Measuring the Efficiency of Using Raman Photoexcitation to Generate Singlet Oxygen in Distilled Water

Abstract

We determine the efficiency of generating singlet oxygen molecules through Raman excitation in distilled water. Focused nanosecond light pulses in the spectral blue region induce a Raman transition toward the singlet oxygen state, generating a Stokes signal in the red spectral region. The signal is proportional to the number of photons corresponding to the number of excited oxygen molecules. We calculate the efficiency by dividing the number of generated singlet oxygen molecules by the number of incoming pump photons, determining an efficiency of (8 ± 2) × 10⁻⁵ for water when pumping at 410 nm with a pulse energy of 13 mJ. We demonstrate that the Raman method results in no photobleaching, a phenomenon typically observed when photosensitizers are used. Thanks to this property, Raman excitation can continue for as long as the sample is irradiated, generating more singlet oxygen molecules over time than the photosensitization method.

1. Introduction

Singlet oxygen (¹O₂) is the lowest electronically excited quantum state of molecular oxygen. In this state, the molecule is highly electrophilic, facilitating its broad application in photochemistry, photobiology, and photomedicine [1,2,3,4,5,6,7,8]. Quantum mechanics forbids direct one-photon excitation toward the ¹O₂ state, resulting in a low transition probability. Photosensitization is the most common method of efficient ¹O₂ photoproduction [8,9,10,11,12,13]. A large photosensitizer molecule absorbs visible (VI) light, accumulating the absorbed energy on a triple metastable level. From there, the energy is transmitted to surrounding oxygen molecules, exciting them to the ¹O₂ state. We recently demonstrated a photosensitizer-free method of ¹O₂ photoexcitation based on the Raman effect [14,15,16]. A pump photon produces a Raman transition from the ground to the ¹O₂ level, emitting a Stokes photon of lower energy. The difference between the pumping and Stokes photon energies is equal to the energy needed for the transition. The absence of a photosensitizer simplifies the procedure and prevents potential secondary effects, such as excessive photosensitivity or unexpected phototoxicity. In a recent publication, we demonstrated the detection of phosphorescence at 1270 nm from Raman photoexcited ¹O₂ [16]. Here, we determine the efficiency of the process (Φ_Δ), an aspect that was missing in our previous study. We define Φ_Δ as the relation between the number of Raman-generated ¹O₂ molecules and the total number of pump photons. We measure a Φ_Δ of (8 ± 2) × 10⁻⁵ in aerated distilled water (H₂O) when pumping photons at 410 nm with nanosecond pulses of 13 mJ.

Furthermore, we compare the Raman approach with the photosensitization method using the well-known photosensitizer, Rose Bengal. The ¹O₂ chemical trap, uric acid, is used to evaluate both methods. We show that, over time, the Raman method generates more ¹O₂ molecules than the photosensitization method. Rose Bengal exhibits near-full photobleaching after sixty minutes of irradiation, interrupting ¹O₂ photoproduction. In the Raman method, photobleaching does not take place. Consequently, Raman ¹O₂ generation continues without interruption even after hours of irradiation.

2. Theoretical Considerations

Figure 1a shows a simplified schematic of ¹O₂ Raman excitation. A pump photon induces a Raman transition from the ground to the ¹O₂ level, emitting a Stokes photon. For example, when pumping at 410 nm in H₂O, the ¹O₂ Stokes component occurs around 605 nm. The magnitude of this signal yields a direct measurement of the number of excited ¹O₂ molecules. Other Stokes photons related to the excitation of solvent vibrational modes and their overtones are generated. When pumping at 410 nm, the H₂O stretching vibrational mode produces a Stokes component at 475 nm. An additional overtone of this Stokes signal is observed at 566 nm. The difference in wavelength between the Stokes components allows for their identification and separation using optical filters. The experiments described below demonstrate that most of the energy involved in the Raman process originates from the solvent stretching modes. Despite this fact, ¹O₂ Stokes photons can still be detected, showing that a significant number of ¹O₂ molecules are still being generated.

