Polymethine materials with solid-state third-order optical susceptibilities suitable for all-optical signal-processing applications

Abstract

Judicious substitution of chalcogenopyrylium-terminated polymethine dyes with sterically demanding groups ameliorates the deleterious effects of aggregation on the optical properties of these materials in the solid state, facilitating high-number-density films that exhibit an unprecedented combination of nonlinear optical properties with low linear and nonlinear losses.

---

Conceptual insights

This work provides a conceptual advance in the field of nonlinear optics by demonstrating how molecular engineering can be used to obtain materials that exhibit the requisite linear and nonlinear optical properties for phase-based all-optical signal processing (AOSP) at telecommunications wavelengths. Although cyanine-like polymethine dyes have previously been identified as suitable candidate chromophores on the basis of their molecular properties in dilute solution, translating those properties to high-chromophore density films or solutions, as would be required to make AOSP a reality, has proved problematic owing to the tendency of these chromophores to aggregate. The conceptual advance here over previous approaches to this problem, such as those involving dendrimers, is the use of several moderately sized rigid substituents that necessarily project above and below the plane of the π-conjugated chromophore to suppress aggregation. This strategy leads to solution-like absorption spectra in films, without excessively diluting the active chromophore; in addition to affording materials with an unprecedented combination of linear and nonlinear optical properties suitable for AOSP, the same approach may be applicable to other areas in photonics and electronics where solution-like optical and electronic properties are required in solid films.

The increasing demand for high-speed data processing and advances in photonic device platforms that can support all-optical signal processing (AOSP) applications, such as silicon–organic hybrid waveguide devices^1,2 and liquid-core optical fibers,³ provide a great opportunity for third-order nonlinear optical (NLO) materials to play a role in future telecommunication technologies. Candidate materials for AOSP applications based on the nonlinear refractive index (such as all-optical switching) must, at the wavelength of interest, meet a number of requirements. Firstly, the nonlinear refractive index, n₂, which is proportional to the real part of the third-order susceptibility, Re(χ⁽³⁾), should be large in magnitude; secondly, optical losses from two-photon absorption (2PA) should be low, specifically sufficiently low that the 2PA figure-of-merit (FOM), |Re(χ⁽³⁾)/Im(χ⁽³⁾)|, exceeds ca. 12;⁴ thirdly, linear optical losses from one-photon absorption (1PA) and/or scattering should also be low; and, finally, the materials should be highly photostable.⁵ Although there is a paucity of materials with a suitable combination of these properties for AOSP at telecommunications wavelengths (1300–1550 nm), the molecular optical properties of cyanine-like polymethines are particularly promising. They exhibit the largest magnitudes of the molecular third-order polarisability, |Re(γ)|, for a given conjugation length,⁶ and their relatively narrow, widely spaced 1PA and 2PA bands allow for near-resonant enhancement of Re(γ), without incurring large linear or nonlinear losses.

We have recently demonstrated that chalcogenopyrylium-terminated cyanine-like polymethines such as 1 and 2 (Fig. 1) exhibit solution values of |Re(γ)| and |Re(γ)/Im(γ)| at telecommunication wavelengths⁷ that, if retained in high-number-density materials, would satisfy the |Re(χ⁽³⁾)| and 2PA-FOM criteria for AOSP. However, translating molecular optical properties to highly concentrated solutions or films is challenging: the electronic structure of cyanine-like polymethines responsible for their large |Re(γ)| values also leads to extremely large linear polarisabilities, and, in turn, strong London dispersion forces, which can drive interchromophore aggregation. Aggregation can dramatically modify the electronic structure, introducing new states arising from intermolecular coupling, which can lead to linear and nonlinear absorption spectra no longer resembling those of the monomeric species, reduce |Re(γ)|, increase linear loss, and, through narrowing or closing of the 2PA transparency window, decrease the 2PA-FOM.⁸ Functionalisation of “end” (R, Fig. 1) and/or “front” (R′) positions of thiopyrylium-terminated polymethines with Fréchet-type dendrons has been found to suppress aggregation effects to some extent; however, films still exhibited unacceptably large linear losses in the telecommunications region.⁹

Fig. 1 Chemical structures of chalcogenopyrylium heptamethines previously reported to exhibit promising molecular-level optical properties for AOSP applications (1 and 2), synthesised with the aim of reducing aggregation effects in the solid state by introduction of bulky out-of-plane substituents into the “end” (R), “front” (R′), and/or “back” (R″, R‴) positions (3–9), and studied in molecular dynamics simulations (M1 and M2).

