Synthesis, solid-state fluorescence properties, and computational analysis of novel 2-aminobenzo[4,5]thieno[3,2-d]pyrimidine 5,5-dioxides

New fluorescent compounds, benzo[4,5]thieno[3,2-d]pyrimidine 5,5-dioxides (3a–g), 2-amino-4-methylsulfanylbenzo[4,5]thieno[3,2-d]pyrimidine (6), and 2-amino-4-methylsulfanyl-7-methoxybenzo[4,5]furo[3,2-d]pyrimidine (7), were synthesized in good yields from heterocyclic ketene dithioacetals (1a–c) and guanidine carbonate (2a) or (S)-methylisothiourea sulfate (2b) in pyridine under reflux. Among the fused pyrimidine derivatives, compound 3c, which has an amino group at the 2-position and a benzylamino group at the 4-position of the pyrimidine ring, showed the strongest solid-state fluorescence. The absorption and emission properties of the compounds were quantitatively reproduced by a series of ab initio quantum-chemical calculations.


Introduction
Solid-state fluorescent compounds are currently attracting considerable interest from both theoretical and practical standpoints [1][2][3][4]. Recently, we prepared new pyrimidine derivatives, which have solid-state blue fluorescence, by means of a one-pot synthesis, involving the reaction of ketene dithioacetals, amines, and guanidine carbonate in pyridine [5]. We have also reported the one-pot synthesis of a new, fluorescent, fused pyrimidine derivative 2,4-diaminoindeno [1,2-d]pyrimidin-5-one; this pyrimidine derivative, which was synthesized by heating a ketene dithioacetal, 2-[bis(methylsulfanyl)methylidene]indan-1,3-dione, under reflux with amine and amidine derivatives in pyridine solution, showed blue-green fluorescence in the solid state [6]. These one-pot synthetic methods are also promising for the preparation of other new solid-state fluorescent pyrimidine derivatives containing polycyclic heterocycles. In this paper, we report the synthesis of new fluorescent 2-aminobenzo [4,5]thieno [3,2-d]pyrimidine 5,5-dioxides and related fused pyrimidine derivatives. It is thought that these derivatives show strong fluorescence as a result of the presence of a hetero-π-electron conjugated system [7]. The electronic and emission spectra of these new compounds were analyzed computationally by using a series of ab initio quantum-chemical calculations.

Comparison of predicted and experimental absorption spectra
The UV-vis spectra of the compounds exhibited peaks (λ max ) at 280-360 nm. The spectra possessed multiple subpeaks around λ max . The experimental and computational λ max values are given in Table 1. The theoretical and experimental λ max values are in fairly good agreement, within 50 nm deviation. The redshifts observed for 3d, 3e, and 6 relative to 3b were well reproduced by the TDDFT computations. The sidebands around λ max were not reproduced by our TDDFT computations, which predicted single-peak maxima for all the compounds. The subpeaks are considered to be vibronically assisted absorptions, because the extended π-systems of the molecules have large degrees of vibrational freedom effectively coupled with their electronic states. As a representative example of the electronic structures, the HOMO and LUMO of 3a are depicted in Figure 1. The λ max was assigned as the HOMO-LUMO π-π* excitation (configuration weight = 0.697) with an oscillator strength of 0.096 at 335 nm. The next theoretical peak appeared  at 290 nm, derived from (HOMO − 1) to LUMO excitations, considerably separated from the first peak. The HOMO and LUMO delocalize on the whole system, indicating that modest intramolecular electron transfer from the methylthio and amino groups on the pyrimidine ring (HOMO) to the phenyl moiety (LUMO) is expected upon S 0 →S 1 transition. The transition character is also rationalized with the transition dipole moment directed along the long molecular axis. The electron density difference between HOMO and LUMO is shown in Figure 2; this reflects the balanced electron redistribution between the pyrimidine and phenyl moieties.

