phd thesis on organic semiconductors

Help | Advanced Search

Condensed Matter > Materials Science

Title: phd thesis - 'electronic processes at the interface of organic semiconductors'.

Abstract: The thesis focuses on study of optical and photovoltaic properties of iso-type and aniso-type heterostructures based on photosensitive layers of organic semiconductors. The primary attention has been paid on determination of optimal conditions for preparation of organic thin-film structures of methyl substituted perylene pigment, hexathiopentacene and SnCl2 phthalocyanine. These structures are perspective for fabrication of photosensitive heterostructures and can be used for development of effective photoconverters of solar energy. The influence of annealing and substrate temperature during thermal deposition of the films on morphology and surface structure, optical and photovoltaic properties of thin films of methyl substituted perylene pigment, hexathiopentacene and SnCl2 phthalocyanine was studied. The surface and bulk parameters of considered films were determined from the results obtained here. The photovoltaic and optical properties of organic iso-type and aniso-type heterostructures prepared on the substrates with different temperatures were studied. The obtained data for heterostructures were analyzed using Anderson model and Van Opdorp model, respectively. This analysis showed that organic aniso-type heterostructures based on methyl substituted perylene pigment and iso-type heterostructures based on pentacene can be described by the models proposed and developed for inorganic semiconductor heterojunctions. The optimal conditions were determined for preparation of organic photosensitive heterostructures, which absorb most part of solar radiation in visible and near infrared range, as well as effectively generate and separate charge carriers on the interface of heterojunction. The aniso-type and iso-type heterostructures prepared under optimal conditions are perspective elements for development of organic photoconverters, including organic solar cells.

Submission history

Access paper:.

• Other Formats

References & Citations

• Google Scholar
• Semantic Scholar

BibTeX formatted citation

Bibliographic and Citation Tools

Code, data and media associated with this article, recommenders and search tools.

• Institution

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs .

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• My Account Login
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Open access
• Published: 23 April 2021

Active discovery of organic semiconductors

• Christian Kunkel   ORCID: orcid.org/0000-0002-0612-1706 1 ,
• Johannes T. Margraf   ORCID: orcid.org/0000-0002-0862-5289 1 ,
• Ke Chen   ORCID: orcid.org/0000-0003-0807-1930 1 ,
• Harald Oberhofer   ORCID: orcid.org/0000-0002-5791-6736 1 &
• Karsten Reuter   ORCID: orcid.org/0000-0001-8473-8659 1 , 2  

Nature Communications volume  12 , Article number:  2422 ( 2021 ) Cite this article

14k Accesses

66 Citations

102 Altmetric

Metrics details

• Computational chemistry
• Organic molecules in materials science

The versatility of organic molecules generates a rich design space for organic semiconductors (OSCs) considered for electronics applications. Offering unparalleled promise for materials discovery, the vastness of this design space also dictates efficient search strategies. Here, we present an active machine learning (AML) approach that explores an unlimited search space through consecutive application of molecular morphing operations. Evaluating the suitability of OSC candidates on the basis of charge injection and mobility descriptors, the approach successively queries predictive-quality first-principles calculations to build a refining surrogate model. The AML approach is optimized in a truncated test space, providing deep methodological insight by visualizing it as a chemical space network. Significantly outperforming a conventional computational funnel, the optimized AML approach rapidly identifies well-known and hitherto unknown molecular OSC candidates with superior charge conduction properties. Most importantly, it constantly finds further candidates with highest efficiency while continuing its exploration of the endless design space.

Similar content being viewed by others

Interpretable discovery of semiconductors with machine learning

Organic materials repurposing, a data set for theoretical predictions of new applications for existing compounds

An artificial intelligence-aided virtual screening recipe for two-dimensional materials discovery

Introduction.

The sheer vastness of chemical spaces 1 has long motivated prior-to-synthesis virtual discovery. In corresponding work, promising candidate molecules or materials for refined study are often searched and identified on the basis of a small number of quantities that are deemed representative for the targeted application 2 , 3 , 4 . Prevalent for first-principles computational screening approaches is to calculate such descriptors at predictive quality through electronic structure theory for every candidate in a somehow enumerated chemical space or otherwise given database. Initially performed for small focused libraries, the screening is now extended to search spaces of ever increasing size and—since discovery is limited to the explicitly considered molecules or materials—to ever more systematic and exhaustive enumerations within these spaces.

Unfortunately, the combinatorial explosion characteristic for chemical versatility quickly leads to intractable numbers of candidates for such exhaustive first-principles screenings, even if based on computationally comparably undemanding descriptors. A common strategy to tackle this problem is a computational funnel 5 . Here, the exhaustive screening is only performed for computationally least-demanding descriptors or even less demanding estimates thereof. Subsequently, the large candidate set is narrowed in staged filtering and the calculation of other descriptors is only performed for smaller and smaller subsets which appear promising in terms of the previously calculated descriptors. Unfortunately, chemical diversity suggests the multi-objective (descriptor) landscape spanned over the search space to be quite rugged 6 , with molecular or materials sub-classes likely constituting separate funnels and related analogs leading to multiple local minima. This raises concerns whether the true optimum candidates can reliably be identified through such computational funneling.

An ever more appealing alternative is therefore to completely abandon the original idea to exhaustively screen a once defined chemical space or database. Instead, the explicit first-principles computation of the descriptors is restricted to candidates emerging in an iteratively refining search 7 , 8 , 9 . In the context of data science, this is afforded by several learning concepts, which additionally allow to even avoid predefining or a priori enumerating the search space itself. Examples include (semi-)supervised learning, meta-, transfer-, or few-shot learning and generative models 10 , 11 . For drug-discovery tasks 12 , 13 , such concepts have already been successfully employed to further accelerate molecular de novo design 14 and drive autonomous discovery 15 . For materials discovery based on first-principles descriptors, in particular active machine learning (AML) 16 has been explored as a most data-efficient method 17 , 18 , 19 , 20 , 21 , 22 .

In AML, the acquired knowledge in form of explicitly calculated descriptors is used to successively establish a surrogate model of larger and larger regions of the rugged descriptor landscape. In an iterative procedure, the predictive-quality calculations for new candidates can then also be balanced between exploitation and exploration. In exploitation, the global insight provided by the current surrogate model is used for a targeted identification of new promising candidates. In exploration, descriptors for new candidates are specifically calculated to refine and extend the surrogate model. For this, we here employ Gaussian Process Regression (GPR) and use high values of its inherent Bayesian uncertainty estimate to flag candidates (or regions in chemical space) for which an explicit descriptor calculation will maximally contribute new information.

We pursue this concept for the efficient virtual discovery of organic semiconductors (OSCs) for electronic applications. Used in organic field effect transistors (OFETs), 23 photovoltaics (OPVs), 24 or light emitting diodes (OLEDs), 25 OSCs offer great versatility and novel materials’ properties, paired with a low ecologic and economic footprint. Typical OSC-constituting molecules are, however, of considerable size (e.g., 22 or 42 non-hydrogen atoms in the classic examples pentacene or rubrene, respectively) and the spanned electronic property landscapes are known to be highly sensitive even to small molecular substitutions. 26 , 27 , 28 A vast number of ~10 33 similar-sized molecules is estimated to be synthesizable 1 , raising the suspicion that presently known well-performing OSC molecular materials are not even the tip of the iceberg. This has motivated a number of preceding exhaustive screening or virtual discovery studies in more or less restricted closed subspaces. 3 , 5 , 29 , 30 , 31 , 32 , 33 , 34 .

In this work we first analyze a diverse set of OSC molecules to derive clear molecular-construction rules that allow to generate an in principle unlimited OSC chemical space. This space is then successively explored by the AML discovery strategy, rapidly identifying molecular candidates that are superior to well-known OSC materials in terms of their molecular electronic descriptors assessing efficient charge injection and charge mobility. Deep methodological insight is gained by analyzing and visualizing the AML exploration inside a chemical space network (CSN) containing only a subset of the design space, limited to allow its full enumeration. Even inside this truncated chemical space the AML-discovery clearly outperforms a conventional funnel approach.

Morphing based generation of an unlimited OSC search space

The basis for our efficient AML exploration of an a priori unlimited molecular search space is the development of a concise set of molecular construction rules that allow to generate this space by iterative application. To establish a diverse, but problem-specific chemical space, we resort to existing domain knowledge and analyze the building blocks and motives contained in molecules constituting a number of well-performing crystalline OSC molecular materials. For this analysis, we exploit the fact that most functionalized organic molecules can be unambiguously fragmented into a molecular backbone (of one or more cores), linkers (that connect cores) and side groups (attached to cores) as illustrated in Fig.  1 . Without loss of generality, we correspondingly fragment 30 prominent π-conjugated molecules that belong to a variety of important molecular families 23 (Acenes, Thienoacenes, TTF-derivatives, Carbazoles, Triphenylamines, Diimides, Quinacridones and Azaacenes) and consist of the most common organic elements C, H, N, O and S. Figure  1 highlights some of these peer molecules and the full set is given in the SI in Supplementary Fig.  1 . Intriguingly, the richness of chemical building blocks identified in this way can be exhaustively generated by a set of only 22 simple molecular morphing operations starting from the smallest aromatic building block benzene. As illustrated in Fig.  1 these morphing operations each act on a molecule’s individual atomic sites or fragments, each time adding, modifying or removing fragments. These morphing operations should be seen as alchemical transformations to navigate between molecules, while applying organic synthesis steps could be a viable alternative. 35 Even though at a first glance rather unintuitive for the generation of successively larger or complex molecules, we also note that the inclusion of every morphing operation in a backwards step, i.e., resubstituting a fragment substructure, is crucial to increase the interconnectivity of the forming chemical space, see Supplementary Fig.  3 .

a Important π-conjugated molecular families and examples of well-performing OSC-molecules therein. Molecular morphing operations are designed such that the generated OSC space includes these families. b Schematic overview of the molecular generation process. Starting from benzene, diverse molecules are created by iterative application of up to 22 morphing operations. The first generation resulting from the 8 morphing operations applicable to benzene is fully shown. Molecules in further generations are only shown as examples, but every operation type is depicted at least once, see also Supplementary Fig.  2 for an extended depiction. c Fragment-definitions used throughout the text exemplified for the molecule BDTTE. Connected aromatic ring structures are cores. Linkers and sidegroups both branch from a core structure with a single bond, but are either connecting to at least two core structures or only bonded to one core fragment. d Concepts for symmetry detection used throughout the molecular generation process. e Modified molecular morphing step, adapted to the symmetry constraints imposed on candidate molecules.

The generic nature of the morphing operations identified through the fragmentation ansatz is not only a stepping stone for the efficient AML exploration. It also provides a blueprint for future variations of the present search space or the generation of different search spaces for other applications. Additional morphing operations will lead to more general search spaces and could be automatically extracted from a diverse chemical database 36 , while deliberate suppression of morphing operations can be used to focus on molecular sub-classes. Ring-annelation type morphing operations as well as biphenylic addition are for example essential for the iterative construction of core Acene fragments, such as in Pyrene or DPA. To build structures like Thienoacenes, Azaacenes or Carbazoles, ring contractions that lead to 5-membered rings are included as intermediates for heteroaromatic ring construction. This, though, comes at the cost of potentially yielding pericyclically reactive molecules, as discussed further below. Similarly, two types of linker operations are included to access the family of Triphenylamines. Further examples together with a detailed description of every morphing operation are provided in Supplementary Note  1 . Considering their known OSC tuning potential, 28 , 37 , 38 we note that in particular the augmentation of the present backbone-oriented set of construction rules by specific morphing operations for side groups or additional functional groups is expected to lead to an important extension of the here showcased search space.

The construction rules may also be modified to incorporate further prior knowledge about the OSC design problem. Here, we notably include constraints on molecular symmetry. Molecular symmetry may be beneficial for synthetic accessibility. Furthermore, it can mitigate mobility reducing charge localization 27 and in particular in monomolecular crystals often favors charge percolation pathways 3 , 39 , 40 (albeit its role can be intricate 41 ). We correspondingly prune the construction rules for the present OSC context to enforce 2D graph symmetries expected to provide a prosymmetry for the 3D case. Specifically, generated molecules are only considered for further morphing, if they fall into three types of symmetry classes as explained in Fig.  1 d, e: They (1) exhibit a full graph-symmetry, with all atomic environments appearing at least twice. (2) An asymmetric part in the molecule made of one or more fragments is symmetrically substituted by an even number of similar fragments, or (3) a molecule is prosymmetric such that it has atomic sites on which a single substitution operation could lead to a molecule of class (1) or (2). Further details on symmetry detection are provided in Supplementary Note  2 . As always, incorporation of any such domain-specific heuristics like symmetry is thereby a double-edged sword, possibly generating more meaningful search spaces as much as introducing a limiting bias. AML is particularly appealing in this respect. Any such rules can readily be added or dropped without incurring excessive computational costs as in exhaustive screenings of predefined search spaces.

