6dbe0eba8d------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 12:39:00 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@selomanethato780/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[When Chemistry Meets Artificial Intelligence]]></title>
            <link>https://medium.com/@selomanethato780/when-chemistry-meets-artificial-intelligence-5109e50f5f3c?source=rss-be6dbe0eba8d------2</link>
            <guid isPermaLink="false">https://medium.com/p/5109e50f5f3c</guid>
            <category><![CDATA[future]]></category>
            <category><![CDATA[writing]]></category>
            <category><![CDATA[researchers]]></category>
            <category><![CDATA[chemistry]]></category>
            <dc:creator><![CDATA[Thato Selomane]]></dc:creator>
            <pubDate>Fri, 02 Oct 2026 05:49:37 GMT</pubDate>
            <atom:updated>2026-10-02T05:49:37.231Z</atom:updated>
            <content:encoded><![CDATA[<p>Can AI help chemists design materials that have never existed before?</p><p>One of the things I find interesting about scientific research is that a good research question does not always lead directly to an answer. Sometimes it leads to another question that changes the direction of the investigation. After considering how materials can be deliberately designed around particular properties, I began thinking about the amount of work involved in turning such a design into something that can actually be produced and tested. A researcher may have an idea of the structure required, but there could still be thousands of possible chemical arrangements, compositions and synthesis conditions worth investigating. Exploring all of them experimentally would be unrealistic. This is where I think artificial intelligence could become more than a tool used to analyse scientific information. It could become part of the way a research project is planned, conducted and improved.</p><p>In Article 7, I introduced the growing relationship between artificial intelligence and materials discovery. The purpose of this discussion is different. I want to examine what happens when AI is placed inside the research strategy itself, where it does not simply provide information about existing materials but helps researchers decide what to investigate, which experiments to conduct and what direction to take when results do not match expectations. The distinction matters because there is a considerable difference between using a computer to predict a material&#39;s properties and developing a research system in which predictions, chemical reasoning and experimental evidence continually influence one another.</p><p>A useful starting point is the work of Cheng et al. (2026), whose review in Nature Materials examines the development of AI-driven approaches to materials design. Their discussion includes evolutionary algorithms, reinforcement learning and deep generative models, showing how research is moving beyond screening existing candidates towards generating possible structures according to desired properties. This is particularly relevant to the direction I am considering because it raises a question about the role of the researcher. Rather than manually selecting every possible composition for investigation, a researcher could define the scientific requirements and use computational methods to propose candidates that deserve further attention (Cheng et al., 2026).</p><p>Generative models are especially interesting in this context. Most people encounter generative AI through systems that produce written responses, images or computer code, but the underlying concept can also be applied to chemical structures and materials. A generative model learns patterns from examples and uses those patterns to propose new candidates. In materials research, these candidates may be represented as molecular structures, atomic arrangements, crystal structures or other descriptions of a material. The important point is that the model is not restricted to selecting an entry from a list of materials already known to the researcher. Depending on its architecture and training data, it can propose structures that have not previously appeared in the dataset.</p><p>However, producing a new structure computationally does not automatically mean that a new material has been discovered. A proposed structure must still satisfy chemical and physical requirements. Its composition must be meaningful, its atomic arrangement must be plausible, and its predicted properties must justify further investigation. It must also be possible, at least in principle, to synthesise it. This distinction is important because a computer can generate a candidate that looks convincing mathematically but presents serious difficulties when a chemist attempts to produce it in the laboratory. Cheng et al. (2026) discuss several generative approaches, including variational autoencoders and diffusion models, while also identifying the remaining challenges associated with realistic materials generation.</p><p>The development of generative models for crystalline materials illustrates how far this research has progressed. Zeni et al. (2025), in a study published in Nature, presented a generative model for inorganic materials design. Their work demonstrates the potential of generative approaches to propose inorganic compounds and explore chemical possibilities beyond straightforward searches through existing databases. What makes this direction significant is not simply the ability to produce hypothetical structures, but the possibility of using computational methods to explore chemical spaces that would otherwise be difficult to investigate systematically (Zeni et al., 2025).</p><p>I find this particularly interesting when considering how a researcher might approach a problem involving porous materials. Instead of beginning with a particular material and testing whether it has the required properties, a research project could begin by defining the chemical environment and the function that the material must perform. The researcher could then consider the possible structural arrangements, compositions and functional groups that might support the required behaviour. A generative model could help propose candidate structures, while other computational methods could examine their stability and predicted properties. The purpose would not be to allow AI to decide what chemistry should look like, but to use it to investigate possibilities that a researcher might not have considered independently.</p><p>There is another part of this process that deserves attention: deciding what to do after the first predictions have been made. A model might generate hundreds of candidates, but a research team cannot necessarily synthesise every one of them. Some candidates may be too unstable, others may be difficult to produce, and some may have predicted properties that are not sufficiently different from existing materials. The research team therefore needs a way of selecting which candidate would provide the most useful next step. This is where active learning becomes relevant.</p><p>Active learning is an approach in which a model helps select the next data point or experiment from which it expects to learn something useful. Rather than simply choosing the candidate with the highest predicted performance, the system can consider where its knowledge is uncertain or where a new experiment could improve its understanding of the problem. For a researcher, this is a valuable distinction. An experiment does not have to produce the best-performing material to be worthwhile. It may be valuable because it reveals that an assumption was incorrect, identifies an unexpected relationship or helps distinguish between competing explanations. Recent work on active learning in materials discovery has examined how experiments can be selected according to the information they are expected to provide, particularly when experimental resources are limited (Warren et al., 2026).</p><p>I think this is one of the most important ways AI could change the research process. In conventional experimental planning, researchers often decide on the next experiment using previous experience, theoretical understanding and the results already obtained. These remain essential, but an active-learning system could provide an additional way of deciding which experiment deserves priority. If a model is uncertain about how a particular change in composition affects a material&#39;s behaviour, it may recommend an experiment that helps resolve that uncertainty. The result can then be added to the available data, allowing the model to update its predictions before another experiment is selected.</p><p>This creates a relationship in which the computer does not simply produce an answer for the researcher. It participates in determining what information should be collected next. The researcher remains responsible for evaluating whether the proposed experiment makes chemical sense, whether the method is appropriate and whether the result can be interpreted correctly. In this arrangement, AI becomes part of the reasoning process without replacing the scientific judgement needed to conduct meaningful research.</p><p>A particularly interesting development is the movement towards closed-loop experimentation. In this approach, computational predictions and laboratory experiments are connected through a repeated feedback process. A model proposes a candidate or experimental condition, the laboratory carries out the work, the resulting measurements are analysed, and the information is returned to the model. The next decision is then informed by the outcome of the previous experiment. The process continues until the research team has gathered sufficient evidence or reached the objectives of the investigation.</p><p>Angello et al. (2024) demonstrated an important aspect of this approach in their study, Closed-loop transfer enables artificial intelligence to yield chemical knowledge, published in Nature. Their work combined automated experiments, physics-based feature selection and supervised learning to investigate the factors influencing the photostability of molecules used in organic electronics. The researchers explored only a small proportion of the theoretical chemical space while still obtaining information about the chemical factors influencing the property being investigated. This is significant because the purpose of their system was not restricted to finding a favourable experimental outcome. It also aimed to extract scientific understanding from the process.</p><p>That distinction is worth considering. An AI system that identifies a promising candidate may help researchers achieve a particular objective, but a system that also helps explain why the candidate behaves as it does could contribute to a deeper understanding of chemistry. Scientific research should not become a competition to produce the highest-performing material without understanding the reasons behind its behaviour. If we cannot explain the relationship between a material&#39;s structure and its measured properties, we may struggle to reproduce the result or apply the knowledge to another system. Angello et al. (2024) provide an example of how experimental feedback and scientific interpretation can be brought together rather than treated as separate activities.</p><p>The development of autonomous laboratories takes this idea further. These laboratories combine computational systems with automated equipment capable of carrying out selected experimental procedures. In a research environment of this kind, AI may assist with planning, while robotic equipment performs particular stages of synthesis, measurement or analysis. The system can then use the experimental results to guide subsequent work. The idea is not that a laboratory becomes independent of human researchers, but that certain repetitive and highly controlled procedures can be connected to computational decision-making.</p><p>Szymanski et al. (2023) demonstrated this approach through the A-Lab, an autonomous laboratory for the synthesis of inorganic materials. Their system integrated computational predictions, information extracted from scientific literature, machine learning, active learning and robotic equipment. During 17 days of continuous operation, the laboratory carried out 353 experiments and synthesised 36 compounds from 57 targeted materials. The work also examined unsuccessful synthesis attempts, using those results to identify possible improvements to experimental procedures. This example is important because it shows that the relationship between AI and chemistry can extend beyond predicting a material&#39;s properties. Computational systems can also contribute to deciding how a material might be synthesised and how experimental procedures could be improved (Szymanski et al., 2023).</p><p>The unsuccessful experiments are particularly interesting to me. In research, a failed synthesis is often treated as a disappointing result because the intended material was not obtained. However, if the conditions and observations are recorded properly, that experiment can still provide useful information. It may indicate that a reaction temperature was unsuitable, that a precursor combination did not produce the expected phase, or that a proposed synthesis route requires modification. When such information is incorporated into a computational system, the next recommendation can be informed by more than successful experiments alone. The system can begin to learn from the limitations encountered during actual laboratory work.</p><p>This is an important consideration for any research strategy involving AI. A model trained only on successful outcomes may develop an incomplete understanding of the problem. Experimental failures, unexpected products and inconsistent measurements can contain information about the boundaries of a chemical system. They should not automatically be discarded simply because they do not support the original hypothesis. The quality of an AI-assisted research process will therefore depend partly on the quality and completeness of the experimental records supplied to it.</p><p>There is also a practical question about how much independence such systems should be given. Automated equipment can carry out carefully defined procedures, but scientific research involves decisions that are not always straightforward to reduce to numerical targets. A reaction may produce an unexpected result that requires a researcher to reconsider the original explanation. A computational prediction may appear promising while relying on assumptions that are inappropriate for the material being investigated. The researcher must still decide whether the result is scientifically meaningful, whether further testing is necessary and whether the model is being asked to answer the correct question.</p><p>The broader literature on self-driving laboratories recognises this as an important part of their development. Tom et al. (2024), in their review of self-driving laboratories for chemistry and materials science, discuss the integration of automated experimentation, data processing and decision-making. Their work highlights the potential of these systems while also showing that building an effective autonomous research environment requires more than connecting a machine-learning model to laboratory equipment. Experimental design, instrumentation, data handling, reproducibility and human oversight all matter.</p><p>For the research direction I am beginning to develop, this raises a more specific question. If I were investigating materials for applications involving water and energy, how could AI be incorporated into the actual process of selecting and evaluating candidates? I would not want to begin by asking a model to generate a material simply because it is new. I would first need a clearly defined research problem, supported by chemical reasoning and an understanding of the properties that matter. The computational system would then have a meaningful task rather than an open-ended instruction to produce something impressive.</p><p>For example, a research project could investigate whether a particular class of porous materials can be modified to obtain a specific combination of chemical and structural characteristics. A generative model might propose possible modifications, while computational calculations could help assess their predicted stability and interactions. An active-learning method could then identify which candidates would provide the most useful information if synthesised. Once the material has been produced, experimental measurements could be compared with the predictions, and the results could guide the next stage of investigation. This would create a research process in which chemical knowledge determines the problem, AI helps explore the possibilities, and experimental chemistry establishes what actually happens.</p><p>I would also want the system to account for the fact that a material can have several competing requirements. A candidate with a favourable predicted property may be difficult to synthesise, while another may be easier to produce but less effective for the intended application. These differences should influence which experiments are selected. This is not simply a matter of asking AI to identify the material with the highest predicted value. It is about using computational tools to help researchers understand the available choices and the consequences of pursuing them.</p><p>There is still a considerable gap between generating a hypothetical material and demonstrating that it can be produced reliably. Generative models may propose structures that satisfy certain computational conditions, but real materials can contain defects, disorder and variations introduced during synthesis. These features may influence their properties in ways that are not fully represented in an idealised model. A computational candidate must therefore pass through further assessment before it can be regarded as a genuine materials discovery. Its structure, composition, stability and experimentally measured behaviour all need to be investigated.</p><p>The limitations of AI should therefore remain part of the discussion rather than being treated as minor obstacles. Models depend on the data and assumptions used to develop them, and predictions may become unreliable when they are applied to systems that differ significantly from their training examples. Warren et al. (2026) discuss several barriers to implementing AI and machine-learning methods in materials advancement, including data availability, experimental integration, model reliability and the development of autonomous research systems. These issues are important because an impressive computational result is not necessarily evidence that the proposed material will behave as expected outside the model.</p><p>A further concern is whether the research community can reproduce the results produced by AI-assisted systems. If a model recommends a particular synthesis route, researchers need access to the relevant conditions, measurements and reasoning behind that recommendation. If the system produces a result that cannot be independently checked, its usefulness becomes limited. For this reason, I believe transparency should be considered part of the research strategy. The aim should not be to create a system that produces answers which are difficult to question, but one that helps researchers investigate chemical problems more systematically.</p><p>What I find most promising is the possibility of developing a research process in which AI does not operate as an isolated tool. It becomes connected to the decisions made before an experiment, the measurements obtained during the experiment and the interpretation that follows. The researcher defines the scientific objective and provides the chemical understanding, while computational methods help explore candidate possibilities and identify useful experiments. Laboratory evidence then provides the information needed to correct predictions and improve the next stage of investigation.</p><p>This is the kind of relationship I want to understand more deeply as I continue exploring materials science. I am not suggesting that every chemistry laboratory should become fully autonomous, or that AI will eventually make experimental researchers unnecessary. There are still many questions about the reliability, accessibility and practical implementation of these systems. However, the published work shows that AI-assisted research is moving towards more integrated forms of investigation, where the process of designing, testing and learning is connected rather than divided into unrelated stages.</p><p>For me, the most important question is therefore not whether artificial intelligence can generate a material that has never existed before. It is whether we can develop a research strategy in which a scientifically meaningful problem guides the computational search, the proposed material can be tested through chemistry, and the experimental evidence can improve what we attempt next. That would make AI part of the scientific method being developed for a particular research problem, rather than simply another technology mentioned in an article about materials discovery. I am still developing my understanding of how such a system could be applied to the materials that interest me. But I can see why the connection between chemistry, computational design and experimental feedback deserves serious attention. The opportunity is not just to produce more candidate materials. It is to organise research in a way that allows each calculation and experiment to contribute to the next decision.</p><p>Perhaps the most interesting possibility is that a research project could gradually become more informed as it progresses. The first prediction would not need to be perfect, and the first experiment would not need to solve the entire problem. Each result would provide another piece of evidence, allowing the research strategy to become more focused. In that sense, the value of AI may lie not only in what it can generate, but also in how it can help a researcher learn from the process of trying to create something new.</p><p>The question I want to take into the next article is what happens when this research strategy has to deal with several competing requirements at once. A material may need to be chemically stable, affordable, structurally suitable and capable of performing more than one function. Can a computational system help researchers navigate those competing demands without losing sight of the chemistry that makes the material possible? That is where I want to take this investigation next.</p><h3>References</h3><ul><li>Angello, N.H. et al. (2024) Closed-loop transfer enables artificial intelligence to yield chemical knowledge, Nature. Available at: [https://doi.org/10.1038/s41586-024-07892-1](https://doi.org/10.1038/s41586-024-07892-1).</li><li>Cheng, M. et al. (2026) Artificial intelligence-driven approaches for materials design and discovery, Nature Materials, 25, pp. 174–190. Available at: [https://doi.org/10.1038/s41563-025-02403-7](https://doi.org/10.1038/s41563-025-02403-7).</li><li>Szymanski, N.J. et al. (2023) An autonomous laboratory for the accelerated synthesis of inorganic materials, Nature, 624, pp. 86–91. Available at: [https://doi.org/10.1038/s41586-023-06734-w](https://doi.org/10.1038/s41586-023-06734-w).</li><li>Tom, G. et al. (2024) Self-driving laboratories for chemistry and materials science, Chemical Reviews, 124(16), pp. 9633–9732.</li><li>Warren, J.A. et al. (2026) Overcoming roadblocks for implementing AI/ML methods for materials advancement, Annual Review of Materials Research, 56, pp. 1–24. Available at: [https://doi.org/10.1146/annurev-matsci-072924-102951](https://doi.org/10.1146/annurev-matsci-072924-102951).</li><li>Zeni, C. et al. (2025) A generative model for inorganic materials design, Nature, 639, pp. 624–632.</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5109e50f5f3c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Can We Design Materials Instead of Finding Them by Trial and Error?]]></title>
