Zhaoyang Loua,
Yingqi Cuia,
Mingli Yang*a and
Jun Chen*b
aInstitute of Atomic and Molecular Physics, Key Laboratory of High Energy Density Physics and Technology of Ministry of Education, Sichuan University, Chengdu 610065, China. E-mail: myang@scu.edu.cn; jun_chen@iapcm.ac.cn
bBeijing Institute of Applied Physics and Computational Mathematics, Beijing 100081, China
First published on 15th May 2015
When an amino acid-capped quantum dots solution meets 2,4,6-trinitrotoluene (TNT), it changes from colorless to red and its fluorescence is quenched. This is a recently developed technique for the detection of TNT in trace amounts. However, what causes the changes in coloration and fluorescence remains controversial. Using density functional theory calculations, we studied the structures and optical properties of the products of TNT reacting with cysteine. Two compounds, namely, a Meisenheimer complex and a TNT anion, which are obtained from an addition reaction and an acid–base reaction respectively, were characterized, but neither of them could be used solely to interpret the experimental results. Our calculations proposed the possibility of their coexistence in the solution from their similar thermodynamic stability and their predicted absorption and vibrational spectra. The superposition of their calculated optical absorption spectra produces band distributions similar to those of the experiments. Moreover, the measured Raman spectra that had been used to characterize the formation of the Meisenheimer complex cannot exclude the formation of the TNT anion whose characteristic vibrations are buried by those of the former. Our calculations also revealed that in the Meisenheimer complex the electron delocalization in the phenyl ring of TNT is blocked by the attached cysteine, while for the TNT anion, the removal of the hydrogen atom enhances the electron delocalization and leads to a redshift of its first excitation in comparison with that of the Meisenheimer complex. Therefore, the key to TNT detection is to control the size of the QDs so as to adjust their emission to a wavelength around the absorption bands of either the Meisenheimer complex or the TNT anion that are formed with TNT and the amino acid.
To date, TNT detection is mainly based on fluorescence,7–21 Raman scattering22–29 and electrochemical30–36 techniques and among these the fluorescence detection technique has attracted considerable attention recently owing to its strengths of high signal output, sensitivity and simplicity.37–40 The rationale of fluorescence detection is that the fluorescence of a reagent changes, shifts or is quenched when it meets TNT. TNT is electron-deficient due to the presence of three nitro groups and tends to interact with electron-rich materials such as some amino acids through a charge-transfer interaction or reaction.41 Many experimental works have reported the formation of charge-transfer complexes that are responsible for the fluorescence shift or quenching.11,15,42–44 However, what the formed complexes are remains controversial. For example, Dasary et al.26 reported a highly selective and ultra-sensitive, cysteine (Cys) modified gold nanoparticle based label-free surface enhanced Raman spectroscopy (SERS) probe for TNT detection at a 2 picomolar level in aqueous solution. The TNT solution varied in coloration from colorless to red when Cys was added. The authors believed the formation of a Meisenheimer7,41,45–47 complex through covalent addition of nucleophiles to a ring carbon atom of the electron-deficient aromatic substrates. The Meisenheimer complex has optical absorptions at about 520 nm and 630 nm, appearing red in solution. However, the absorption at 630 nm has not been observed by other authors for similar systems. Tu et al.15 prepared Mn2+-doped ZnS nanocrystals with a cysteamine-capping layer and used them for the fluorescence detection of ultratrace TNT by quenching their strong orange Mn2+ photoluminescence. Two absorptions at 465 nm and 515 nm were observed when cysteamine was added to the TNT solution, and the solution became red. The formation of TNT anions through an acid–base reaction was proposed. The overlap of the absorption band of the TNT anions with the emission of the Mn-doped ZnS QDs leads to the fluorescence quenching of the nano solution systems. Zou et al.42,43 synthesized Cys modified Mn2+-doped and Co2+-doped ZnS quantum dots to image ultratrace TNT in water and observed absorptions at 465 nm and 512 nm after adding TNT to the prepared QD solution. Interestingly, the authors proposed the formation of Meisenheimer complexes that quench the fluorescence emitted by the Mn2+- or Co2+-doped ZnS QDs.