Charge-conduction based fitness

In the spanned search space, we assess the suitability of candidate molecules for OSC applications by two descriptors known to probe two important and complementary aspects related to the conduction of charge. One concerns the efficient injection of charge from a contacting electrode into the OSC material. The other assesses the required high charge mobility inside the OSC bulk. For predominantly p -type OSC materials 23 a detrimentally high barrier for a corresponding hole injection from a standard gold electrode is readily probed by a level-alignment descriptor ϵ align  =  ∣ ϵ HOMO  − Φ Au ∣ , 42 which evaluates the energetic mismatch between the Au work function Φ Au  = − 5.1 eV 43 and the energetic position of the highest occupied molecular orbital (HOMO) ϵ HOMO as a common approximation of the material’s ionization potential. 44 , 45 Adapting this descriptor to other electrode materials or to n -type OSC materials (then involving the energetic position of the lowest unoccupied molecular orbital, LUMO) is straightforward. As an equally established descriptor for the bulk charge mobility we employ the intra-molecular (hole) reorganization energy λ h , which measures the cost of accommodating a new charge state after the carrier has moved to the next molecular site. 46 , 47 As molecular properties, both ϵ HOMO and λ h can be determined by efficient first-principles calculations as detailed in Supplementary Note  2 , where the density-functional theory (DFT) B3LYP 48 , 49 , 50 level of theory constitutes a well established accuracy standard 27 , 31 , 39 , 40 , 51 , matching experimental data 44 , 52 . We emphasize though that using the lowest-energy gas-phase conformer for the descriptor calculation disregards packing-effects in the molecular crystal 53 , 54 , 55 and we further discuss the influence of conformers on descriptor values in Supplementary Note  3 .

To evaluate molecular fitness and prioritize candidates during AML discovery, both objectives are combined in a scalarized fitness function

which an ideal candidate molecule will maximize. 56 Here, the weight vector w  = (1.0, 0.7) ⊤ accommodates the generally different absolute scales of the two descriptors, with the value of 0.7 chosen to yield an essentially Ohmic alignment with the electrode of ∣ ϵ align ∣  < 0.3 eV if λ h  falls into the range of commonly known OSCs. We note, though, that the exact choice of weights is rather unimportant for the performance of the AML search, as it only linearly biases F towards either of the descriptors, as further detailed below. With the currently chosen weight and at the DFT-B3LYP level of theory, pentacene and rubrene – materials that have been contacted by gold electrodes before 57 , 58 – will feature F values of −0.16 and −0.2, respectively. A threshold F ≥ −0.2 will therefore later on be used to measure discovery success of the AML.

AML: design and search strategy

By successively querying the explicit first-principles calculation of the descriptors for identified candidate molecules, the AML algorithm establishes an ever improving surrogate model of the fitness function F over the search space. Out of a manifold of in principle possible surrogate models, we found GPR to already achieve outstanding performance at very moderate amounts of data. In brief, the employed model uses circular Morgan fingerprints 59 to compare the structural similarity of not yet explicitly calculated molecules with the hitherto acquired ones. Specifically, counts of substructures that can be extracted by moving up to two bonds away from each central atom are generated. The similarity between two molecules is then measured with a substructure count kernel. A full account of the GPR learning through log-marginal likelihood maximization is provided in Supplementary Note  2 . A central advantage of GPR for the AML context is that it not only provides a prediction for the targeted fitness function F , but also the corresponding predictive uncertainty σ from the Gaussian variance. Balancing between exploitation and exploration, the AML algorithm can thus query new candidate molecules either because they are highly promising in terms of a maximum predicted fitness F or because they exhibit a high uncertainty σ such that their explicit calculation will maximally improve the surrogate model. Practically, molecules are thereby chosen according to an upper confidence bound acquisition function

This represents a simple, well-tested strategy in Bayesian optimization 60 , 61 , 62 or active-search 63 , 64 with GPRs, which contains only one hyperparameter κ to balance exploration and exploitation.

Multiple possibilities arise how to actually execute the iterative AML process. After initializing the surrogate model by training on a defined number N initial of molecules, central questions concern the acquisition of new data before the surrogate model is retrained. Compatible with super-computing resources that encourage a parallel first-principles evaluation of the descriptors for multiple molecules, we opt for a batch-based learning where N batch molecules with maximum F acq are queried and the model is then retrained on the basis of the accumulated new descriptor data. Future improvements could include an additional enforcement of diversity in the prioritized batch. 18 , 21 , 65 , 66 In an in principle infinite chemical space, another central AML design choice regards the extent over which new molecules are practically assessed with the established, conceptually global surrogate model. Aiming for high-performance OSC molecules of tractable size and complexity, we here opt for a single tree expansion that limits the candidates to those in the vicinity of already sampled ones 67 .

In a most straightforward realization and if all molecules for which first-principles descriptors have already been computed define the current population at step n of the AML search, then the N batch molecules for the next step n  + 1 are identified in the search space formed by all molecules that can be generated by one-time application of any of the morphing operations to every molecule in the current population. While this nicely exploits the evolutionary pressure contained in the current population of size N pop  =  N initial  +  n  ×  N batch , the search space for step n  + 1 could also be systematically increased by exhaustive multiple-time application of the morphing operations. As illustrated below by comparing a corresponding search depth of one- or two-time application, this may help to overcome local funnels and navigate more efficiently through chemical space. On the other hand and regardless of the actual search depth d search , the continuously growing population size will at later learning steps n inevitably lead to a combinatorial explosion of new candidates for any such exhaustive enumeration. Eventually, this requires to decrease the resolution in the ever increasing search space. Note that precisely this combinatorial explosion also precludes popular supervised machine learning approaches that exhaustively learn molecular properties in a closed chemical space, possibly followed by some form of data mining 3 .

A decreasing resolution in the AML search space can for instance be achieved by imposing additional heuristic selection criteria, e.g., selectively suppressing certain morphing operations for increasing search depths, or other more sophisticated tree-search policies 68 also employed in reinforcement learning 35 , 69 . Here, we realize deeper partial expansions of the search tree up to a search depth d search by applying the molecular morphing operations only to a fixed number of N deep molecules selected first from the current population and then subsequently from those molecules that were created by the previous morphing operations. By each time selecting the N deep molecules through fitness-rank based roulette-wheel selection, i.e., by assigning higher selection probabilities to molecules with high F acq values, the search tree is thus preferentially expanded into regions of the OSC space that the surrogate model anticipates to be rewarding (either in terms of exploitation or exploration).

Hyperparameter optimization

The thus defined AML approach contains a number of hyperparameters that may critically affect its performance. Most notably, these are κ that balances exploration and exploitation in the acquisition function, N batch the size of the prioritized batch in each learning step, as well as d search the depth of the search space in terms of the number of applied morphing operations. The decreased resolution strategy additionally requires the specification of the fixed subset size of N deep molecules to which morphing operations are applied. Less decisive is the initial number of molecules N initial used for the first training of the surrogate model, which defines only an insignifiant part of the total executed first-principles calculations and which should only be large enough to somehow kick-start the AML process. Here, we suitably set N initial to the 179 unique molecules that result in the first two generations when applying all morphing operations up to two times starting from the simplest building block benzene, cf. Fig.  1 .

In order to explore the effect of the other hyperparameters and optimize them for first-principles OSC discovery, we consider the finite subspace formed of all molecules up to a maximum size of 4 rings, 4 heteroatoms and 2 linkers that are generated by exhaustive application of all morphing operations up to 14 times, see Supplementary Note  2 . With 65.552 unique molecules this subspace is already representative for the design problem and contains many and diversely structured high-performing molecules as illustrated in Fig.  2 . At the same time, the still tractable size of the finite test space allows for the exhaustive calculation of all molecular descriptors with van der Waals (vdW) corrected density functional tight-binding (DFTB). 70 While this semi-empirical level of theory is not fully quantitative, it provides a sufficiently realistic account of the descriptor landscape for the intended method testing as analyzed in detail in Supplementary Fig.  4 . Further details on molecular test space generation and descriptor calculation are provided in the Supplementary Note  2 .

Left panel: Chemical space network (CSN) representation of the finite OSC test space of 65.552 unique molecules generated by exhaustive application of all morphing operations up to 14 times. Each molecule is surrounded by morphing-related analogs (see text). Benzene as the smallest base molecule is colored in blue. All other molecular nodes are colored according to their fitness function F as calculated at the semi-empirical density-functional tight-binding level. 2438 red nodes form the target discovery group of top-performing molecules with high fitness F  ≥ −0.2. Right panel: Example molecules from the top-performing group, chosen randomly from different areas of the CSN to illustrate the structural diversity contained in the test space.

The finite test space contains a total of 2438 top-performing molecules with a high fitness F  ≥ −0.2. As a quantitative benchmark, we thus measure the discovery success S ( N ) as the fraction of these molecules that are identified after the descriptors of N molecules have been queried. With 179 queries used for the initialization, see above, the final measure S (5179) thus evaluates the discovery success after n  = 50 learning steps when using N batch  = 100. Supplementary Fig.  6 compiles the corresponding success curves S ( N ), when systematically combining N batch  = 50, 100, or 200 with κ values in half-integer steps between 0 and 5, as well as for a search depth of one- or two-time exhaustive application of all morphing operations. Fortunately, we find the AML search to be highly robust with respect to the choice of N batch and κ . Only a small variation of 0.71 <  S ( N  = 5179) < 0.80 is obtained over all tested combinations for a search depth of one, meaning that 70–80% of the top-performing molecules are consistently found after descriptors for less than 8% of the entire test space have actually been computed. For a search depth of two, this success rate becomes slightly higher, reaching up to 85% as compiled in Supplementary Fig.  7 . Generally, larger batch sizes seem to implicitly increase the explorative behavior, such that an almost indistinguishably optimum performance is obtained for larger N batch in combination with successively smaller exploration weights κ in the acquisition function, cf. Eq. ( 2 ). For too small κ , the success curves become stepped though, indicating that temporarily the mainly exploitative algorithm then only meanders through identified sub-pockets of the test space. Too large κ , on the other hand, diminish the initial success of a then too explorative algorithm in the first learning steps. Overall, an intermediate value pair ( N batch , κ ) = (100, 2.5) thus provides a robust setting and is henceforth employed in all AML runs. For these values of ( N batch , κ ), we also performed a sensitivity analysis with regard to the employed weight vector w in Eq. ( 1 ) and the bond radius in the Morgan fingerprints used to assess molecular similarity. The results are summarized in Supplementary Figs.  8 and 9 , respectively, and again demonstrate a high robustness with respect to these parameters.

The higher success rate for d search  = 2 indicates that it is generally advantageous to further expand the search space away from the known topologies of the current population. Assessing the dependence of the decreased resolution AML algorithm on its two additional hyperparameters, Supplementary Table  1 summarizes the corresponding discovery successes when systematically combining a varying subset size N deep = 100, 250, 500 and 1000 with search depths d search = 1, 2, 3, 4, 5 and 10. Again, we find the algorithm to be quite robust, with higher d search compensating smaller N deep . Within the finite test space, many combinations thus saturate at success rates around 82–83%. This is essentially as good as the best performance of the previous exhaustive enumerations, but comes at the advantage of a controlled growth of the search space at later learning steps. For the first-principles AML discovery in the virtually unlimited OSC space below we correspondingly employ this decreased resolution search strategy with a top-performing hyperparameter combination ( d search , N deep ) = (3, 500).

Visualizing AML at work

The finite test space can also be viewed as a chemical space network (CSN), in which the morphing operations establish a total of 315.451 directed connections between the constituting molecules. This allows us to visualize the space in form of a 2D graph structure, in which the molecules are mutually repelling nodes, while morphing relationships between them lead to attractive edges 71 , see Supplementary Note  1 for details. In such a representation each molecule is thus spatially surrounded by morphing-related analogs. Figure  2 shows the resulting graph, in which the individual nodes are colored according to their DFTB calculated fitness. As expected, the target group for discovery in form of the 2438 top-performing molecules is widely scattered over disjoint parts of chemical space, with ensembles of related molecules often clustered in sub-pockets.

