1. Introduction

A variety of noninvasive optical microscopy and spectroscopy techniques have been developed for in situ and time-resolved characterization of chemicals and biological systems. Among these spectroscopic techniques, Raman and photoluminescence spectroscopies are particularly suited for the characterization of chemical and biological aerosol particles [1–4]. However, photo-luminescence (fluorescence or phosphoresce) is not omnipresent in bio-aerosol particles and often is weak unless excited in the UV. By contrast, Raman scattering occurs in most materials excited at wavelengths from UV to infrared. Because Raman scattering is sensitive to both the chemical constituents and the secondary to tertiary structures [5–7], it has been used to identify chemical and biological species, or to monitor the reaction of aerosol particles with the surrounding chemicals or particles. However, when laser excitation is close to an electronic transition, Raman scattering frequently coincides with fluorescence. Since fluorescence is typically several orders of magnitude stronger than Raman scattering and covers a broad spectral range, the fluorescence from the target material or even traces of fluorescent impurities [8] can hinder the measurement of the Raman spectrum of the major material. Hence, it is essential to develop strategies to avoid fluorescence in Raman spectroscopy. To avoid electronic excitation and consequent fluorescence, one strategy is to employ a near infrared (NIR) or infrared (IR) laser as the Raman excitation for bio-aerosols. The use of an NIR/IR instead of UV excitation has adverse effects, e.g., reduced Raman scattering efficiency and scarcity of high-quantum-efficiency IR CCD for the detection. In principle, the signal-to-noise ratio of Raman signal can be improved with increasing the excitation intensity, provided that the signal ratio of Raman scattering to fluorescence remain the same or even improve with increasing photoexcitation becaused of photobleaching of fluorescent impurities. However, many bio-materials are adversely affected by heating and photodegradation, which prevent the use of excessive excitation intensity or power. Therefore, it is useful to identify optimal ranges of Raman excitation wavelengths and intensities for studies of bio-materials. Other strategies to enhance Raman scattering relative to fluorescence, e.g., resonance Raman scattering [6, 9–11] and surface enhanced Raman scattering (SERS) [12–14], are beyond the scope of this work.

Most of the UV-excited intrinsic fluorescence from proteins is due to the photoexcitation of tryptophan (Trp) [15], tyrosine (Tyr) [16] and phenylalanine (Phe) [17]. These aromatic amino acids and related compounds and peptides have been the basis for studies of photochemical, photoionization and photodissociation processes in proteins [15–17]. The absorption and emission spectra, and fluorescence quantum yields and lifetimes, of Trp, Tyr, and Phe are sensitive to the local structure and polarity in the surrounding micro-environment [18, 19]. However, the fluorescence spectra of proteins containing Trp, Tyr and Phe typically resemble that of Trp because of Föester resonance energy transfer (FRET) from Phe to Tyr and from Tyr to Trp, and fluorescence quenching in Tyr [20]. Furthermore, Trp is a common aromatic amino acid in proteins, present at a concentration of about 0.1 to 1 mol %. Hence, fluorescence from Trp has been used to monitor conformation and folding of proteins [18,19]. In addition, indole [21], the chromophore of Trp, can be found in biological molecules such as melatonin, auxin, and the neurotransmitter serotonin. Studies of the optical response of Trp and its photochemical and photothermal transformations (typically photodegradations) are essential in developing a comprehensive understanding of molecular fluorescence and applying it in biological systems [22].

The Raman spectrum of a protein provides information on the molecular structure and how it is modified in interactions with other molecules. The structural information can be extracted from the Raman spectrum by identifying and interpreting a selected set of marker bands that reflect the structure and interaction in the protein. Raman spectra of Trp in solution and dry particles have been measured using UV or near-IR excitation [10, 23–27]. The Trp side chain has been demonstrated as a Raman structural marker for the conformation, hydrogen bonding, hydrophobic interaction, and cation–π interaction of the indole ring of Trp [28]. For example, Trp has been used as a Raman marker in the non-invasive in vivo optical Raman spectroscopy for the detection of oxidation in proteins where Trp is a target for oxidation [29]. Tryptophan has strong absorption for λ ≲ 290 nm (4.3 eV), and a solvatochromic emission peak position ranging between approximately 300 to 350 nm depending on the polarity of the local environment [15, 18].