There is no doubt that a complex is formed when TNT meets Cys. The complex may absorb the QDs' fluorescent emission and/or lead to the variation in color. However, there are two explanations of the mystery complex: a Meisenheimer complex through covalent bonding between the electron-deficient TNT11,48–50 and the electron-rich Cys or a TNT anion through an acid–base reaction between the acidic Cys and the basic TNT.15,43,51 Stimulated by the experimental works, we studied the interaction of molecular TNT with Cys and the structure and optical absorption of the TNT–Cys complex by means of density functional theory (DFT) calculations, aiming to make an interpretation of the experimental observations including the color change and the fluorescence quenching of TNT–Cys-QD solutions. Understanding the TNT detection rationale is certainly important to the development of its detection techniques and to the preparation of new semi-conducting QDs with proper fluorescence emissions.
Fig. 1 TDDFT predicted absorption spectra of TNT calculated at the M06-2X (cyan), ωB97XD (green), PBE0 (purple), and B3LYP (red) levels with the 6-311++G(d,p) basis set. |
The optimized structures of S1 and S2 are shown in Fig. 2, while the other less stable isomers are presented in the ESI.† Table 1 lists the selected bond lengths and bond angles of S1 and S2 obtained with M06-2X/6-311++G(d,p), as well as those of TNT and Cys for comparison. A covalent bond about 1.56 Å in length and 0.69 in bond order was predicted between Cys–N and TNT–C1, indicating a typical C–N single bond. Accordingly, the conjugated phenyl ring is partially deconstructed. C1–C2, C1–C6, and C1–C7 elongate remarkably, whereas the other C–C bonds vary by about ±0.02 Å. Comparing with the TNT structure, C1 changes from sp2 to sp3 hybridization. The conjugation path is thus blocked at C1. The bond lengths and bond angles near C1 are subjected to greater distortion than the others, as reflected by the geometrical parameters in Table 1. In addition, the amine-H is about 1.95 Å from one of the nitro-O, which is a typical HB length.
Bond | TNT | S1 | S2 | S3 | S4 | |||||
---|---|---|---|---|---|---|---|---|---|---|
BL | WBO | BL | WBO | BL | WBO | BL | WBO | BL | WBO | |
C1–C2 | 1.40 | 1.36 | 1.51 | 0.98 | 1.48 | 1.05 | 1.40 | 1.36 | 1.48 | 1.06 |
C2–C3 | 1.38 | 1.40 | 1.36 | 1.52 | 1.37 | 1.47 | 1.39 | 1.39 | 1.37 | 1.54 |
C3–C4 | 1.38 | 1.39 | 1.40 | 1.27 | 1.40 | 1.31 | 1.38 | 1.40 | 1.40 | 1.25 |
C4–C5 | 1.38 | 1.39 | 1.40 | 1.27 | 1.41 | 1.23 | 1.38 | 1.37 | 1.40 | 1.25 |
C5–C6 | 1.38 | 1.40 | 1.36 | 1.52 | 1.36 | 1.56 | 1.38 | 1.41 | 1.37 | 1.54 |
C6–C1 | 1.40 | 1.36 | 1.52 | 1.00 | 1.48 | 1.05 | 1.40 | 1.35 | 1.48 | 1.06 |
C1–C7 | 1.50 | 1.03 | 1.54 | 0.98 | 1.34 | 1.79 | 1.50 | 1.04 | 1.35 | 1.77 |
C2–N | 1.48 | 0.90 | 1.43 | 1.01 | 1.42 | 1.04 | 1.48 | 0.90 | 1.44 | 0.98 |
C4–N | 1.47 | 0.91 | 1.43 | 1.01 | 1.43 | 1.01 | 1.47 | 0.91 | 1.41 | 1.04 |
C6–N | 1.48 | 0.90 | 1.44 | 1.00 | 1.44 | 0.98 | 1.48 | 0.89 | 1.44 | 0.98 |
Dihedral angle | ||||||||||
C7–C1–C2–C3 | 174.6 | 118.3 | 150.4 | 178.3 | 161.6 | |||||
O–N–C2–C3 | 38.3 | 9.7 | 11.1 | 35.1 | 14.1 | |||||
O–N–C6–C5 | −38.3 | −9.4 | −26.2 | −53.5 | −14.0 |
In S2 the removal of one proton from the methyl leads to a considerable change in the conjugated phenyl ring. First, the methyl-C changes from sp3 to sp2 hybridization, shortening the C1–C7 from a single bond to a double bond. C7 remains in the phenyl plane. Accordingly, the aromatic phenyl ring changes into a quinoid structure with alternating short and long C–C bonds. C2–C3 and C5–C6 are apparently shorter than C1–C2, C1–C6, C3–C4 and C4–C5.