Apart from providing a bird’s eye view of the design problem, the CSN representation also affords a direct visual access to the AML process. Plotting the evolving population N over subsequent learning steps n reveals how much a chosen AML strategy is able to focus its exploration onto the interesting regions of chemical space and how efficiently it prioritizes OSC molecules with desired properties. Figure  3 illustrates this for the determined optimum hyperparameters and contrasts the learning for exhaustive searches with depths of one or two, with the decreased resolution strategy where the searches partially expand subsets of N deep  = 500 molecules at search depth three. For the exhaustive search with d search  = 1, the discovery is centered to more morphing-related top-performing molecules all more or less located in the core region of the CSN. In contrast, for the deeper exhaustive search, the algorithm also successfully identifies top-performing molecules in the periphery of the network that are topologically quite disconnected from the initial population. The downside is a rapidly increasing size of the search space that in the present case is only bounded by the finiteness of the considered test space. This is largely mitigated by the decreased resolution search, which nevertheless equally successfully identifies top-performing molecules at the CSN periphery.

The same CSN representation of the OSC test space as in Fig.  2 is shown in gray. Superimposed are the target group of 2438 top-performing molecules in red. Each panel shows the discovery success after n learning steps with the color of all identified top-performing molecules changed to blue and the search space for the next learning step n  + 1 colored in dark gray. Left upper panels: Steps n  = 10, 30, 50 for an exhaustive search with search depth of one. Left middle panels: Steps n  = 10, 30, 50 for an exhaustive search with search depth of two. Left lower panels: Steps n  = 10, 30, 50 for a decreased resolution search ( N deep  = 500) with search depth of three (see text). Supplementary Movies  1 – 3 provide the detailed, full trajectory of all three AML discovery runs over learning steps 1–50. Right centered panel: Discovery success of a conventional computational funnel after computing an equal number of descriptors (5179) as after 50 learning steps, and anticipating that knowledge of 13.755 molecules with optimum ∣ ϵ align ∣  < 0.3 eV is present (see text).

To put this performance of the AML searches into perspective, we also contrast them in Fig.  3 with the result of a conventional computational funnel. For the latter we pretend that the calculation of ϵ HOMO  has a negligible computational cost and the value of this descriptor is known for every molecule in the test space. This allows to identify a subset of 13.755 promising molecules for which ∣ ϵ align ∣  < 0.3 eV and which contains all previously considered 2438 top-performing molecules. The computational funnel approach would then focus the explicit calculation of the more demanding λ h  descriptor to molecules in this subset. To enable a direct comparison with the preceding AML assessment, a random selection of 5179 molecules out of this subset would then lead to a success rate of S (5179) ≈ 0.4. Even in this finite test space, where the AML algorithm can not even unfold its real strength, less than half of the top-performing molecules are thus found by this prevalent computational screening strategy after spending the same amount of CPU time (assuming that the exhaustive calculation of 65.552  ϵ HOMO  descriptors for the entire test space would constitute an insignificant computational effort).

First-principles AML discovery in a virtually unlimited OSC chemical space

Based on the gathered methodological understanding and optimized algorithmic settings ( N batch  = 100,  κ  = 2.5,  d search  = 3,  N deep  = 500) we now proceed to first-principles AML discovery at the vdW-corrected DFT-B3LYP level of theory. This is a truly challenging endeavor, considering the vastness of the OSC design space. While the space of molecules that can be generated through the morphing operations is in principle unbounded, we here restrict it to the realm of “small molecules” containing a maximum of 100 atoms (including H atoms). This realm appears as a first, more practical target for synthesis and crystallization, also considering that essentially all known top-performing OSC molecules to date fall into this size range. Estimated to surpass a size of 10 30 molecules, see Supplementary Note  2 , the corresponding chemical space is nevertheless virtually unlimited for all practical purposes and would defy any conventional exhaustive computational screening. While an iterative search as with AML is thus the only tractable means to explore this space at predictive quality, an additional technical aspect emerges that did not yet play a role in the analysis of the finite test space at the semi-empirical level before. It concerns the typically massively parallel processing on the required high-performance computing (HPC) infrastructure. As a result of queuing or down-times, as well as convergence behavior of the first-principles calculations, the results for the N batch descriptor calculations can become available at quite different times (or in rare cases of failed convergence or system instabilities may not become available at all). A practical way to avoid long waiting times before the last calculations are ready is to initially select a larger batch size for descriptor calculation and then continue with the forthcoming learning steps whenever the desired number of N batch molecules has been processed (successfully or unsuccessfully). We found this strategy to afford an efficient and continuous HPC workflow, here initially submitting the 200 molecules with highest F acq values for descriptor calculations. These are continuously processed on the HPC system by 40–100 parallel worker processes, to reach the targeted batch size N batch  = 100, while for a retraining of the surrogate model only successfully processed cases are included. In this respect, the above determined robustness of the AML performance with regard to the exact batch size also constitutes an important asset for such HPC operation.

Figure  4 summarizes the results of the AML discovery run over its first 15 learning steps. Gratifyingly, the algorithm quickly stabilizes into a highly efficient mode of operation while simultaneously meandering deep into unknown chemical space. Already after five learning steps even the median fitness of the entire prioritized batch exceeds the threshold value F  ≥ −0.2 for the first time, reflecting top-performing molecules. However, as clearly seen from the violin plots of the F distribution over the batches in Fig.  4 b, this high efficiency does not simply result from the algorithm just exploiting its established knowledge. Even at later learning steps, the algorithm steadily queries quite unfavorable molecules with a fitness worse than F  < −0.3. While such exploratory queries can either be based on high model uncertainty or induced by model prediction errors, they serve to continuously improve the surrogate model also outside the already considered search space. As a result, at each later learning step, the algorithm keeps on identifying top-performing molecules at a stable, high rate.

a Median values of molecular fitness F over the prioritized N batch molecules at the different learning steps (step 0 shows the median of the initial population N initial ). b Corresponding violin plot showing the (kernel-density estimated) distribution of molecular fitness F over the batch. These smooth kernel-density estimated distributions can slightly extend beyond the true range of F values as indicated by the explicit values marked by blue crosses. The number of queries leading to favorable and unfavorable molecules is indicated next to each violin. Due to descriptor calculation failures (see text) these numbers do not always add up to N batch  = 100. c Examples of top-performing molecules identified at various learning steps (see text for an explanation of the different color-highlighted geometric motifs). An extended list of the 4 top-performing molecules of each learning step is shown in Supplementary Fig.  10 .

After 15 learning steps and a corresponding calculation of first-principles descriptors for 1680 molecules (and only 35 unsuccessfully terminated calculations), a total of 900 molecules with molecular fitness F  ≥ −0.2 have been found. A relative success rate of 54%, i.e., essentially every second first-principles calculation yields a promising molecule and this without any a priori knowledge of the vast OSC space. A second AML discovery run described in Supplementary Note  4 confirms the robustness of this high performance. Notably, due to the random nature in our search strategy, significantly different, but equally favorable molecules are identified in this run. This performance becomes even more impressive from the viewpoint that these molecules are true discoveries, as essentially none of them are contained in existing focused libraries assembled in previous screening studies 3 , 31 , 32 , 33 , 34 . With typically ~10 5  − 10 6 entries, these data sets reflect the wealth of our existing knowledge and synthesis efforts, but simply do not even scratch the surface of the true OSC design possibilities. To this end, the negligible overlap with the top-performing molecules identified in these previous studies also has to do with molecular size. Within the first learning steps, the average size in the prioritized batch quickly rises to around 90 atoms, which is at the edge of the limit currently imposed on our search and in a size regime that could barely be addressed by the previous exhaustive enumeration studies. At the same time, even archetypical and acclaimed molecular OSC materials like DNTT (C 22 H 12 S) or rubrene (C 42 H 28 ) approach this size regime, with many other experimentally tested candidates falling right into it 23 . The preferred prioritization of such larger molecules is thereby to some extent likely simply a result of the combinatorially exploding phase space. On the other hand, another physical factor could be that the AML algorithm learns and exploits the tendency of λ h to decrease with increasing molecular size 3 as a consequence of a larger hole delocalization (which even at the hybrid DFT-B3LYP level of theory may be slightly overestimated 72 ). The inclusion of molecular coupling-sensitive descriptors into the fitness function is therefore certainly a promising topic for future studies.

The discovered molecules exhibit a diverse set of structures, incorporating distinct core fragments and the full set of allowed heteroatoms and linkers. Figure  4 c illustrates this with the best-performing molecules identified at selected learning steps, and an extended list being compiled in Supplementary Fig.  11 . This diversity indicates that the AML algorithm successfully explored topologically widely differing areas of the OSC space and did not get stuck in one or a few subpockets. Nevertheless, some commonalities can be spotted, like the recurrent presence of phenylamine linker motifs (marked in orange in the best-performing molecule of learning step 1 in Fig.  4 c). Similarly, more complex ring systems emerged at later learning stages (marked in blue and green in the most favorable molecule of step 3 and 9, respectively) and are from thereon quite pronounced among well-performing molecules. While a diverse molecular space is searched, the AML discovery thus automatically identifies and prioritizes privileged design motifs. After harvesting a larger number of molecules in further learning steps, an exciting prospect for future studies is therefore to mine the accumulating data set and systematically extract this implicit knowledge for rational design. To this end, the trained surrogate model can also be used to quickly assess the suitability of such manually constructed molecules or of deliberate modifications of the here identified ones. The latter could be particularly appealing in view of long-term device-stability or synthetic accessibility. We note that certainly not all identified molecules are suitable in this regard. For instance, the 5-membered unsaturated rings of the displayed compound of learning step 1 (marked in red) in Fig.  4 c could be problematic as they might undergo Diels-Alder type reactions, and we attribute the appearance of such ring motives as the algorithm’s intent to provide intermediates on the way to the later explored, more stable 5-membered heterocycles. Nonetheless, multiple of the favorable molecules are symmetric and composed of standard building blocks that should be easily accessible through short and reliable synthesis routes, with the surrogate model furthermore available to gauge the effect of stabilizing modifications.

In our view, active machine learning based on first-principles descriptors constitutes a most promising route to prior-to-synthesis virtual discovery. Its iterative refinement allows to most efficiently focus the data-generating calculations and meaningfully explore the vastness of chemical spaces at predictive quality and without a priori specifications, enumeration or reliance on empirical descriptors with limited validity range. In this work we have established such an AML discovery approach for molecular OSC materials through versatile molecular morphing operations and based on charge injection and conduction querying descriptors. Fortunately and with a view on explainable ML models, our systematic assessment within a finite test space suggests the approach to be quite robust with respect to the algorithmic hyperparameters. Most promising to further increase its already high efficiency and prevent an over-exploitation of particular structural motifs, is likely to additionally enforce structural diversity among the N batch molecules selected at each learning step, instead of the present purely fitness-ranked roulette-wheel selection.

Central to assess this performance and enable an unbiased and systematic comparability of different AML approaches will be the establishment of well-designed, balanced and freely available benchmark platforms for unlimited search spaces. As clear from the present work, already within the here pursued single-tree expansion there are multiple design strategies and concomitant algorithmic parameters. While we have explored these in a truncated test space, AML only unfolds its full potential in the exploration of unlimited spaces. Representative and standardized benchmark platforms as already available for drug-design tasks 13 will therefore be pivotal to truly compare various learning concepts that work without a priori enumeration or pre-definition of the search problem.

Further challenges and advancements in the physico-chemical domain comprise the adaption and extension of the molecular morphing operations to tailor the OSC search space. The present set derived from literature domain knowledge spans a design space geared towards flexible, π-conjugated molecules. Ultimately, a generic, but chemically-valid creation of morphing operations could drive discovery of many novel structural motives. Heavier requirements on the surrogate GPR-model in such cases could then be tackled with improved covariance functions for 2D molecular graphs 73 or conformer-specific 3D coordinates 74 , while alleviating the limited scaling by sparse approximations 75 , or application of alternative models 76 , 77 , 78 , 79 .

