Trap Polarity and the p/n Asymmetry in Oxidised DNTT: A Frontier-Shift Rule
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThis is an interesting article that computationally screens many (39) possible defects that could occur in DNTT. They find a clear sign rule for traps: pi-donating hydroxyl raises HOMO and yields a hole trap, whereas pi-accepting carbonyl or quinone lowers energy levels and gives deep electron trap. These results explain the observed asymmetry of hole and electron transport in air exposed DNTT: hole traps are shallower and thus do not strongly impact hole mobility whereas the electron traps are significantly deeper and thus would destroy electron mobility. While the semi-empirical method used is not very accurate, I agree that it is valuable for the trends. The authors extensively discuss the limits of the calculations but also point out the value of their calculations. I conclude that the methodology is appropriate for this work.
This work provides a better understanding of how various degradation products result in traps of varying depth, which in turn nicely explains the asymmetry effect of degradation on electron and hole transport. The results thus fill a gap between experimental results (where for e.g. we don’t know why electron mobility is destroyed), and device-specific theoretical results, which tend to focus on a few traps as opposed to all possible ones. It fills a gap of knowledge about traps, which is important for the design of better organic semiconductors for transistors.
I think this is original and I am currently not aware of other papers addressing this. The conclusions are consistent with the results and the authors pinpoint and discuss the limits of their calculations in great length. Thus, they do not claim anything more than what the results show. The figures and Tables are clear and the references are appropriate. Overall, I believe it is a good fit for Materials and should be published after addressing these comments:
They are tentatively assigning an experimental hole trap based on their semi-empirical studies. I would recommend doing a higher-level calculation for the suspected assignment to confirm. That one calculation does not seem outside the scope of this paper.
The presence of results in the Materials and Methods section is a little strange and not what I would expect. I would recommend rewriting such that the results section presents the results, as is usual in scientific literature.
Author Response
Comment 1: They are tentatively assigning an experimental hole trap based on their semi-empirical studies. I would recommend doing a higher-level calculation for the suspected assignment to confirm. That one calculation does not seem outside the scope of this paper.
Response: We agree, and we have carried out the calculation rather than argue that it lies outside the scope. Single-point hybrid DFT (B3LYP-D3(BJ)/def2-TZVP, with the RIJCOSX approximation) was performed on the GFN2-xTB geometries of fourteen representative closed-shell defects, including the peri hydroxyl (12-OH) that constitutes the suspected assignment of the measured hole trap. The DFT calculations reproduce the sign of the frontier-level shift and the ordering of the defects: 12-OH remains a hole trap (ΔHOMO = +0.14 eV at DFT, against +0.18 eV at xTB), and the quinones and the diketone remain deep electron traps. For the LUMO shifts the two methods correlate with a Pearson coefficient r = 0.99 across the donor/acceptor defects, so that the sign rule and the p/n asymmetry are confirmed to be robust to the level of theory. This validation is described in the new Section 4.4 (“Methodological Standing: Validity and Limits of the Screening”), with the comparison summarised in Figure 7 and Table 2; the DFT protocol has been added to Section 2.3, and corresponding statements to the Abstract, Introduction and Conclusions. In the interest of full transparency we also report the one class (the endoperoxides) for which the semi-empirical and DFT pictures disagree systematically.
Comment 2: The presence of results in the Materials and Methods section is a little strange and not what I would expect. I would recommend rewriting such that the results section presents the results, as is usual in scientific literature.
Response: We agree and have restructured accordingly. The experimental result - the DLTFS-measured activation energy Ea = 0.255 ± 0.011 eV and the corresponding Arrhenius plot (now Figure 2) - has been moved to a new Results subsection, “Experimental Observation: A Hole Trap Resolved by DLTFS.” The Materials and Methods section (Section 2.2) now contains only the measurement protocol (device, temperature range, bias, filling-pulse and rate-window settings, and how Ea is extracted), with a forward reference to the Results.
Reviewer 2 Report
Comments and Suggestions for AuthorsThis manuscript by Martin Weis and coworkers tries to connect atmospheric oxidation products of DNTT with the polarity of charge traps. The main idea is that hydroxyl-type defects can raise the HOMO, and therefore behave as hole traps, while carbonyl/quinone-type defects lower the frontier levels and behave as deep electron traps. Overall, the conceptual framework is useful. Here are several comments for improving the manuscript.
1. The main conclusions rely heavily on GFN2-xTB single-molecule, gas-phase calculations. The authors themselves note that the absolute orbital energies are not reliable and that only relative shifts are meaningful. It would be important to validate at least the key defects, such as hydroxyl, quinone, diketone, and hydroperoxide species, using a higher-level DFT method or a simple crystal-embedding/polarization model. At least one example would be very helpful.
2. The DLTFS experiment gives a hole-trap activation energy of 0.255 eV, but this alone cannot identify the chemical origin of the trap. A further discussion on this issue should be provided.
Author Response
Comment 1: The main conclusions rely heavily on GFN2-xTB single-molecule, gas-phase calculations. […] It would be important to validate at least the key defects, such as hydroxyl, quinone, diketone, and hydroperoxide species, using a higher-level DFT method or a simple crystal-embedding/polarization model. At least one example would be very helpful.