In the photosensitization method, the photosensitizer’s quantum yield measures the efficiency of ¹O₂ photoproduction [17,18,19,20,21]. The photosensitizer’s quantum yield is defined as the number of excited oxygen molecules divided by the number of photons absorbed by the photosensitizer. The commonly used photosensitizer Rose Bengal exhibits a quantum yield value of 0.76 at 532 nm in H₂O [22,23]. As discussed above, for the Raman method, Φ_Δ is defined by the following equation

$${\mathrm{\Phi }}_{\mathrm{\Delta }}={\displaystyle \frac{N_{\mathrm{\Delta }}}{N_p}}$$

(1)

where N_Δ and N_p are the numbers of ¹O₂ Stokes and pump photons, respectively. The number of photons can be estimated by measuring the signal energy corresponding to each process divided by the energy of one photon. Then, we can write

$${\mathrm{\Phi }}_{\mathrm{\Delta }}={\displaystyle \frac{{\overline{\lambda }}_{\mathrm{\Delta }}\times E_{\mathrm{\Delta }}}{{\lambda }_p\times E_p}}$$

(2)

where ${\overline{\lambda }}_{\mathrm{\Delta }}$ is the averaged ¹O₂ Stokes wavelength, E_Δ is its energy, λ_p is the pump wavelength, and E_p is its energy.

3. Materials and Methods

Figure 1b shows the experiment setup. An optical parametric oscillator (OPO, OPOTEK, Carlsbad, CA, USA) provides 6-nanosecond excitation in the blue-green region (410–520 nm) with an average energy per pulse of 13 mJ. A beam splitter, B, directs part of the light toward a reference detector. A 20 cm focal length lens, L₁, focuses the pump light onto a 10 cm path-length glass cuvette containing the sample. A neutral filter (NDF1) placed at the exit of the sample significantly reduces the power to avoid optical damage to the filters. A lens, L₂, focuses the resulting signal onto the detector. An additional neutral density filter, NDF2, is placed at the entrance of the detector, reducing the signal intensity to avoid detector saturation. To detect the signal, we use a silicon detector (Thorlabs DET100A2, Newton, NJ, USA) with a responsivity value of 0.2 A/W at 410 nm and 0.4 A/W at 610 nm and a rise time of 35 ns.

The silicon detector signal is sent without further amplification toward a digital oscilloscope (TDS 3052, Tektronix, Beaverton, OR, USA) for averaging and display. The oscilloscope averages 512 pulses, providing a stable signal proportional to the emitted luminescence. Due to their Raman character, the Stokes signals follow the time dependence of the pump pulse. Since the detector risetime is longer than the excitation pulse, it limits the signal time evolution. To calibrate the signal provided by the silicon sensor, we use calibrated detectors. To measure pulse energies above 100 μJ, we use a pyroelectric energy meter (PE25BF-DIF-C, Ophir USA, Newport Corporation, North Logan, UT, USA). To measure energies below 100 μJ, we use a diode power meter (Thorlabs PM100, Newton, NJ, USA).

We detected the pump signal using the corresponding interference filter. When pumping at 410 nm, we used an interference filter at (410 ± 10) nm. The ¹O₂ Stokes signal was measured using a long-pass filter. When pumping at 410 nm, we used a long-pass filter with a cutoff at 610 nm. Below, we show that the collected signal corresponded to the number of ¹O₂ molecules. The Stokes signals from the solvent stretching modes were evaluated in a similar manner. We chose 410 nm as a pumping wavelength to illustrate the procedure for determining the efficiency of the Raman method. Different pumping wavelengths can be used. Previous publications show that ¹O₂ Raman photogeneration is possible for different wavelengths [14,15,16]. This would require conducting experiments similar to those discussed in the Results section and the corresponding use of different sets of optical filters.