Another major area for development concerns the first-principles descriptors entering the employed multi-objective fitness function. Devising such suitable descriptors has evolved into an important research area of its own 80 , 81 , 82 , 83 , independent of the present AML and OSC context. With the presently employed level-alignment descriptor ϵ align and the hole reorganization energy λ h our search readily identified a diverse range of hitherto unknown molecular candidates. Just as in conventional computational screening, there are numerous possibilities to refine the underlying candidate evaluation through additional (or alternative) descriptors. In the exemplified OSC context, obvious avenues could be to explicitly consider synthetic accessibility 84 , electronic coupling and charge-transport networks in the molecular solid 46 , 51 , 85 , 86 or electron-phonon coupling 87 . In view of the high data efficiency of the AML approach, one may also drop the present focus on computationally least-demanding descriptors, originally dictated by the excessive queries in conventional exhaustive screening work. More elaborate descriptors like structural interfacing with electrode materials 88 could therefore routinely (or at least occasionally) be requested. Eventually, one could even think of incorporating experimental feedback from self-driving laboratories 89 . The prospects are thus as manifold as exciting. Regardless of the specific road chosen, it is conceptually clear that autonomously operating workflows like the present AML approach offer an unparalleled means to accelerate the discovery and design of viable future materials like the high-mobility organic semiconductors featured in this work.

Data availability

The source data necessary to reproduce the main figures of the manuscript is provided in the supplementary materials of this article.  Source data are provided with this paper.

Code availability

The code used to run AML discovery is available at https://doi.org/10.5281/zenodo.4554331

Polishchuk, P. G., Madzhidov, T. I. & Varnek, A. Estimation of the size of drug-like chemical space based on gdb-17 data. J. Comput. Aided Mol. Des. 27 , 675–679 (2013).

Article   ADS   CAS   PubMed   Google Scholar  

Jorgensen, W. L. The many roles of computation in drug discovery. Science 303 , 1813–1818 (2004).

Schober, C., Reuter, K. & Oberhofer, H. Virtual screening for high carrier mobility in organic semiconductors. J. Phys. Chem. Lett. 7 , 3973–3977 (2016).

Article   CAS   PubMed   Google Scholar  

Pulido, A. et al. Functional materials discovery using energy–structure–function maps. Nature 543 , 657–664 (2017).

Article   ADS   CAS   PubMed   PubMed Central   Google Scholar  

Pyzer-Knapp, E. O., Suh, C., Gómez-Bombarelli, R., Aguilera-Iparraguirre, J. & Aspuru-Guzik, A. What is high-throughput virtual screening? a perspective from organic materials discovery. Annu. Rev. Mater. Res. 45 , 195–216 (2015).

Article   ADS   CAS   Google Scholar  

Gaspar, H. A., Baskin, I. I., Marcou, G., Horvath, D. & Varnek, A. Chemical data visualization and analysis with incremental generative topographic mapping: big data challenge. J. Chem. Inf. Model. 55 , 84–94 (2015).

Reymond, J.-L., van Deursen, R., Blum, L. C. & Ruddigkeit, L. Chemical space as a source for new drugs. Med. Chem. Commun. 1 , 30–38 (2010).

Article   CAS   Google Scholar  

Devi, R. V., Sathya, S. S. & Coumar, M. S. Evolutionary algorithms for de novo drug design – a survey. Appl. Soft Comput. 27 , 543–552 (2015).

Article   Google Scholar  

Le, T. C. & Winkler, D. A. Discovery and optimization of materials using evolutionary approaches. Chem. Rev. 116 , 6107–6132 (2016).

Sanchez-Lengeling, B. & Aspuru-Guzik, A. Inverse molecular design using machine learning: generative models for matter engineering. Science 361 , 360–365 (2018).

Elton, D. C., Boukouvalas, Z., Fuge, M. D. & Chung, P. W. Deep learning for molecular design – a review of the state of the art. Mol. Syst. Des. Eng. 4 , 828–849 (2019).

Reker, D. & Schneider, G. Active-learning strategies in computer-assisted drug discovery. Drug Discov. Today 20 , 458–465 (2015).

Article   PubMed   Google Scholar  

Brown, N., Fiscato, M., Segler, M. H. & Vaucher, A. C. Guacamol: benchmarking models for de novo molecular design. J. Chem. Inf. Model. 59 , 1096–1108 (2019).

Schneider, G.  De novo Molecular Design (Wiley, 2013) https://books.google.de/books?id=ZxlrmwEACAAJ .

Jensen, K. F., Coley, C. W. & Eyke, N. S. Autonomous discovery in the chemical sciences part i: progress. Angew. Chem. Int. Ed. 59 , 22858–22893 (2019).

Google Scholar  

Settles, B. Active learning literature survey (University of Wisconsin, Madison, 2010).

Lookman, T., Balachandran, P. V., Xue, D. & Yuan, R. Active learning in materials science with emphasis on adaptive sampling using uncertainties for targeted design. Npj Comput. Mater. 5 , 21 (2019).

Article   ADS   Google Scholar  

Smith, J. S., Nebgen, B., Lubbers, N., Isayev, O. & Roitberg, A. E. Less is more: sampling chemical space with active learning. J. Chem. Phys. 148 , 241733 (2018).

Article   ADS   PubMed   CAS   Google Scholar  

Häse, F., Roch, L. M., Kreisbeck, C. & Aspuru-Guzik, A. Phoenics: a bayesian optimizer for chemistry. ACS Cent. Sci. 4 , 1134–1145 (2018).

Article   PubMed   PubMed Central   CAS   Google Scholar  

Vandermause, J. et al. On-the-fly active learning of interpretable bayesian force fields for atomistic rare events. Npj Comput. Mater. 6 , 20 (2020).

Janet, J. P., Ramesh, S., Duan, C. & Kulik, H. J. Accurate multiobjective design in a space of millions of transition metal complexes with neural-network-driven efficient global optimization. ACS Cent. Sci. 6 , 513–524 (2020).

Article   CAS   PubMed   PubMed Central   Google Scholar  

Bisbo, M. K. & Hammer, B. Efficient global structure optimization with a machine-learned surrogate model. Phys. Rev. Lett. 124 , 086102 (2020).

Wang, C., Dong, H., Hu, W., Liu, Y. & Zhu, D. Semiconducting π -conjugated systems in field-effect transistors: a material odyssey of organic electronics. Chem. Rev. 112 , 2208–2267 (2012).

Lin, Y., Li, Y. & Zhan, X. Small molecule semiconductors for high-efficiency organic photovoltaics. Chem. Soc. Rev. 41 , 4245–4272 (2012).

Xu, R.-P., Li, Y.-Q. & Tang, J.-X. Recent advances in flexible organic light-emitting diodes. J. Mater. Chem. C 4 , 9116–9142 (2016).

Geng, H. et al. Theoretical study of substitution effects on molecular reorganization energy in organic semiconductors. J. Chem. Phys. 135 , 104703 (2011).

Uejima, M., Sato, T., Tanaka, K. & Kaji, H. Vibronic coupling density analysis for the chain-length dependence of reorganization energies in oligofluorenes: a comparative study with oligothiophenes. Phys. Chem. Chem. Phys. 15 , 14006–14016 (2013).

Wilbraham, L., Smajli, D., Heath-Apostolopoulos, I. & Zwijnenburg, M. A. Mapping the optoelectronic property space of small aromatic molecules. Commun. Chem. 3 , 14 (2020).

Gryn’ova, G., Lin, K.-H. & Corminboeuf, C. Read between the molecules: computational insights into organic semiconductors. J. Am. Chem. Soc. 140 , 16370–16386 (2018).

Saeki, A. & Kranthiraja, K. A high throughput molecular screening for organic electronics via machine learning: present status and perspective. Jpn. J. Appl. Phys. 59 , SD0801 (2019).

Matsuzawa, N. N. et al. Massive theoretical screen of hole conducting organic materials in the heteroacene family by using a cloud-computing environment. J. Phys. Chem. A 124 , 1981–1992 (2020).

Nematiaram, T., Padula, D., Landi, A. and Troisi, A. On the largest possible mobility of molecular semiconductors and how to achieve it. Adv. Funct. Mater . 30 , 2001906 (2020).

Atahan-Evrenk, S. & Atalay, F. B. Prediction of intramolecular reorganization energy using machine learning. J. Phys. Chem. A 123 , 7855–7863 (2019).

Cheng, C. Y., Campbell, J. E. & Day, G. M. Evolutionary chemical space exploration for functional materials: computational organic semiconductor discovery. Chem. Sci. 11 , 4922–4933 (2020).

Segler, M. H. S., Preuss, M. & Waller, M. P. Planning chemical syntheses with deep neural networks and symbolic AI. Nature 555 , 604–610 (2018).

Segler, M. H. S. & Waller, M. P. Neural-symbolic machine learning for retrosynthesis and reaction prediction. Chem. Eur. J. 23 , 5966–5971 (2017).

Yao, Z.-F., Wang, J.-Y. & Pei, J. Control of π – π stacking via crystal engineering in organic conjugated small molecule crystals. Cryst. Growth Des. 18 , 7–15 (2018).

Kunkel, C., Schober, C., Margraf, J. T., Reuter, K. & Oberhofer, H. Finding the right bricks for molecular legos: a data mining approach to organic semiconductor design. Chem. Mater. 31 , 969–978 (2019a).

Stehr, V., Pfister, J., Fink, R. F., Engels, B. & Deibel, C. First-principles calculations of anisotropic charge-carrier mobilities in organic semiconductor crystals. Phys. Rev. B 83 , 155208 (2011).

Li, P., Cui, Y., Song, C. & Zhang, H. Electronic and charge transport properties of dimers of dithienothiophenes: effect of structural symmetry and linking mode. RSC Adv. 5 , 50212–50222 (2015).

Ren, L. et al. Critical role of molecular symmetry for charge transport properties: a paradigm learned from quinoidal bithieno[3,4-b]thiophenes. Chem. Mater. 29 , 4999–5008 (2017).

Ishii, H., Sugiyama, K., Ito, E. & Seki, K. Energy level alignment and interfacial electronic structures at organic/metal and organic/organic interfaces. Adv. Mater. 11 , 605–625 (1999).

Michaelson, H. B. The work function of the elements and its periodicity. J. Appl. Phys. 48 , 4729–4733 (1977).

Schwenn, P., Burn, P. & Powell, B. Calculation of solid state molecular ionisation energies and electron affinities for organic semiconductors. Org Electron. 12 , 394 – 403 (2011).

Bhandari, S., Cheung, M. S., Geva, E., Kronik, L. & Dunietz, B. D. Fundamental gaps of condensed-phase organic semiconductors from single-molecule calculations using polarization-consistent optimally tuned screened range-separated hybrid functionals. J. Chem. Theory Comput. 14 , 6287–6294 (2018).

Oberhofer, H., Reuter, K. & Blumberger, J. Charge transport in molecular materials: an assessment of computational methods. Chem. Rev. 117 , 10319–10357 (2017).

Nelsen, S. F., Blackstock, S. C. & Kim, Y. Estimation of inner shell marcus terms for amino nitrogen compounds by molecular orbital calculations. J. Am. Chem. Soc. 109 , 677–682 (1987).

Becke, A. D. Density-functional exchange-energy approximation with correct asymptotic behavior. Phys. Rev. A 38 , 3098–3100 (1988).

Lee, C., Yang, W. & Parr, R. G. Development of the colle-salvetti correlation-energy formula into a functional of the electron density. Phys. Rev. B 37 , 785–789 (1988).

Stephens, P. J., Devlin, F. J., Chabalowski, C. F. & Frisch, M. J. Ab initio calculation of vibrational absorption and circular dichroism spectra using density functional force fields. J. Phys. Chem. 98 , 11623–11627 (1994).

Moral, M., Garzón-Ruiz, A., Castro, M., Canales-Vázquez, J. & Sancho-García, J. C. Virtual design in organic electronics: screening of a large set of 1,4-bis(phenylethynyl)benzene derivatives as molecular semiconductors. J. Phys. Chem. C 121 , 28249–28261 (2017).

Kera, S. et al. Experimental reorganization energies of pentacene and perfluoropentacene: effects of perfluorination. J. Phys. Chem. C 117 , 22428–22437 (2013).

Stuke, A. et al. Atomic structures and orbital energies of 61,489 crystal-forming organic molecules. Sci. Data 7 , 58 (2020).

Mitzel, N. W. & Rankin, D. W. H. Saracen – molecular structures from theory and experiment: the best of both worlds . Dalton Trans . 3650–3662 (2003).