In this study, we investigate the effects of different photoexcitation energy (wavelenght) and intensity on the Raman and fluorescence spectra of dry L-Trp particles in the size range ∼5 to 500 μm. Raman and fluorescence spectra are generated using 488-nm (2.54 eV) or 633-nm (1.96 eV) excitation. We describe the appearance of bright orange and red fluorescence with an emission peak near 580–600 nm in dry Trp particles when excited by a 488 nm laser with an intensity greater than approximately 1 kW/cm² for a few minutes. We attribute the orange/red fluorescence to a transformation of Trp induced by a combination of photochemical and photothermal effects. In contrast to pristine dry Trp particles, the photochemical products emit orange (∼600 nm) and red (∼700 nm) light when excited at 488 nm and 633 nm, respectively. To the best of our knowledge, these orange and red fluorescent products generated from Trp by photochemical and photothermal processes have not been described previously.

2. Experiments

Raman scattering and fluorescence experiments are performed in a standard micro-photoluminescence setup using a 488 nm Argon-ion laser (Lexel Lasers, 95-SHG CW UV Ion laser) or a 632.8 nm HeNe laser as the photoexcitation source. The excitation laser beam is introduced at an oblique incident angle of approximately 60° with respect to the silicon substrate and is focused to a spot on the targeted sample with dimensions ∼ 42 × 20 μm (1/e² diameter) for the 488 nm excitation and ∼ 52 × 26 μm for the 633 nm excitation. Raman scattering signals and fluorescence emissions are measured in a backscattering geometry through an objective lens (Mitsutoyo M Plan Apo, 50×, NA = 0.42), an f = 120 mm imaging lens, and an imaging spectrometer (Acton SpectraPro 2300) equipped with an EM-CCD (Princeton Instruments, ProEM:1600, 1600×200 pixels, pixel size = 16 μm×16 μm, TE-cooled to −65 °C). The spectral resolution for Raman spectra is ≈1.3 cm⁻¹ for 488 nm excitation, and ≈0.8 cm⁻¹ for 633 nm excitation when using a 1200 grooves/mm grating. L-tryptophan (BioUltra grade, ≥99.5% purity) is purchased from Sigma-Aldrich and used without further purification. Solid powder of L-tryptophan is deposited on a silicon substrate for optical studies of individual ∼5–500 μm diameter particles. All measurements are conducted at room temperature in air.

3. Results

3.1. Raman characterization of dry L-tryptophan

Previously, Raman spectra of Trp were mostly measured in solution under UV excitation [10, 26, 28]. Similar to this work, Chuang et al. [27] reported Raman spectra of dry Trp using a 488-nm laser excitation. In this work, we compare Raman scattering from dry solid particles of LTrp in the spectral region from about 350 to 3500 cm⁻¹ using 488-nm and 633-nm laser excitation (Fig. 1). Previously, Takeuchi et al. [28] reported that certain Raman bands were missing in the Raman spectra of Trp solution excited with visible lasers. By contrast, the Raman spectra measured here using both 488-nm and 633-nm excitation reproduce mostly all Raman bands previously observed using UV excitation.

 

Fig. 1 Raman spectra of solid particles of L-tryptophan. The laser excitation is at 488.0 nm (blue curve) or 632.8 nm (red curve). The excitation powers (intensities) are 2 mW (0.30 kW/cm²) for 488 nm and 5.6 mW (0.53 kW/cm²) for 633 nm, respectively. The Stokes and anti-Stokes Raman scattering peaks at 755 cm⁻¹ are identifiable under both the 488-nm and 633-nm laser excitation, and can be used to determine the effective lattice (phonon) temperature, T^*. The mode at ±523 (521) cm⁻¹ is from the silicon substrate. T^* ≈ 350–400 K in the powder particles of Trp, and T^* ≈ 320–340 K in the silicon substrate. Note that the detection of Raman 521 cm⁻¹ mode from the silicon substrate suggests that the Trp particle is largely transparent under this relatively low photoexcitation intensity. The spectra shown here have not been corrected for the spectral responses of the optical setup, including the collection optics, spectrometer, and CCD, and this may result in relatively small errors in the estimation of the lattice temperature.