Response: We have carried out exactly this validation, and for considerably more than one example. Single-point B3LYP-D3(BJ)/def2-TZVP calculations were performed on fourteen representative closed-shell defects spanning all of the classes named by the reviewer - hydroxyls (five, including the peri 12-OH), the hydroperoxide, four quinone/diketone species and epoxides - together with the endoperoxides. Across the donor/acceptor defects the DFT and xTB frontier-level shifts correlate closely (r = 0.99 for the LUMO shift), and the polarity classification is reproduced throughout: every hydroxyl remains a hole trap and every quinone and the diketone a deep electron trap. We additionally verified basis-set consistency (def2-SVP versus def2-TZVP). This is the subject of the new Section 4.4, Figure 7 and Table 2 (method in Section 2.3). We report transparently that the endoperoxides are the one motif for which xTB and DFT disagree systematically, and we stress that it is the relative shifts - not the absolute levels - that are validated, consistent with the framing of the paper.
Comment 2: The DLTFS experiment gives a hole-trap activation energy of 0.255 eV, but this alone cannot identify the chemical origin of the trap. A further discussion on this issue should be provided.
Response: We fully agree, and this point is central to our framing. The manuscript deliberately withholds a unique microscopic assignment; the dedicated subsection “The Experimental Anchor, and Why a Unique Identification Is Withheld” (Section 4.2) discusses precisely this issue. There we note that a single activation energy - without its capture cross-section, and without independent chemical or structural characterisation of the film - cannot single out one candidate from the dense band of hydroxyl-related states that the map predicts near this energy, and that structural or contact-related origins cannot be excluded. Accordingly, the measured trap is described only as “consistent with a hydroxyl-related origin,” and the map is said to constrain the polarity and plausible chemistry of the trap rather than its precise identity. This caution is stated consistently in the Abstract, the Results, Section 4.2 and the Conclusions.
Reviewer 3 Report
Comments and Suggestions for AuthorsThe manuscript entitled “Trap Polarity and the p/n Asymmetry in Oxidised DNTT: A Frontier-Shift Rule” addresses a potentially interesting topic for the organic electronics community, particularly for researchers working on organic field-effect transistors. Overall, the study appears to be carefully conducted. However, several aspects should be improved to make the manuscript more accessible, self-explanatory, and easier to follow.
- The topic addressed in this work is highly specific. Therefore, schematic illustrations of the oxidation processes, defect formation mechanisms, or representative molecular interactions would significantly improve the readability of the manuscript. Such graphical information would help readers understand the physical and chemical processes being discussed.
- The study focuses exclusively on DNTT. However, current research in organic electronics increasingly uses functionalized DNTT derivatives, such as C10-DNTT and S-DNTT, because they generally offer improved processability, film formation, and device performance. The authors should briefly discuss whether the proposed frontier-shift rule and the conclusions obtained for DNTT could also be applicable to these technologically relevant derivatives.
- The molecular symmetry of DNTT should also be presented more clearly. In the current manuscript, the carbon atoms are numbered from 1 to 12. However, because of the molecular symmetry, several carbon positions are chemically equivalent, and only six distinct carbon environments are present. The authors should consider revising the numbering scheme to reflect this equivalence. For example, symmetry-related atoms could be labelled as 1 and 1′, 2 and 2′, and so forth. The same notation should be consistently applied throughout the manuscript.
- Representative chemical structures of the hydroxyl, epoxide, endoperoxide, quinone, and diketone defects should be included in Table 1 and/or Figure 6. Each structure should be assigned a clear identifier that can be directly related to the corresponding data points, calculations, and discussion. Although the nomenclature used may be familiar to synthetic chemists, not all researchers working in organic electronics will immediately recognize these structures. Their explicit representation would therefore make the manuscript considerably more accessible to a broader readership.
- The calculations demonstrate how different defects affect the electronic properties of DNTT. However, not all defects are necessarily equally probable or energetically favourable. The authors should discuss the relative formation energies, thermodynamic stability, or expected occurrence of the investigated oxidised structures. Without this information, it is difficult to assess which defects are most likely to be experimentally relevant and which should mainly be regarded as theoretical limiting cases.
Author Response
Comment 1: Schematic illustrations of the oxidation processes, defect formation mechanisms, or representative molecular interactions would significantly improve the readability of the manuscript.
Response: The manuscript already contains two such schematics, which we now signpost more clearly: Figure 3 presents the photo-oxidation pathway and the defect-formation mechanism (the Type I / Type II branches and the peri-centred sequence), and Figure 5 is a schematic level diagram of the sign rule (a π-donor raising the HOMO versus a π-acceptor lowering the levels). We have additionally added representative chemical structures of the defect classes as a new Figure S2 (see our reply to the next comment).
Comment 2: The study focuses exclusively on DNTT. […] The authors should briefly discuss whether the proposed frontier-shift rule and the conclusions obtained for DNTT could also be applicable to these technologically relevant derivatives (e.g. C10-DNTT and S-DNTT).