Blomeyer, S. et al. Intramolecular π – π interactions in flexibly linked partially fluorinated bisarenes in the gas phase. Angew. Chem. Int. Ed. 56 , 13259–13263 (2017).

Besnard, J. et al. Automated design of ligands to polypharmacological profiles. Nature 492 , 215–220 (2012).

Takeya, J. et al. Very high-mobility organic single-crystal transistors with in-crystal conduction channels. Appl. Phys. Lett. 90 , 102120 (2007).

Jurchescu, O. D., Baas, J. & Palstra, T. T. M. Effect of impurities on the mobility of single crystal pentacene. Appl. Phys. Lett. 84 , 3061–3063 (2004).

Rogers, D. & Hahn, M. Extended-connectivity fingerprints. J. Chem. Inf. Model. 50 , 742–754 (2010).

Srinivas, N., Krause, A., Kakade, S. M. & Seeger, M. W. Information-theoretic regret bounds for gaussian process optimization in the bandit setting. IEEE Trans. Inf. Theory 58 , 3250–3265 (2012).

Article   MathSciNet   MATH   Google Scholar  

Auer, P. Using confidence bounds for exploitation-exploration trade-offs. J. Mach. Learn. Res. 3 , 397–422 (2002).

MathSciNet   MATH   Google Scholar  

Srinivas, N., Krause, A., Kakade, S. & Seeger, M. Gaussian process optimization in the bandit setting: no regret and experimental design. In Proceedings of the 27th International Conference on International Conference on Machine Learning, ICML’10, 1015–1022 (Omnipress, Madison, WI, USA, 2010).

Vanchinathan, H. P., Marfurt, A., Robelin, C.-A., Kossmann, D. & Krause, A. Discovering valuable items from massive data. In Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD ’15 , 1195–1204 (Association for Computing Machinery, New York, NY, USA, 2015) https://doi.org/10.1145/2783258.2783360 .

Ma, Y., Huang, T.-K. & Schneider, J. Active search and bandits on graphs using sigma-optimality. In Proceedings of the 31st Conference on Uncertainty in Artificial Intelligence, UAI 2015 , 542–551 (2015).

Pinsler, R., Gordon, J., Nalisnick, E. & Hernández-Lobato, J. M. Bayesian batch active learning as sparse subset approximation. In Advances in Neural Information Processing Systems 32 , (eds Wallach, H., Larochelle, H., Beygelzimer, A., d’Alché-Buc, F., Fox, E. & Garnett, R.) 6359–6370 (Curran Associates, Inc., 2019).

Ash, J. T., Zhang, C., Krishnamurthy, A., Langford, J. & Agarwal, A. Deep batch active learning by diverse, uncertain gradient lower bounds. In International Conference on Learning Representations (2020).

Madhawa, K. & Murata, T. A multi-armed bandit approach for exploring partially observed networks. Appl. Netw. Sci. 4 , 26 (2019).

Browne, C. et al. A survey of monte carlo tree search methods. IEEE Trans. Comput. Intell. AI Games 4 , 1–43 (2012).

Zhou, Z., Kearnes, S., Li, L., Zare, R. N. & Riley, P. Optimization of molecules via deep reinforcement learning. Sci. Rep. 9 , 10752 (2019).

Article   ADS   PubMed   PubMed Central   CAS   Google Scholar  

Grimme, S., Bannwarth, C. & Shushkov, P. A robust and accurate tight-binding quantum chemical method for structures, vibrational frequencies, and noncovalent interactions of large molecular systems parametrized for all spd-block elements (z = 1–86). J. Chem. Theory Comput. 13 , 1989–2009 (2017).

Kunkel, C., Schober, C., Oberhofer, H. & Reuter, K. Knowledge discovery through chemical space networks: the case of organic electronics. J. Mol. Model. 25 , 87 (2019b).

Brückner, C. & Engels, B. A theoretical description of charge reorganization energies in molecular organic p-type semiconductors. J. Comput. Chem. 37 , 1335–1344 (2016).

Article   PubMed   CAS   Google Scholar  

Ralaivola, L., Swamidass, S. J., Saigo, H. & Baldi, P. Graph kernels for chemical informatics. Neural Netw. 18 , 1093 – 1110 (2005).

Himanen, L. et al. Dscribe: Library of descriptors for machine learning in materials science. Comput. Phys. Commun. 247 , 106949 (2020).

Quinonero-Candela, J. A unifying view of sparse approximate gaussian process regression. J. Mach. Learn. Res. 6 , 1939–1959 (2005).

Beluch, W. H., Genewein, T., Nürnberger, A. & Köhler, J. M. The power of ensembles for active learning in image classification. In 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition (2018) pp. 9368–9377.

Kendall, A. & Gal, Y. What uncertainties do we need in bayesian deep learning for computer vision? In Advances in Neural Information Processing Systems (2017).

Zhang, Y. & Lee, A. A. Bayesian semi-supervised learning for uncertainty-calibrated prediction of molecular properties and active learning. Chem. Sci. 10 , 8154–8163 (2019).

Wenzel, F. et al. How good is the bayes posterior in deep neural networks really? In International Conference on Machine Learning (2020).

Curtarolo, S., Hart, G. L. W., Nardelli, M. B., Mingo, N., Sanvito, S. & Levy, O. The high-throughput highway to computational materials design. Nat. Mater. 12 , 191–201 (2013).

Pracht, P., Bauer, C. A. & Grimme, S. Automated and efficient quantum chemical determination and energetic ranking of molecular protonation sites. J. Comput. Chem. 38 , 2618–2631 (2017).

Andersen, M., Levchenko, S. V., Scheffler, M. & Reuter, K. Beyond scaling relations for the description of catalytic materials. ACS Catal. 9 , 2752–2759 (2019).

Cubuk, E. D., Sendek, A. D. & Reed, E. J. Screening billions of candidates for solid lithium-ion conductors: a transfer learning approach for small data. J. Chem. Phys. 150 , 214701 (2019).

Coley, C. W., Rogers, L., Green, W. H. & Jensen, K. F. Scscore: synthetic complexity learned from a reaction corpus. J. Chem. Inf. Model. 58 , 252–261 (2018).

Ishii, H. et al. Charge mobility calculation of organic semiconductors without use of experimental single-crystal data. Sci. Rep. 10 , 2524 (2020).

Friederich, P. et al. Molecular origin of the charge carrier mobility in small molecule organic semiconductors. Adv. Funct. Mater. 26 , 5757–5763 (2016).

Landi, A. & Troisi, A. Rapid evaluation of dynamic electronic disorder in molecular semiconductors. J. Phys. Chem. C 122 , 18336–18345 (2018).

Egger, A. T. et al. Charge transfer into organic thin films: a deeper insight through machine-learning-assisted structure search. Adv. Sci. 7 , 2000992 (2020).

MacLeod, B. P. et al. Self-driving laboratory for accelerated discovery of thin-film materials . Sci. Adv . 6 , eaaz8867 (2020).

Download references

Acknowledgements

C.K. and J.T.M are grateful for support by Deutsche Forschungsgemeinschaft (DFG) through TUM International Graduate School of Science and Engineering (IGSSE), GSC 81. C.K., H.O. and K.R. gratefully acknowledge support from the Solar Technologies Go Hybrid initiative of the State of Bavaria. K.C. acknowledges funding from the China Scholarship Council. We also thankfully acknowledge computational resources provided by the Leibniz Supercomputing Centre. J.T.M. would like to acknowledge illuminating discussions on synthesizability and chemical stability with Tobias Schaub.

Open Access funding enabled and organized by Projekt DEAL.

Author information

Authors and affiliations.

Chair for Theoretical Chemistry and Catalysis Research Center, Technische Universität München, Garching, Germany

Christian Kunkel, Johannes T. Margraf, Ke Chen, Harald Oberhofer & Karsten Reuter

Fritz-Haber-Institut der Max-Planck-Gesellschaft, Berlin, Germany

• Karsten Reuter

You can also search for this author in PubMed   Google Scholar

Contributions

C.K, H.O., J.T.M and K.R. conceived the idea. C.K. implemented the algorithms in code and carried out the calculations. Methodological details were thereby worked out by C.K., K.C. and J.T.M. C.K, H.O., J.T.M and K.R. wrote the manuscript.

Corresponding author

Correspondence to Karsten Reuter .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Additional information

Peer review information Nature Communications thanks Graeme Day and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary information, description of additional supplementary files, supplementary movie 1, supplementary movie 2, supplementary movie 3, source data, source data, rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Kunkel, C., Margraf, J.T., Chen, K. et al. Active discovery of organic semiconductors. Nat Commun 12 , 2422 (2021). https://doi.org/10.1038/s41467-021-22611-4

Download citation

Received : 29 September 2020

Accepted : 15 March 2021

Published : 23 April 2021

DOI : https://doi.org/10.1038/s41467-021-22611-4

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

This article is cited by

New luminescent pyridine-based disc type molecules: synthesis, photophysical, electrochemical, and dft studies.

• Deepak Devadiga
• T. N. Ahipa
• Sandeep Kumar

Journal of Fluorescence (2023)

Charge carrier mobilities of organic semiconductors: ab initio simulations with mode-specific treatment of molecular vibrations

• Sebastian Hutsch
• Michel Panhans
• Frank Ortmann

npj Computational Materials (2022)

Predicting binding motifs of complex adsorbates using machine learning with a physics-inspired graph representation

• Mie Andersen

Nature Computational Science (2022)

Human- and machine-centred designs of molecules and materials for sustainability and decarbonization

• Daniel Schwalbe-Koda
• Yang Shao-Horn

Nature Reviews Materials (2022)

Synthesis, structural characterization, therotical and electrical properties of novel sulpho-coumarin based methacrylate polymer

• Mücahit Özdemir
• Fatih Biryan
• Ahmet Orhan Görgülü

Journal of Polymer Research (2022)

By submitting a comment you agree to abide by our Terms and Community Guidelines . If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Sameer Patwardhan

Organic semiconductors (phd thesis work).

The field of organic electronics has been thriving for the last decades due to growing commercial interest. One of the advantages of using organic materials as semiconductors is the possibility to tune their optoelectronic properties in a number of ways. The chemical structure and organization of the individual molecules in the solid state determine the charge and energy transport properties. The chemical structure can be modified by synthesis, whereas the local and long-range arrangement of the individual molecules can be controlled using the principles of molecular self-assembly.

I studied the charge-carrier and excited-state properties in relation to the chemical and supramolecular structure in over hundred organic materials. The charge transport properties were studied by pulse radiolysis time-resolved microwave conductivity (PR-TRMC) measurements and theoretical calculations. The excited state properties were investigated by combining optical spectroscopy with exciton theory calculations. We made significant progress in developing chlorophyll-based p-type biomaterials and perylene diimide n-type materials.

In general, our experimental results indicate that the performance of existing organic electronic devices is not limited by the intrinsic charge transport properties of the active materials. Instead, it is determined by the device parameters such as the choice of the electrodes and the active material, and the contact between these materials. Therefore, there is considerable scope for improving the performance of these devices.

2016  High charge carrier mobility and efficient charge separation in highly soluble perylenetetracarboxyl-diimides

2012  Efficient Charge Transport in Semisynthetic Zinc Chlorin Dye Assemblies

2012  Biosupramolecular Nanowires from Chlorophyll Dyes with Exceptional Charge-Transport Properties

2011  Delocalization and Mobility of Charge Carriers in Covalent Organic Frameworks

One thought on “ Organic Semiconductors (PhD thesis work) ”

this work is important to me

Leave a Reply Cancel reply

Your email address will not be published. Required fields are marked *

Save my name, email, and website in this browser for the next time I comment.

• Help & FAQ

An electrically driven organic semiconductor laser

• School of Physics and Astronomy

Student thesis : Doctoral Thesis (PhD)

• Organic LEDs
• Semiconductor lasers

Access Status

• Full text embargoed until
• 14 August 2026

DOI: https://doi.org/10.17630/sta/582

Full text available in St Andrews Research Repository subject to embargo

Related content

Datasets/software, an electrically driven organic semiconductor laser (thesis data).

Dataset : Thesis dataset

Molecular Spintronics

From Organic Semiconductors to Self-Assembled Monolayers

• © 2016
• Marta Galbiati 0

Institute of Molecular Science, University of Valencia, Paterna, Spain

You can also search for this author in PubMed   Google Scholar

• Nominated as an outstanding PhD thesis by the Unité Mixte de Physique CNRS/Thales, France
• Includes a detailed review of the spinterface field
• Presents the first evidence of spinterface effects at room temperature in organic spin-valves
• Includes a detailed review of transport models through self-assembled monolayers and reports on spin-dependent transport in molecular magnetic tunnel junctions
• Includes supplementary material: sn.pub/extras

Part of the book series: Springer Theses (Springer Theses)

9227 Accesses

4 Citations

This is a preview of subscription content, log in via an institution to check access.