Download Full Size | PDF

The Raman spectra shown in Fig. 1 are spatially integrated over an effective area of 3×20 μm² across the laser focal spot on a sample particle labeled as SA. Under a laser excitation of 488 nm, fluorescence is expected to be minimal in Trp, allowing for the measurement of Raman spectra with a high Raman scattering efficiency. Indeed, Raman spectra under a 488-nm (energy E_b = 2.54 eV) laser excitation reveals Raman peaks previously identified with a resonant UV excitation [28] or similar 488 nm laser excitation [27]. However, we observe significant fluorescence with a spectral peak near 600 nm even at a photoexcitation intensity as low as 0.3 kW/cm². Such spectrally broad fluorescence can be suppressed (photobleached) by exposing the samples to 488 nm excitation with an intensity of ∼1 kW/cm² for less than five minutes. The signal ratio between Raman and fluorescence as well as the spectral resolution are both improved when a 633 nm (energy E_r = 1.96 eV) laser is used as the Raman excitation source. Aside from that the Raman scattering efficiency is decreased by a factor of three (≈ (E_b/E_r)⁴), the 633 nm laser excitation gives similar Raman scattering peaks with slight variations of frequency shifts.

Below we discuss selected Raman bands following the assignments in the previous works [27, 28]. The Raman spectra reflect the characteristic indole-stretching mode ν(NH) at 3402 cm⁻¹, and the aromatic and aliphatic ν(CH) vibrations above and below 3056 cm⁻¹. Most other intense Raman scattering signals are associated with the bending of the C–H bonds on the indole ring, the indole-ring stretching, and the deformation of the indole ring. These bands appear at ∼1556, ∼1423, ∼1010, ∼876, and ∼757 cm⁻¹. The bands at ∼1423 and ∼1615 cm⁻¹ are attributed to the symmetric and asymmetric stretching of the COO^– group. Other bands related to the ${\text{NH}}_3^{+}$ vibrations are at ∼1575, ∼1119, ∼1105, ∼1068, and ∼840 cm⁻¹. In the spectral region below 760 cm⁻¹, most of the weak bands are due to the vibrations associated with the deformation of the benzene ring [27].

In the following discussion, we focus on selected bands that are used as markers for conformation, hydrophobic interaction, van der Waals interaction, and hydrogen bonding.

Conformation marker: 1550 cm⁻¹ band (denoted W3)

The Raman band at 1550 cm⁻¹ is attributed to the C₂=C₃ stretch arising from an indole ring vibration (denoted W3) [28]. The wavenumber of W3 (νW3) is expected to be affected by the torsion angle χ^(2,1) about the C₂=C₃–C_β–C_α linkage. The relationship between νW3 and χ^(2,1) is approximated by νW3 = 1542+6.7(cos3|χ^(2,1)| +1)^1.2 [cm⁻¹]. Based on the measured wavenumber 1556 cm⁻¹ (1557 cm⁻¹ under 633-nm excitation), the torsion angle χ^(2,1) ≈ 110°, which agrees with the value (106°) found in the crystal structure.

Hydrophobic interaction marker: 1340/1360 cm⁻¹ doublet (denoted W7)

The relative intensity of the doublet near 1340 and 1360 cm⁻¹ (I₁₃₆₀/I₁₃₄₀) are indicative of the hydrophobicity [23]. The 1360 cm⁻¹ band is stronger in hydrophobic solvents (I₁₃₆₀/I₁₃₄₀ > 1.1), while the 1340 cm⁻¹ is stronger in hydrophilic environment (I₁₃₆₀/I₁₃₄₀ < 0.9). Here, I₁₃₆₀/I₁₃₄₀ is measured to be more than 1.5 in dry tryptophan where water or solvent is absent.

van der Waals interaction marker: 1010 cm⁻¹ band

The 1010 cm⁻¹ band is sensitive to the strength of van der Waals interactions of the indole ring of the Trp with surrounding residues. A Raman shift near or above 1012 cm⁻¹ indicates strong van der Waals interactions. The measured Raman shift of 1007–1010 cm⁻¹ reflect the weak or absence of van der Waals interactions between Trp and residual impurities.