            <link>https://medium.com/@selomanethato780/can-we-design-materials-instead-of-finding-them-by-trial-and-error-6e4433b4c3bf?source=rss-be6dbe0eba8d------2</link>
            <guid isPermaLink="false">https://medium.com/p/6e4433b4c3bf</guid>
            <category><![CDATA[future]]></category>
            <category><![CDATA[writing]]></category>
            <category><![CDATA[chemistry]]></category>
            <category><![CDATA[materials-science]]></category>
            <dc:creator><![CDATA[Thato Selomane]]></dc:creator>
            <pubDate>Wed, 09 Sep 2026 04:28:47 GMT</pubDate>
            <atom:updated>2026-09-09T04:28:49.402Z</atom:updated>
            <content:encoded><![CDATA[<p>This takes us into rational materials design, structure–property relationships, computational chemistry, molecular modelling and prediction. Research consideration: Instead of asking: What material can we find? we begin asking: What properties do we need, and what structure could produce those properties?</p><p>There is a question I have started asking myself as I read more about materials science: are we really limited to searching through materials until we happen to find one that works, or can we eventually design the material we need before we make it?</p><p>For a long time, materials development has depended heavily on experimentation. A researcher proposes a composition, synthesises it, characterises it and tests its properties. If the result is not good enough, another composition or structure is tried. The process is repeated until something useful appears. There is nothing wrong with this approach because many important discoveries have come from experiments that could not have been predicted completely beforehand. The problem is that the number of possible materials is enormous. We cannot realistically synthesise and test every possible combination of elements, structures and processing conditions. This is where the idea of rational materials design becomes interesting.</p><p>Instead of beginning with a material and asking what it can do, we can begin with a desired property and work backwards. If we need a material that can selectively capture a particular molecule, for example, what type of pore size, surface chemistry and interaction would make that possible? If the target is energy storage, what combination of composition, structure and ion movement could provide the required performance? If the application involves water, how would the material behave when several contaminants are present rather than only the one used during a controlled laboratory experiment?</p><p>The question changes from “What material can we find?” to “What material would have the properties we need?”</p><p>That may sound like a small change in wording, but scientifically it represents a very different way of approaching materials research.</p><p>The reason this is possible at all is that materials do not have their properties by accident. Their behaviour is strongly connected to their composition and structure. The arrangement of atoms, the nature of chemical bonds, defects, surfaces, interfaces, pore structures and electronic behaviour can all influence what a material is capable of doing. Research in computational materials science has shown that understanding and predicting these relationships can help researchers investigate materials before they are physically synthesised (Oganov <em>et al</em>., 2019; Louie <em>et al</em>., 2021).</p><p>This gives chemistry a different role. Chemistry is not only about producing a substance and measuring what happens. It can also help us understand why a particular structure produces a particular behaviour. Once that relationship becomes sufficiently understood, it becomes possible to think about changing the structure deliberately in order to change the property.</p><p>That is the foundation of a structure–property relationship.</p><p>Consider a porous material. Its usefulness may depend on much more than simply having pores. The size and distribution of those pores can influence which molecules enter. The chemical groups on the internal surface can determine whether certain molecules interact strongly or weakly with the material. The surrounding framework can influence stability, diffusion and regeneration. If we understand these relationships, then modifying the material no longer has to be completely based on trial and error. We can begin making informed predictions.</p><p>This is where computational chemistry becomes valuable. Techniques such as density functional theory, molecular dynamics and other forms of modelling allow researchers to investigate structures and interactions at scales that are difficult to observe directly. These calculations can be computationally demanding, but they provide a way of exploring possible materials before committing to every candidate experimentally. Computational approaches are now regarded as an important part of modern materials research alongside experimental and theoretical methods (Louie <em>et al</em>., 2021).</p><p>The advantage becomes clearer when the number of possibilities increases.</p><p>Imagine that I want to investigate a family of porous materials. Instead of synthesising hundreds or thousands of possible structures individually, I could potentially use calculations to eliminate candidates that are unlikely to possess the desired properties. The remaining candidates could then be examined experimentally. This does not remove experiments. It makes the experiments more targeted.</p><p>High-throughput computational methods have already demonstrated this principle. Curtarolo et al. (2013) described computational materials design as an approach that combines electronic-structure calculations, thermodynamic methods, databases and large-scale computing to investigate large numbers of possible materials.</p><p>That is an important development because it changes the relationship between computation and the laboratory. Computation does not necessarily replace experimental chemistry. Instead, it can help decide where experimental chemistry should look.</p><p>The same idea has developed further through machine learning.</p><p>Machine learning models can be trained using existing materials data and then used to identify relationships between composition, structure and properties. Reviews of machine-learning-based materials research describe applications ranging from property prediction and materials screening to the design of new materials (Butler <em>et al</em>., 2018; Liu <em>et al</em>., 2020).</p><p>What interests me most is not simply that a computer can make predictions. It is what those predictions could change about the research process.</p><p>Traditionally, we might start with a collection of available materials and test them until one performs well enough. In a more design-oriented approach, we start with the performance we want and search backwards for structures that could potentially provide it.</p><p>That is sometimes described as inverse design.</p><p>The distinction becomes important when we think about the problems I have been exploring throughout this series. If the goal is to develop a material for water purification, for example, we could define the requirements before choosing the material. We might need strong interaction with a particular contaminant, selectivity in the presence of competing substances, chemical stability in water, regeneration capability and reasonable production requirements. Those requirements become the starting point.</p><p>The material becomes the answer we are trying to construct. That way of thinking also changes how failure is interpreted. If an experimentally produced material does not work, the result is not necessarily wasted effort. It can tell us that a particular structural feature did not produce the expected property. That information can be fed back into the next design cycle.</p><p>The research process becomes iterative. Predict. Design. Synthesise. Test. Learn. Then design again.</p><p>This is very different from simply producing material after material and hoping that one performs well.</p><p>However, I do not think we should romanticise computational design either. A prediction is not the same thing as an experimental result. Models depend on the quality of the data, the assumptions built into the calculations and the range of materials represented in the training data. Machine-learning models can also struggle when asked to predict materials that are significantly different from those on which they were trained. Recent reviews continue to identify data quality, interpretability and generalisation as major challenges in machine-learning-driven materials research.</p><p>This matters because it prevents the idea of rational design from becoming another form of overconfidence.</p><p>We are not at a point where we can simply tell a computer what material we want and expect it to produce the perfect answer.</p><p>The chemistry still matters.</p><p>The experiment still matters. The unexpected result still matters. What is changing is the amount of information that can be used before an experiment is performed.</p><p>This is perhaps the part of materials science that interests me most. We are moving towards a situation where chemistry, computation and experimentation can inform each other rather than functioning as completely separate activities. A computational model can identify promising structures, an experimentalist can synthesise and test them, and the resulting data can then improve the model. The process can continue.</p><p>This also connects with the way I have been thinking about advanced materials since the previous articles. In Article 11, I considered the problem of materials that perform exceptionally well under controlled laboratory conditions but struggle when exposed to real environments. In Article 12, I moved further and asked whether a high-performing material is actually better if its synthesis, raw materials, regeneration or end-of-life create other problems.</p><p>Those questions should not disappear when we start designing materials. In fact, they should become part of the design process.</p><p>If I eventually want to design a material for a real application, I should not ask a computer to maximise one property while ignoring everything else. A theoretically excellent material may depend on an element that is difficult to obtain. Another candidate may have impressive predicted performance but require a complicated synthesis. Another may perform slightly less well but be far easier to produce and regenerate.</p><p>This means that rational materials design should eventually become more than property prediction. It should become constraint-aware design.</p><p>The material needs to perform, but it also needs to satisfy the conditions of the problem. This is where I see a possible connection between the ideas developed in the previous articles and the research direction I am trying to build. The goal would not simply be to discover a new material because it has never been reported before. Novelty matters, but novelty by itself is not enough. I would rather ask whether a material can be deliberately designed around a real need and whether its structure can explain why it performs that function. That distinction is important to me.</p><p>There are millions of possible chemical structures. Finding something new is not necessarily the hardest part anymore. The more interesting challenge may be finding something new for a reason.</p><p>For example, instead of randomly modifying a porous framework and testing whether its adsorption improves, we could first identify the chemical interaction required to capture a particular target. We could then consider which functional groups might produce that interaction, which pore environment would support it, and which structural characteristics would maintain stability. Computational modelling could help compare possible structures before synthesis. The experiment would then test a hypothesis rather than simply search blindly.</p><p>That is what I mean when I talk about moving away from trial and error. I do not mean eliminating experimentation. I mean making experimentation more intelligent.</p><p>This distinction has become increasingly relevant as materials databases and computational methods have grown. The Materials Genome Initiative, launched in the United States in 2011, was built around the broader goal of accelerating materials innovation by integrating computation, experiments and data rather than relying entirely on conventional development cycles. Later work in machine-learning-driven materials discovery has continued developing this idea, including property prediction, high-throughput screening and inverse design. The direction is clear even though many difficulties remain.</p><p>Materials science is beginning to move from a world where we mainly ask what materials exist towards one where we can increasingly ask what materials could exist and what they might be capable of doing. That is a much larger scientific space.</p><p>It also raises a question that I think will become increasingly important in the future: if we can design materials for specific properties, how many properties should we design for at the same time?</p><p>A material for water purification might need adsorption capacity, selectivity, stability and regeneration. A material for energy storage might need energy density, cycle life, safety and resource availability. A material intended for both environmental and energy applications would introduce even more constraints.</p><p>The problem becomes an optimisation problem rather than a search for a single best material. And perhaps this is where my own research thinking is heading.</p><p>I am becoming less interested in the idea of simply finding a material that performs well and more interested in whether we can understand the relationship between structure, chemistry and function well enough to design the material deliberately.</p><p>That does not mean I already know what the final material should look like. I do not.</p><p>And I think that is exactly where research begins.</p><p>The important thing at this stage is identifying the question.</p><p>Can we start with a real-world problem, define the properties required to address it, identify the structures and chemical interactions that could produce those properties, use computation to narrow the possibilities, and then return to the laboratory to determine whether the predictions survive reality?</p><p>If that cycle becomes increasingly reliable, materials science could become less dependent on discovering useful materials by accident and more capable of creating them with purpose.</p><p>For me, that is a much more exciting idea than simply searching through an enormous list of existing materials.</p><p>It suggests that the future researcher may not always begin with a material. They may begin with a problem. Then a required property. Then a structure. Then a prediction. And finally, an experiment.</p><p>That is a different way of thinking about chemistry, and it is one of the reasons I am becoming increasingly interested in the point where chemistry, materials science, computation and experimental research meet.</p><p>The question I want to carry forward is therefore no longer simply: What material can we find?</p><p>It is: If we know what the world needs, can we learn enough chemistry to design the material that could help provide it?</p><p>I think that question is worth pursuing.</p><h3><strong>References</strong></h3><ul><li>Butler, K.T., Davies, D.W., Cartwright, H., Isayev, O. and Walsh, A. (2018) <em>Machine learning for molecular and materials science</em>, Nature, 559, pp. 547–555.</li><li>Curtarolo, S., Hart, G.L.W., Nardelli, M.B., Mingo, N., Sanvito, S. and Levy, O. (2013) <em>The high-throughput highway to computational materials design</em>, Nature Materials, 12, pp. 191–201.</li><li>Liu, Y., Zhao, T., Ju, W. and Shi, S. (2020) <em>Machine learning in materials genome initiative: A review,</em> Journal of Materials Science &amp; Technology, 57, pp. 113–122.</li><li>Louie, S.G., Chan, Y.-H., da Jornada, F.H., Li, Z. <em>et al</em>. (2021) <em>Discovering and understanding materials through computation,</em> Nature Materials, 20, pp. 728–735.</li><li>Nematov, D. and Hojamberdiev, M. (2025) Machine learning-driven materials discovery: Unlocking next-generation functional materials – <em>A review,</em> Computational Condensed Matter, 45, e01139.</li><li>Oganov, A.R., Pickard, C.J., Zhu, Q. and Needs, R.J. (2019) <em>Structure prediction drives materials discovery, </em>Nature Reviews Materials, 4, pp. 331–348.</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6e4433b4c3bf" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Hidden Cost of Advanced Materials]]></title>
            <link>https://medium.com/@selomanethato780/the-hidden-cost-of-advanced-materials-500a89df2705?source=rss-be6dbe0eba8d------2</link>
            <guid isPermaLink="false">https://medium.com/p/500a89df2705</guid>
            <category><![CDATA[future]]></category>
            <category><![CDATA[researchers]]></category>
            <category><![CDATA[chemistry]]></category>
            <category><![CDATA[materials]]></category>
            <dc:creator><![CDATA[Thato Selomane]]></dc:creator>
            <pubDate>Thu, 03 Sep 2026 20:32:32 GMT</pubDate>
            <atom:updated>2026-09-03T20:32:34.065Z</atom:updated>