Access this book

• Available as EPUB and PDF
• Read on any device
• Instant download
• Own it forever
• Compact, lightweight edition
• Dispatched in 3 to 5 business days
• Free shipping worldwide - see info
• Durable hardcover edition

Tax calculation will be finalised at checkout

Other ways to access

Licence this eBook for your library

Institutional subscriptions

Table of contents (8 chapters)

Front matter, introduction to organic and molecular spintronics, introduction to spintronics.

Marta Galbiati

Why Bring Organic and Molecular Electronics to Spintronics

State of the art in organic and molecular spintronics, self-assembled monolayers for molecular spintronics, introduction to self-assembled monolayers, sams based device fabrication and characterization, magneto-transport results in sam based mtjs, room temperature spin injection in organic semiconductors, state of the art in alq3-based spintronic devices, magneto-transport results in alq3 based osvs, back matter.

• Interface hybridization
• Magnetic devices
• Magnetic tunnel junctions
• Magnetoresistance
• Molecular spintronics
• Organic semiconductors
• Organic spintronics
• Self-assembled monolayers
• Spin valves
• Spinterface

About this book

This thesis targets molecular or organic spintronics and more particularly the spin polarization tailoring opportunities that arise from the ferromagnetic metal/molecule hybridization at interfaces: the new concept of spinterface. Molecular or organic spintronics is an emerging research field at the frontier between organic chemistry and spintronics.

The manuscript is divided into three parts, the first of which introduces the basic concepts of spintronics and advantages that molecules can bring to this field. The state of the art on organic and molecular spintronics is also presented, with a special emphasis on the physics and experimental evidence for spinterfaces.

The book’s second and third parts are dedicated to the two main experimental topics investigated in the thesis: Self-Assembled Monolayers (SAMs) and Organic Semiconductors (OSCs). The study of SAMs-based magnetic tunnel nanojunctions reveals the potential to modulate the properties of such devices “at will,”since each part of the molecule can be tuned independently like a “LEGO” building block. The study of Alq3-based spin valves reveals magnetoresistance effects at room temperature and is aimed at understanding the respective roles played by the two interfaces. Through the development of these systems, we demonstrate their potential for spintronics and provide a solid foundation for spin polarization engineering at the molecular level.

Authors and Affiliations

About the author, bibliographic information.

Book Title : Molecular Spintronics

Book Subtitle : From Organic Semiconductors to Self-Assembled Monolayers

Authors : Marta Galbiati

Series Title : Springer Theses

DOI : https://doi.org/10.1007/978-3-319-22611-8

Publisher : Springer Cham

eBook Packages : Chemistry and Materials Science , Chemistry and Material Science (R0)

Copyright Information : Springer International Publishing Switzerland 2016

Hardcover ISBN : 978-3-319-22610-1 Published: 04 November 2015

Softcover ISBN : 978-3-319-36964-8 Published: 23 August 2016

eBook ISBN : 978-3-319-22611-8 Published: 15 October 2015

Series ISSN : 2190-5053

Series E-ISSN : 2190-5061

Edition Number : 1

Number of Pages : XIX, 183

Topics : Organic Chemistry , Surface and Interface Science, Thin Films , Physical Chemistry , Magnetism, Magnetic Materials

• Publish with us

Policies and ethics

• Find a journal
• Track your research

The search for singlet fission: non-radiative loss pathways in organic semiconductors

--> Hook, Daniel (2024) The search for singlet fission: non-radiative loss pathways in organic semiconductors. PhD thesis, University of Sheffield.

In this thesis, we present in-depth investigations into three materials which have been predicted to undergo singlet fission via density functional theory calculations. The first is Pigment Red 254, presented by Padula et al., a diketopyrrolopyrrole derivative which shows great stability in the solid state and is also very cheap to produce, both of which are useful for device fabrication. The second and third are BoDiPy 6 and 7, derivatives of boron dipyrromethane. Singlet fission is a phenomenon which occurs in a small number of organic semiconductor materials that permits the generation of two triplet excitons from one singlet exciton, which promises to greatly increase the efficiency of solar cell devices. In the search for more singlet fission materials, being able to predict the occurrence of singlet fission com- putationally rather than relying on high-cost, time consuming experimental procedures, it would greatly expedite the process of improving the efficiency of photovoltaics. However, in this work we demonstrate that none of these materials unequivocally show singlet fission, finding in the case of Pigment Red 254 evidence for rapid non-radiative decay to the ground state which out-competes this process. We perform in-depth infra- red investigations of this material, and demonstrate that this decay takes place via a hydrogen bond mediated proton transfer process, resulting in its non-radiative loss mechanism. In the case of the BoDiPy dyes, we show that intersystem crossing takes place preferentially to singlet fission, taking place in the solution phase and showing a strong dependence on the heavy atom substituents in the BoDiPy dyes. We highlight here the need to account for well known structural markers of singlet fission loss mechanisms when attempting to predict efficient singlet fission materials, including hydrogen bonding and the heavy atom effect.

--> Final eThesis - complete (pdf) -->

Embargoed until: 23 April 2025

Please use the button below to request a copy.

Filename: Daniel_Hook___Thesis-finalapprovedsubmission.pdf

Request a copy

Embargo Date:

Please use the 'Request a copy' link(s) in the 'Downloads' section above to request this thesis. This will be sent directly to someone who may authorise access. You can contact us about this thesis . If you need to make a general enquiry, please see the Contact us page.

Academia.edu no longer supports Internet Explorer.

To browse Academia.edu and the wider internet faster and more securely, please take a few seconds to  upgrade your browser .

Enter the email address you signed up with and we'll email you a reset link.

• We're Hiring!
• Help Center

PhD thesis - 'Electronic Processes at the Interface of Organic Semiconductors

The thesis focuses on study of optical and photovoltaic properties of iso-type and aniso-type heterostructures based on photosensitive layers of organic semiconductors. The primary attention has been paid on determination of optimal conditions for preparation of organic thin-film structures of methyl substituted perylene pigment, hexathiopentacene and SnCl2 phthalocyanine. These structures are perspective for fabrication of photosensitive heterostructures and can be used for development of effective photoconverters of solar energy. The influence of annealing and substrate temperature during thermal deposition of the films on morphology and surface structure, optical and photovoltaic properties of thin films of methyl substituted perylene pigment, hexathiopentacene and SnCl2 phthalocyanine was studied. The surface and bulk parameters of considered films were determined from the results obtained here. The photovoltaic and optical properties of organic iso-type and aniso-type heterostr...

RELATED PAPERS

International Journal of Engineering Research and Technology (IJERT)

IJERT Journal

Laurence Rubenstein

Revista Panamericana de Salud Pública

Adriana Troyo

Digital Health

Jennifer Drews

Chandan Roy

Naomi Weidner

Facultad de Ciencia Política y Relaciones Internacionales

Soledad Bravo

Health Care and Informatics Review Online [serial on the Internet]

Mehnaz Adnan

EDUKATIF : JURNAL ILMU PENDIDIKAN

dwi wedayanthi

Dwight Howell

Revista Medica De Chile

Isidora Doggenweiler

Chemosphere

Wade Tucker

HAL (Le Centre pour la Communication Scientifique Directe)

Caterina Tiazzoldi

Clément Venco

Kaddour Meribout

Martine Roussel

Annals of K-Theory

Guihua Gong

AJHSSR Journal

Plastic and Aesthetic Research

Keith Casper

ghjtftgd ghhtrf

Vladimiro Miranda

Red Shine Publication

PARIJAT CHAKRABARTY

University of Stavanger, Norway

Kolace Silwimba

RELATED TOPICS

•   We're Hiring!
•   Help Center
• Find new research papers in:
• Health Sciences
• Earth Sciences
• Cognitive Science
• Mathematics
• Computer Science
• Academia ©2024

PhD Theses (Chemistry)

Recent submissions.