Table 2 lists the computed energy and Gibbs free energy changes (ΔE and ΔG) of Reactions I and II. The predicted ΔE values are rather small (less than 3.4 kcal mol−1 in absolute value at the M06-2X level and 5.2 kcal mol−1 at the PBE0 level) for both reactions. The opposite signs of ΔE for Reaction I, negative (−3.3 kcal mol−1) by M06-2X and positive (4.1 kcal mol−1) by PBE0, are a reflection of their small magnitudes around zero. Both reactions have small, positive ΔG values, indicating that extra energy is required for their proceeding. Such small amounts of energy can be served either by heating or illumination in experiments. The formation of both S1 and S2 is therefore feasible under ambient conditions, although they are thermodynamically less stable than the reactants. On the other hand, S1 and S2 have similar ΔE and ΔG values (the differences are only 5.5 kcal mol−1 for ΔE and 1.9 kcal mol−1 for ΔG), indicating that they have similar thermodynamic stability. It is then possible that both S1 and S2 coexist in the system.
Reaction | M06-2X | PBE0 | ||
---|---|---|---|---|
ΔE | ΔG | ΔE | ΔG | |
I | −3.3 | 15.8 | 4.1 | 22.5 |
II | 2.2 | 17.7 | 5.1 | 20.1 |
ΔE (S0 − S1) | ΔE (S0 − S2) | |||
---|---|---|---|---|
λ | f | λ | f | |
TNT | 310 | 0.017 | 237 | 0.249 |
Cys | 203 | 0.041 | 178 | 0.016 |
S1 | 451 | 0.219 | 340 | 0.388 |
S2 | 508 | 0.155 | 408 | 0.366 |
S3 | 311 | 0.010 | 277 | 0.014 |
S4 | 523 | 0.157 | 404 | 0.348 |
First, both TNT and Cys have their absorption in the ultraviolet region, 310 nm for TNT and 203 nm for Cys, which have no effect on the color of solutions. One more possibility was considered here. TNT and Cys may form a complex via hydrogen bonds (S3 in Fig. 2). As shown in Fig. S3 in the ESI,† two HBs are formed between TNT and Cys with ΔG(298 K) = 3.5 kcal mol−1. Its predicted first excitation is at 311 nm in water. Therefore, the assembly of TNT with Cys via HBs should not be responsible for the observed fluorescence quenching or color change in the visible region either.
The computed first peak of S1 is at 451 nm, lying in the visible region. The second peak is at 340 nm, lying in the ultraviolet region. For S2, its first two peaks are at 508 nm and 408 nm. Neither S1 nor S2 have the absorption band distribution matching the observed spectra in the experiments. In Zou's50 measurement, two new bands, a strong one at 465 nm and a weak one at 516 nm, were observed when TNT was added to a Cys-modified ZnS QD aqueous solution. Similar spectra were also obtained by Tu et al.15 It is interesting to note that although the computed absorption spectra of S1 and S2 do not comply with the measurements, they possess most of the measured bands when they come together. As our calculations on the thermodynamic properties of S1 and S2 suggest the possibility of their coexistence in the solution, it is interesting to see the combination of their absorption spectra. Fig. 3 displays the optical absorption spectra of S1, S2 and their combination at a 1:1 ratio. The superposition of the spectra of S1 and S2 comes to a curve, which is very close to the observed one15,50,51,63 in band locations and shapes. The two bands at 447 nm and 506 nm are in accordance with the observed ones15,50 at about 465 nm and 515 nm.
Fig. 3 Comparison among the experimental (black line) and predicted absorption spectra of S1 (grey dashed line), S2 (grey solid line) and the combination of S1 and S2 at a 1:1 ratio (red line). |
Both TNT and Cys-capped QD solutions are colorless because they do not have absorption in the visible region. However, S1 and S2 are formed when these two solutions come together. S1, the Meisenheimer complex, is responsible for the absorption around 460 nm, whereas S2 is responsible for the absorption around 520 nm. This is why the mixture becomes red. The absorption and emission of Cys-capped QDs vary with the QD size, the capping ligand and the dielectric constant of the solvent. The QDs in the measurements are about 3–50 nm in diameter, with emission at 400–600 nm.15,17,42,50,51,62 It is evident that the size of QDs can be controlled to have emissions close to the absorption bands of S1 and S2. This is why the fluorescence of Cys-capped QDs is quenched in TNT solution. Our calculations revealed that the key to TNT detection is to control the QD size so as to adjust their emission to a wavelength around the absorption bands of the complexes formed with TNT and the amino acid.
In addition, we computed the optical absorption of the TNT anion (S4 in Fig. 2) in the absence of the protonated Cys. Its first two peaks are at 523 and 404 nm, which are very close to those of S2, indicating that the electron excitations in S2 are mainly contributed by the TNT anion moiety and the hydrogen bonding that binds the TNT anion and protonated Cys together has rather limited effect on the absorption spectra.