Table 1. The experimental parameters of the ¹O₂ Raman photogeneration experiment.

Pump wavelength	410 nm
Pulse width	6 ns
Repetition rate	10 Hz
Pulse energy	13 mJ
Beam area at waist	8.5 × 10−7 cm2
Pulse intensity at waist	2.5 × 1012 W/cm2
Cuvette path length	10 cm
Focusing lens focal length	20 cm
Detector Type	DET100 A2 (Thorlabs)
Detector rise time	35 ns

shows the main experimental parameters used for the ¹O₂ Raman photoexcitation procedure.

Aerated distilled H₂O in 10 cm pathlength glass cuvettes was used as a sample. We used solutions of sodium bisulfite (NaHSO₃) as an oxygen quencher to aid in identifying the oxygen contribution [24]. The concentration of dissolved oxygen in H₂O decreased when the NaHSO₃ concentration increased. Water solutions of NaHSO₃ exhibited a reduced ¹O₂ Stokes signal, corresponding to a reduced oxygen concentration. To measure the amount of oxygen in the solutions, we used an oxygen optical sensor (Vernier Optical DO Probe, Vernier, Beaverton, OR, USA).

We used uric acid as an ¹O₂ chemical trap to compare the efficiency of the photosensitization and Raman methods. Despite its limitations, uric acid is commonly used to quantify ¹O₂ generated through photosensitizer reactions [25]. Parabanic acid has been identified as the singlet oxygen specific oxidation product of uric acid [26]. The uric acid UV absorbance at 294 nm is a direct measurement of the amount of ¹O₂ being generated [27,28]. We use Rose Bengal as a photosensitizer. A solution of 10 μM Rose Bengal in H₂O was prepared.

To collect the spectra of the Stokes signals, we use a spectrometer (Red Tide USB 650, Ocean Optics, Orlando, FL, USA).

4. Results and Discussion

The solvent Stokes contributions are significant, but can be eliminated using optical filters. Figure 2a shows the spectra collected using the Ocean Optics spectrometer after passing through neutral filters with a total optical density of six when pumping distilled water at 410 nm. The first peak at 410 nm corresponds to the pumping field. The peak at 475 nm corresponds to the H₂O stretching mode at a frequency of 3350 cm⁻¹ [29,30,31]. This mode is the dominant part of the total Stokes field.

Figure 2b shows the spectrum after using a neutral-density filter with an optical density of two and a long-pass filter with a cutoff at 550 nm. A diminished Stokes peak at 475 nm and a peak at 566 nm are observed. The last peak corresponds to the first overtone of the H₂O vibrational mode at 6700 cm⁻¹. When using a neutral-density filter with an optical density of one and a long-pass filter with a cutoff at 610 nm, the solvent Stokes components at 475 nm and 566 nm are fully depleted (see Figure 2c). The arbitrary units in all sub-figures, Figure 2a–c, are the same. Thus, the signal depicted in Figure 2c is four orders of magnitude smaller than the signal shown in Figure 2a and one order of magnitude smaller than the signal shown on Figure 2b. To demonstrate that the curve shown in Figure 2c is oxygen-dependent, we reduced the oxygen concentration using the oxygen quencher NaHSO₃.