Hydrogen bond marker: 876 cm⁻¹

The 876 cm⁻¹ band (denoted as W17) is an indole ring vibration mode associated with a displacement of the N₁H group nearly along the N₁–H bond. Its wavenumber (νW17) is sensitive to hydrogen bonding. Under a UV laser excitation, νW17 was found to decrease in the strongly-hydrogen-bonded state, and exhibit a linear correlation with the νNH (3400 cm⁻¹ band) of the N₁–H stretching vibration mode [28]. Here, we find νNH = 3403 cm⁻¹ (488 nm excitation) or 3406 cm⁻¹ (633 nm excitation), which are similar to the value reported in Ref. [28]. However, we find that νW17 deviates from the previously measured linear correlation [25] by about −2 cm⁻¹ under the 488 nm excitation and −5 cm⁻¹ under the 633 nm excitation, respectively.

In addition to W17, the W4 (≈1490 cm⁻¹) and W6 (≈1430 cm⁻¹) modes have been found to be sensitive to hydrogen bonding as well. In contrast to W17, the wavenumbers of W4 and W6 modes increase with an increase of the strength of hydrogen bonding because of the contributions from the N₁–H bending. Using the 488 nm excitation, we measure νW4 and νW6 to be 1423 and 1486 cm ⁻¹, respectively.

Overall, the values of νNH, νW4, and νW6 measured here are about 5–6 cm ⁻¹ less than previously reported values [27]. We suspect that the correlation between νW17 and νNH found under a UV excitation is not applicable for visible laser excitation as used in this work.

3.2. Lattice temperature – Stokes vs. anti-Stokes Raman scattering

The ratio of Stokes and anti-Stokes Raman scattering mode intensity can be used to monitor the substrate and lattice temperature T^*. The lattice temperature is related to the Stokes (I_S) and anti-Stokes (I_AS) Raman mode intensity through the equation I_S/I_AS = exp(2ΔE/k_BT^*)), where ΔE is the Raman shift and k_B the Boltzmann constant. The lattice temperature of a LTrp powder particle is estimated from the 756 cm⁻¹ mode (Fig. 1). The Trp temperature is approximately 350 K under a 2 mW (0.3 kW/cm² 488 nm excitation and 400 K under a 5 mW (0.5 kW/cm²) 633 nm excitation. On the other hand, using the 521 cm⁻¹ mode from the silicon substrate, we determine the substrate temperature to be approximately 340 K under a 2 mW 488 nm excitation and 380 K under a 5 mW 633 nm excitation, respectively. These temperatures are above room temperature, but are still below the temperature that can cause rapid thermal degradation of tryptophan (≳500 K) [30]. When the intensity of 488 nm excitation is increased to more than 1 kW/cm², orange to red fluorescence overwhelms the Raman scattering, and we are unable to determined the lattice temperature.

3.3. Emergence of red fluorescence

Tryptophan has three absorption bands in the UV region, specifically singlet states ¹B_b at ∼220 nm, ¹L_a at ∼275 nm, and ¹L_b at ∼290 nm [10, 15, 31, 32]. For a far-off resonance excitation at 488 nm, the absorbance and fluorescence in Trp is expected to be negligible. Nonetheless, spectrally broad emission appears in dry Trp powder particles even at a photoexcitation intensity as low as 0.3 kW/cm² (^{Figs. 1–2}), indicative of weak absorption at 488 nm (2.54 eV) in L-Trp samples being studied. This weak absorption is attributed to the transition to the spectrally broad triplet state centering around 430 nm (2.88 eV) [15, 33]. The triplet state is in principle a ‘forbidden’ transition; however, the weak mixing of singlet and triplet state via spin-orbit coupling and symmetry breaking in Trp powder particles can result in weak absorption at 488 nm [15, 34]. Moreover, the fluorescence intensity increases by more than two orders of magnitude after certain duration of exposure (∼30 min under a photoexcitation of 2 kW/cm², see ^{Figs. 3–4}) when the photoexcitation is increased to 1 kW/cm² or more. An irreversible photochemical transformation occurs. The products from the photochemical transformation of tryptophan particles are highly absorptive for visible light, and generate orange to red (550–750 nm) fluorescence under 488-nm or 633-nm laser excitation. To characterize such photo-chemical transformation and consequent products, we examine the spatially resolved Raman and fluorescence spectra with the varying photoexcitation intensity and duration of exposure (exposure time Δt) using a 488-nm or 633-nm laser excitation.