The absorption band shifts of S1 and S2 can be interpreted from their electronic structures. Our calculations revealed that the first excitations of both S1 and S2 correspond to the transitions from their highest occupied molecular orbitals (HOMO) to their lowest unoccupied molecular orbitals (LUMO). The HOMO-LUMO gaps are 5.04 eV for S1 and 4.62 eV for S2, complying with the first peak locations in their absorption spectra. Fig. 4 compares the HOMO and LUMO contours of TNT, S1 and S2. Both the HOMO and LUMO of TNT are mainly contributed by the conjugated phenyl ring, while the bonded nitro and methyl groups are also partially involved in the transition. The reaction with Cys, by either Path I or Path II, destabilizes the TNT structure and shifts its electronic excitation to longer wavelength. In S1, the conjugation in the phenyl ring was partially broken by the Cys addition. Its HOMO and LUMO are localized in the conjugated moiety. In S2, the proton removal changes the phenyl from an aromatic structure to a quinoid structure in which the conjugation path remains. Its HOMO covers all the atoms in the quinoid structure and its LUMO is also contributed from part of the atoms in the quinoid structure. The HOMO-to-LUMO transition in S2 is therefore energetically more favorable than that in S1. This is why S2 has its first absorption at a longer wavelength than S1 does. It is also revealed from Fig. 4 that the Cys part is not itself involved in the electronic excitations, but it affects the excitation by changing the electronic structure of the TNT part.
Even though significant variation in the spectra among TNT, S1 and S2 was noted, most of the bands of TNT can be traced in the S1 and S2 spectra. In addition to the characteristic peaks of the nitro groups, the aromatic CC stretching (at about 1700 cm−1), the C6H2–NO2 stretching (979 cm−1 in TNT, 989 cm−1 in S1 and 992 cm−1 in S2) and the C6H2–C stretching (1257 cm−1 in TNT and 1080 cm−1 in S1) can be clearly identified in the spectra. The strong bands in Cys are associated with C–H, N–H and S–H stretching. These bands become very weak in S1 and S2 in comparison with the strong bands of TNT.
The formation of S1 is featured by the C1–N bond, which is at 857 cm−1 as shown in Fig. 5c. Upon its formation, the C1–CH3 bond is weakened from 1257 cm−1 to 1080 cm−1, and the N–H stretching in Cys shifts from 3542 cm−1 to 3354 cm−1. The formation of S2 is featured by the newly formed C1CH2 bond, which peaks at 1714 cm−1 and has similar vibrational frequency with the other CC stretching vibrations in the aromatic ring. The N–H stretching vibration in the newly formed NH3+ group is at 3269 cm−1.
In the Raman spectra measured by Dasary et al.,26 a strong broadband was observed at 2900 cm−1, which was assigned to the NH2 symmetric stretching, C–H stretching and CH2 asymmetric stretching in the Meisenheimer complex. The TNT peaks are at 1615 cm−1 for the CC aromatic stretching vibration, at 1360 cm−1 for the NO2 symmetric stretching vibration, 1210 cm−1 for the C1–CH3 vibration, etc. Because the measurement was conducted in the presence of gold nanoparticles, the observed bands should be to somewhat different from our calculations. The shifts are about 50–110 cm−1. However, one can still note that all the new vibrational bands featured by S1 and S2 can be associated with the observation. Both the C1–N vibration in S1 and the C1CH2 vibration in S2 can be assigned in the regions of around 850 cm−1 and around 1650 cm−1, respectively. Moreover, the broad band at 2900 cm−1 may cover the newly formed stretching vibration of NH3+ in S2. Therefore, although the observed Raman spectra did not provide adequate evidence to confirm the coexistence of S1 and S2, the observation does not exclude the possibility that both S1 and S2 are formed simultaneously when TNT meets a Cys-capped QD solution.
It should be mentioned that the effect of QDs on the structure and optical absorption of S1 and S2 was not considered in this study. The QDs are usually capped with Cys via bonding with the sulfur atom. Two saturated carbon atoms stand between the sulfur atom and the nitrogen atom that interacts with TNT. Therefore, the sulfur atom and the QD have limited influence on the electronic structures of S1 and S2. Our computations on the TNT–Cys systems are reasonable approximations for studying the mechanism of TNT detection with amino-capped QDs.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra07088k |
This journal is © The Royal Society of Chemistry 2015 |