Figure 3a shows the singlet peak obtained when using the long-pass filter at 610 nm for different concentrations of NaHSO₃. The blue line corresponds to the condition under which no NaHSO₃ is added. As the NaHSO₃ concentration increases, the signal decreases. At a concentration of 40 mg/mL, the signal was significantly reduced. Using the optical oxygen sensor, we measured the oxygen concentration in each sample solution. Figure 3b shows how the oxygen concentration decreased as the NaHSO₃ concentration increased. A nearly linear dependence is observed for concentrations below 10 mg/mL. Figure 3 shows that the observed Stokes signal collected using the 610 nm long-pass filter is oxygen-dependent. Its amplitude provides a direct measurement of the number of ¹O₂ molecules generated through the process. Using a calibrated power sensor, we estimated the energy corresponding to the number of excited ¹O₂ molecules. We used the same detector to measure the energy of the incoming pumping beam. Dividing both magnitudes indicates the efficiency of the process. This measurement approach is more convenient than detecting ¹O₂ phosphorescence at 1270 nm. The phosphorescence signal is weak, several orders of magnitude smaller than the Stokes signal shown in Figure 3a. Sensitive photon-counting detection methods are generally applied for this purpose. The phosphorescence method measures the number of ¹O₂ molecules independently from the excitation mechanism. In this study, the Stokes signal detected can only be due to the Raman mechanism employed. Measuring the Stokes signal produced via ¹O₂ photoexcitation is a more direct and precise method that does not require complex calibration procedures. We note that the H₂O Stokes components are not affected by the presence of NaHSO₃.

Figure 4a shows the pulse corresponding to the excitation pump light (blue light). The signal was collected using the diode detector and energy sensor for calibration purposes. We used an interference filter at 410 ± 10 nm to select only the pump light. Using an interference filter at 475 ± 10 nm, we measured a Stokes signal corresponding to the H₂O stretching mode at 3350 cm⁻¹ (red line in Figure 4a). A significant portion of the pump field, 23 ± 1%, was converted into this Stokes component. We also measured the contribution from the overtone at 566 nm using the corresponding filter, which accounted for 0.25 ± 0.01% of the total incoming light. The ¹O₂ Stokes contribution is significantly smaller. Figure 4b shows the ¹O₂ Stokes signal and the pumping signal on a logarithmic scale. We estimated a Stokes pulse energy of 0.7 ± 0.2 μJ. Under these conditions, Equation (2) provides Φ_Δ = (8 ± 2) × 10⁻⁵.

Table 2. The energy, efficiency, and relative intensity of the Stokes signals from the H₂O vibrational mode and its overtone, as well as the ¹O₂ Stokes when pumping at 410 nm.

Stokes Signal	Signal Energy (mJ)	Efficiency	Relative Intensity
H2O mode at 475 nm	3.2 ± 0.14	0.23 ± 0.01	1
H2O overtone at 566 nm	(35 ± 1.4) × 10−3	0.0025 ± 0.0001	10−2
1O2	(7 ± 2) × 10−4	(8 ± 2) × 10−5	2.2 × 10−4

summarizes the results showing the efficiency calculated using Equation (2), the relative intensity and energy of the H₂O Stokes vibrational mode at 475 nm and its overtone at 566 nm, and the ¹O₂ Stokes values when pumping at 410 nm. The standard deviation is estimated over ten experiments performed under the same conditions. Due to the definition of the efficiency as the ratio between the number of ¹O₂ molecules excited and the number of incoming pump photons, there is no need to estimate losses due to reflection, scattering, or other processes. From the Φ_Δ value, we estimate that each pulse of pumping light generates approximately 4 × 10^{11 1}O₂ molecules. This is equivalent to a concentration of 3.6 nM of ¹O₂ molecules per pulse in the volume defined by the pump pulse inside the sample.

To understand the potential practical applications of the Raman method, we compared it with a conventional photosensitization method. We used the well-known photosensitizer Rose Bengal for this purpose. We prepared solutions of Rose Bengal in H₂O at a concentration of 10 μM. We also used uric acid at a concentration of 120 μM as a ¹O₂ chemical trap to monitor the number of ¹O₂ molecules generated via each method. We irradiated the Rose Bengal and uric acid solution at 532 nm in five-minute intervals. At the end of each interval, we measured the UV spectrum of the solution to monitor the peak amplitude at 294 nm.