Here, we describe an approach to reduce aggregation in chalcogenopyrylium-terminated heptamethines: “end”, “front” and/or “back” positions (Fig. 1) are substituted with groups that (i) are moderately-sized, (ii) are fairly rigid and attached to the chromophore without intervening flexible groups, and (iii) project above and below the plane of the polymethine π-system,¹⁰ in order to prevent close approaches between the π-systems of adjacent chromophores. We discuss a series of such chromophores with different substitution patterns (Fig. 1), compare their linear and nonlinear optical properties, and describe how this strategy has led to chromophores with molecular optical properties that can be successfully translated to high-concentration films (containing 50–100% chromophore by weight) with large |Re(χ⁽³⁾)| values, good 2PA-FOM, and low linear optical losses at 1550 nm.

The “end”, “front”, and “back” groups chosen are shown in Fig. 1. We initially focused on thiopyrylium, rather than selenopyrylium, dyes, for synthetic convenience. The [B(3,5-C₆H₃(CF₃)₂)₄]⁻ counterion was used because it renders chalcogenopyrylium polymethines highly soluble in many solvents and in polymer host materials, and facilitates their purification.⁹ Moreover, ion pairing, which can result in loss of cyanine-like optical properties, can be minimised by using large weakly polarising anions, such as tetraarylborates.¹¹ The synthesis of the dyes is described in the ESI.†

Fig. 2 shows the 1PA spectra for some of the dyes in dilute solution and in films at 50 wt% in an amorphous polycarbonate (APC) host; spectra of additional examples are shown in Fig. S3 in the ESI.† The thin-film spectrum of 2/APC exhibits broad features significantly displaced to both the low- and high-energy sides of the solution absorption band; batho- and hypsochromic shifts are generally attributed to J- and H-aggregation, respectively,^12,13 while the large bandwidths seen here likely reflect a distribution of aggregate geometries. Replacement of the phenyl “end” (R) groups of 2 with out-of-plane, non-conjugated tert-butyl groups results in a low-energy absorption maximum for 3/APC similar to that observed in the solution spectrum, while substantially suppressing the low-energy tail. Use of the large out-of-plane 3,6-di(tert-butyl)carbazol-9-yl group in the “front” (R′) position tends to suppress broadening on the high-energy side of the film spectra (4vs.2 in Fig. 2; see also 5vs.3 in Fig. S3†). In the case of 4/APC (R = Ph), there is also a pronounced low-energy shoulder rather than the broad tail of 2/APC; we attribute this to a more well-defined J-aggregate geometry in 4/APC. Bulky groups in the “back” position(s) (R″, R‴) primarily affect the film spectra on the high-energy side of the main peak (6–9). In summary, bulky “end” groups drastically reduce the J-aggregate-like absorption and the “front” groups (and to a lesser extent “back” groups) reduce the H-aggregate-like absorption. Importantly, when bulky R, R′, and R″ groups are all present, the spectra of the APC blend films are narrowest and most solution-like. Similar trends are observed in the spectra of films of the neat chromophores (Fig. S3†).

Fig. 2 Normalised absorption spectra of chalcogenopyrylium heptamethines in (a) dilute chloroform solution and (b) in 50 wt% blends with APC.

Atomistic molecular dynamics simulations of the amorphous bulk polymethine/counterion systems (see ESI† for details) give insight into the effect of bulky substituents on the polymethine aggregate geometries. These geometries were characterised by considering the relative positions and orientations of the molecular long axes in polymethine pairs. For pairs with intermolecular backbone-to-centre distances (radial distances) less than 6 Å, the offset of the molecular centres along the long axis of one polymethine and the torsion angle between the long axes were considered (Fig. 3a). This approach distinguishes between perpendicular aggregates (large torsion angles) and parallel aggregates in either H-like (small offset) or J-like (large offset) aggregate geometries.

Fig. 3 (a) Schematic showing geometric parameters describing the polymethine–polymethine interaction geometries and probability distribution of aggregate geometries for M1 (b) and M2 (c) from molecular dynamics simulations. Two molecules are considered to aggregate when their radial distance is below 6 Å. The colour scale (far right) indicates the probability of finding aggregates, with white indicating no aggregates and darker colors corresponding to larger aggregate probabilities; a probability of one corresponds to the average bulk density of polymethine pairs.