            <content:encoded><![CDATA[<p>Does advanced automatically mean better?</p><p>The more I read about advanced materials, the more I have started questioning something that initially seemed obvious to me: if a material performs better than the one it replaces, why would we not consider it better? In materials science, performance is often one of the first things we look at. We measure adsorption capacity, conductivity, strength, selectivity, catalytic activity, energy density or stability and compare the results with existing materials. A material that produces a higher value can immediately appear more promising. But the further I have gone into the subject, the less convincing that way of thinking has become. A material can be scientifically impressive and still be a poor choice when its entire life is considered.</p><p>This is an important distinction because the word <em>advanced</em> describes what a material can do, but it does not necessarily tell us how responsibly that performance was achieved. A material may have an exceptional structure and excellent properties while requiring expensive precursors, hazardous chemicals, large amounts of energy or scarce elements during its production. It may work well during its first application but be difficult to regenerate, recycle or safely dispose of afterwards. These factors are not separate from materials science. They are part of the material’s overall usefulness. Recent work on safe-and-sustainable-by-design approaches makes this point increasingly clear. Caldeira et al. (2024), after reviewing 155 sustainability assessment frameworks, found that sustainability assessment has developed considerably but that important dimensions, particularly social and economic aspects, are still not consistently quantified. Their review also emphasises the importance of considering sustainability throughout a material’s life rather than assessing only one stage of its development.</p><p>This changes the question I would ask when evaluating a new material. Instead of asking only, How well does it perform?, I would also want to know: What was required to produce it? Where do its raw materials come from? How much energy is involved? What chemicals are consumed? What happens when the material reaches the end of its useful life? Can it be regenerated or reused? And, perhaps most importantly, does the improvement in performance justify the additional burden created elsewhere?</p><p>The question of raw materials is particularly difficult for technologies connected to the energy transition. The International Energy Agency reported that global lithium demand increased by nearly 30% in 2024, while demand for nickel, cobalt, graphite and rare earth elements also increased. Much of this growth was associated with electric vehicles, batteries, renewable energy and electricity networks (IEA, 2025).</p><p>This does not mean that materials containing these elements are inherently unsustainable. That would be too simplistic. Rather, it shows why the availability and sourcing of elements have to become part of the design discussion. If demand for a particular element grows rapidly while supply is geographically concentrated or difficult to expand, depending heavily on that element can introduce another vulnerability into a technology.</p><p>For me, this is where materials chemistry becomes more interesting. The challenge is no longer simply finding a composition that produces the desired property. It becomes a question of whether the composition is sensible when viewed within a much larger system. The same principle applies to the synthesis of advanced materials. Laboratory research can make a synthesis route appear straightforward because experiments are often performed on a relatively small scale. A researcher may obtain a highly ordered structure using carefully selected solvents, catalysts, ligands, additives or reaction conditions. The resulting material may then show excellent performance. But every one of those inputs has an environmental and economic footprint.</p><p>A recent review of bioderived advanced materials illustrates why even apparently sustainable material choices need careful assessment. Kaksonen et al. (2024) found substantial variation in the environmental performance of bioderived advanced materials depending on factors such as the production pathway, chemical inputs, energy requirements and other stages of the life cycle. In other words, using a renewable or biological starting material does not automatically make the final material sustainable.</p><p>That observation is important because it prevents sustainability from becoming another label that we attach to materials without examining the chemistry behind them. A material derived from biomass might reduce dependence on fossil resources, but if its conversion requires large quantities of solvents, energy or other chemicals, some of the expected environmental advantage may be reduced. The correct answer therefore cannot come from the identity of the starting material alone. This is also why I think regeneration deserves more attention when we discuss materials for applications such as water treatment. An adsorbent that removes a contaminant effectively during its first cycle is useful, but that is only part of the story. If the material becomes exhausted rapidly and cannot be regenerated economically or safely, its practical value changes. The same material may look excellent in a single laboratory experiment but become much less attractive when repeated cycles, regeneration chemicals, energy requirements and waste streams are considered.</p><p>The idea can be extended beyond water treatment. In energy storage, for example, improving energy density is valuable because it can increase the amount of energy stored relative to the mass or volume of a device. But energy density alone cannot determine whether a material represents a better technology. Safety, durability, raw-material requirements, manufacturing, cost and end-of-life management also influence the usefulness of the system. This is why I am beginning to see materials development less as a search for one superior property and more as a balancing problem.</p><p>A material may be highly conductive but expensive. Another may be cheap but unstable. One may have excellent adsorption capacity but require difficult regeneration. Another may be less impressive in a single performance test but use abundant raw materials, survive many cycles and be much easier to manufacture. The best material therefore depends on what we are actually trying to optimise.</p><p>Researchers are increasingly responding to this problem through approaches described as Safe-and-Sustainable-by-Design (SSbD). The basic idea is significant because safety and sustainability are considered during the development of a material rather than being treated as problems to solve after the material has already been designed. Kostapanou et al. (2024), for example, describe SSbD as an approach for reconsidering the advanced-materials lifecycle while maintaining the required functionality and performance. I find this shift particularly interesting because it changes the role of the materials scientist. Instead of developing a material first and asking later whether it is safe, affordable or sustainable, those constraints can become part of the design problem from the beginning.</p><p>That does not mean sacrificing performance. It means recognising that performance has more than one dimension. If I were beginning to evaluate a material for one of the research areas that interest me, I would want to look beyond the headline number reported in a paper. I would want to consider at least five broad questions: <em>how well does it perform, how affordable is it, how sustainable is its production and use, how readily available are the resources required to make it, and how stable is it during its intended lifetime?</em></p><p>I am not suggesting that these five considerations form a universal scientific equation for judging materials. They are better understood as the criteria I am beginning to think should sit together when I consider what makes a material genuinely useful. A material that performs exceptionally well but fails badly on several of these other dimensions may not represent the improvement that its performance initially suggests. This way of thinking also creates an interesting connection with computational materials science and artificial intelligence. If materials discovery is increasingly becoming a problem of searching through enormous numbers of possible compositions and structures, then perhaps the objective should not be to predict only one property. A genuinely useful design process could eventually consider several constraints at once. Recent work has already begun discussing how artificial intelligence could connect materials performance with life-cycle assessment and sustainability metrics across design, use and recycling (Han <em>et al</em>., 2026).</p><p>That possibility changes the type of question I want to ask in my own development as a materials scientist. Instead of asking only whether we can discover materials with higher performance, I am becoming more interested in whether we can design materials whose performance is achieved without creating equally serious problems somewhere else.</p><p>This becomes especially important for countries and communities that cannot afford technologies designed around expensive or highly specialised supply chains. A material intended to address a problem such as water purification or energy storage cannot be judged entirely by what happens inside a well-equipped laboratory. Its value ultimately depends on whether the chemistry can make sense under the conditions in which the material is expected to exist. The more I think about this, the more I realise that the word <em>advanced</em> can be misleading. It can make us focus on novelty, complexity and impressive performance while overlooking simpler questions about usefulness. Perhaps the real advancement is not producing the most complicated material or achieving the highest number in a performance test. Perhaps it is learning how to achieve the required function with fewer compromises. That is a much harder scientific problem.</p><p>It requires chemistry, materials science, engineering and environmental thinking to interact rather than operate independently. It also requires researchers to think about a material before it enters the laboratory, while it is being produced and used, and after its useful life has ended. Life-cycle assessment and related sustainability frameworks are increasingly being used for precisely this reason: the environmental performance of an advanced material cannot reliably be understood by looking at one stage in isolation (Caldeira <em>et al</em>., 2024; Kaksonen <em>et al</em>., 2024).</p><p>For me, this is where the idea of advanced materials becomes more meaningful. I no longer want to think of an advanced material simply as something with a more sophisticated structure or a better performance value. I want to understand whether the material represents an improvement when the entire chain of consequences is considered. A material should be able to do what we need it to do. But ideally, it should also be possible to make it responsibly, obtain its resources reasonably, use it safely, maintain its performance, recover value from it where possible and manage what remains at the end.</p><p>That changes the question from <strong>Can we make a better material?</strong> to something much more demanding: <strong>Can we design a material that is better in the way that actually matters?</strong></p><p>That is the question I want to carry into the next stage of this research journey. Because if materials science is going to help solve problems in water, energy and other areas of human need, then discovering what a material can do may only be the beginning. The harder task may be designing something that performs well without passing its hidden costs on to someone else, somewhere else, or some point in the future.</p><h3>References</h3><ul><li>Caldeira, C., Abbate, E., Moretti, C., Mancini, L. and Sala, S. (2024) <em>Safe and sustainable chemicals and materials: a review of sustainability assessment frameworks</em>, Green Chemistry, 26, pp. 7456–7477.</li><li>Han, N., Guo, H., Zhang, B. et al. (2026) <em>Artificial intelligence as a driver of sustainable materials and circularity</em>, Nature Reviews Materials, 11, pp. 510–512.</li><li>International Energy Agency (IEA) (2025) Global Critical Minerals Outlook 2025. Paris: IEA.</li><li>Kaksonen, A.H. et al. (2024) <em>Life cycle sustainability assessment of bioderived advanced materials: A state-of-the-art review</em>, Journal of Environmental Chemical Engineering, 12, 114387.</li><li>Kostapanou, A., Chatzipanagiotou, K.-R., Damilos, S., Petrakli, F. and Koumoulos, E.P. (2024) <em>Safe-and-Sustainable-by-Design Framework: (Re-)Designing the Advanced Materials Lifecycle,</em> Sustainability, 16(23), 10439.</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=500a89df2705" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[When a Material Looks Perfect on Paper]]></title>
            <link>https://medium.com/@selomanethato780/when-a-material-looks-perfect-on-paper-bc359169a879?source=rss-be6dbe0eba8d------2</link>
            <guid isPermaLink="false">https://medium.com/p/bc359169a879</guid>
            <category><![CDATA[chemistry]]></category>
            <category><![CDATA[future]]></category>
            <category><![CDATA[researchers]]></category>
            <dc:creator><![CDATA[Thato Selomane]]></dc:creator>
            <pubDate>Thu, 03 Sep 2026 09:53:57 GMT</pubDate>
            <atom:updated>2026-09-03T09:53:57.383Z</atom:updated>
            <content:encoded><![CDATA[<p>Why can a material that performs exceptionally well in the laboratory still fail in the real world?</p><p>There is something fascinating about materials science. A scientist can take a material that has never existed in that particular form before, synthesise it in a laboratory, characterise its structure, and discover that it performs exceptionally well. It may remove pollutants from water faster than an existing adsorbent, store more energy than a conventional electrode material, separate one molecule from another with remarkable selectivity, or catalyse a reaction more efficiently. At that moment, it is easy to become excited about what the material could become.</p><p>But there is a question that comes after the excitement: <em>Will it still work when the conditions are no longer perfect?</em></p><p>This question is becoming increasingly important as materials research moves towards more sophisticated structures and increasingly ambitious applications. A material can look outstanding in a research paper and still be a poor candidate for practical use. The problem is not necessarily that the science was wrong. Rather, the conditions under which the material was tested may have been very different from the conditions in which it would eventually have to operate.</p><p>Consider an adsorbent developed for water purification. In a laboratory experiment, a researcher may prepare a solution containing a known concentration of a particular contaminant and then introduce a carefully measured amount of the adsorbent. The pH can be controlled. Temperature can be controlled. Contact time can be selected. The concentration of the contaminant is known, and there may be very few other substances competing for the available adsorption sites.</p><p>Under those conditions, the material may produce an impressive removal percentage or adsorption capacity.</p><p>Actual wastewater is rarely that simple.</p><p>Water coming from an industrial process, mine, agricultural area or municipal system can contain a mixture of contaminants rather than one isolated substance. Different ions and organic molecules can compete for active sites. Changes in pH can alter the charge of both the pollutant and the material surface. Natural organic matter can interfere with adsorption. The concentration of the target contaminant may also be much lower than the concentration used during laboratory testing. Researchers studying adsorption have therefore emphasised factors such as pH, temperature, contaminant concentration, adsorbent dosage and competing conditions because these variables can significantly influence performance.</p><p>This creates an important distinction between demonstrating that a material can work and demonstrating that the material can work reliably in a real system. The difference may appear small when written on paper, but scientifically it is enormous. A similar problem exists in energy storage. A new electrode material might demonstrate excellent capacity during an initial laboratory test. Its theoretical properties may suggest that it could outperform an existing battery material. Yet a battery is not simply a powder placed inside a container. The material must be manufactured consistently, combined with other components, processed into an electrode, and operated repeatedly under realistic conditions.</p><p>Scaling a material from a few grams in a laboratory to kilograms or tonnes introduces problems that may not be visible during early experiments. Research on lithium-based battery materials has highlighted challenges involving synthesis yield, impurities, quality control, processing and manufacturing consistency. Something that can be produced successfully in a small laboratory experiment may therefore behave very differently when the same process is expanded. This is one of the reasons I have started to think differently about the word performance. Performance is not one number.</p><p>A material’s performance might include its adsorption capacity, but it could also include how quickly it works, how selective it is, whether it can be regenerated, how stable it remains, how much material is required, and what happens to it after use. For a battery material, performance could involve energy density, power capability, cycle life, safety, stability, manufacturing compatibility and cost. The best material, therefore, is not necessarily the material with the highest value in one category. It may be the material that maintains a useful balance across several categories.</p><p>This becomes particularly clear when regeneration is considered in water treatment. Imagine developing an adsorbent that captures a pollutant extremely effectively. If the material can only be used once, the story does not end with the removal of the pollutant. The exhausted material now has to be regenerated, reused or disposed of. Otherwise, part of the environmental problem has simply been transferred from contaminated water to contaminated solid waste. Reviews of adsorbent regeneration have identified recovery, reuse and sustainable management of spent adsorbents as important challenges for moving adsorption technologies towards broader application. This changes the research question.</p><p>Instead of asking only:</p><p><strong><em>How much contaminant can this material remove?</em></strong></p><p>we should eventually be asking:</p><p><strong><em>How much contaminant can this material remove over its useful lifetime, under realistic conditions, at an acceptable cost and environmental burden?</em></strong></p><p>That is a much harder question. It is also a much more useful one. The same thinking applies to energy materials. Suppose a new electrode has a very high theoretical capacity. That sounds impressive, but the material still has to survive repeated charging and discharging. It has to interact properly with the electrolyte and other battery components. Its synthesis has to be reproducible. Its properties must remain sufficiently stable over time. If the material depends on expensive or difficult-to-source elements, its practical value may also be limited.</p><p>This is where laboratory science begins to meet engineering and economics.</p><p>A recent discussion of techno-economic analysis in clean technologies argues that translating laboratory innovations towards the market requires attention to both technical and economic performance. Cost models and process modelling can help researchers understand whether a promising innovation remains competitive when scaled up.</p><p>That idea is important because scientific research can sometimes concentrate heavily on proving that something works while leaving the question of whether it can ever be made practically for much later.</p><p>I do not think this means that every early-stage experiment needs to be commercially ready. Fundamental research has a different purpose, and many important discoveries begin with very small laboratory experiments. The point is not to demand a finished product from the first experiment. The point is to understand the eventual constraints early enough that they can influence the direction of the research.</p><p>For someone interested in materials for both water and energy applications, this becomes even more important. Water treatment and energy storage are very different technological fields, but they share some underlying materials challenges. In both cases, researchers care about stability, transport, surface interactions, selectivity, reproducibility and the ability to control material properties. In both cases, a material may perform well under carefully selected conditions but encounter a much more complicated environment during actual operation.</p><p>That makes the environment in which a material functions part of the scientific problem itself. If a material is intended for wastewater treatment, perhaps it should be tested in increasingly complex water matrices rather than only in highly purified laboratory solutions. If it is intended for energy storage, perhaps its behaviour should be studied not only during an initial performance test but also over extended cycling and under conditions closer to practical operation. The experimental question becomes more realistic as the material moves closer to its intended application.</p><p>There is another issue that is becoming especially interesting to me: the connection between materials discovery and artificial intelligence.</p><p>AI can help researchers search enormous chemical and materials spaces, predict properties and identify promising candidates. But a predicted material is not automatically a useful material. Recent work in materials science has stressed that data scarcity, synthesis complexity and chemical knowledge remain important limitations. Computational predictions ultimately have to be connected to real chemistry and experimental validation.</p><p>This creates what I see as an important chain:</p><p><strong>prediction → synthesis → characterisation → testing → feedback → improvement.