Response: We have added a discussion of this point to the Implications (Section 4.6). Because the sign rule is governed by the donor/acceptor character of the oxygen functionalisation acting on the conjugated core, it is expected to carry over to the side-chain- and core-modified derivatives: in the dialkyl Cn-DNTTs and the diphenyl variant the substituents govern solubility and film packing but leave the π-core essentially intact, while the heavier-chalcogen (seleno) analogues preserve the same donor/acceptor logic, with only their absolute frontier levels shifted. The polarity classification should therefore transfer to these materials, although the precise trap depths would have to be recomputed in each case.
Comment 3: The molecular symmetry of DNTT should be presented more clearly. […] only six distinct carbon environments are present. The authors should consider revising the numbering scheme to reflect this equivalence (e.g. 1 and 1′), applied consistently throughout.
Response: We have clarified the symmetry in the caption of Figure 1: DNTT belongs to the C2h point group (with an inversion centre), so that the ring C–H positions occur in symmetry-equivalent pairs and only six distinct C–H environments exist. These six symmetry-unique sites are precisely the ones enumerated by the frontier-reactivity analysis (Figure 4), so the equivalence the reviewer notes is already reflected in the analysis. We have chosen to retain the standard IUPAC locants rather than introduce a 1/1′ scheme, because the IUPAC numbering is unambiguous, is applied consistently throughout the text, tables and figures, and links each defect directly to its data; the new symmetry statement conveys the equivalence without the potential for confusion that two parallel numbering schemes could introduce. We hope the reviewer finds this an acceptable resolution.
Comment 4: Representative chemical structures of the hydroxyl, epoxide, endoperoxide, quinone, and diketone defects should be included in Table 1 and/or Figure 6. Each structure should be assigned a clear identifier that can be directly related to the corresponding data points, calculations, and discussion.
Response: We have added a new Figure S2 in the Supplementary Materials showing representative chemical structures of pristine DNTT and of each closed-shell class — the peri hydroxyl (12-OH), the peri hydroperoxide (12-OOH), the 7,12-para-quinone, the 5,12-diketone, the 9,10-epoxide and the 7,12-endoperoxide. Each structure carries the same identifier used in Table 1, Figure 6 and the numerical map (Table S1), so that it can be related directly to the corresponding data point and discussion. A pointer to Figure S2 has been added to the caption of Table 1. We placed the structures in the Supplementary Materials to avoid crowding Table 1 and Figure 6, while keeping them explicitly linked to both.
Comment 5: The authors should discuss the relative formation energies, thermodynamic stability, or expected occurrence of the investigated oxidised structures.
Response: We have added a paragraph (Section 4.1) on the relative likelihood of the mapped defects. The frontier-reactivity index provides an a priori ordering of the points of attack: the peri positions carry by far the largest HOMO population and are therefore the most susceptible to electrophilic oxidation, so that the peri hydroxyl and the peri-derived carbonyl/quinone - the species that anchor the analysis - are expected to dominate the early-stage defect population, whereas the multiply-oxidised species and the high-locant single defects are best regarded as later-stage or less-probable products that delineate the boundaries of each electronic regime. We note explicitly that a quantitative formation-energy ranking would require explicit reference states for O2 and H2O together with a treatment of the reaction barriers, which lies outside the scope of the present frontier-orbital screening; the reactivity index is therefore used only as an ordinal proxy for susceptibility.
Round 2
Reviewer 3 Report
Comments and Suggestions for AuthorsThe authors have satisfactorily addressed most of my previous concerns and effectively resolved the main weaknesses identified in the earlier version of the manuscript. Overall, the revisions have significantly improved the quality and clarity of the work, and I therefore recommend the manuscript for publication.
I have only one minor suggestion regarding Figure 2. The current flow diagram would be sufficiently informative for a manuscript aimed primarily at chemists. However, given the broader scope and target audience of the present work, I recommend including the chemical structures of the molecules involved. This addition would make the figure more self-explanatory and facilitate its interpretation by readers from different scientific backgrounds.
Author Response
Comment: I have only one minor suggestion regarding Figure 2. The current flow diagram would be sufficiently informative for a manuscript aimed primarily at chemists. However, given the broader scope and target audience of the present work, I recommend including the chemical structures of the molecules involved. This addition would make the figure more self-explanatory and facilitate its interpretation by readers from different scientific backgrounds.
Response: We thank the reviewer for this helpful suggestion, with which we fully agree. We have revised Figure 2 by adding, to each node of the scheme, the chemical structure of the corresponding species alongside the existing label and carbon-position annotation. Pristine DNTT, the peroxyl and aryloxyl/carbonyl radicals, the hydroperoxide, the hydroxyl, the 7,12-endoperoxide, the 7,12-para-quinone and the 5,12-diketone are now shown explicitly, so that the site of oxidation and the nature of each product can be recognised at a glance. The colour legend (hole trap, electron trap, gap-widener, transient radical) has been retained. We hope that, in this form, the figure is self-explanatory for readers from different scientific backgrounds.