Model compound M1 forms many aggregate structures, each polymethine having, on average, 2.6 neighboring chromophores within a 12 Å centre–centre offset along the length of the polymethine chain. The most common aggregate geometries are parallel pairs with short offsets, as indicated by the darker region near the bottom left corner of Fig. 3b. This is consistent with the H-aggregate-like film maximum seen in film spectra of 2 (Fig. 2 and S3†). In contrast, M2 forms essentially no aggregates, as indicated by the nearly white areas throughout most of Fig. 3c, consistent with the effectiveness of the bulky substituents of 6 largely suppressing the effects of aggregation on its spectra in films.

Table 1 compares the solution molecular NLO parameters for 3–9 at 1550 nm. Compound 2 possesses large |Re(γ)| and |Re(γ)/Im(γ)|.⁹ At the concentrations employed (<5 mM), aggregation is not expected to be significant and so variations in Re(γ) and Im(γ) can largely be ascribed to substituent effects on the S₀ → S₁ (1PA) band position. Re(γ) is primarily influenced through near-resonance enhancement, increasing with reduced detuning between the excitation photon energy and the energy of the S₁ state. The effective conjugation length is somewhat reduced on replacing the aryl “end” groups with alkyl groups;¹⁴ accordingly λ_max and |Re(γ)| are largest for 2 and 4 and smallest for 5–7. |Re(γ)/Im(γ)| is also sensitive to the S₀ → S₁ transition energy. Although 2PA into the S₁ state of cyanine-like dyes is electronically forbidden, significant vibronically assisted 2PA into S₁ is generally observed, with the peak transition energy ca. 0.2 eV higher than the 1PA peak energy:^7,15 the hypsochromic shift of λ_max seen for 3 and 5–8 relative to 2 brings this 2PA peak closer to twice the energy corresponding to the 1550 nm laser wavelength, resulting in a reduction in |Re(γ)/Im(γ)|. Thus, although changing “end” groups from phenyl to tert-butyl significantly improves the linear spectra of films, it also leads to poorer molecular 2PA-FOMs for 3 and 5–8 than for 2 and 4. However, the S₀ → S₁ absorptions of selenopyrylium dyes are bathochromically shifted relative to those of their thiopyrylium analogues;¹⁶ thus, the 2PA-FOM of the selenopyrylium dye 9 is larger than that of its thiopyrylium analogue 8 and is similar to those for 2 and 4, while the film spectra of 9 demonstrate considerably reduced aggregation relative to 2 and 4 due to the bulkier substitution pattern.

Table 1 Linear and nonlinear^a optical properties of chalcogenopyrylium-terminated heptamethines at 1550 nm in chloroform and in films (50 wt% in APC unless stated otherwise)

	Solution			Film
	λ max/nm	Re(γ)/10^−32 esu	|Re(γ)/Im(γ)|	Re(χ^(3))ex^b/10^−11 esu	Re(χ^(3))/10^−11 esu	|Re(χ^(3))/Im(χ^(3))|	Linear loss^c/dB cm^−1
a Determined using the fs Z-scan technique. Experimental uncertainties are ±8% for Re(γ), ±11% for |Re(γ)/Im(γ)|, ±13% for Re(χ^(3)) and ±18% for |Re(χ^(3))/Im(χ^(3))|. b Estimated by using the solution value of Re(γ) and the concentration in the film. c Determined using prism coupling. The loss measured for a pure APC film is 1.9 dB cm^−1. d Im(χ^(3)) not determined due to significant linear absorption.
2	1067	−3.2	35	−4.2	−1.5		>20
3	990	−1.8	11	−2.9	−2.4	4.6	5.6
4	1078	−2.4	30	−3.2	−8.6	11	8.5
5	997	−1.8	13	−2.5	−2.6	8.6	10
6	988	−1.6	9	−2.1	−2.5	6.0	4.1
7	997	−1.7	13	−2.3	−2.7	12	3.5
8	1008	−2.0	14	−2.1	−2.4	13	3.4
9	1055	−2.2	36	−2.6	−3.3	12	4.4
9 (neat)				−5.2	−5.1	21	8.2