</strong></p><p>If one part of that chain is weak, the whole process can suffer. A model might identify a material with an attractive predicted property, but if the material cannot be synthesised reliably, the prediction has limited practical value. A material might be synthesised successfully but lose its predicted properties because of defects, impurities or differences between the ideal computational structure and the real sample. Even when the material works, it may still be too expensive or unstable for its intended application.</p><p>This does not make computational materials science less valuable. In my view, it makes the connection between computation and experiment even more important.</p><p>The real opportunity may not be to use AI to replace laboratory research, but to make the laboratory research more intelligent.</p><p>Instead of synthesising hundreds of materials without a clear strategy, researchers could use computational methods to narrow the possibilities, experimentally test selected candidates, feed the results back into the models, and continue the cycle. The material that eventually emerges would not simply be the result of computation or experimentation alone. It would be the product of an interaction between the two.</p><p>This also changes how I think about the future of porous materials.</p><p>Zeolites taught me that structure matters. Moving beyond zeolites into MOFs, COFs and other advanced porous systems has shown me that researchers can manipulate pore environments, surface chemistry and composition to obtain particular properties. But Article 10 also raised a larger question: <strong>why continue developing increasingly complicated materials if we cannot understand what makes them useful outside the laboratory?</strong></p><p>Perhaps the next step in materials chemistry is not simply to make materials more sophisticated. Perhaps it is to make our understanding of their behaviour more sophisticated.</p><p>A material with a slightly lower maximum adsorption capacity might ultimately be more valuable if it is inexpensive, stable, regenerable and effective in real wastewater. A battery material with a lower theoretical energy density might still be attractive if it can be manufactured safely from abundant elements and maintain its performance for thousands of cycles. A porous material with an extraordinary predicted structure may be scientifically interesting, but its importance depends partly on whether that structure can actually be produced and maintained.</p><p>This is where I think the meaning of “better material” becomes complicated.</p><p>Better according to what?</p><p>Higher capacity?</p><p>Greater selectivity?</p><p>Lower cost?</p><p>Longer lifetime?</p><p>Safer chemistry?</p><p>Lower environmental impact?</p><p>Easier manufacturing?</p><p>The answer may be all of them, but improving one property can sometimes make another worse. Increasing complexity can increase synthesis cost. Increasing surface functionality can alter stability. A material designed for exceptional selectivity may become more difficult to regenerate. A material made from highly specialised components may perform brilliantly but be difficult to produce at scale. There is no universal definition of the perfect material. There are only materials that are more or less suitable for a particular purpose and set of conditions.</p><p>That thought may seem obvious, but it has consequences for how research questions are designed. If I eventually work on advanced porous materials for water purification or energy storage, I do not want to begin with the assumption that the material with the most impressive laboratory result is automatically the material worth pursuing. I would rather begin with the problem and work backwards.</p><p><em>What environment will the material encounter?</em></p><p><em>What property is actually needed?</em></p><p><em>Which chemical interactions produce that property?</em></p><p><em>How stable must the material be?</em></p><p><em>Can it be regenerated or reused?</em></p><p><em>What happens after its useful life?</em></p><p><em>Can it be synthesised consistently?</em></p><p><em>What resources are required?</em></p><p><em>And perhaps most importantly, can the material continue to perform when the conditions become less than ideal?</em></p><p>These questions do not make materials research less exciting. They make it more demanding. They also open another possibility for the research direction I am beginning to imagine. If advanced materials can be described in terms of their structures, compositions, surface properties, stability and performance, then perhaps computational methods and artificial intelligence could eventually help researchers identify materials that do not merely maximise one property, but balance several requirements at once.</p><p>Instead of asking an algorithm to find the material with the highest adsorption capacity, perhaps we could ask it to search for materials that combine adsorption performance with regeneration ability, chemical stability, low-cost components and suitability for realistic water conditions.</p><p>Likewise, instead of searching only for a battery material with maximum theoretical capacity, we could consider capacity alongside cycle life, safety, resource availability, manufacturing requirements and environmental impact.</p><p>That is a different way of thinking about materials discovery.</p><p>It moves the goal from finding the best property towards finding the best combination of properties for a real problem.</p><p>I am still a student, so I do not have the answer to whether such a research approach can eventually produce materials capable of addressing interconnected problems in water and energy. I also think it would be premature to claim that one material should solve everything. Materials science is more complicated than that. But I increasingly believe that the conditions surrounding a material should be considered part of its design. A material should not only be designed to perform well. It should be designed for the environment in which it will actually be used.</p><p>That principle may eventually become one of the most important ideas behind the research direction I want to explore. If the future of materials discovery is going to involve advanced porous structures, computational modelling, artificial intelligence and experimental chemistry, then perhaps the objective should not simply be to discover materials that look extraordinary on paper.</p><p>The objective should be to discover materials that remain useful when the paper ends and the real world begins.</p><h3>References</h3><ul><li>Bagotia, N., Sharma, A.K., Kumar, S. <em>et al.</em> (2024) <em>Regeneration strategies for exhausted adsorbents used in water treatment – A critical review</em>, Journal of Water Process Engineering, 62, 105560.</li><li>Cheng, Y. et al. (2026) <em>Artificial intelligence-driven approaches for materials design and discovery,</em> Nature Materials.</li><li>Qureshi, T.I. (2025) <em>Understanding the scaling up of future innovations with techno-economic analysis,</em> Nature Reviews Clean Technology, 1, p. 232.</li><li>Smit, B. and Garcia, S. (2026) <em>The data-only illusion in materials discovery,</em> Nature Materials, 25, pp. 1290–1292.</li><li>Xiao, J. et al. (2023) <em>From laboratory innovations to materials manufacturing for lithium-based batteries</em>, Nature Energy, 8, pp. 329–339.</li><li>Zeng, L. (2025) <em>Techno-economic analysis for lithium-ion battery manufacturing and recycling,</em> Nature Reviews Clean Technology, 1, p. 114.</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bc359169a879" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Beyond Zeolites: The Rise of Advanced Porous Materials]]></title>
            <link>https://medium.com/@selomanethato780/beyond-zeolites-the-rise-of-advanced-porous-materials-6829ca5c5e2e?source=rss-be6dbe0eba8d------2</link>
            <guid isPermaLink="false">https://medium.com/p/6829ca5c5e2e</guid>
            <category><![CDATA[chemistry]]></category>
            <category><![CDATA[researchers]]></category>
            <category><![CDATA[water]]></category>
            <category><![CDATA[zeolite]]></category>
            <category><![CDATA[writing]]></category>
            <dc:creator><![CDATA[Thato Selomane]]></dc:creator>
            <pubDate>Thu, 20 Aug 2026 11:21:39 GMT</pubDate>
            <atom:updated>2026-08-20T11:21:39.954Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Why are scientists developing new porous materials?</em></p><p>For many years, zeolites have been some of the most useful porous materials in science and industry. They have been applied in areas such as catalysis, adsorption, ion exchange and separation. However, science rarely stops at a material simply because it works. As new problems appear, scientists begin asking whether existing materials can do more. This is one of the reasons research has moved beyond traditional porous materials towards structures such as metal-organic frameworks (MOFs), covalent organic frameworks (COFs), porous organic polymers and other advanced materials.</p><p>The important point is that scientists are not developing new porous materials because zeolites have suddenly become useless. Instead, the limitations of existing materials become more obvious when they are placed in more demanding situations. Modern applications may require a material to recognise one molecule from a mixture of many others, capture a particular gas, remove a specific contaminant from water, or work repeatedly without losing its structure. These requirements have encouraged researchers to design materials whose properties can be controlled much more precisely.</p><h3>The problem with using one material for every purpose</h3><p>One of the biggest challenges in materials science is that different applications need different properties. A material that works well for separating small molecules may not be suitable for capturing larger molecules. Similarly, a material that performs well under dry laboratory conditions may behave differently when exposed to water, impurities, high temperatures or corrosive chemicals.</p><p>This is where traditional porous materials can become limiting. Their pore structures are not always easy to modify for a particular application. Scientists therefore want materials where the size, shape and chemical environment of the pores can be deliberately changed rather than simply accepting what the material naturally provides.</p><p>In simple terms, imagine having a storage room with shelves that cannot be moved. It may work perfectly for one type of object, but once the objects change, the space becomes less useful. Advanced porous materials are being developed more like adjustable storage systems. Researchers can modify their internal spaces so that they interact more effectively with particular molecules.</p><p>Song and Ma (2025) explain that the ability to control pore size, shape, distribution and surface functionality is one of the major reasons MOFs and COFs have attracted so much attention. This level of control allows scientists to design materials for specific tasks rather than relying mainly on naturally available structures.</p><h3>From simply having pores to designing the pores</h3><p>The word “porous” by itself does not tell us whether a material is actually good for a particular application. What matters is what those pores can do.</p><p>A porous material may have a large internal surface area, but if the target molecule cannot enter the pores, that surface area is not very useful. The chemical environment inside the pores also matters. Certain molecules may be attracted strongly to a surface while others pass through. Therefore, researchers are increasingly interested not only in creating pores, but in controlling what happens inside them.</p><p>MOFs are a good example. They are constructed from metal ions or metal clusters connected to organic molecules. By changing these building blocks, scientists can produce frameworks with different pore sizes, shapes and chemical properties. COFs take another approach because they are constructed from organic building blocks connected through covalent bonds. Their structures can also be designed to provide specific pore environments (Song and Ma, 2025).</p><p>This is an important change in the way scientists think about porous materials. Instead of asking, “How porous is this material?” researchers can ask a much more useful question: “Can I design the pores to perform the job I need?”</p><h3>Why selectivity matters</h3><p>One of the major reasons for developing advanced porous materials is selectivity.</p><p>In real life, substances rarely exist alone. Carbon dioxide, for example, may be mixed with nitrogen and other gases. Pollutants in water may exist together with salts, organic matter and other contaminants. A useful material therefore needs to distinguish between molecules instead of simply absorbing everything.</p><p>This is where highly tunable porous materials become interesting. By adjusting pore dimensions and chemical groups on the pore surfaces, scientists can make a material interact more strongly with a particular molecule. Research on MOFs and COFs has shown their potential for selective gas separation, catalysis and environmental remediation (Song and Ma, 2025).</p><p>This idea connects directly with the problems discussed in the previous article on water purification. Removing a contaminant from water is not always as simple as placing an adsorbent into the water and waiting. The material has to function in a complicated environment. Selectivity, regeneration and stability all become important. Advanced porous materials are being investigated partly because their structures can be modified to address these challenges rather than treating every contaminant in exactly the same way.</p><h3>Beyond adsorption: materials that can do more</h3><p>Another reason for developing new porous materials is that researchers want materials with several functions at once.</p><p>A porous material does not necessarily have to act only as a sponge. It can also provide a surface for chemical reactions, transport ions, detect molecules or assist in storing gases. MOFs and COFs, for example, have been investigated for catalysis, chemical sensing, gas storage and separation, environmental applications and energy-related technologies (Song and Ma, 2025).</p><p>This makes advanced porous materials particularly interesting because their pores can become reaction spaces rather than simply empty spaces.</p><p>For example, a catalyst can be designed so that molecules enter its pores, interact with active sites and undergo a chemical reaction. The product can then leave while the material remains available for another reaction. In this way, the internal architecture of the material becomes part of the chemistry itself. That is a major reason researchers continue looking beyond conventional porous materials. The aim is no longer just to store or trap something. The aim is to create materials that can recognise, separate, transform and sometimes transport molecules in a controlled way.</p><h3>But newer does not automatically mean better</h3><p>It would be easy to look at MOFs and COFs and assume that they have solved all the problems associated with porous materials. They have not.</p><p>One of the challenges is stability. Some MOFs can lose their structure when exposed to water, corrosive chemicals or other harsh conditions. Their performance in a controlled laboratory environment may therefore not always translate directly into an industrial setting. Researchers have been studying the chemical stability of MOFs because applications involving impure or corrosive gas streams can place considerable stress on these frameworks (Rieth <em>et al</em>., 2019).</p><p>There are also practical issues involving production, cost, mechanical strength and regeneration. A material may perform exceptionally well in a laboratory test but still be difficult to manufacture on a large scale. Recent research has also highlighted challenges in turning MOF and COF powders into mechanically robust macroscopic structures that can be used effectively in practical devices (Joss <em>et al</em>., 2025).</p><p>This is an important lesson in materials science: <em>high performance is only one part of the problem</em>. A material also has to survive real conditions, be produced reasonably, be reusable and ultimately make sense outside the laboratory.</p><h3><strong>The rise of materials with combined properties</strong></h3><p>Because individual materials often have strengths and weaknesses, scientists are also exploring hybrid materials.</p><p>MOFs and COFs can be combined so that the resulting material takes advantage of properties from both systems. Research into MOF–COF hybrids has grown because combining different porous frameworks can create materials with properties that may not be achievable from either material alone (Zhang <em>et al</em>., 2021).</p><p>This approach is interesting because it shows where materials science is heading. Scientists are not necessarily searching for one “perfect” material. Instead, they can combine different structures and functions to produce a material suited to a particular problem.</p><p>It is similar to engineering in general. If one tool cannot perform every task, we do not necessarily throw the tool away. We look for a better combination of tools. Materials scientists are applying a similar idea at the molecular level.</p><h3>So, why are scientists developing new porous materials?</h3><p>The answer is not simply because scientists want to discover something new. There are practical reasons behind the research.</p><p>The world is dealing with increasingly complicated problems involving clean water, carbon dioxide emissions, energy storage, chemical production and the separation of valuable substances from mixtures. These problems require materials that can perform specific tasks under difficult conditions.</p><p>Traditional porous materials remain valuable, but they cannot always provide the level of control that some modern applications demand. Advanced porous materials give scientists more freedom to manipulate pore size, shape, surface chemistry and functionality. This creates opportunities for materials that are more selective and better suited to particular applications.</p><p>At the same time, researchers have realised that designing a material in the laboratory is only the beginning. Stability, cost, regeneration, scalability and environmental impact have to be considered before a material can become genuinely useful.</p><p>For me, this is what makes the development of advanced porous materials interesting. It is not really a story about replacing old materials with new ones. It is a story about understanding the weaknesses of what we already have and then trying to build something that can deal with problems that the older materials were not designed for.</p><p>Zeolites opened an important chapter in porous materials science. The development of MOFs, COFs and other advanced structures represents another step in that journey. The direction is becoming clearer: scientists are moving from materials that simply possess pores towards materials whose pores can be deliberately designed for a purpose.</p><p>And perhaps that is the bigger idea behind materials science. Progress does not always mean throwing away what came before. Sometimes it means taking what already works, understanding where it falls short, and asking a better question about what could come next.</p><h3>References</h3><ul><li>Joss, L., Wuttke, S., Dincă, M. and Arjmand, M. (2025) Advancing from MOFs and COFs to functional macroscopic porous constructs, <em>Advanced Materials</em>, 37(52), Article 2411617.</li><li>Rieth, A.J., Wright, A.M. and Dincă, M. (2019) Kinetic stability of metal–organic frameworks for corrosive and coordinating gas capture, <em>Nature Reviews Materials</em>, 4, pp. 708–725.</li><li>Song, Y. and Ma, S. (2025) Pore engineering in metal–organic frameworks and covalent organic frameworks: strategies and application, <em>Chemical Science</em>, 16, pp. 11740–11767. https://doi.org/10.1039/D5SC01635E.</li><li>Zhang, Y., <em>et al</em>. (2021) Hybrid porous crystalline materials from metal-organic frameworks and covalent organic frameworks, <em>Advanced Materials</em>, 33.</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6829ca5c5e2e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Future of Water Purification Begins with Materials Chemistry]]></title>
            <link>https://medium.com/@selomanethato780/the-future-of-water-purification-begins-with-materials-chemistry-d3ee030318bc?source=rss-be6dbe0eba8d------2</link>
            <guid isPermaLink="false">https://medium.com/p/d3ee030318bc</guid>
            <category><![CDATA[industry]]></category>
            <category><![CDATA[future]]></category>
            <category><![CDATA[writing]]></category>
            <category><![CDATA[researchers]]></category>
            <dc:creator><![CDATA[Thato Selomane]]></dc:creator>
            <pubDate>Mon, 17 Aug 2026 21:10:25 GMT</pubDate>