Solid-state fluorescence
Tris(8-hydroxyquinolinato)aluminium (Alq 3 ) was used as the standard for fluorescence spectrum measurements [17]. We analyzed the solid-state fluorescence emission spectra of 3a-g, 6 and 7 at room temperature. The fluorescence maxima (λ em,max ) and relative fluorescence intensities (RI) of these compounds are listed in Table 2. The λ em,max values of 3a and 3c-g were in the range 430-462 nm. The fluorescence of the N-unsubstituted diamino compound 3b exhibited a large bathochromic shift, and green fluorescence was observed (λ em,max = 529 nm). Compound 3a, which has an amino group at the 2-position in the pyrimidine ring, showed stronger fluorescence than did compound 3g, which has a methylsulfanyl group. This result was in agreement with the previous findings that an amino group at the 2-position of an indenopyrimidine influences the fluorescence intensity [6]. 2,4-Diaminopyrimidine derivatives (3b-e) also showed stronger fluorescence than did 3g. Compound 3c, which has an amino group at the 2-position and a benzylamino group at the 4-position of the pyrimidine ring, showed the strongest fluorescence (RI = 1.95). This value was larger than those of indenopyrimidine derivatives (RI = 0.01-0.73). We speculated that the effect of the π-π-electron conjugated system in the sulfonyl group of benzothienopyrimidine 5,5-dioxide is stronger than that of the carbonyl group of the indeno pyrimidines.
Solid-state fluorescence is strongly influenced by intermolecular steric hindrance. In general, the fluorescence intensity can be enhanced by minimizing intermolecular interactions through the introduction of bulky substituents. Compound 3c, which has a bulky benzylamino moiety shows the strongest fluorescence of these molecules ( Figure 2). The emission from 3e with a bulky phenylamino moiety, however, is relatively weak. This contradiction reflects the complex emission mechanism, which is driven not only by intermolecular stacking effects but also by the electronic structures. The weak fluorescence of 7 is assumed to be dominated by the inductive effect of the methoxy group, rather than by stacking effects.
To elucidate the S 1 nature of the compounds in question, we carried out ab initio molecular-orbital calculations, focused on 3a ( Figure 3). The computed key bond lengths are shown in Table 3. Comparison of the S 0 (DFT) and S 1 (CIS) bond lengths shows that the S 0 →S 1 (dominant HOMO→LUMO) transition is reflected in the bond-length variations. The C4-C5 bond, with a bonding lobe in S 0 but an antibonding one in S 1 , is significantly lengthened, by 0.045 Å, as a result of the S 0 →S 1 transition. In contrast, the C5-C8 bond, with an antibonding lobe in S 0 but a bonding one in S1, is considerably shortened, by 0.081 Å, on S 0 →S 1 excitation. No significant deviation from planarity was observed in the fused-ring structure associated with the S 0 →S 1 excitation. The geometrical discrepancies between DFT and CAS for S 0 and CIS and CAS for S 1 indicate that the significant roles of dynamic electron correlations are inadequate in the CAS method, which consistently predicted longer bond lengths. These disagreements could be resolved by geometric optimizations using MS-CASPT2 including dynamic correlation, but this is not feasible in the present study owing to the huge computational burden.  The theoretical fluorescence λ max values of 3a, obtained using several quantum-chemical methods, are summarized in Table 4.
The CIS-predicted peak shows a significant blue-shift relative to the experimental value; this is well known, and is a result of the lack of multi-excitation character. The TDDFT peaks partially improve the gap, with the CIS optimized geometry for the S 1 state not being of trustworthy quality. The CASSCF peak, in contrast, overshoots the experimental peak. The best prediction was obtained by using the most elaborate method, MS-CASPT2, which can take the majority of the electronic correlations into account. The MS-CASPT2 result indicates that  the excitation character includes significant contributions from the HOMO→LUMO double excitation. This implies that the predictive abilities of CIS and other single-configuration-referenced methods, which cannot handle multi-excitation features, are limited.

Conclusion
New solid-state fluorescent compounds, benzo [4,5] were also synthesized by one-pot reactions between 2a and 1b or 1c, under the same conditions. The products 6 and 7 showed solid-state fluorescence. Compound 3c, which has an amino group at the 2-position and a benzylamino group at the 4-position of the pyrimidine ring, showed the strongest solidstate fluorescence. These results indicated that heteroatoms in the ring moieties have a strong influence on the solid-state emissions. The absorption and emission spectra of the compounds were computationally analyzed by means of a series of ab initio quantum-chemical calculations. The theoretical analyses quantitatively reproduced the S 0 →S 1 and S 1 →S 0 transitions. The associated HOMO/LUMO distributions and the transition characters were also elucidated theoretically.

Experimental General
Identifications of compounds and measurements of properties were carried out by general procedures employing the following equipment: All melting points were determined with a Mitamura Riken Kogyo Mel-Temp apparatus or a Laboratory Devices Mel-Temp II apparatus and were uncorrected. Infrared (IR) spectra were recorded in potassium bromide pellets on a JASCO 810 or Shimazu IR-460 spectrometer. Ultraviolet (UV) absorption spectra were determined in 95% ethanol on a Hitachi 323 spectrometer. Fluorescence spectra were determined on a Shimazu RF-1500. Nuclear-magnetic-resonance (NMR) spectra were obtained on Gemini 300NMR (300 MHz) and 500NMR (500 MHz) spectrometers with tetramethylsilane as an internal standard. Mass spectra (MS) were recorded on a JOEL DX-303 mass spectrometer. Microanalyses were performed by H. Mazume on a Yanaco M-5 at Nagasaki University. All chemicals were reagent grade and used without further purification unless otherwise specified.

Method of measurement of fluorescence
A powder sample of the subject compound was heaped in the tray. After the sample was covered with a quartz plate, this part was fixed in the fluorescence spectrometer. After fixing the fluorescence wavelength, the excitation spectrum was determined by scanning with the fluorescence wavelength. Similarly, the fluorescence spectrum was obtained after scanning with the excitation wavelength. After obtaining these results, the excitation wavelength was decided and the fluorescence spectrum was measured. The relative intensity of fluorescence was determined by using Alq 3 as the standard sample. The fluorescence of the standard sample and all subject compounds was measured at 272 nm excitation.