---

Table 1 also includes linear optical loss and NLO data for films composed of 50 wt% chromophore/APC blends at 1550 nm. The low-energy tail in the spectrum of 2/APC, attributed to J-like aggregation, results in exceedingly large linear losses; these are reduced considerably through “end” group substitution (e.g., 3/APC), while “front” and “back” substitution further reduce the optical losses (6–9/APC). H-type aggregates in the 2/APC film result in a Re(χ⁽³⁾) value significantly smaller than that extrapolated from dilute solution, Re(χ⁽³⁾)_ex.⁹ Conversely, for many of the films where “front” substitution effectively inhibits formation of H-type aggregates (5–9/APC), the Re(χ⁽³⁾) values are fairly similar to the values of Re(χ⁽³⁾)_ex. Blend films with thiopyrylium chromophores having out-of-plane R, R′, and R″ substituents also exhibit 2PA-FOMs very similar to those observed for dilute solutions.¹⁷ Remarkably, films of 4/APC exhibit nonlinearities far exceeding those extrapolated from solution; we attribute this to the strong J-aggregate-like low-energy absorption (see above and Fig. 2), which, moreover, does not impact the linear optical loss as severely as the less well-defined low-energy absorption seen for 2/APC. Films of 9/APC exhibit the expected increase in |Re(χ⁽³⁾)| compared to its thiopyrylium analogue, 8/APC, while simultaneously exhibiting a 2PA-FOM suitable for AOSP and low optical losses. Interestingly, a neat film of 9 gives a 2PA-FOM (ca. 21) even larger than 9/APC (ca. 12). Although comparable or even higher values of |Re(χ⁽³⁾)| have been demonstrated in other organic materials at telecommunications wavelengths, including polyacetylene derivatives¹⁸ and polymethines with dioxaborine¹⁹ and tricyanofuran²⁰ termini, these have been accompanied by larger linear and/or nonlinear losses. Conversely, an arylethynyl-aryltetracyanobutadiene exhibits excellent loss parameters, but smaller |Re(χ⁽³⁾)| than found for the present compounds.²¹ Films of the 7–9/APC blends and of neat 9 show an exceptional combination of large |Re(χ⁽³⁾)|, 2PA-FOM that meets or exceeds the required value for AOSP, low linear optical loss, and, importantly for device applications, solution processability.

In conclusion, through a design strategy aimed at reducing interchromophore aggregation without introducing excessive dilution, cyanine-like dyes can be engineered to afford the requisite optical properties for phase-based AOSP: large |Re(χ⁽³⁾)|, acceptable 2PA-FOMs, and reasonably low linear losses. Moreover, the minimal 1PA and 2PA absorption at 1550 nm, coupled with the low energies and, therefore, presumably limited reactivity of the S₁ states, suggest the possibility of good photostability. In the case of 4/APC, control of the nature of the aggregation led to a larger |Re(χ⁽³⁾)| value than anticipated from solution data without incurring unacceptable loss. Our strategy provides a path to organic films with high nonlinearity, 2PA-FOM, and linear losses that would, if methods of integration with silicon photonic device platforms could be developed, enable AOSP applications such as low-power all-optical switching or modulation, efficient four-wave mixing frequency conversion, and, more specifically, the generation of optical frequency combs. Moreover, although polymethine salts are typically poorly soluble in the low-polarity solvents, such as CCl₄ and CS₂, that have suitably high refractive indices and low optical loss at telecommunications wavelengths for use in liquid-core optical fibers (LCOFs), the bulky substitution patterns of some of the compounds reported here impart good solubility in these solvents, opening the possibility of such LCOF applications. In particular, the negative sign of Re(γ) for these chromophores suggests their use in cancelling the positive Re(χ⁽³⁾) of the solvent to achieve LCOFs with near-zero Re(χ⁽³⁾) at 1550 nm, which could have possible applications in high-power fiber lasers and in telecommunications.^22–24 Work to demonstrate the feasibility of this concept is in progress. Thus, the molecular engineering approach to the inhibition of deleterious intermolecular interactions that we have described represents a critical step in the development of organic third-order NLO materials. Similar approaches may be useful for controlling intermolecular interactions in other areas of organic photonics and electronics.

Acknowledgements

This work was supported by the Air Force Office of Scientific Research through the COMAS MURI program (Agreement no. FA9550-10-1-0558).