            <atom:updated>2026-08-17T21:10:25.477Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Why do new materials matter for clean water?</em></p><p>Water is one of those things that is easy to take for granted until it becomes difficult to obtain. We turn on a tap, fill a glass and continue with our day without thinking much about what had to happen before that water became safe enough to use. Behind that simple moment is a much more complicated process involving treatment plants, filtration systems, chemicals, membranes, energy and continuous monitoring. As populations grow and industrial activity increases, the challenge is no longer simply finding water. Increasingly, the challenge is finding ways to make contaminated water safe, while doing so efficiently, affordably and without creating another environmental problem in the process.</p><p>The importance of this problem is difficult to overstate. The World Health Organization describes safe and readily available water as fundamental to public health, while the United Nations continues to identify water scarcity, pollution and unequal access as major challenges to sustainable development (WHO, 2024; UN-Water, 2024). Water quality is also becoming more complicated because contamination is not limited to the substances that conventional treatment systems were originally designed to remove. Industrial chemicals, pharmaceuticals, pesticides, heavy metals and other emerging contaminants can occur at relatively low concentrations while still creating concerns for ecosystems and human health. Recent reviews of wastewater treatment therefore increasingly focus on technologies capable of targeting contaminants that conventional approaches may struggle to remove effectively (Li <em>et al</em>., 2024).</p><p>This is where materials chemistry becomes interesting to me. When I first started learning about materials such as zeolites, I mainly thought about them as substances with useful properties. The deeper I look into materials science, however, the more I realise that the important question is not simply whether a material can remove something from water. The real question is <em>why it can remove it, how selectively it can do so, how long it can continue working, whether it can be regenerated, and what happens to the material afterwards</em>. That is where chemistry begins to determine the usefulness of a technology.</p><h3>The problem is not simply dirty water</h3><p>It is tempting to think of water purification as a simple process: identify the contaminant and remove it. In reality, water is chemically complicated. A real wastewater stream may contain several contaminants at the same time, dissolved salts, organic matter, microorganisms and particles that can interfere with treatment. A material that performs extremely well in a controlled laboratory experiment may behave differently when exposed to an actual wastewater sample.</p><p>This difference between laboratory performance and real-world performance is one of the issues that I think deserves much more attention. A material may show a high adsorption capacity under carefully controlled conditions, but that does not automatically mean it will work economically in a treatment plant. It may lose efficiency when competing ions are present. It may be difficult to regenerate. It may require expensive synthesis. It may produce secondary waste. It may even release components of its own into the treated water.</p><p>A 2024 review of wastewater treatment technologies made a similar point from a broader perspective: no single treatment technology can effectively address every contaminant, and combinations of treatment methods can be more promising than relying on one process alone (Al-Mamun <em>et al</em>., 2024). This suggests that the future of water treatment may not be about finding one perfect material. It may be about designing materials that perform specific functions and then integrating them intelligently into treatment systems.</p><h3>Why materials chemistry changes the problem</h3><p>Materials chemistry gives scientists a way of designing matter around a particular function. Instead of accepting a material exactly as it occurs naturally, researchers can modify its composition, surface chemistry, pore structure, charge and other properties.</p><p>That matters because contaminants do not all behave in the same way. Some molecules interact strongly with particular surfaces. Some ions can be attracted or repelled depending on electrical charge. Some contaminants are too large to pass through certain pores, while others are small enough to move through them. In membrane systems, even subtle differences in molecular size, charge and chemical interactions can influence whether a substance passes through or remains behind. This is why the chemistry of a material can be more important than simply calling it an advanced material.</p><p>For example, researchers are increasingly investigating membranes with specific chemical binding sites that can distinguish between ions with similar sizes and charges. Liu and colleagues (2024) describe how angstrom-scale confinement and carefully selected binding sites could improve ion-selective transport. Their discussion also points toward the use of machine learning to help identify chemical groups that could provide desirable ion-binding properties. That is a fascinating direction because it connects directly with the previous article in this series. AI can help search for the material. Chemistry explains why the material works.</p><h3>From simply filtering water to designing selectivity</h3><p>Traditional filtration can be thought of as separating things based largely on size. But the future of water purification may require much greater selectivity.</p><p>Imagine having a membrane that does not simply separate large from small, but can distinguish between chemically similar ions because of how they interact with specific groups inside the membrane. That is a much more sophisticated problem. Modern membrane research is moving toward this type of controlled separation. Pressure-driven membrane technologies such as reverse osmosis and nanofiltration are already important for desalination and water treatment, but researchers continue to work on improving membrane performance, energy efficiency, fouling resistance and selectivity (Liu <em>et al</em>., 2024).</p><p>The chemistry becomes particularly important when researchers begin modifying the membrane itself. Functional groups can be introduced to change interactions with particular contaminants. Pore sizes can be controlled. Different materials can be combined to produce structures that would not be possible using one material alone. In other words, the membrane is no longer just a barrier. It becomes a chemically engineered system.</p><h3>The rise of advanced porous materials</h3><p>This brings us back to the materials that first attracted my interest: porous materials. Zeolites are one example, but the field has expanded considerably. Researchers are also investigating metal-organic frameworks, covalent organic frameworks, graphene-based materials, MXenes, engineered polymers and different nanocomposites for water treatment. The reason is relatively straightforward. Porosity creates a large internal surface area and provides places where molecules and ions can interact with the material. But having pores is not enough. What matters is the size, shape, connectivity and chemical environment of those pores.</p><p>A 2024 review of metal-organic-framework-based membranes, for example, highlighted mechanisms including adsorption, electrostatic exclusion and size-based exclusion in the removal of emerging contaminants from water (Choi <em>et al</em>., 2024). Similarly, research into graphene oxide membranes has focused on how their structure can be engineered for desalination and purification applications (Tiwary <em>et al</em>., 2024).</p><p>The field is therefore moving from simply asking, What material can absorb this contaminant? toward a more sophisticated question:</p><p>Can we design a material whose structure and chemistry are deliberately matched to the contaminant we want to remove?That is a materials chemistry problem.</p><h3>The contaminants we are beginning to worry about</h3><p>One reason new materials are receiving so much attention is that the definition of water pollution continues to expand. Heavy metals and industrial chemicals have been studied for decades, but researchers are also increasingly concerned about pharmaceuticals, pesticides, personal-care chemicals, microplastics and persistent compounds such as PFAS.</p><p>These substances create difficult treatment problems because some occur at very low concentrations, while their chemical properties can make them difficult to remove using conventional processes. Reviews published in 2024 describe emerging contaminants as an increasing concern and discuss adsorption, membrane separation, advanced oxidation and biological approaches among the strategies being investigated (Li <em>et al</em>., 2024; Al-Mamun <em>et al</em>., 2024).</p><p>This is one area where materials chemistry could become extremely important.</p><p>A material could potentially be designed to recognise a particular chemical structure or interact strongly with a specific class of contaminant. Molecularly imprinted polymers, for instance, are being investigated because their structures can be designed around target molecules, potentially improving selective detection and removal of emerging contaminants (Li <em>et al</em>., 2024).</p><p>The concept is powerful: rather than treating every contaminant as if it were the same, we begin designing materials around the chemistry of individual pollutants.</p><h3>But better materials are not automatically better solutions</h3><p>There is something I think researchers need to be careful about. The word advanced can make a material sound better than it actually is. A material can remove contaminants extremely efficiently in a laboratory and still fail as a practical technology. It might require difficult synthesis conditions, expensive raw materials or large quantities of energy. It might not be reusable. It might perform poorly in real wastewater. And if the material itself becomes a waste product after treatment, we may simply be moving the problem from one place to another. This is why sustainability needs to become part of materials design from the beginning.</p><p>A successful water-treatment material should ideally be effective, stable, reusable, affordable and safe. Its production should also be considered. Researchers are increasingly discussing the need for scalable synthesis, recyclability, reduced secondary pollution and performance in complex water matrices when evaluating advanced materials (Alyasi <em>et al</em>., 2024).</p><p>For me, this changes the definition of a better material. Better does not necessarily mean the material removes the highest percentage of a contaminant in one experiment. Better could mean that it performs well for years, can be regenerated, requires less energy, uses abundant raw materials and can actually be manufactured at a reasonable cost.</p><h3>Where Africa fits into this research</h3><p>Water research is global, but the challenges are not identical everywhere.</p><p>Different regions experience different combinations of water scarcity, industrial pollution, infrastructure limitations, agricultural runoff and treatment capacity. This means that materials designed for one environment may not automatically be the best solution somewhere else. This creates an important opportunity for African researchers. Instead of simply importing technologies developed elsewhere, researchers can investigate materials and treatment systems according to the conditions found in African water systems. That could include locally available raw materials, region-specific contaminants, lower-energy treatment approaches and systems suitable for communities where large centralized treatment infrastructure is difficult to maintain.</p><p>I do not think the goal should be to create science that only benefits Africa. The better goal is to develop science from Africa that can contribute to the world. A material developed to treat a contaminant in a South African wastewater system could potentially have applications elsewhere. A low-cost synthesis method developed because resources are limited could become useful internationally. Research driven by local problems can sometimes produce solutions with much wider relevance. That is one of the reasons I am becoming increasingly interested in materials chemistry.</p><h3>The connection between water, energy and AI</h3><p>At this point, the subjects we have discussed in the previous articles begin to connect.</p><p>Article 6 looked at batteries and the need for better energy-storage materials. Article 7 looked at artificial intelligence and its potential to accelerate materials discovery. Now water purification introduces another major application.</p><p>These areas may initially appear unrelated, but they share a common foundation: materials must be designed to perform specific chemical functions.</p><p>A membrane for water purification needs selective transport.</p><p>A battery material needs controlled ion movement and electrochemical stability.</p><p>A catalyst needs to accelerate a reaction without being consumed.</p><p>A porous adsorbent needs to interact selectively with molecules or ions.</p><p>AI can potentially help researchers search through possible material structures, while chemistry provides the understanding needed to determine whether those structures make sense. This is where I see the future becoming particularly interesting. Perhaps the most valuable materials of the future will not be designed for only one application. Perhaps researchers will discover materials whose properties can be adapted for water treatment, energy technologies, catalysis or environmental applications. That possibility is still an open research question.</p><h3>What I think the future should focus on</h3><p>The future of water purification should not simply be a competition to discover the material with the highest removal efficiency. It should be a search for materials that work outside the laboratory.</p><p>We need materials that can operate in complicated water environments. We need membranes that resist fouling. We need adsorbents that can be regenerated. We need selective materials that target contaminants without creating additional environmental risks. We need manufacturing methods that are affordable and scalable. And we need to think about the entire life cycle of the material.</p><p>Where do the raw materials come from? How much energy is required to make it? Can it be reused? What happens when it reaches the end of its useful life? Can the material be produced at a scale large enough to matter?</p><p>Those questions may not sound as exciting as announcing a new material with record-breaking laboratory performance, but they are the questions that determine whether a scientific discovery becomes a useful technology.</p><h3>Perhaps the question worth asking are:</h3><p>This is where I believe the most interesting work is still ahead.</p><p>Can we design porous materials that selectively remove difficult emerging contaminants while remaining inexpensive and reusable? Can zeolite-based or other porous systems be modified to target specific pollutants rather than simply relying on broad adsorption? Can artificial intelligence help researchers predict which pore structures and surface functional groups will interact most effectively with particular contaminants? Can water-treatment materials be designed using abundant and locally available resources without sacrificing performance? Can one material platform be adapted for both water purification and energy-related applications?</p><p><em>And perhaps the biggest question:</em></p><p>Can materials chemistry help us move from treating polluted water after contamination has occurred to designing systems that prevent, capture and recover pollutants more intelligently? I do not think we have all the answers yet. That is precisely why the field interests me.</p><h3>Conclusion</h3><p>The future of clean water will not be solved by chemistry alone. Infrastructure, policy, economics, engineering and responsible water management will remain essential. But materials chemistry can change what is technically possible. The materials we use determine what we can separate, what we can capture and what we can recover. That is why I believe the future of water purification begins, at least partly, at the molecular level. We are moving toward a period where scientists are not simply searching for materials that happen to work. We are learning to design materials around the problem itself. The challenge now is to make those materials practical. A material that works beautifully inside a laboratory flask is only the beginning. The real achievement is creating something that can leave the laboratory, enter the real world, operate reliably, remain affordable and continue working without creating another problem behind it. That is where materials chemistry becomes more than an academic subject. It becomes a way of asking whether we can design matter differently in order to live differently. And perhaps that is one of the most important questions chemistry can help us answer.</p><h3>References</h3><ul><li>Alyasi, H., Wahib, S., Gomez, T.A. <em>et al</em>. (2024) The power of MXene-based materials for emerging contaminant removal from water – A review, <em>Desalination,</em> 586, 117913.</li><li>Al-Mamun, A., <em>et al</em>. (2024) Trends in effective removal of emerging contaminants from wastewater: A comprehensive review, <em>Desalination and Water Treatment</em>, 317, 100258.</li><li>Choi, J.S., Jung, B., Kim, H.-H. <em>et al</em>. (2024) A review of metal–organic framework-based membranes for the removal of emerging contaminants from water<em>,</em> <em>Journal of Water Process Engineering</em>, 68, 106456.</li><li>Li, F., Yue, S., Zhao, Z., Liu, K., Wang, P. and Zhan, S. (2024) Application of molecularly imprinted polymers in the water environmental field: A review on the detection and efficient removal of emerging contaminants<em>,</em> <em>Materials Today Sustainability</em>, 27, 100904.</li><li>Liu, W., Livingston, J.L., Wang, L. <em>et al</em>. (2024) Pressure-driven membrane desalination, <em>Nature Reviews Methods Primers.</em></li><li>Liu, Y., et al. (2024) Designing membranes with specific binding sites for selective ion separations, <em>Nature Water</em>, 2, pp. 706–718.</li><li>Tiwary, S.K., Singh, M., Chavan, S.V. <em>et al</em>. (2024) Graphene oxide-based membranes for water desalination and purification, <em>npj 2D Materials and Applications</em>, 8, 27.</li><li>UN-Water (2024) UN World Water Development Report 2024: Water for Prosperity and Peace. <em>United Nations.</em></li><li>WHO (2024) Guidelines for drinking-water quality: small water supplies. <em>Geneva: World Health Organization</em>.</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d3ee030318bc" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Artificial Intelligence Is Changing How We Discover New Materials]]></title>
            <link>https://medium.com/@selomanethato780/artificial-intelligence-is-changing-how-we-discover-new-materials-c2a7c94eca0c?source=rss-be6dbe0eba8d------2</link>
            <guid isPermaLink="false">https://medium.com/p/c2a7c94eca0c</guid>
            <dc:creator><![CDATA[Thato Selomane]]></dc:creator>
            <pubDate>Thu, 13 Aug 2026 10:54:32 GMT</pubDate>
            <atom:updated>2026-08-13T10:54:32.423Z</atom:updated>