 

Fig. 2 Imaging Raman spectra. (a–c) sample SA and (d–f) sample SB. The laser excitation spot (∼ 40×20 μm) is ∼2–3 times smaller than the lateral dimensions of sample SA and slightly larger than those of sample SB. (a)&(d) Raman/fluorescence images. The white rectangles represent the region of interest area as defined by the entrance slit of the spectrometer. (b)&(e) Imaging spectra. (c)&(f) Cross-sectional profiles along the vertical direction (Y-position) for selected Raman modes as labeled.

Download Full Size | PDF

 

Fig. 3 (a) Temporal evolution of Raman and fluorescence spectra under 488 nm excitation at P = 14.6 mW (2.2 kW/cm²) (sample SC). The interference fringes in the spectra are due to the notch filter used to reject the scattered excitation laser light. The gap from 2.5 to 2.6 eV in the spectra is due to the notch filter and the scattered laser light is removed for clarity. (b) Spectrally integrated intensity vs. exposure time (Δt). Measurements of two separate particles are shown here: (i) sample SC for P = 14.6 mW (solid circles), and (ii) sample SD for P = 17.5 mW (open circles).

Download Full Size | PDF

 

Fig. 4 Raman scattering and Fluorescence as a function of excitation power P (intensity): P ∼2 mW (0.3 kW/cm²) to 25 mW (3.8 kW/cm²) (sample SB) (a) emission spectra for selected excitation power and exposure time. (b) Spectrally integrated fluorescence intensity vs. P. For P ≳ 5 mW, fluorescence intensity increases with the exposure time (Δt). The arrows indicate the sequence of the measurements for laser excitation spot fixed in a location on SB. There is an approximately 5 min break to keep the particle in darkness when P is increased. The spectrally integrated fluorescence intensity increases nonlinearly with P and Δt.

Download Full Size | PDF

3.3.1. Light scattering in Trp particles

First, we describe laser light scattering in and through the solid Trp powder particles, as revealed by the spatially resolved Raman spectra in two individual Trp particles (samples SA and SB) (Fig. 2). The lateral dimensions of sample SA are ∼2–3 times larger than the laser excitation spot (∼ 40 × 20 μm). The Raman scattering and background fluorescence are most visible near the central laser focal spot. However, the fluorescence/Raman microscope image of sample SA reveals additional spatially non-uniform signals across the entire particle, in particular around the edges [Fig. 2(a)]. A spatially resolved imaging spectrum vertically across the particle within a horizontal dimension of 3 μm is shown in Fig. 2(b). Raman scattering and fluorescence are present across the particle with lateral dimensions of tens to hundred microns, significantly larger than the initial laser excitation spot. In Fig. 2(c), we compare the vertical cross-sectional profile of fluorescence/Raman intensity at wavenumbers of three Raman modes:(i) ∼520 cm⁻¹ mode from the silicon substrate, (ii) ∼757 cm⁻¹ mode in Trp, and (iii) ∼3056 cm⁻¹ νCH mode in Trp. The presence of a silicon Raman signal across the particle suggests that the laser excitation diffusely scatters when propagating into and through the particle with its irregular surfaces and internal voids. Nonetheless, the intensity ratio of the Trp:757 cm⁻¹ or νCH mode to the Si:523 cm⁻¹ mode increases by a factor of approximately two at the edges. The measured background fluorescence also increases at the edges; therefore, the ratio of Raman scattering signal to background fluorescence is not necessarily improved at the edges of the particle. We further examine a separate sample SB with lateral dimensions slightly smaller than the laser excitation spot. In this case, the intensity profile of the silicon 523 cm⁻¹ Raman mode is nearly a Gaussian distribution, following the focused excitation laser beam profile. By contrast, the measured Trp Raman signals at the sample edges are two to three times higher than that at the center of the focal spot. The increased Raman and fluorescence intensities at the edges of the particle are likely due to an increased leakage and multiple light scattering caused by the particle’s rough surface, rather than an actual enhancement of Raman scattering.