Notes and references

1. M. Hochberg, T. Baehr-Jones, G. X. Wang, M. Shearn, K. Harvard, J. D. Luo, B. Q. Chen, Z. W. Shi, R. Lawson, P. Sullivan, A. K. Y. Jen, L. Dalton and A. Scherer, Nat. Mater., 2006, 5, 703.
2. C. Koos, P. Vorreau, T. Vallaitis, P. Dumon, W. Bogaerts, R. Baets, B. Esembeson, I. Biaggio, T. Michinobu, F. Diederich, W. Freude and J. Leuthold, Nat. Photonics, 2009, 3, 216.
3. K. Kieu, L. Schneebeli, R. A. Norwood and N. Peyghambarian, Opt. Express, 2012, 20, 8148.
4. V. Mizrahi, K. W. DeLong, G. I. Stegeman, M. A. Saifi and M. J. Andrejco, Opt. Lett., 1989, 14, 1140.
5. J. M. Hales, S. Barlow, H. Kim, S. Mukhopadhyay, J.-L. Brédas, J. W. Perry and S. R. Marder, Chem. Mater., 2014, 26, 549.
6. F. Meyers, S. R. Marder, B. M. Pierce and J.-L. Brédas, J. Am. Chem. Soc., 1994, 116, 10703.
7. J. M. Hales, J. Matichak, S. Barlow, S. Ohira, K. Yesudas, J.-L. Brédas, J. W. Perry and S. R. Marder, Science, 2010, 327, 1485.
8. S. Mukhopadhyay, C. Risko, S. R. Marder and J. L. Brédas, Chem. Sci., 2012, 3, 3103.
9. A. Scarpaci, A. Nantalaksakul, J. M. Hales, J. D. Matichak, S. Barlow, M. Rumi, J. W. Perry and S. R. Marder, Chem. Mater., 2012, 24, 1606.
10. Out-of-plane substituents have also been found to help suppress aggregation in other materials systems. For example, incorporation of a CMe₂ group into the donor portion of dipolar merocyanines, so that methyl groups inevitably project above and below the plane of the donor, has been shown to result in a dramatically reduced equilibrium constant for the formation of antiparallel dimers relative to analogues with more planar donors and improved photorefractive performance: F. Würthner, R. Wortmann and K. Meerholz, ChemPhysChem, 2002, 2, 17.
11. P.-A. Bouit, C. Aronica, L. Toupet, B. L. Guennic, C. Andraud and O. Maury, J. Am. Chem. Soc., 2010, 132, 4328.
12. F. Würthner, T. E. Kaiser and C. R. Saha-Möller, Angew. Chem., Int. Ed., 2011, 50, 3376.
13. W. West and S. Pearce, J. Phys. Chem., 1965, 69, 1894.
14. J. Panda, P. R. Virkler and M. R. Detty, J. Org. Chem., 2003, 68, 1804.
15. J. Fu, L. A. Padilha, D. J. Hagan, E. W. Van Stryland, O. V. Przhonska, M. V. Bondar, Y. L. Slominsky and A. D. Kachkovski, J. Opt. Soc. Am. B, 2007, 24, 67.
16. A. I. Tolmachev and M. A. Kudinova, Khim. Geterotsikl. Soedin., 1974, 49.
17. Film 2PA-FOMs are generally slightly lower than found in solution. The vibronic shoulders of the thin-film spectra are stronger than those seen in solution; this may be related to increased vibronically assisted 2PA in the films.
18. S.-H. Chi, J. M. Hales, C. Fuentes-Hernandez, S.-Y. Tseng, J.-Y. Cho, S. A. Odom, Q. Zhang, S. Barlow, R. R. Schrock, S. R. Marder, B. Kippelen and J. W. Perry, Adv. Mater., 2008, 20, 3199.
19. J. M. Hales, S. Zheng, S. Barlow, S. R. Marder and J. W. Perry, J. Am. Chem. Soc., 2006, 128, 11362.
20. Z. Li, Y. Liu, H. Kim, J. M. Hales, S.-H. Jang, J. Luo, T. Baehr-Jones, M. Hochberg, S. R. Marder, J. W. Perry and A. K.-Y. Jen, Adv. Mater., 2012, 24, OP326.
21. B. Esembeson, M. L. Scimeca, T. Michinobu, F. Diederich and I. Biaggio, Adv. Mater., 2008, 20, 4584.
22. A. Grudinin, Nat. Photonics, 2013, 7, 846.
23. A. R. Chraplyvy, J. Lightwave Technol., 1990, 8, 1548.
24. G. P. Agrawal, Fiber Optic Communication Systems, Wiley, New York, 2010.

---

Footnotes

† Electronic supplementary information (ESI) available: Materials and methods, including characterising data for new compounds, details of linear and nonlinear optical characterisation, and theoretical methodology; additional linear spectra, including those for neat chromophore films; and representative Z-scan spectra. See DOI: 10.1039/c4mh00068d

‡ These authors contributed equally.

---