            <content:encoded><![CDATA[<h3>Purpose</h3><p>To examine how artificial intelligence and machine learning are changing materials discovery, explain how these technologies complement chemistry and experimental science, and consider whether AI could help researchers develop better materials for batteries, water treatment, energy conversion, and other technologies.</p><h3>Introduction</h3><p>For much of scientific history, discovering a new material has depended on a familiar process: a scientist identifies a promising chemical composition, prepares the material, measures its properties, modifies the formulation and repeats the experiment. Sometimes this process produces a breakthrough. More often, it produces another experiment that does not work. Materials science has therefore always involved a difficult combination of knowledge, intuition, computation and trial and error. The problem is not that scientists lack ideas. It is that the number of possible combinations of elements, structures and processing conditions is extraordinarily large. Artificial intelligence is beginning to change that situation.</p><p>Machine-learning systems can examine large collections of existing materials data, learn relationships between composition, structure and properties, and identify candidates that researchers may wish to investigate experimentally. Modern approaches are moving beyond simply predicting properties: researchers are increasingly exploring systems capable of proposing entirely new materials and helping determine how they might be synthesized.</p><p>This does not mean that chemistry is becoming unnecessary. Quite the opposite. The emergence of AI in materials science may make chemical knowledge more important, because predictions still have to be interpreted, tested and converted into real substances. The important question is therefore not whether artificial intelligence will replace materials scientists.</p><p>The more interesting question is:</p><ul><li>Can artificial intelligence help scientists search the enormous chemical space of possible materials more intelligently?</li></ul><h3>The Problem: There Are Too Many Materials to Test</h3><p>Imagine trying to find a material with a very specific combination of properties. Suppose a researcher wants a material that is:</p><ol><li>inexpensive</li><li>chemically stable</li><li>electrically conductive</li><li>environmentally acceptable</li><li>resistant to high temperatures</li><li>easy to manufacture</li><li>suitable for use in an energy-storage device.</li></ol><p>Finding such a material is not simply a matter of choosing one element from the periodic table.</p><p>The final properties of a material depend on composition, crystal structure, bonding, defects, surfaces, processing conditions and interactions between its components. The possible combinations become enormous. Traditional experimental research remains indispensable, but testing every possible candidate physically would be impossible. Computational chemistry and materials modelling have already helped researchers narrow this search, but even sophisticated calculations can become expensive when the number and complexity of candidate materials increase. Recent reviews emphasize that machine learning can help accelerate the prediction of material stability and properties where conventional calculations or experiments would be too costly to perform across the entire search space. This is where AI becomes particularly interesting.</p><p>Instead of asking researchers to examine every possibility, machine-learning models can be trained using existing data and then used to prioritize promising candidates. The laboratory still matters. AI simply changes where scientists look first.</p><h3>What Does AI Actually Do in Materials Science?</h3><p>The phrase artificial intelligence can sound almost magical when used in scientific discussions. In reality, many materials applications involve more specific techniques collectively described as machine learning.</p><p>A model may be trained using information about known materials and their measured or calculated properties. Once trained, it can attempt to identify relationships that allow researchers to estimate the properties of previously unexplored candidates. For example, a model might learn relationships between chemical composition and:</p><ul><li>electronic properties</li><li>mechanical strength</li><li>thermal stability</li><li>ionic conductivity</li><li>magnetic behaviour</li><li>formation energy.</li></ul><p>This creates a different research workflow. Instead of:</p><p><em>Design → Synthesize → Test → Fail → Modify → Repeat</em></p><p>researchers can increasingly work toward:</p><p><em>Data → Prediction → Candidate selection → Synthesis → Characterization → New data</em></p><p>The second process does not eliminate experimentation. It makes experimentation more targeted.</p><p>This distinction is important because a computer prediction is not the same thing as a successful material. A predicted structure may be unstable under real conditions. A material that is theoretically stable may be extremely difficult to synthesize. A substance with impressive laboratory properties may be too expensive or toxic to manufacture at scale. AI can narrow the search. Chemistry must determine whether the answer survives contact with reality.</p><h3>From Prediction to Discovery</h3><p>One of the most significant developments in this field is the movement from property prediction toward materials generation. Earlier computational approaches often asked questions such as:</p><p>&quot;Given this material, what properties should we expect?&quot;Newer approaches increasingly ask:</p><ul><li><em>“Given the properties we want, what materials might produce them</em>?”This is sometimes described as inverse design.</li></ul><p>A conventional approach starts with a material and predicts its behaviour. Inverse design starts with a desired behaviour and searches backward for a material capable of producing it.</p><p>A 2026 review in Nature Materials describes this development as a broader transition from high-throughput screening toward AI-driven inverse design, including approaches based on reinforcement learning and generative models. This could become important for fields such as energy storage.</p><p>Instead of searching through existing battery materials alone, researchers could specify desired characteristics such as high ionic conductivity, chemical stability and low cost and use computational methods to identify potential candidates. The result is not a finished battery. It is a hypothesis. And hypotheses still need experiments.</p><h3>The GNoME Example</h3><p>One of the most striking demonstrations of AI-assisted materials discovery came from the development of Graph Networks for Materials Exploration, known as GNoME. In research published in Nature in 2023, Merchant and colleagues reported that their deep-learning system identified approximately 2.2 million crystal structures that were computationally stable relative to the reference data used in the study. More than 380,000 were identified as particularly promising stable candidates.</p><p>The scale is difficult to imagine. Researchers have traditionally built knowledge about new materials through experimental work and computational calculations. GNoME demonstrated how machine learning could dramatically expand the number of candidate structures that can be examined computationally.</p><p>The system used graph neural networks to represent relationships between atoms and crystal structures. Its predictions were repeatedly checked using density functional theory calculations, and the resulting information was fed back into subsequent rounds of training. This iterative process is an example of active learning.</p><p>Some of the predicted materials were also independently synthesized by external research groups. According to Google DeepMind, 736 of the predicted structures had been independently realized experimentally at the time of its report. The significance is not that a machine suddenly &quot;invented chemistry.&quot;The significance is that AI helped researchers explore a chemical space at a scale that would have been extremely difficult to search using conventional approaches alone.</p><h3>What Does This Mean for Batteries?</h3><p>This brings us back to Article 6. Better batteries require better materials. Researchers need electrodes that can store and release ions efficiently, electrolytes that permit ion transport while remaining stable, interfaces that resist degradation, and materials that can survive see of charging cycles. These requirements create a complicated materials-design problem.</p><p>AI could assist by screening candidate materials for characteristics relevant to battery performance. The GNoME research, for example, identified hundreds of predicted lithium-ion conductors among its discoveries. The researchers suggested that such materials could be relevant to future rechargeable battery technologies. But this should not be misunderstood. A computationally predicted material is not automatically a superior battery material.</p><p>It still needs to be:</p><p>1. synthesized<br>2. structurally characterized<br>3. electrochemically tested<br>4. evaluated for stability<br>5. tested repeatedly<br>6. assessed for safety<br>7. considered from an economic and environmental perspective.</p><p>The distance between &quot;predicted&quot; and &quot;commercially useful&quot; is one of the most important issues in AI-assisted materials science.</p><h3>AI and Porous Materials</h3><p>This is where the subject becomes particularly interesting for researchers working with materials such as zeolites, metal-organic frameworks and other porous structures. Porous materials contain networks of cavities, channels or pores that can interact selectively with molecules and ions.</p><p>Their properties can make them useful for applications including:</p><ul><li>adsorption</li><li>separation</li><li>catalysis</li><li>gas storage</li><li>water treatment</li><li>potentially energy-related technologies.</li></ul><p>The enormous number of possible structures creates an obvious opportunity for computational discovery. AI can potentially help researchers identify relationships between a porous material’s structure and properties such as adsorption capacity, selectivity or stability.</p><p>Research has already explored machine learning for the accelerated discovery of metal-organic frameworks and other porous materials. The Materials Project and related computational databases also provide increasingly important resources for materials modelling and machine-learning applications. This does not mean that every zeolite or MOF predicted by a computer will be useful. It means that researchers can begin asking a different question:</p><ul><li>Instead of discovering useful materials one experiment at a time, can we build computational systems that help us decide which experiments are worth performing? That is a very different philosophy of research.</li></ul><h3>From the Computer to the Laboratory</h3><p>There is another problem. Even if AI can predict thousands of promising materials, scientists still have to make them. This creates what might be called the prediction-to-synthesis gap. A material can look excellent on a computer and fail in the laboratory.</p><p>Its predicted structure may be difficult to obtain. The required reaction conditions may be unrealistic. Impurities may appear. The material may transform into another phase. Or the theoretical advantage may disappear when the material is incorporated into a real device. Researchers are therefore developing systems that connect computation directly with automated experimentation.</p><p>A notable example is the A-Lab, an autonomous laboratory described in Nature. The system combined computational predictions, machine learning, literature-derived information, robotics and active learning to plan and perform inorganic-material synthesis. Over 17 days of continuous operation, it reported realizing 36 compounds from 57 targeted materials. This represents an important change in scientific workflow. The laboratory is no longer simply a place where humans execute predetermined experiments.</p><p>It can become part of a feedback system:</p><p><em>AI prediction → experiment → measurement → learning → improved prediction → new experiment</em></p><p>The cycle continues.</p><h3>But AI Has a Serious Weakness: Data</h3><p>It is tempting to assume that more data automatically means better scientific predictions. Materials science is not that simple. High-quality experimental data can be difficult to obtain. Different laboratories may use different instruments, measurement conditions and protocols. Some datasets are small, incomplete or biased toward materials that have already received considerable scientific attention. A machine-learning model can therefore produce confident predictions without genuinely understanding the chemistry behind them.</p><p>Recent work has emphasized the small-data problem in materials machine learning, while a 2026 Nature Materials commentary argued that materials discovery cannot rely on data alone because synthesis complexity and limited data require deep chemical knowledge to be integrated into AI systems.</p><p>This is one of the reasons I would be cautious about describing AI as a replacement for scientists. AI is powerful at recognizing patterns. Chemists understand that a pattern may have a physical reason or may simply be an artifact of the data. That distinction matters.</p><h3>The Scientist of the Future May Need Two Languages</h3><p>The emergence of AI may therefore change what it means to be a materials scientist. Future researchers may need to understand two different languages.</p><p>The first is the language of chemistry:</p><p>atoms, bonds, structures, reactions, thermodynamics, kinetics and mechanisms.</p><p>The second is the language of data:</p><p>algorithms, models, databases, uncertainty, statistics and computation.</p><p>Neither is sufficient on its own. An AI researcher who does not understand the chemistry may generate predictions that are difficult to interpret or synthesize. A chemist who ignores computational methods may overlook opportunities to explore enormous datasets and chemical spaces. The most productive research may occur where the two meet. This is why interdisciplinary materials science is becoming increasingly important.</p><h3>What This Could Mean for Global Energy and Water</h3><p>The potential applications extend beyond batteries. If AI can accelerate the discovery of materials with specific chemical and physical properties, the same general approach could contribute to technologies involving:</p><ul><li>Energy</li><li>batteries</li><li>fuel cells</li><li>solar materials</li><li>hydrogen technologies<br>catalysts.</li><li>Water</li><li>membranes</li><li>adsorbents</li><li>ion-selective materials</li><li>catalysts for water treatment.</li><li>Environment</li><li>carbon capture</li><li>pollutant removal</li><li>resource recovery</li><li>sustainable chemical processes.</li><li>Industry</li><li>catalysts</li><li>lightweight structural</li><li>materials</li><li>corrosion-resistant materials</li><li>advanced coatings.</li></ul><p>The Materials Project explicitly identifies applications ranging from batteries and solar energy to water splitting and catalysis among the areas that computational materials discovery can support. The underlying idea is remarkably simple:</p><ul><li>If we can search for materials more intelligently, we may be able to solve technological problems more intelligently.</li></ul><h3>What AI Cannot Do Alone</h3><p>There is a danger in making AI sound more powerful than it actually is. AI does not remove the laws of chemistry. It cannot make an unstable compound stable simply because a model predicts that it should exist. It cannot automatically determine whether a laboratory synthesis is economically viable. It cannot decide whether mining the required elements is environmentally responsible. And it cannot replace experimental evidence. The strongest future systems will therefore probably be hybrid systems.</p><p>They will combine:</p><p><strong>Artificial intelligence + chemical knowledge + physics + computation + experimentation + human judgment.</strong></p><p>Recent research on physics-informed AI reflects this direction, arguing that combining physical laws with machine learning can help address the limited data available for complex materials problems. This is perhaps a more realistic vision of AI in science than the idea of a machine independently discovering everything.</p><h3>The Research Oppor</h3><p>For someone entering chemistry today, this field presents an unusual opportunity. A student does not necessarily need to choose between chemistry and computational science. There is an expanding space between them.</p><p>A chemist could learn enough programming and machine learning to work with materials databases. A computational researcher could develop a deeper understanding of chemical synthesis. An experimental scientist could use AI to prioritize which materials should be synthesized. And a materials researcher could eventually combine all three.</p><p>This is particularly relevant to emerging research environments where resources must be used carefully. If a computational model can reduce the number of unsuccessful experiments, it could potentially help laboratories direct limited time, chemicals and equipment toward the most promising candidates. That does not make experimentation less important. It makes each experiment potentially more informed.</p><h3>Research Questions Worth Asking</h3><p>The most interesting part of this field may not be what AI can already accomplish. It is what remains unresolved.</p><ul><li>Can AI predict materials that have never been imagined by human researchers?</li><li>Can machine-learning models reliably identify materials that are not only theoretically stable but also realistically synthesizable?</li><li>Can AI help design safer and more sustainable battery materials?</li><li>Can machine learning accelerate the discovery of porous materials for water purification?</li><li>How can small experimental datasets be used without producing misleading predictions?</li><li>Can AI integrate chemical knowledge, physical laws and experimental evidence into one research workflow?</li><li>Could autonomous laboratories eventually perform much of the repetitive experimental screening while scientists concentrate on higher-level scientific questions? These are not merely technological questions. They are questions about how science itself may be conducted in the future.</li></ul><h3>Conclusion: The Machine Is Not the Scientist</h3><p>The arrival of artificial intelligence in materials science does not mark the end of chemistry. It may mark the beginning of a different way of doing chemistry.</p><p>For centuries, scientists have explored matter through observation, experimentation and theory. Computational science expanded that capability by allowing researchers to simulate systems that were difficult to study directly. Machine learning now adds another layer: the ability to learn from enormous collections of scientific information and use those patterns to guide the search for new possibilities. The transformation is already visible.</p><p>AI systems have generated millions of candidate crystal structures. Autonomous laboratories have begun connecting prediction with physical synthesis. Materials databases are becoming increasingly important scientific infrastructure. But the central role of the scientist remains. A predicted material must still be understood. A chemical structure must still be synthesized. A promising property must still survive experimental testing. And a technological breakthrough must ultimately prove that it can be produced safely, sustainably and economically.</p><p>Perhaps, then, the future of materials discovery will not belong to humans or machines alone. It will belong to the partnership between them. The chemist brings understanding of matter. The computer brings scale. The laboratory provides evidence. And science decides whether the prediction survives reality. That may be the real revolution not replacing the scientist, but giving the scientist a much larger map of what is possible.</p><h3>References</h3><ul><li>Butler, K.T., Davies, D.W., Cartwright, H., Isayev, O. and Walsh, A. (2018) <em>Machine learning for molecular and materials science</em>, <em>Nature</em>, 559, pp. 547–555.</li><li>Cheng, M., Fu, C.-L., Okabe, R. <em>et al</em>. (2026) <em>Artificial intelligence-driven approaches for materials design and discovery,</em> <em>Nature Materials</em>, 25, pp. 174–190.</li><li>Merchant, A., Batzner, S., Schoenholz, S.S. <em>et al</em>. (2023) <em>Scaling deep learning for materials discovery</em>, <em>Nature</em>, 624, pp. 80–85.</li><li>Xu, P., Ji, X., Li, M. <em>et al</em>. (2023) <em>Small data machine learning in materials science</em>, <em>npj Computational Materials</em>, 9, 42.</li><li>Hippalgaonkar, K., Li, Q., Wang, X. <em>et al</em>. (2023) <em>Knowledge-integrated machine learning for materials: lessons from gameplaying and robotics</em>, Nature Reviews Materials, 8, pp. 241–260.</li><li>Raabe, D., Rezaei Mianroodi, J. and Neugebauer, J. (2023) <em>Accelerating the design of compositionally complex materials via physics-informed artificial intelligence</em>, Nature Computational Science, 3, pp. 198–209.</li><li>Gregoire, J.M., Zhou, L. and Haber, J.A. (2023) <em>Combinatorial synthesis for AI-driven materials discovery</em>, Nature Synthesis, 2, pp. 493–504.</li><li>Szymanski, N.J. <em>et al</em>. (2023) <em>An autonomous laboratory for the accelerated synthesis of inorganic materials</em>, Nature.</li><li>Smit, B. and Garcia, S. (2026) <em>The data-only illusion in materials discovery</em>, Nature Materials, 25, pp. 1290–1292.</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c2a7c94eca0c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why Better Batteries Matter More Than Ever: The Science Behind the Global Energy Revolution]]></title>
            <link>https://medium.com/@selomanethato780/why-better-batteries-matter-more-than-ever-the-science-behind-the-global-energy-revolution-627616287fc1?source=rss-be6dbe0eba8d------2</link>
            <guid isPermaLink="false">https://medium.com/p/627616287fc1</guid>
            <category><![CDATA[researchers]]></category>
            <category><![CDATA[chemistry]]></category>
            <category><![CDATA[battery]]></category>
            <category><![CDATA[future]]></category>
            <dc:creator><![CDATA[Thato Selomane]]></dc:creator>
            <pubDate>Wed, 05 Aug 2026 18:01:20 GMT</pubDate>