3.3.2. Maturation – exposure time dependence

The fluorescence from the focal spot of the laser excitation increases with the duration of exposure time Δt and can rapidly overwhelm the Raman signal when the photoexcitation intensity is elevated to above 1–2 kW/cm². In Fig. 3, we study the temporal evolution of the emission spectra measured at the laser focal spot for an excitation intensity of 2.2 kW/cm² (power P = 14.6 mW, sample SC). Spectrally broad emission emerges and overwhelms the Raman signals with increasing exposure time. Here, the interference fringes in the spectra are due to the notch filter used to reject the scattered excitation laser light. Moreover, the fluorescence spectral peak shifts progressively to the red from approximately 570 to 630 nm with the increasing Δt (Fig. 3(a)). Such emergence of orange/red fluorescence previously unseen in Trp is indicative of the formation of new chemical products. The overall fluorescence intensity increases by more than two orders of magnitude over the exposure time of about 50 min. A dramatic increase of fluorescence over 1–2 min is observed near Δt ≈ 40 min. The transient fluorescence increase is followed by a sudden decrease and slow increase toward a saturation intensity (Fig. 3(b)). The spectrally integrated fluorescence intensity in a separate particle (sample SD) under a slightly higher excitation intensity (2.6 kW/cm²) is also shown in Fig. 3(b). The fluorescence intensity also exhibits a dramatic increase and slight decrease to a saturation intensity with Δt. The initially higher photoexcitation intensity used for SD leads to a faster transformation occurring at Δt ≈ 20 min. Note that the temporal evolution of fluorescence is qualitatively similar in terms of the intensity increase with Δt and spectral distributions for all particles studied; however, a quantitative measurement of the quantum efficiency and Δt required to complete the transformation is not available due to the aforementioned surface roughness and scattering in Trp particles with varying sizes and shapes used in this work.

3.3.3. “Brightness” and Photostability – photoexcitation intensity dependence

Next, we investigate the Raman and fluorescence spectra as a function of photoexcitation intensity. Figure 4 shows the emission spectra in sample SB [see also Fig. 2(d–e)] for selected excitation power/intensity (P). The fluorescence increases nonlinearly by four orders of magnitude when P is increased by about a factor of ten. Because the fluorescence intensity also increases with the exposure time (Δt), particularly for P > 1 kW/cm², we also show measurements for selected Δt to illustrate the range of intensity variation for a specific P. The overall fluorescence increase with P is highly nonlinear. Note that the spectra shown in ^{Figs. 3–4} are collected from an effective rectangular area of approximately 3 × 20 μm² and do not reveal the spatial inhomogeneity of the fluorescence distribution resulting from the photochemical transformation of Trp. Note that the integrated fluorescence intensity does not necessarily increase linearly with the photoexcitation dose P × Δt. The lack of a linear dependence on P × Δt suggests that the photochemical transformation is a nonlinear process which might be caused by photogeneration of products and impurities with stronger absorption than Trp, or in which multiple excited Trp molecules are involved, or in which photothermal effects play a role.

In Fig. 5, we describe the conversion process of Trp to red-emitting fluorescent products using bright-field and Raman/fluorescence microscope images. The photochemical transformation is more evident around the laser focal spot under 488-nm excitation with P ≈ 17.5 mW [Fig. 5(i–j)]. Although the fluorescence at the central focal spot increases with Δt, as shown previously, the fluorescence intensities peak at certain locations ∼20 μm away from the center of the focal spot. The conversation is irreversible and generally reaches a nearly steady state after ∼40–50 min under a moderate excitation intensity of 1–2 kW/cm². The bright-field microscope image of the sample after such a transformation reveals an oval dark area matching the beam profile of the excitation laser incident on the sample (Fig. 5b). When the same sample is excited with 488-nm at a much lower power [Fig. 5(h)], P = 2.1 mW (0.3 kW/cm²), the ring-shaped fluorescence pattern remains with intensity overwhelming the Raman scattering signals. Surprisingly, the fluorescence from the product(s) is also intense under a 633-nm excitation [Fig. 5(e)]. The presence of intense fluorescence in the product(s) from photochemical transformation of Trp under 633-nm and 488-nm with a low excitation intensity (<0.5 kW/cm²) when thermal heating is most likely negligible suggests that the conversion is permanent and produces stable fluorescent products.