• No Thumbnail Available Item Synthesis of Furan, Pyran, Pyrrolidine, and Piperidine Scaffolds via Tandem Prins Cyclization Reactions ( 2023 ) Shit, Sudip Show more The content of this thesis has been divided into five chapters on the basis of results of experimental work performed during the complete course of the PhD tenure. The chapter 1 describes the tandem Prins cyclization reactions and their mechanisms to construct Furan, Pyran, pyrrolidine and piperidine scafolds in brief. The chapter 2 deals with the stereoselective synthesis of hexahydrofuro[3,4-b] furan-4-ol and its dimer via tandem Prins and pinacol rearrangement. The dimer was conveniently converted into its corresponding monomer using aqueous zinc(II) chloride in THF in quantitative yields. Chapter 3 describes synthesis of spiro[furan-2,1′-isoindolin]-3′-ones from 2‑(4- hydroxybut-1-yn-1-yl)benzonitriles and aryl aldehydes under the action of triflic acid. The plausible mechanism of the reactipon has been drawn on the basis of control experiments and literature evidence. The synthetic utility of the reaction was performed using Sonogashira reaction and click reaction conditions. In chapter 4, nitrile stabilized synthesis of pyrrolidine and piperidine derivatives via tandem alkynyl aza-Prins-Ritter reactions is described. In chapter 5, regio- and chemoselective synthesis of 3-(dihydrofuran-3(2H)-ylidene)isobenzofuran-1(3H)-imines via tandem alkynyl Prins- and intramolecular oxycyclization reaction is disclosed. The methodology was extended towards synthesis of its pyran derivatives. The post synthetic applications of the reaction were extended towards synthesis of furanylidene-isobenzofuranones in excellent yields. The mechanistic investigation of the reaction was performed on the basis of controlled experiment. Show more
• No Thumbnail Available Item Design, Synthesis and Investigations of Liquid Crystalline Organic Semiconductors ( 2023 ) Vishwakarma, Vinod Kumar Show more This thesis entitled “Design, Synthesis and Investigations of Liquid Crystalline Organic Semiconductors” describes simple and straightforward synthetic pathways, characterizations and applications of new LC/non-LC organic materials, with their potential applications in acid sensing, optoelectronic devices mainly OLEDs and OFETs applications. A brief overview of the chapters is given. Chapter 1, gives a general introduction to liquid crystals, characterization techniques and their applications. Chapter 2a addresses the synthesis and characterization of pyrazino[2,3-g]quinoxaline (PQ) derivatives where the central pyrazino[2,3-g]quinoxaline central core is substituted with eight peripheral flexible tails of varying lengths. The compounds with very short/branched peripheral chains did not stabilize any liquid crystalline phase, while the medium to long-chain homologs exhibited columnar phases. All the compounds exhibited a high molar extinction coefficient and bright greenish-yellow emission behavior in solution and solid state. One of the columnar liquid crystalline materials was used in the fabrication of host-guest OLED exhibited higher efficiency and bright green emission. Chapter 2b describes the sensing ability of the pyrazino[2,3-g]quinoxaline derivative to volatile trifluoroacetic acid in trace amounts (in parts per billion levels). The reversible detection of acid-sensing behavior was visually perceivable in both solutions as well as in the drop-casted film on a TLC paper-based strip. Chapter 3 explains new design strategies for the synthesis of donor-acceptor-donor pyrazino[2,3-g]quinoxaline carbazole derivatives with straight chain and branched alkyl peripheral chains. The electron-donating carbazole moieties that have been placed in the periphery showed a strong electron-donating ability, high molar extinction coefficients and lower band-gap, bright emission behavior in solution and solid-state, along with positive solvatochromism. One of the branched chain compounds was used in yellow and white OLEDs device fabrications. Chapter 4 describes the detailed synthesis of naphthalene and perylene-based imidazole derivatives with a new donor-π-acceptor-π-donor architecture containing ten flexible chains for solution-processable organic field effect transistor (OFETs) applications. These compounds showed a wide range of columnar LC behavior. One of these naphthalene and perylene bisimide derivatives were evaluated for their application in OFET devices exhibited high hole mobility values. Show more
• No Thumbnail Available Item Stimuli-responsive Turn-on Fluorogenic Donors of Hydrogen Sulfide (H2S) and the Prodrugs of Anti-cancer Compounds ( 2023 ) Mahato, Sulendar K Show more The dissertation entitled “Stimuli-responsive Turn-on Fluorogenic Donors of Hydrogen Sulfide (H2S) and Prodrugs of Anti-Cancer Compounds” consists of four chapters based on the results of experimental works performed during the complete course of Ph.D. research tenure. Show more
• No Thumbnail Available Item Stereospecific Opening and Cyclization of the Strained Ring Systems for the Synthesis of Six-Membered Heterocycles ( 2022 ) Das, Bijay Ketan Show more The thesis is divided into four chapters. The first chapter describes the general introduction to the opening and cyclization of the strained ring system for the synthesis of six-membered heterocycles. The second chapter focuses on the synthesis of piperazines and tetrahydropyrazines through stereospecific ring-opening and cycloisomerization of aziridines with N-propargylamines. Third chapter covers stereospecific synthesis of substituted 1,4-oxazine via Zn/Ag Relay catalyzed ring opening/hydroalkoxylation of oxiranes with N-propargylamines. Chapter four deals with the stereospecific assembly of tetrahydroquinolines via tandem ring-opening/oxidative cyclization of donor-acceptor cyclopropanes with N-alkyl anilines. Show more
• No Thumbnail Available Item Application of Modified Smooth Exterior Scaling Method to Study Auger and Shape Resonances in Different Atomic and Molecular Systems ( 2024 ) Banuary, Mwdansar Show more This thesis focuses on the application of modified smooth exterior scaling (MSES) as an efficient method to impose outgoing boundary conditions in e-atom and e-molecule scattering resonances. This is the first time that the MSES method has been applied to calculate energies and widths of Auger and shape resonances in three-dimensional many-body electron systems. MSES converts the divergent resonance wave functions into square integrable ones thereby making the study of temporary bound states (resonance states) amenable to bound state electronic structure methods. The main objective of this thesis is to formulate the MSES method in bivariational SCF and electron propagator methods. Show more
• No Thumbnail Available Item N- and O- Donor Ligands for Fluorometric and Colorimetric Detection of Metal Ions ( 2024 ) Bhattacharya, Araghini Show more This Thesis contains five chapters. Chapter 1 is the introduction which describes the contribution of metals in our daily lives and their adverse effects on human beings when exposed to excess amounts along with a brief elaboration of different detection techniques. Some recent developments in devising fluorescent and colorimetric probes are also discussed. In Chapter 2 the 2,4,5-tris(2-pyridyl)imidazole (L1H) molecule has been evaluated as a probe for dual sensing of Hg2+ and Cu2+ ions in EtOH/HEPES buffer medium (5 mM, pH = 7.34, 1:1, v/v). Probe L1H shows a good sensitive and selective turn off response in the presence of both Hg2+ and Cu2+ ions, which is comprehensible under long UV light. Its sensitivity was evaluated in different pH medium and in presence of other metal ions. Paper strip experiments and in-vitro cell imaging was done to know the sensitivity of the probe towards the metal ions in different environment. Chapter 3 describes the probe 2,6-di(2-pyridyl)-1,5-dihydroimidazo[4,5-f]benzimidazole (L2H2) and its sensing application towards metal ions. This probe could detect Zn2+ and Cd2+ ions in three different aqueous systems viz., water, DMSO/HEPES buffer (1:1, pH = 7.34, rt), and DMSO/water (1:1, rt). In water a “turn-on” response was observed for both metal ions, whereas in the latter two solvent systems, a ratiometric change in fluorescence maximum was observed. The detection limit of this probe was as low as 0.3 μM and 0.62 μM (in water) for Zn2+ and Cd2+ ions, respectively. In Chapter 4 synthesis and evaluation of a novel probe N'-(pyren-1-ylmethylene)benzo[d]imidazo[2,1-b]thiazole-2-carbohydrazide (L3H) as metal ion sensor was explored. It was found to be able to nanomolar detection of Pd2+ and Ni2+ ions by colorimetric change. This probe was also able to detect the presence of Pd2+ ions in drug samples and APIs without any major interference. LOD values were calculated to be 46.1 to 93.9 nM (4.9 to 6.0 ppb) for PdCl2 and 10.6 to 19.6 nM (1.1 to 2.1 ppb) for Pd(PPh3)2Cl2 and 9.301 nM for NiCl2 solutions. It also showed extremely good recovery of Pd2+ in presence of all the drug molecules. In Chapter 5 another novel ligand L4H based on 2,7-dichlorofluorescein was synthesized and evaluated for sensing applications. It was found to be able to detect Co2+ and Cu2+ ions differentially by different colouration of the solution in presence of these two metal ions. In situ Cu-complex of the ligand was utilised for quantification of amino acids like BSA and HSA proteins. Its differential selectivity paved the for molecular logic gate application. Show more
• No Thumbnail Available Item Computational investigation of excited state processes in ESIPT-based systems and vinylene-linked thiophene pyrrole ( 2024 ) Mawa, Ibanrishisha Show more The thesis focuses on understanding the mechanistic pathway in systems undergoing excited state intramolecular proton transfer and cis-trans isomerization. Unveiling the mechanism of these processes at an atomistic scale is of utmost importance as it would add to our understanding and assist in designing materials with better performance. These kinds of processes are observed in our everyday life such as the vision process in retinal chromophores, vitamin D production in humans on exposure to sunlight and mutation during DNA replication, etc. The application part of systems undergoing photoinduced processes are realized in the design and development of certain materials such as optoelectronic devices. The thesis has three working chapters. The first work is based on 1-hydroxy-2-acetonaphthone (HAN) due to the unsettled issues regarding the proton transfer process. In addition, the process of full photocycle including the non-radiative relaxation pathways is proposed. The second work highlights the effect of implicit solvents on the photoinduced processes in nitrile-substituted 2-(oxazolinyl)-phenols. Additionally, the mechanisms behind these two regiomers’ weakly emissive properties in the solvent phases are investigated. My last work involves the exploration of photoisomerization pathways in vinylene-linked thiophene-pyrrole system. Considering the computational cost for the dynamics study in the excited state, we have employed single-reference method such as time-dependent density functional theory (TDDFT) and algebraic diagrammatic construction scheme of second order (ADC(2)). However, multi-reference studies are also incorporated in our study wherever the single-reference methods fail. Show more
• No Thumbnail Available Item Properties and Potential Applications of Biomimetic and Bio-derived Nanofluidic Systems ( 2021 ) Konch, Tukhar Jyoti Show more The branch of fluid dynamic that explore the flow of liquid in structure constrained to nanometer size regime (1-100nm) is defined as nanofluidic. Fluidic transport in and around nanofluidic structures is dominated by interactions of otherwise weak effects such as the formation of electrical double layers (EDL), attractive or repulsive forces of charged species, and entropic barriers. Typically, transport of charged species through nanometer-sized channels are dominated by the overlapping electrical double layers. One of the major difficulties in designing nanofluidic devices is the inherent complexity. The overall transport characteristics are determined by the interplay of various nanoscale or even molecular level physical, geometric, and chemical factors. Biological ion channels, however, are known for their capability of elaborately manipulating these factors to regulate the transmembrane ionic flow, which plays a crucial role in a number of physiological processes. Mimicking the biological systems researchers has tried to demonstrate its artificial counterparts. In light of this feature, various ion-channel-mimetic smart 1D nanofluidic systems have been developed that can reproduce functions analogous to its parent biological systems. Although systematic research in single-pore devices makes the physical picture of this nanofluidic process much clear, it is still far from competent for practical applications. Toward practical applications, one major challenge is to extrapolate individual nanofluidic devices to macroscopic platform in a cost-efficient way. Interestingly solution to the above mentioned dilemma was also resolved from natural inspirations in the form of lamellar microstructure of nacre, in which soft materials (polysaccharides and proteins) are sandwiched between hard inorganic layers (aragonite platelets), forming an alternatively arranged layered structure. This novel method of material designing and large-scale integration of individual artificial nanofluidic channels into a macroscopic platform give birth a new research filed known as the 2D nanofluidics. Via a simple vacuum filtration process, colloidal dispersions of individual 2D nanosheets can be reassembled into a densely stacked multi-layered structure. The interstitial space between opposite 2D nanosheets can be treated as lamellar channels for mass and charge transport. Show more
• No Thumbnail Available Item Transition-Metal Catalyzed Regioselective C-C/C-Heteroatom Bond Formations: Access to Functionalized Arenes and Heterocycles ( 2022 ) Sarkar, Tanumay Show more The thesis is divided into four chapters. The first chapter illustrates a Ru(II)-catalyzed siteselective C-H acyloxylation of N-aryl-2-pyrimidines with carboxylic acids as the acyl source. The second chapter describes a Ni(II)-catalyzed oxidative C-H heteroarylation of arenes with azoles utilizing a removable oxazoline-based directing auxiliary. The third chapter deals with the Bi(III)- catalyzed annulation of 2-naphthols with N-sulfonylaziridines. The fourth chapter demonstrates (3+3)-cycloaddition of aziridines with diaziridines for the stereospecific synthesis of triazines under Fe(III)-catalysis. Show more
• No Thumbnail Available Item Ipso Nucleophilic Substitution on Electron Deficient Arene Systems ( 2024 ) Mondal, Sandip Show more The thesis entitled, “Ipso Nucleophilic Substitution on Electron Deficient Arene Systems” mainly focused on the development of greener and transition metal free methodologies for various alkylation reactions. The contents of the thesis have been divided into five chapters based on the results of experimental works performed during the research period. Show more