            <atom:updated>2026-08-05T18:01:20.582Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Purpose:</strong> To explain why batteries have become one of the most important scientific challenges of the twenty-first century while introducing the critical role of chemistry and materials science in developing safer, more efficient, and more sustainable energy storage technologies.</p><p>Rechargeable batteries have become a <strong>cornerstone of modern technology and the transition to clean energy.</strong> They power billions of personal devices, enable electric vehicles (EVs), and stabilize power grids with intermittent renewables. Thanks to pioneering work by researchers like M. Stanley Whittingham, John B. Goodenough, and Akira Yoshino – awarded the 2019 Nobel Prize in Chemistry – today’s lithium-ion batteries deliver far higher energy density than early cells. Yet lithium-ion technology is reaching its practical limits. Fundamental constraints in electrode materials, charging speed, lifespan, and reliance on scarce elements (Li, Co, Ni) mean that “<strong>better batteries</strong>” are an urgent global scientific challenge. This report explains <strong>why battery science matters now more than ever</strong>: the history of battery innovation, the basic chemistry behind how batteries work, current limitations of lithium-ion systems, and the most promising future technologies (sodium-ion, solid-state, lithium–sulfur, lithium–air, and others). It highlights the role of advanced materials – including porous frameworks – and artificial intelligence in accelerating discovery. Key research questions are identified to guide the next decade of battery research. In short, advancing battery materials is central to achieving sustainable, abundant energy worldwide, and research in this area will continue to be highly funded and of enormous global interest.</p><h3><strong>Introduction: Why Batteries Matter Today</strong></h3><p>Nearly every electronic device or clean-energy system we use depends on batteries. Uninterruptible power for homes and hospitals, <strong>renewable energy</strong> from solar and wind, <strong>electric transportation</strong>, portable electronics, and even critical medical implants all rely on efficient energy storage. As the IEA notes, batteries are now the <strong>fastest-growing energy technology in the world</strong>: in 2023 alone, <strong>42 GW</strong> of new battery storage was installed globally. Lithium-ion batteries – first commercialized in 1991 – now power roughly 45 million EVs and over 85 GW of grid storage, and they dominate battery applications thanks to steep cost reductions (≈90% drop since 2010) and improvements in energy density. In 2023 the global stock of EVs reached 45 million, displacing billions of barrels of oil per day under net-zero scenarios. Yet meeting Paris-Agreement goals means tripling renewable capacity and <strong>sextupling battery deployment by 2030</strong>. Put simply: <strong>the clean-energy revolution depends on better batteries</strong>.</p><p>Batteries also matter for <strong>Africa and developing regions</strong>. One-fifth of the African population now uses solar home systems or mini-grids with battery storage. Affordable, safe batteries could accelerate rural electrification, power industry without fossil fuels, and enable mobile telecommunications. At the same time, Africa has abundant renewable resources (sun, wind, hydro) and emerging technology talent. African scientists and engineers can help develop the next generation of battery materials and systems, for the benefit of both the continent and the global community.</p><h3>A Brief History of Battery Innovation</h3><p><strong>Batteries have a long history</strong>: Alessandro Volta’s voltaic pile (1800) was the first true battery, and Gaston Planté’s lead–acid battery (1859) was the first rechargeable system used at scale. For most of the 20th century, nickel-cadmium and lead-acid batteries filled niche roles (portable power tools, car starters). A paradigm shift came in the 1970s–80s when researchers began exploring <strong>lithium-based chemistries</strong>. Whittingham first demonstrated a working lithium-ion cell (using TiS₂ cathode) in 1976, but it was John Goodenough’s 1980 discovery of a cobalt-oxide cathode (LiCoO₂) that <strong>dramatically increased energy density</strong>. Akira Yoshino then paired this with a safe carbon (graphite) anode to create the first fully rechargeable lithium-ion battery in 1985, commercialized by Sony in 1991. These advances <strong>revolutionized portable electronics</strong> – the Nobel Prize cited Goodenough’s cathode as “a decisive step towards the wireless revolution”. Today’s lithium-ion chemistry (with variations like LiFePO₄, NMC (LiNiMnCoO₂), and others) still stems from those breakthroughs.</p><p><strong>Figure 1</strong> (below) highlights major milestones. Early batteries (Voltaic pile, lead–acid) gave way to nickel-metal hydride and then lithium-ion. In 2019, Goodenough, Whittingham, and Yoshino shared the Nobel Prize for lithium-ion innovations. Since then, research has surged on “<strong>post-lithium-ion</strong>” systems (sodium-ion, solid electrolytes, metal–sulfur cells, etc.), marking a new era of battery discovery.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*oTa32Ux38IDeql3xvlO_yw.jpeg" /></figure><h3><strong>How Batteries Work (Chemistry Basics)</strong></h3><p>A battery cell consists of two electrodes (anode and cathode) separated by an electrolyte. During discharge, ions (e.g. Li⁺ in lithium-ion batteries) shuttle through the electrolyte while electrons flow through the external circuit, powering devices. In a typical Li-ion cell, the <strong>cathode</strong> is a lithium-containing metal oxide (e.g. LiCoO₂, NMC, LiFePO₄) and the <strong>anode </strong>is usually graphite (carbon) or silicon. When you charge the battery, lithium ions move from cathode to anode (intercalating into the graphite). When discharging, they move back to the cathode, releasing energy. The <strong>electrolyte</strong> (often a lithium salt in organic solvent) provides the ionic pathway.</p><p>The performance of a battery depends critically on its materials: the crystal structure of the cathode and anode, the composition of the electrolyte, and how these materials change over many charge cycles. For example, Goodenough’s LiCoO₂ cathode enabled more lithium to be stored and released without degrading the crystal structure. Yoshino’s innovation was the petroleum-coke anode that safely hosted lithium ions. These foundational components defined the lithium-ion cell, and millions of researchers since have been tweaking chemistries and microstructures to squeeze out performance.</p><h3><strong>Why Current Batteries Are Not Enough</strong></h3><p>Even with these successes, today’s lithium-ion batteries face <strong>serious limitations</strong>:</p><ul><li><strong>Energy Density</strong>: Consumer expectations and EV range requirements keep rising. Lithium-ion cells now reach ~250 Wh/kg at the cell level, but further large gains are hard to achieve. Fundamental limits of intercalation chemistry mean new electrode materials or architectures (e.g. Li metal anodes) are needed for significantly higher density.</li><li><strong>Charging Speed vs. Lifespan</strong>: Fast charging is in demand, but quick charge/discharge accelerates battery degradation and generates heat. Optimizing charging algorithms and materials is an ongoing challenge.</li><li><strong>Cycle Life</strong>: Each charge/discharge cycle causes microscopic changes (particle cracking, side reactions) that gradually reduce capacity. Extending lifetime (especially under harsh conditions) remains an active research area.</li><li><strong>Safety</strong>: Most Li-ion batteries use flammable liquid electrolytes. Thermal runaway and fires are well-known hazards. Improving safety (through solid electrolytes, flame-retardant additives, or intrinsically safer chemistries) is critical.</li><li><strong>Materials Criticality</strong>: Lithium, cobalt, nickel, and graphite – key Li-ion components – are finite and often geographically concentrated. For example, cobalt mining raises ethical concerns. As Chacana-Olivares <strong>et al</strong>. note, these materials “<strong>are not only finite but also heavily concentrated in a limited number of countries</strong>”, creating supply risks. High demand from EVs is straining supply; China dominates processing of lithium and cobalt.</li><li><strong>Recycling and Sustainability</strong>: Ideally, battery materials should be recovered at end-of-life. However, current recycling processes are energy-intensive and not yet widespread. The IEA warns that insufficient recycling capacity could become a bottleneck. Chacana-Olivares <strong>et al</strong>. project <strong>14 million EV batteries</strong> retiring per year by 2040, underscoring the urgency of developing efficient recycling technologies.</li></ul><p>In short, although lithium-ion batteries have propelled the clean-energy transition, <strong>they may become the limiting factor </strong>if these constraints are not addressed. Recognizing this, governments and industry are heavily funding R&amp;D in new battery chemistries and materials (for example, China tripled its battery storage investments in 2023). The stakes are high: to meet climate goals, battery deployment needs to <strong>increase sevenfold by 2030</strong>.</p><h3><strong>The Next Frontier: Beyond Lithium-Ion</strong></h3><p>Researchers worldwide are exploring <strong>post-lithium-ion batteries</strong>. Some of the most promising directions include:</p><ul><li><strong>Sodium-Ion Batteries</strong>: These use abundant sodium instead of lithium. Because Na is far more plentiful and cheaper than Li, Na-ion promises lower cost and more sustainable supply. In the past few years, labs have improved sodium-ion cathodes (layered oxides, polyanion phosphates, Prussian-blue analogues) and anodes (hard carbon). Companies like CATL have begun mass-production of Na-ion cells for low-range EVs in 2026 (targeting ~1 GWh). <strong>Trade-off</strong>: Na-ion currently lags Li-ion in energy density (roughly 60–80% of Li-ion), due to the heavier Na ion and fewer high-voltage cathodes available. Safety and cycle life are promising (Na salts can be fluorine-free for moisture tolerance), and by 2030 pack costs may compete with low-cost Li-ion variants. Overall, sodium-ion is moving from laboratory to pilot manufacturing quickly.</li><li><strong>Solid-State Batteries (All-Solid Batteries)</strong>: Replacing the liquid electrolyte with a solid ceramic, polymer, or glass electrolyte can dramatically improve safety (no flammable solvent) and enable use of lithium-metal anodes (higher capacity). Solid-state cells <strong>could</strong> achieve energy densities well above current Li-ion. Recently, composite solid electrolytes using porous fillers (e.g. 3D zeolite networks) have shown enhanced Li⁺ conduction, and industry players (Toyota, QuantumScape, Solid Power) are racing to commercialize such cells for EVs. Key challenges are brittle interfaces and cost of solid electrolytes, but AI-driven materials search (see next section) is rapidly advancing candidate identification.</li><li><strong>Lithium–Sulfur Batteries</strong>: Li–S batteries use lithium metal anodes and sulfur cathodes. Sulfur is extremely abundant and has a theoretical capacity ten times higher than conventional cathodes. In principle, Li–S cells could reach ~400–500 Wh/kg (versus ~250 Wh/kg for Li-ion). However, they face major hurdles: soluble lithium polysulfide intermediates cause “shuttle” (loss of active material) and lithium metal can form dangerous dendrites. Solid-state versions (with sulfide or oxide electrolytes) are being studied to address these issues. Recent reviews note that while Li–S promises “<strong>significantly enhanced energy density and improved safety</strong>”, achieving high power, fast charging, and long cycle life is an open challenge.</li><li><strong>Lithium–Air (Metal–Air) Batteries</strong>: These have the highest theoretical energy density (oxygen from air is the cathode), potentially rivaling gasoline. They remain largely in the research phase. Key obstacles include controlling reactions with O₂ (moisture and CO₂ sensitivity) and ensuring reversibility. If solved, Li–air or Zn–air batteries could be game-changers for long-range EVs and grid storage.</li><li><strong>Multivalent Batteries (Mg, Al, etc.)</strong>: Using multivalent ions (Mg²⁺, Al³⁺) could, in theory, double or triple charge per ion versus Li⁺. For example, magnesium–sulfur or aluminum–graphene concepts are under study. These offer potentially higher energy density and use more abundant metals (Mg and Al are far more common than Li). However, their development has been very slow because suitable electrolytes and electrodes that enable fast Mg or Al diffusion are hard to find. At best, these remain <strong>long-term research prospects</strong>.</li><li><strong>Advanced Porous Materials (Zeolites, MOFs, COFs)</strong>: Innovations in materials science open new possibilities. Porous crystalline materials like zeolites, metal–organic frameworks (MOFs), and covalent organic frameworks (COFs) have well-defined channels at the nanoscale. They can be engineered for specific functions in batteries: for example, as solid electrolyte scaffolds, ion sieves, or to immobilize polysulfides in Li–S batteries. Recent studies have begun incorporating <strong>3D zeolite networks into polymer electrolytes</strong>, showing continuous Li⁺ conduction pathways and improved stability. While still early, this cross-pollination between porous materials and energy storage is a fertile area: these materials were originally developed for catalysis and adsorption, but their tunability makes them attractive for next-generation batteries.</li></ul><p><strong>Table 1 : </strong>below compares key figures of merit for these battery types. (All data are approximate and depend on specific designs.)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/708/1*WvL8i0Bbn9_aTa59uqwo2A.jpeg" /></figure><p><strong>(Data sources: IEA 2024; Nazakat 2026; Wang 2025; plus public roadmaps.)</strong></p><p>This table shows that no current alternative exceeds Li-ion on all counts. Sodium-ion and LFP batteries score well on cost and safety but have lower energy density. Solid-state and Li–S promise much higher energy but are far from commercial. Lithium–air and multivalent batteries remain scientific challenges. In this landscape, <strong>materials chemistry determines the outcome</strong>: designing new electrode compounds, electrolytes, and interfaces is key to overcoming the trade-offs in the table.</p><h3>Artificial Intelligence and the Battery Revolution</h3><p>Traditionally, discovering a new battery material could take decades of trial-and-error. Now, <strong>artificial intelligence (AI) and machine learning</strong> are accelerating that process. AI can rapidly screen millions of hypothetical materials for desirable properties (voltage, stability, conductivity) before any lab work is done. For example, a recent Nature study used deep-learning on crystal structure databases to predict millions of new stable compounds. The authors noted that <strong>AI training at scale led to a ~10× expansion in known stable materials</strong>, with many promising candidates for <strong>layered materials and solid electrolytes </strong>(relevant to batteries).</p><p>A 2025 review in <strong>Advanced Functional Materials</strong> details how AI/ML is transforming battery research end-to-end. Key points include:</p><ul><li>AI accelerates <strong>materials discovery </strong>, finding new electrode and electrolyte compounds faster than human intuition.</li><li>AI optimizes <strong>synthesis conditions</strong> and manufacturing parameters for scale-up.</li><li>AI improves <strong>property prediction </strong>(e.g. battery lifetime, conductivity) to guide experimental design.</li><li>In battery cell production, AI can implement closed-loop control for uniform quality, though data fragmentation and standardization remain obstacles.</li></ul><p>In short, AI is becoming a powerful research <strong>tool</strong> in battery labs. It does not replace chemists, but magnifies their reach. For example, instead of testing 1000 candidate cathodes by hand, an AI can screen 100,000 virtually and narrow to the most promising dozen. This synergy – chemistry + computation – is expected to shorten innovation cycles significantly. As Li <strong>et al.</strong> put it: “<strong>AI-driven approaches enhance discovery efficiency for electrode materials</strong>”. However, the field is still maturing; “data fragmentation” and a lack of standardized protocols are cited barriers.</p><p>More broadly, several reports emphasize that <strong>materials informatics</strong> is key for future breakthroughs. The battery sector’s tight feedback loop (from fundamental research to commercialization) is well-suited for AI integration. As one source concludes, the “<strong>AI revolution in battery research is still in its early stages, and progress on multiple fronts is needed to realize its full potential</strong>.” (NAE report). We should note research questions like “<strong>Which new materials will AI help us find?</strong>” and “<strong>How to build better datasets for battery AI?</strong>” as active topics.</p><h3>Global Impact and African Perspective</h3><p>Batteries are a <strong>global research priority</strong>. Governments see battery innovation as critical to meeting climate targets. In 2023, nearly 200 countries at COP28 committed to tripling renewables and accelerating efficiency, goals that hinge on battery deployment. Public funding and private investment in battery tech are surging worldwide. Companies like Tesla, CATL, LG Chem, and startups alike pour billions into development and new factories. Meanwhile, multilateral agencies (IEA, World Bank) urge expansion of energy storage infrastructure, especially in emerging economies.</p><p><strong>Africa’s role</strong>: As noted, Africa’s vast renewables could be tied together by batteries, enabling rural electrification and industry. African scientists and engineers have growing opportunities: universities are beginning research in materials science and energy storage, and international collaborations (for example through JICA, the European Union, and U.S. NSF grants) are increasing. Africa can supply critical minerals (e.g. cobalt, graphite, and potentially lithium), but can also move toward “<strong>value-added</strong>” participation by developing recycling and manufacturing capacities. Future African research can focus on low-cost, locally abundant materials (like sodium or iron-based systems) and on portable solar+storage systems suited to African grids.</p><p>At present, battery manufacturing is dominated by East Asia (China: ~85% of cells) and, to a lesser extent, the U.S., EU, Japan, and Korea. This concentration is a strategic risk. Diversifying research and supply chains – including involving African research institutions – could improve global resilience. In practice, this means supporting African universities in energy materials, creating battery test-bed facilities, and fostering public-private partnerships for battery projects. Science is international, and solving the battery challenge will require contributions from every region.</p><h3>Research Questions Worth Asking</h3><p>Instead of ending with firm conclusions, it is useful to spotlight <strong>open scientific questions</strong>. These guide future work and show readers that science is dynamic. Some key questions include:</p><ul><li><strong>How can battery chemistries use more Earth-abundant elements?</strong> For example, can sodium or aluminum fully replace lithium in large-scale storage without sacrificing performance?</li><li><strong>Which new electrode or electrolyte materials could dramatically boost energy density or lifespan?</strong> Advanced materials like 2D crystals, novel polymers, or porous networks might hold surprises.</li><li><strong>How can we design batteries for rapid charging with minimal degradation?</strong> Understanding the mechanisms of fast charging and finding self-healing or adaptive materials is a major challenge.</li><li><strong>What is the role of AI and automation?</strong> How do we build the high-quality databases and robotic labs needed to let machine learning truly accelerate discovery?</li><li><strong>Can we crack the Li–sulfur or Li–air puzzle? </strong>These systems promise high energy but have struggled with instabilities. What breakthroughs (e.g., solid electrolytes, catalysts) would make them practical?</li><li><strong>How to achieve a truly circular battery economy?</strong> What design innovations (modular cells, alternative chemistries) will make recycling efficient and economically viable?</li><li><strong>What should Africa (and other developing regions) focus on in battery research?</strong> Perhaps decentralized storage, low-cost flow batteries, or solar–battery hybrids tailored to local needs.</li></ul><p>By framing the article with such questions, we emphasize that the story is <strong>not over</strong>. Each answer found leads to new questions in the race to better batteries.</p><h3>Conclusion</h3><p>Batteries sit at the heart of the 21st-century energy revolution. They are as fundamental to a renewable future as the turbines and solar panels that generate power. Chemistry and materials science are the core of battery innovation: by designing new crystal structures, electrolytes, and interfaces, researchers aim to push beyond current limitations. The impact is immense: safer, higher-capacity batteries could enable truly electric transport and smooth integration of green energy, ushering in a “<strong>wireless, fossil fuel-free society</strong>” as noted by the Nobel committee.</p><p>For a young chemist or materials scientist, batteries offer a wealth of opportunity. Each incremental improvement requires deep understanding of materials and creative problem-solving. Over the next decades, as Africa and the world work toward sustainable development, battery research will remain at the forefront. Ultimately, better batteries aren’t just about longer-lasting phones or cars; they are about redefining how we power our lives on a changing planet.</p><p><strong>References</strong>: Landmark and review articles (Goodenough et al. 2019 Nobel news, IEA battery report 2024, REN21 2024, Li <strong>et al</strong>. 2025 (AI in batteries), Merchant <strong>et al</strong>. 2023 (AI materials discovery), Nazakat <strong>et al</strong>. 2026 (sodium-ion), Wang <strong>et al.</strong> 2025 (Li–S review), Chacana-Olivares <strong>et al</strong>. 2025 (recycling)).</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=627616287fc1" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Kind of Scientist I Want to Become]]></title>