 

Fig. 5 Fluorescence/Raman microscope images of a dry Trp particle (sample SD). (a–b) Bright-field microscope image under white-light illumination of (a) pristine Try particle (sample SD), and (b) same particle after the photochemical transformation by the exposure to 488-nm excitation at P = 17.5 mW for 50 min. (c) 633-nm and (f) 488-nm laser excitation profiles as measured by the emission from the silicon substrate. (d–e) Emission images under 633-nm excitation at P = 3.2 mW: (d) before, and (d) after the photochemical transformation. (g–h) Emission images under 488-nm excitation at P = 2.1 mW: (g) before, and (f) after the photochemical transformation. (i–j) Fluorescence images under 488-nm excitation at P = 17.5 mW: (i) exposure time Δt ∼ 1 min, and (j) Δt ∼ 50 min.

Download Full Size | PDF

3.3.4. Red fluorescence

To further characterize the fluorescence from the product(s), we measure spatially resolved fluorescence at selected locations across the exposed area. Fig. 6 shows spectra measured at three selected locations as labeled by the rectangular strips (≈3×20 μm²) on the fluorescence images of sample SC. Under the 488-nm excitation with P = 2.1 mW (0.3 kW/cm²), orange fluorescence with emission peaks around 570–620 appears across the area previously exposed to ∼ 15 mW 488-nm excitation for about 60 min. The fluorescence intensity distribution peaks at certain locations around the perimeter of the excitation spot. Under a 633-nm excitation with P = 5.6 mW, red fluorescence appears with spectral peaks near 700–750 nm. The red fluorescence intensity is nearly as intense as that under a 488-nm excitation. Here, we observe two main types of spectra: one with a relatively flat spectral intensity from 650–750 nm (1.9–1.7 eV), and one with a clear peak near 720–740 nm (1.72–1.68 eV). Two or more products are possibly produced in the photochemical transformation of dry Trp, albeit we are unable to identify the molecular structures of the products yet.

 

Fig. 6 Spatially resolved fluorescent spectra of the products of dry tryptophan (sample SC) after exposure of 488-nm excitation at P = 14.6 mW (2.2 kW/cm²) for 60 min. Laser excitation are at P = 5.6 mW for 633 nm and P = 2.1 mW for 488 nm, respectively. The spectra are integrated over an effective area of 3×20 μm² as illustrated by the rectangles labeled as L, C, and R on the inset fluorescence images. Insets: false-color optical fluorescence and bright filed microscope images of sample SC. (Top) 488-nm excitation, P = 2.2 mW, (Middle) 633-nm excitation, P = 5.6 mW, and (Bottom) white-light illumination.

Download Full Size | PDF

4. Discussion

The photochemical and photophysical properties and related photodegradation of tryptophan under UV radiation or thermal heating have been studied [31, 35–38]. Selected main products identified from the UV irradiation of tryptophan are hydroxytryptophan, N-formylkynurenine, N′-formylkynurenine, kynurenine, residues consistent with the dehydration of kynurenine, and hydroxykynurenine. When solutions of proteins are exposed to sunlight [39], the tryptophans can be photo-oxidized to N′-formylkynurenine. Then kynurenine can be released from these photo-oxidized proteins [39]. The emission peaks for these products are as follows: 350–370 nm (5-hydroxytryptophan), 420 nm (N-formylkynurenine), 440 nm (N′-formylkynurenine), 460 nm (kynurenine), and 495 nm (3-hydroxykynurenine). None of these products have been reported to produce red fluorescence (λ > 600 nm) with a visible light excitation (∼450–650 nm).