Figure 5a shows the UV-VI absorbance spectra of the Rose Bengal solution containing uric acid at a concentration of 120 μM, measured for different irradiation times. We observe a reduction in the uric acid peak at 294 nm, corresponding to the depletion of ¹O₂ generated by the uric acid. The peak depletes by a factor of four after twenty minutes but does not entirely disappear. The spectra also show the Rose Bengal absorbance peak at around 540 nm. This peak also decreases, indicating photobleaching. The photobleaching of Rose Bengal halts the process of ¹O₂ generation. Figure 5b shows the results obtained for a solution of uric acid with no photosensitizer in H₂O. The experiments shown in Figure 5a,b were conducted under identical irradiation and sample geometry conditions. In this experiment, irradiation was performed at 410 nm over longer time intervalsof 60, 90, 120, and up to 480 min. Again, we observed depletion of the uric acid peak at 294 nm, evidencing the Raman generation of ¹O₂ molecules. No photobleaching effect was observed in the Raman experiment, resulting in a more complete reduction in the 294 nm peak over time. Figure 5c shows the uric acid absorbance at 294 nm as a function of irradiation time for both experiments. For the Rose Bengal solution (blue corssed squares), the uric acid absorbance at 294 nm decreased from an initial value of 2.3 to 0.3 after 20 min of irradiation. It remained around this value for the rest of the experiment. For the Raman experiment (red crossed circles), the peak reduced from 2.3 to a value around 0.1 after 400 min of irradiation. The Rose Bengal experiment exhibited fast photoproduction of ¹O₂. However, due to photobleaching of the photosensitizer, the process was not fully completed. In the Raman experiment, ¹O₂ photoproduction was less effective. However, because of the absence of photobleaching, ¹O₂ photoexcitation persisted as long as the sample was being irradiated.

5. Conclusions

We estimated the efficiency of Raman ¹O₂ photogeneration in H₂O by measuring the signal amplitude of the ¹O₂ Stokes signal. Defining the efficiency as the number of photogenerated ¹O₂ molecules divided by the number of pump photons, we determined an efficiency of (8 ± 2) × 10⁻⁵ for H₂O at 410 nm. The main reason for this low value is the relatively low number of oxygen molecules in the solution. Most of the Stokes energy is collected by the H₂O molecules, the number of which is more than five orders of magnitude larger than the number of oxygen molecules. We compared the Raman method with photosensitization using Rose Bengal as a photosensitizer. The photosensitizer generates ¹O₂ molecules rapidly, but the process eventually halts due to photobleaching of the photosensitizer. The Raman method generates ¹O₂ molecules at a slower rate than the photosensitization methods. However, because the Raman method is not hampered by photobleaching generates more ¹O₂ molecules over time than the photosensitization method, despite its low efficiency. The efficiency of the Raman method can be enhanced based on the fundamental principles of Raman interactions. Raman lasers, Raman amplifiers, and Raman fiber sensors are based on these principles [32,33,34]. The Stokes signal grows exponentially with distance, and its gain depends on the pumping field intensity, the nonlinear Raman susceptibility, and propagation losses. A system should be designed to provide significant gain for the ¹O₂ Stokes component with low losses, while maintaining low gain and substantial solvent Stokes components losses. The role of resonant effects in Raman excitation is another interesting avenue of exploration. For example, resonant stimulated Raman excitation may occur with the participation of charge-transfer (CT) states [35]. These factors may play a crucial role in improving ¹O₂ Raman photogeneration efficiency for practical applications of the Raman approach. The presence of impurities or ions in the sample may represent an alternative channel producing ¹O₂. However, the efficiency of the Raman method is not expected to be affected since these other channels do not contribute to the ¹O₂ Stokes response. The Raman method may be applied in the purification of water contaminated with viruses or bacteria. However, scattering, absorption, and the presence of quenchers can limit the efficiency of purification. In a previous communication, we demonstrated the deactivation of viruses in an aqueous environment through simple irradiation with blue light, without using photosensitizers [36].