            <link>https://medium.com/@selomanethato780/the-kind-of-scientist-i-want-to-become-976b7233e224?source=rss-be6dbe0eba8d------2</link>
            <guid isPermaLink="false">https://medium.com/p/976b7233e224</guid>
            <category><![CDATA[writing]]></category>
            <category><![CDATA[determination]]></category>
            <category><![CDATA[ambition]]></category>
            <category><![CDATA[chemistry]]></category>
            <dc:creator><![CDATA[Thato Selomane]]></dc:creator>
            <pubDate>Sun, 26 Jul 2026 08:45:03 GMT</pubDate>
            <atom:updated>2026-07-26T08:45:03.552Z</atom:updated>
            <content:encoded><![CDATA[<p><em>There is a question that has quietly followed me for years not simply what career do I want? but what kind of contribution do I want to leave behind?</em></p><p>Many people choose science because they enjoy solving equations, working in laboratories, or understanding how nature works. Those are meaningful reasons. For me, however, science represents something more profound. It is one of humanity&#39;s greatest tools for transforming uncertainty into understanding and knowledge into progress.</p><p>I do not aspire to become a scientist simply to earn a degree, publish papers, or add another title to my name. Those achievements matter, but they are not the destination I seek. My ambition is to become a scientist whose work addresses real problems, creates opportunities for future generations, and contributes to scientific knowledge that benefits not only my country but the world.</p><figure><img alt="Molecules" src="https://cdn-images-1.medium.com/max/1024/0*I-jGzB7EnVKesZgr" /><figcaption>Photo by <a href="https://unsplash.com/@spexypants?utm_source=medium&amp;utm_medium=referral">Ayush Kumar</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h3>Why Chemistry?</h3><p>Chemistry is often called the central science because it connects physics, biology, geology, environmental science, engineering, medicine, and countless other disciplines. Everything around us from the air we breathe and the water we drink to the medicines we take and the materials we build with is influenced by chemistry.</p><p>The more I learn about chemistry, the more I realize that it is not simply about reactions in a laboratory. It is the science of transformation.</p><p><em>Transformation of matter. Transformation of energy. Transformation of possibilities.</em></p><p>Perhaps that is why chemistry continues to inspire me. It reminds me that change is not merely inevitable it can be understood, directed, and used to improve the world.</p><h3>Looking Beyond the Laboratory</h3><p>When people imagine scientific research, they often picture laboratories filled with sophisticated instruments and researchers wearing white coats. While laboratories are where many discoveries begin, science itself extends far beyond those walls.</p><p>Every breakthrough eventually reaches society.</p><p><em>Cleaner water. Better medicines. Sustainable industries. Renewable energy. Safer environments. Improved agriculture. More efficient manufacturing</em>.</p><p>Scientific research has the power to influence how millions of people live. That realization gives purpose to every experiment, every calculation, and every unanswered question.</p><h3>Why Advanced Materials Matter</h3><p>Among the many areas of chemistry, I find myself increasingly drawn toward advanced materials particularly porous materials such as zeolites and related systems.</p><p>These materials may appear ordinary at first glance, yet their microscopic structures allow them to perform extraordinary tasks.</p><p><em>They can separate molecules. Capture pollutants. Purify water. Catalyze industrial reactions. Reduce energy consumption. Support cleaner chemical manufacturing.</em></p><p>Potentially contribute to carbon capture and sustainable energy technologies. What fascinates me most is not simply what these materials already accomplish, but what they may become capable of in the coming decades. The greatest scientific discoveries often begin with asking whether existing materials can be redesigned to solve tomorrow’s problems.</p><h3>Africa’s Scientific Opportunity</h3><p>Africa is frequently discussed in terms of its challenges.</p><p><em>Water scarcity. Energy insecurity. Environmental pollution. Industrial development. Climate change. Limited research infrastructure.</em></p><p>These challenges are real. Yet they also represent opportunities for scientific leadership. Rather than asking why Africa struggles with these issues, I believe we should also ask:</p><p>Why shouldn&#39;t Africa become one of the places where the solutions are created?</p><p>The continent possesses extraordinary natural resources, diverse environments, talented students, and pressing scientific questions that deserve global attention. Scientific innovation should not be concentrated in only a handful of countries. Knowledge grows stronger when every region contributes to it. I hope to become part of that contribution.</p><h3>Learning Before Leading</h3><p>Ambition alone is never enough.</p><p>Meaningful scientific work requires patience, discipline, humility, and years of learning. Every researcher begins by asking questions they cannot yet answer. Every scientist starts by reading papers they do not fully understand.</p><p>Every laboratory skill must first be learned through mistakes. I accept that journey. I know there will be failed experiments.</p><p>Research questions that refuse to cooperate.</p><p><em>Unexpected results. Long nights analyzing data. Ideas that must be abandoned. Hypotheses that prove incorrect.</em></p><p>Science advances not because every experiment succeeds, but because every honest experiment teaches something.</p><h3>Building Bridges Across Continents</h3><p>One of my aspirations is to collaborate with researchers from different parts of the world.</p><p>Africa ~ Europe ~ Asia ~ South America ~ North America.</p><p>Every scientific community brings unique perspectives, technologies, and experiences. The challenges facing humanity do not stop at national borders.</p><p>Neither should scientific collaboration. I hope to learn from researchers around the world while also sharing ideas developed within Africa. Knowledge should travel in every direction.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*EjaxwS7oSo0NhEj_" /><figcaption>Photo by <a href="https://unsplash.com/@iamtru?utm_source=medium&amp;utm_medium=referral">Trust &quot;Tru&quot; Katsande</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h3>A Long-Term Vision</h3><p>My academic journey is not something I see ending with an Honours degree.</p><p>I hope Honours becomes the foundation. A Master’s degree builds upon it. A PhD expands it further.</p><p>Beyond that, I hope to continue contributing through research, collaboration, teaching, and mentoring.</p><p>I want my work to evolve over decades rather than years. The most meaningful scientific careers are rarely defined by a single discovery. They are built through consistent curiosity, careful research, and a willingness to continue learning.</p><h3>Why I Write</h3><p>Some may wonder why I choose to write publicly while still learning. The answer is simple. Writing helps me think. It forces me to ask better questions. It reminds me that science is ultimately about sharing knowledge.</p><p>These articles are not declarations of expertise. They are records of a journey. A place where ideas can grow alongside education. Perhaps years from now I will look back at these words and recognize how much I still had to learn. I hope that happens. Growth should never stop.</p><figure><img alt="Businessman hand putting blank wooden cube block stack on white background with copy space for input text, icon, trend, creative idea, finance, leadership, strategy, business, online marketing concept" src="https://cdn-images-1.medium.com/max/1024/0*SG1N8c0_Cv3aL4uT" /><figcaption>Photo by <a href="https://unsplash.com/@imaginebuddy?utm_source=medium&amp;utm_medium=referral">Imagine Buddy</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h3>The Scientist I Hope to Become</h3><p>If someone were to ask me many years from now what kind of scientist I became, I would not want the answer to begin with awards, publications, or professional titles.</p><p>I would rather they say:</p><p><em>He remained curious. He kept asking difficult questions. He worked with integrity. He collaborated generously. He believed science should improve people’s lives. He never stopped learning.</em></p><p>If my research can help develop cleaner technologies, improve access to safe water, reduce environmental harm, or inspire even one young student to believe that they, too, belong in science, then every hour spent studying will have been worthwhile. Science is not merely a profession. It is a responsibility. And while I have only begun this journey, I hope to spend my life honoring that responsibility through curiosity, persistence, collaboration, and a commitment to advancing chemistry in Africa while contributing to scientific progress across the world.</p><p>The scientist I want to become is still being shaped.</p><p><em>Every book I read. Every experiment I perform. Every question I ask. Every failure I learn from. Every person I meet. Every idea I choose to pursue.</em></p><p>Perhaps becoming a scientist is not about reaching a destination. Perhaps it is about remaining committed to the pursuit of knowledge and using that knowledge to leave the world better than we found it.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=976b7233e224" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why Africa’s Future May Depend on Materials Science]]></title>
            <link>https://medium.com/@selomanethato780/why-africas-future-may-depend-on-materials-science-9fa0ec54afb1?source=rss-be6dbe0eba8d------2</link>
            <guid isPermaLink="false">https://medium.com/p/9fa0ec54afb1</guid>
            <category><![CDATA[africa]]></category>
            <category><![CDATA[early-career-researchers]]></category>
            <category><![CDATA[improvement]]></category>
            <dc:creator><![CDATA[Thato Selomane]]></dc:creator>
            <pubDate>Sat, 25 Jul 2026 20:06:52 GMT</pubDate>
            <atom:updated>2026-07-25T20:06:52.959Z</atom:updated>
            <content:encoded><![CDATA[<p><em>The future is not built by ideas alone. It is built by the materials that make those ideas possible.</em></p><p>When people think about the future of Africa, they often think of politics, economics, education, healthcare, or technology. These are all essential, but beneath each of them lies something that receives far less attention: <strong>materials</strong>.</p><p>Every bridge, every hospital, every solar panel, every water purification system, every electric vehicle, every medical device, every battery, and every satellite depends on materials engineered to perform a specific purpose. Before a technology can transform lives, someone must first discover or improve the material that makes it work. This is why I believe materials science deserves far greater attention across Africa. It is not simply another scientific discipline; it is one of the foundations upon which modern civilization is built.</p><h3>Looking Beyond Today’s Problems</h3><p>Africa is often described through its challenges: water scarcity, energy shortages, pollution, unemployment, climate change, and limited industrial development. These challenges are real, but they should not define the continent&#39;s scientific future.</p><p>History shows that every major leap in civilization has been accompanied by advances in materials. The Stone Age, Bronze Age, and Iron Age were each named after the materials that transformed society. Today, we are living in an era shaped by semiconductors, advanced polymers, nanomaterials, catalysts, and porous materials. The nations that develop these technologies will influence the industries of tomorrow. Africa has an opportunity not only to adopt these innovations but also to contribute to creating them.</p><h3>Why Materials Matter</h3><p>Most people rarely think about the materials surrounding them. We notice the finished products a phone, a car, a building but not the chemistry that made them possible.</p><p>Materials scientists ask questions such as:</p><ul><li>How can we make batteries that last longer?</li><li>How can polluted water be cleaned more efficiently?</li><li>Can industrial emissions be reduced using better catalysts?</li><li>How can solar panels become more affordable?</li><li>Can we design materials that remove carbon dioxide from the atmosphere?</li><li>How do we manufacture products while using less energy?</li></ul><p>These questions may sound technical, yet they are deeply connected to everyday life.</p><h3>Water: One of Africa’s Greatest Challenges</h3><p>Access to clean water remains one of the continent’s most pressing issues. Population growth, industrial pollution, mining activities, agricultural runoff, and climate variability continue to place pressure on freshwater resources. Chemistry offers practical solutions.</p><p>Materials such as activated carbon, membranes, and porous solids can remove contaminants from water. Among these porous materials, <strong>zeolites</strong> are particularly interesting because of their ability to trap certain ions and molecules while allowing others to pass. Researchers continue to investigate how natural and synthetic zeolites can remove heavy metals, ammonium, dyes, and other pollutants. Modified zeolites may also improve wastewater treatment and industrial water recycling.</p><p>Although no single material can solve every water problem, advances in adsorption materials could become an important part of broader water-treatment strategies.</p><figure><img alt="Variety of chemotherapy drugs in vials and an IV bottle." src="https://cdn-images-1.medium.com/max/1024/0*5NwgdJ0bGsyOXa3h" /><figcaption>Photo by <a href="https://unsplash.com/@nci?utm_source=medium&amp;utm_medium=referral">National Cancer Institute</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h3>Energy Is More Than Electricity</h3><p>When people discuss energy, they usually focus on electricity generation. Yet energy also includes transportation, manufacturing, agriculture, and chemical production.</p><p>As countries work toward lower-carbon economies, scientists are searching for materials that can:</p><ul><li>improve hydrogen production.</li><li>Increase battery performance.</li><li>Enhance fuel cells.</li><li>Store renewable energy more efficiently.</li><li>Reduce industrial emissions.</li></ul><p>Catalysts play a central role in many of these technologies because they speed up chemical reactions while reducing energy consumption. Improving catalysts even incrementally can influence industries that produce fuels, fertilizers, plastics, pharmaceuticals, and countless other products.</p><h3>Carbon Capture and Climate Change</h3><p>Climate change is one of the defining scientific challenges of this century.</p><p>Around the world, researchers are developing materials capable of capturing carbon dioxide before it enters the atmosphere or even removing it directly from the air. Porous materials, including zeolites and newer material families, are being studied because their internal structures can selectively adsorb gas molecules under certain conditions.</p><p>Many questions remain:</p><p><em>Can these materials become cheaper? Can they be regenerated repeatedly? Can they perform well under humid conditions? Can captured carbon dioxide be converted into valuable chemicals?</em></p><p>These are scientific questions with significant environmental and economic implications.</p><h3>Why Catalysis Deserves More Attention</h3><p>Many industrial chemical reactions would be impractical without catalysts.</p><p>Catalysts make reactions faster, more selective, and often more energy efficient. The fertilizers that support global food production, the fuels powering transportation, and many essential chemicals depend on catalytic processes.</p><p>Improving catalyst design is therefore not only an academic exercise it has the potential to reduce waste, lower energy consumption, and improve industrial sustainability. This is one reason why materials chemistry and catalysis continue to attract global research interest.</p><h3>Africa’s Untapped Scientific Potential</h3><p>Africa possesses exceptional biodiversity, abundant mineral resources, growing universities, and a rapidly expanding population of young people. Yet scientific success cannot be measured only by natural resources. It depends on investment in laboratories, research infrastructure, collaboration, mentorship, and education.</p><p>There is enormous potential for African scientists to contribute to international research, particularly in fields connected to sustainability, environmental chemistry, advanced materials, renewable energy, and industrial innovation. The goal should not simply be to apply discoveries made elsewhere. It should also be to generate discoveries from within Africa that benefit the world.</p><h3>Science Has No Borders</h3><p>Scientific problems rarely stop at national boundaries. Climate change affects every continent. Water pollution crosses rivers and oceans. Energy security influences economies worldwide.</p><p>Because these challenges are shared, scientific progress increasingly depends on international collaboration. A student beginning research in South Africa today may eventually collaborate with laboratories in Germany, Japan, Brazil, China, Canada, or elsewhere.</p><p>Knowledge grows when scientists exchange ideas, techniques, and perspectives across cultures.</p><p>Science becomes stronger when collaboration replaces isolation.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*ymI1LtVAANE1eqe9" /><figcaption>Photo by <a href="https://unsplash.com/@nejc_soklic?utm_source=medium&amp;utm_medium=referral">Nejc Soklič</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h3>The Importance of Curiosity</h3><p>Every major scientific breakthrough began with someone asking a question.</p><p><em>Why does this happen? Can it be improved? What if we tried something different?</em></p><p>Curiosity is often underestimated because it produces no immediate result. Yet curiosity is the starting point of every experiment, every discovery, and every innovation.</p><p>Before there is a solution, there is usually a question.</p><h3>The Scientist I Hope to Become</h3><p>I do not write these thoughts because I claim to have solved Africa’s challenges. I write because I hope to spend my career asking worthwhile questions.</p><p>I am particularly interested in how chemistry and advanced materials can contribute to cleaner water, more sustainable energy systems, improved catalysis, and environmentally responsible industrial processes. I hope that each stage of my education from Honours to postgraduate research and beyond will deepen that understanding. The path will require learning, mistakes, collaboration, and patience.</p><p>But every meaningful scientific journey begins long before the first published paper.</p><p>It begins with curiosity.</p><h3>A Future Worth Building</h3><p>The future of Africa will not depend on a single invention or a single scientist.</p><p>It will depend on thousands of researchers, engineers, educators, technicians, entrepreneurs, and students working together to solve problems that matter. Materials science is only one part of that future. Yet it is a remarkably important one.</p><p>Every cleaner catalyst, every stronger battery, every safer water filter, every improved adsorbent, and every new functional material has the potential to influence millions of lives. Perhaps that is what makes science so remarkable. Its greatest achievements are often invisible. We do not notice the catalyst inside an industrial reactor. We rarely think about the membrane purifying drinking water. We seldom ask how the battery inside a medical device was designed. Yet these quiet advances shape the quality of human life every day.</p><p>If Africa is to build a more sustainable, innovative, and resilient future, it will require more than ambition. It will require knowledge. It will require research. And it will require a generation of scientists willing to ask difficult questions not only for their own countries, but for the world.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*HAId8N1ovRlXM5m-" /><figcaption>Photo by <a href="https://unsplash.com/@microsoftcopilot?utm_source=medium&amp;utm_medium=referral">Microsoft Copilot</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9fa0ec54afb1" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>