Considering a hypothetical photochemical transformation with N molecules of excited tryptophan Trp^* and a temperature dependent reaction rate, we can approximate the yield of the fluorescent product(s) as ∝ exp(k_BT/Δ_a)[Trp^*]^N, where Δ_a is the activation energy. The number of excited Trp^* does not necessarily increase linearly with the photoexcitation intensity, owing to a possible photothermal effect and associated temperature dependent absorption coefficient. Nonetheless, for simplicity, we assume the number of Trp^* is proportional to the photoexcitation P. Hence, one can envision a highly nonlinear or exponential increase of fluorescence intensity with increasing P (Fig. 4), considering a combination of photochemical and photothermal effects. However, currently we cannot determine the quantum yield and reaction rate, owing to not only the spatial inhomogeneity in dry Trp particles, but also the lack of knowledge about the photochemical reaction process and the chemical structures of the products.

The experiments described here are conducted in dry solid Trp particles deposited on a silicon substrate. We use a silicon substrate because of the absence of luminescence in the visible wavelengths and the presence of the well characterized 520 cm⁻¹ Raman mode that can be used to monitor the substrate temperature at excitation intensities below 0.5 kW/cm². The orange and red fluorescence in photochemical transformation in Trp particles have also been observed on an aluminum-coated microscope slide (not shown), suggesting the conversion of Trp into red-emitting product(s) is independent of the substrate. Hence, the red-emitting products from dry Trp are mostly due to photochemical transformations, possibly expedited by a photothermal effects resulting from weak absorption in Trp particles. We note that spectrally broad photoluminescence has also been observed in defective surfaces of silica and related solid-state optical materials [40,41]. The quasi-continuum luminescence in those wide band gap materials is attributed to photo-active defects.

In addition, when the laser excitation intensity is increased above 10 kW/cm², Raman scattering peaks associated with the G-mode (∼1580 cm⁻¹) and D-mode (∼1360 cm⁻¹) in graphite emerge, indicative of carbonization and formation of nano-crystalline graphite. The ratio R of the intensity of the Raman band at 1360 cm⁻¹ (D-mode) against that at 1580 cm⁻¹ (G-mode) as well as the linewidth can be used to estimate the graphite crystalline sizes [42–44]. The carbonization in this high photoexcitation regime will be reported elsewhere.

One of the critical properties of a fluorescent probe is its photostability (or the absence of photodegradation). Nearly all fluorophores are photobleached upon continuous exposure to photoexcitation, particularly in fluorescence microscopy where the illuminating light intensities are significant. By contrast, red-emitting (∼580–700 nm) products are produced from photochemical transformation of dry tryptophan when exposed to a moderate 488-nm CW laser excitation. Stable red fluorophores are relatively scarce as compared to green fluorophores. In contrast to UV/blue/green photoexcitation, the longer excitation wavelengths (lower energies) used for red fluorophores results in less absorption by the the biological specimen. Consequently, photodegradation (phototoxicity) is suppressed and deeper probe into biological specimen is feasible. Moreover, detectors/CCDs with a nearly unity quantum efficiency, low-cost lasers (e.g., HeNe lasers at 561, 594, and 633 nm), and a wide variety of optical components in the visible optical regime are available. Therefore, it is worth analyzing the chemical structures of the fluorogens from the photochemical transformation of dry Trp for potential applications as a red fluorophore in live-cell imaging.

5. Conclusion

The transformation of dry tryptophan particles into a red fluorescent product occurs at a photoexcitation intensity as low as ∼1 kW/cm² for a 488-nm excitation. To circumvent the fluorescence resulting from such a transformation, it is necessary to measure the Raman spectra of dry tryptophan using near-IR excitation, e.g., wavelengths of 785 nm or 808 nm. It is likely that a NIR laser excitation is also necessary for Raman studies of related solid-state biomaterials or bio-aerosol particles containing Trp and related amino acids. We further characterize orange and red fluorescence in the photo-products during and after the photochemical transformation. When the photoexcitation intensity is increased to 1 kW/cm² or more, Trp particles undergoes an irreversible photochemical transformation, leading to an increase of orange fluorescence intensity of more than two orders of magnitude. Under a 633-nm laser excitation, the products generate emissions ranging from 650 to 800 nm, making it a potential red-emitting fluorophore for biological imaging applications. Our results are potentially relevant to studies of photodegradation or photochemical transformation in tryptophan-containing proteins and bio-materials, e.g., yellowing in wool [36, 38], and aging and cataract formation in human lenses [22, 45, 46].