• No Thumbnail Available Item Design of Coatings Embedded with Tolerant, Tailored and Responsive Underwater Oil Wettability and Oil Adhesion ( 2023 ) Borbora, Angana Show more The anti-oil wettability of various naturally existing underwater creatures has inspired researchers to develop artificial super oil repellent interfaces for multiple applications in engineering, healthcare, and environmental remediation. In the past, several approaches were adopted to artificially fabricate underwater oil-repellent surfaces, formally known as underwater superoleophobicity, by co-optimizing hydrophilic chemical composition and rough micro/nano-structures on their surface. However, the earlier reported approaches in deriving underwater superoleophobicity were unable to associate some other essential properties, such as, physical and chemical durability, adaptive tuning of oil adhesion, and transparency in the prepared surfaces. Here, a facile 1, 4-conjugate addition reaction is exploited to derive covalently crosslinked chemically reactive coatings on various surfaces loaded with residual chemical functionalities that provide the opportunity to embed underwater superoleophobicity through appropriate post-covalent modifications. While the covalent crosslinking tailored mechanical property, the adequate chemical post-modification customized oil adhesion and optical transparency. The thesis entitled “Design of Coatings Embedded with Tolerant, Tailored and Responsive Underwater Oil Wettability and Oil Adhesion” is presented in six chapters. Chapter 1 introduces bio-mimicked underwater superoleophobic surfaces, the existing challenges associated with conventional artificial fabrication approaches, and the objectives of the thesis work. Chapter 2 demonstrates the fabrication of a dually reactive multilayer coating following the 1, 4-conjugate addition reaction and the post-covalent modification of the multilayer coating to immobilize highly sensitive bare micro-meter sized nematic liquid crystal (LC) droplets underwater for single LC droplet based repetitive sensing application. Chapter 3 accounts for the utilization of the dually reactive multilayer coating to develop various responsive underwater superoleophobic surfaces via post-modifications and their adaptive oil adhesion for sensing different amphiphilic (cationic, anionic and facial) molecules. Chapter 4 demonstrates the rational functionalization of the dual reactive multilayer coating to depict the highly selective raising of the oil contact angle (OCA) and rolling of a beaded oil droplet underwater in the presence of targeted and relevant toxic chemicals. Chapter 5 introduces a covalently crosslinked and chemically reactive sol-gel conversion process through the 1, 4-conjugate addition reaction to achieve a substrate-independent, mechanically durable, and optical transparent coating embedded with underwater superoleophobicity. Moreover, this approach allows to modulate mechanical property of highly deformable objects. Chapter 6 provides a brief summary and the future outlook of the work presented here. Show more
• No Thumbnail Available Item (A) Computational Study of human Islet Amyloid Polypeptide Aggregation and its Inhibition ( 2023 ) Roy, Rituparna Show more The aggregation of human islet amyloid polypeptide (hIAPP) stands at the nexus of Type II Diabetes (T2D) pathogenesis. In order to counteract the advancement of this disease, a possible therapeutic avenue is to curb the misfolding and aggregation of hIAPP. Within this thesis, we embark on the intricate journey of hIAPP aggregation, coupled with the myriad classes of compounds harboring the potential to impede this process. In Chapter I, a foundation is laid through the introduction of hIAPP and an array of different categories of inhibitors, each contributing to the modulation of hIAPP aggregation. A brief discussion of the molecular dynamics simulation methodology, which is a vital framework underpinning our study is followed. Thereafter, Chapter II takes the helm into venturing the different conformational states of an amyloid prone fragment of hIAPP, hIAPP20-29, via Markov State Modelling. Here, the transition pathway between the metastable states is analysed, which are crucial for the misfolding of hIAPP. Chapter III explores the influence of two small biological molecules on hIAPP aggregation. In Part (a), we have explored the effect of norepinephrine, which is a common neurotransmitter, on the amyloidogenesis of hIAPP. In Part (b), a new aspect of adenosine triphosphate (ATP), other than being the energy source for biochemical processes, is inquired. This chapter, thus, enlighten us about the diversity of the molecular structures that can modulate the aggregation of hIAPP and the effect of these structures on the activity of the inhibitors. Chapter IV turns the discourse towards peptides and peptidomimetics, probing their roles in shaping the aggregation narrative. Two such inhibitors are investigated, both of which are extracted from the amyloid core region of hIAPP, i.e., N22FGAIL27. In Part (a), this hIAPP fragment is replaced with all D-amino acids, and is used to prohibit the self-assembly of full-length hIAPP. In Part (b), a conformationally restricted element, aminobenzoic acid is incorporated into NFGAIL, by replacing Ile26 and/or Gly24 residues. Here, three different isomers of aminobenzoic acid is used, i.e., (β, γ, δ). β- and γ- containing peptidomimetics successfully prevent the aggregation of hIAPP, but δ- peptidomimetics promote it, highlighting the contrasting behaviour of the isomers. Hence, in this chapter, we have conveyed the effect of stereochemistry of the amino acid residues or modified organic moieties on the inhibitory potential of peptides or peptidomimetics. A novel dimension unfurls in Chapter V, where the alliance between boron nitride nanomaterials and hIAPP aggregation is explored. The curvature of the nanomaterials is observed to have an impact on their interaction site with hIAPP. Finally, Chapter VI unfurls a tapestry of conclusions, weaving together the diverse threads from our journey. In unity, this thesis stands as an ardent exploration, deciphering the aggregation pathway of hIAPP and unveiling a constellation of agents poised to intervene. The information regarding the structure and activity of the various inhibitors provides a holistic comprehension of the crucial molecular scaffolds and properties required to design drugs for combatting T2D's relentless advance. Show more
• No Thumbnail Available Item Effect of the Position of Geminal Di-Substitution of g Amino Acid Residues on their Conformational Preference ( 2023 ) Debnath, Swapna Show more This thesis investigated the role germinal di-substitution at various backbone positions of the gamma amino acid residue on their conformational preferences. The thesis consists of 5 chapters. The first chapter describes the gamma amino acids and their conformations reported in the literature. Chapters 2-5 describe the investigations carried out in this thesis, which includes the incorporation of gamma amino acid residues (g2,2,g3,3 and g4,4) in the peptides. Chapter 2, describes the structures and assemblies in the solid and solution state of different derivatives of gamma amino acid residue. The structures and assemblies in the solid state are reported to be different for the three amino acid residues. The position of the backbone di-substitution is shown to drive the assembly in the solid state but not in solution. In the Chapter 3, three gamma amino acid residues were incorporated in all 􀀀 amino acid containing model helical peptide sequences (tri, hexa and nona petides) and compared their relative helical propensity. The C12 helical conformation diminished as: g4,4 g3,3 g2,2. Helices with a central 􀀀 amino acid residue was shown to adopt mixed 10/12 helices of both handedness (left and right) in both solid and in solution state. Nona peptides containing g3,3 and g2,2 amino acid residues adopted an unusual ambidextrous helical conformation in the solid and in solution state. The ambidextrous conformation was stabilized by a water mediated hydrogen bonding. Ambidexterity was not observed in the nona-peptide containing g4,4 amino acid residues, likely due to the absence of the key water molecule in the structures. Chapter 4 describes the propensity of these three amino acid residues in being able to nucleate an isolated expanded C12 B-turn motif. Chapter 5 studies the ability of these amino acid residues in nucleating 􀀀 hairpin conformation. Both C12 􀀀-turn and 􀀀-hairpin conformation was favoured by g3,3 and g4,4 favoured, whereas g2,2 failed to nucleate either of them due to unfavourable steric contacts. This thesis reported conformational preference of the three differently di-substituted 􀀀 amino acid residues, in the solution and in solid states by primarily using NMR, CD and X ray crystallography. In collaboration with the computational lab, ab initio calculations have also been done to understand the energetics of conformational preference. The conclusions are very well supported by experiments and computations. The thesis showed how a position of disubstitution (in the g amino acid backbone) determines its conformational preference by fine-tuning the energetics. The results are useful for peptidomimetics and rational design of peptides with various architectures. Show more
• No Thumbnail Available Item Creating Life-like Transience in Synthetic Vesicles ( 2022 ) Das, Saurav Show more The thesis "Creating Life-like Transience in Synthetic Vesicles" explores several techniques and approaches for imbuing life-like non-equilibrium features in synthetic vesicular systems and their potential biomimetic applications in laboratory settings. Show more
• No Thumbnail Available Item Effect of Pyridine and Imidazole Functionality on Chiral Resolution, Solution Spin State and Electrochemistry within Ni (II) and Fe (II) Complexes ( 2022 ) Bhattacharya, Sounak Show more This thesis work stems from our quest to find a simple way to recognize an enantiomer from a racemic mixture using coordination bond. To do that, we choose to use Ni (II)(high-spin) and Fe (II) (low-spin) complexes of chiral bidentate Schiff-base ligands. Observations on Fe (II) complexes led to finding complexes that show high- spin <--> low-spin transitions in solution. Digging deeper with more complexes along with a host of electrochemical and spectrometric tools, we ended up finding an intimate relationship between donor groups, redox potential, and spin-state. The effect of replacing pyridine with imidazole on redox and the spin-state properties discussed in the thesis is relevant to biomimetic chemistry. Imidazole group is a part of L- histidine amino acid, ubiquitous in metalloenzyme active sites. On the other hand, pyridine donor is typical in ligands related to biomimetic chemistry. Show more
• No Thumbnail Available Item Selective C-H and C-C Bond Functionalization of Benzo-Fused N-Heteroaromatic Compounds ( 2022 ) Sarmah, Bikash Kumar Show more The present thesis, entitled “Selective C-H and C-C Bond Functionalization of Benzo-Fused N-Heteroaromatic Compounds” is divided into five chapters based on the results obtained from the experimental works during the course of PhD research period. Show more
• No Thumbnail Available Item Exploring the Potential of Homogeneous Ru-SNS/NNS Complexes and Heterogeneous Ru-Hydrotalcite in De(hydrogenative) Transformations ( 2023 ) Sardar, Bitan Show more The contents of the present thesis entitled as “Exploring the Potential of Homogeneous Ru-SNS/NNS Complexes and Heterogeneous Ru-Hydrotalcite in De(hydrogenative) Transformations” have been divided into five chapters. The first chapter contains a brief literature study related to various de (hydrogenative transformations) and the last four chapters were based on the results achieved from the experimental works performed during the entire course of the PhD research program. Chapter 1 contains a brief introduction to the literature review of acceptorless dehydrogenation and borrowing hydrogen reaction of alcohols via homogeneous catalysis and heterogeneous catalysis. In 21st century, the rapid depletion of fossil fuels and growing environmental concerns urges chemists and chemical industries to search for alternative raw materials and to develop new methodologies to produce sustainable chemicals and important building blocks. In this regard, biomass-derived alcohols was found to be best candidate, as they are non-toxic in nature. Moreover, alcohols are considered renewable starting materials that can be used in organic synthesis for various organic transformations and the preparation of commodity chemicals. In this context, “acceptorless dehydrogenation (AD)” and “borrowing hydrogen (BH)” catalysis plays a key role. These approaches are sustainable because this process liberates water and in some cases (i.e., AD) molecular hydrogen as clean by-products. And, these types of reactions could be successfully performed by various types of homogeneous and heterogeneous catalysts. Show more
• No Thumbnail Available Item Interaction and Synchronization of Spiral Waves in a Reaction-Diffusion System ( 2023 ) Kalita, Hrishikesh Show more Over the past few decades, spirals have attracted a lot of interest. From a spinning galaxy to a swarm of honeybees, rotating spirals are widespread in nature. Their widespread presence in nature has made the study of spiral waves relevant across various disciplines. In physical systems like fluid flows, liquid crystals, galactic formations, etc., in biological systems like the heart, chicken retina, neocortex, slime mould, etc., in chemical systems like the Belousov-Zhabotinsky (BZ) reaction system, the Briggs-Rauscher reaction, some simple precipitation processes, the oxidation of CO on platinum surfaces, etc., scientists have observed and studied spiral waves. Despite these studies, the ambiguity of spiral waves has prevented scientists from developing a comprehensive hypothesis Show more
• No Thumbnail Available Item Aggregation Aptitude in Rigid and Flexible Molecular Systems: Comparative Photophysical and Analytical Studies ( 2023 ) De, Sagnik Show more This thesis elucidates the important consequences in comprehension of aggregation outlook of flexible and rigid frameworks and their response towards environmentally and biologically relevant analytes. L1-L3 is designed which shows a comparative aggregation aptitude with chain length variation in amphiphiles. The entire photophysical study on aggregation process is dealt with. Then, these synthesized amphiphiles are used in creating hydrophobic surfaces due to their inherent property of hydrophobicity. Additionally, the concept of Photoinduced Electron Transfer or PET is applied in the detection of nitro antibiotics via fluorescence quenching. This chemo sensing is probed in biofluids viz; simulated gastric and body fluid. Next, a layout is provided where a comparative study between an amphiphile and a non-amphiphile is presented. The compounds designed and synthesized were substituted urea and amide (L4 & L5). Studies on aggregation-induced emission are shown by a binary solvent system DMF-Water. Morphological change is depicted on solvent switching by electronic microscopy imaging. Both solid and solution state emissive property is described. A unique photophysical prospect is shown in this piece of work i.e., light harvesting. Förster resonance energy transfer or FRET mechanism delivers the basis for this light-harvesting phenomenon between the amphiphile and a commercial dye; Rhodamine. Again, PET is applied to detect nitro explosives in water is demonstrated. This detection proceeds via disaggregation of the aggregated state. In the allied chapter, functionalization of amphiphile was done: a comparative outline on substituted urea and thiourea (L4 & L6). Apart from describing aggregational features through spectroscopy and microscopy, an edge on the chemo-sensing property is done. The thiourea selectively recognizes Hg (II) ions in an aqueous solution due to the soft-soft interaction between the sulfur atom and the heavy metal. Turn-On or fluorescence emission enhancement is achieved even in the presence of heavy metal during the chemosensing process. The toxic metal ion interaction causes disaggregation of the aggregated amphiphile confirmed through DLS and FESEM experiments. The chemo-sensing experiments are done in various real samples. Moreover, The Hg(II)-amphiphile ensemble detects sulfide ions in the water among all other sulfur-containing anions and amino acids. Show more
• No Thumbnail Available Item Reactivity Studies of 4-Hydroxydithiocoumarin: Design & Synthesis of Novel Bioactive Molecules ( 2022 ) Mondal, Santa Show more The thesis entitled “Reactivity Studies of 4-Hydroxydithiocoumarin: Design & Synthesis of Novel Bioactive Molecules” has been compiled into six chapters based on the experimental results and findings carried out by me during the entire research period. Chapter 1 provides a brief overview of organosulfur compounds and their importance. In addition, the reason for choosing 4-hydroxydithiocoumarin as the key starting material for the synthesis of novel bioactive molecules. Chapter 2 elaborates on the synthesis of 3-sulfenyl derivatives. Chapter 3 describes the synthesis of α-thiocarbonyl compounds. Chapter 4 illustrates the synthesis of vinyl sulfides and thioethers. Chapter 5 demonstrates the synthesis of 1,4-oxathiin derivatives. Chapter 6 elucidates the synthesis of β-hydroxysulfides and β-aminosulfides by ring-opening reactions. Show more
• 1 (current)