Photomolecular effect: Visible light interaction with air–water interface

Significance

We use 14 different experiments to demonstrate the existence of the photomolecular effect: photons in the visible spectrum cleave off water clusters from air–water interfaces. We use laser to study single air–water interfaces and show polarization, angle of incidence, and wavelength dependent responses, peaking at green where bulk water does not absorb. Raman and infrared absorption spectra and temperature distribution in air show the existence of water clusters under light. We suggest the photomolecular effect provides a mechanism to resolve the long-standing puzzle of larger measured solar absorptance of clouds than theoretical predictions based on bulk water optical constants and demonstrate that visible light can heat up clouds. Our work suggests that photomolecular evaporation is prevalent in nature.

Abstract

Although water is almost transparent to visible light, we demonstrate that the air–water interface interacts strongly with visible light via what we hypothesize as the photomolecular effect. In this effect, transverse-magnetic polarized photons cleave off water clusters from the air–water interface. We use 14 different experiments to demonstrate the existence of this effect and its dependence on the wavelength, incident angle, and polarization of visible light. We further demonstrate that visible light heats up thin fogs, suggesting that this process can impact weather, climate, and the earth’s water cycle and that it provides a mechanism to resolve the long-standing puzzle of larger measured clouds absorption to solar radiation than theory could predict based on bulk water optical constants. Our study suggests that the photomolecular effect should happen widely in nature, from clouds to fogs, ocean to soil surfaces, and plant transpiration and can also lead to applications in energy and clean water.

Evaporation of water is ubiquitous in nature and industrial technologies. The known mechanism for evaporation is “thermal evaporation”, which highlights that the energy input for evaporation is via heat. Due to the weak absorption of water to visible light (1, 2), the first step in using solar energy to evaporate water is usually by converting it into heat through photothermal processes via additional absorbing materials (3–9). The approach of floating a porous material on water surface subjected to solar radiation for evaporation, i.e., solar-interfacial evaporation (6–8), has led to surprising results that the evaporation rate can surpass the thermal evaporation limit (9–12). Based on experiments on hydrogels, we hypothesized that visible light can directly cleave off water molecular clusters and called this process the photomolecular effect (13), in analogy to the photoelectric effect discovered by Hertz (14) and explained by Einstein (15). However, evaporation from porous materials is difficult to quantify due to unknown internal structures. Here, we use a plethora of experimental methods to study visible-light (using mostly lasers) interactions with a single air–water interface. Our experiments show polarization, wavelength, and angle of incidence dependence of the photomolecular effect, which peaks at transverse-magnetic (TM) polarized green light at ~45° angle of incidence. We also show visible light can heat up fog, suggesting that the photomolecular effect provides a theoretical basis for resolving an 80-y-old puzzle on anomalous clouds absorption. Our work solidifies the photomolecular effect as a universal phenomenon happening widely in nature.

In our previous work (13), we made the following unexpected experimental observations on evaporation from hydrogels under visible light illumination (13). a) Partially wetted hydrogels become absorbing in the visible spectral range, where the absorption by both the water and the hydrogel materials is negligible. b) Illumination of hydrogel under solar or visible-spectrum light-emitting-diode (LED) leads to evaporation rates exceeding the thermal evaporation limit, even in hydrogels without additional absorbers. c) The evaporation rates are wavelength-dependent, peaking around 520 nm, despite that the absorptance does not show wavelength dependence in the same spectral range. d) Temperature of the vapor phase becomes cooler under light illumination, and its distribution shows features different from thermal evaporation. And e) vapor phase transmission spectra under the light show new features and peak shifts. We proposed the following possible photomolecular evaporation mechanism (Fig. 1A) to explain experimental observations on hydrogel. First, as shown in Fig. 1A, the density of water changes over a distance of a few angstroms (~3 to 7 $\mathrm{\AA }$) from the liquid to the vapor phase (16, 17). Second, the incident light creates an electrical field gradient over the interfacial region. Based on the macroscopic Maxwell equations, the perpendicular component of the electric displacement field from the electromagnetic wave is continuous at the interface, i.e., ${\epsilon }_1{E}_{1\perp }={\epsilon }_2{E}_{2\perp }$, where ε₁ = 1 for air and ε₂ = 1.8 for water in the visible spectrum, and the subscript $\perp$ emphasizes the direction perpendicular to the interface (18). This condition means that the electrical field in the perpendicular direction is reduced by nearly a factor of two from air to water over a distance of a few angstroms, creating a large field gradient. Third, water molecules are polar and form fluctuating hydrogen bond networks, also called water clusters, although the details of the networks are still under debate (19–23). The dipole moment of a single water molecule is 1.8D and increases with the cluster size (24), which means an effective charge separation of 0.5 $\mathrm{\AA }$ or larger. Hence, the electrical field changes over the distance of a single water cluster at the interface are appreciable. Due to this variation, the force acting on the positive and negative charges of water molecules does not cancel out, leading to a net force exerting on the molecular clusters and pulling them from the interfacial region to the adjacent free space. This is similar in some ways to the surface photoelectric effect, also due to the large electrical field across 1 to 2 $\mathrm{\AA }$ where the electron density changes across the interface (25–27). From the quantum picture of light, the interaction of a photon and a molecular cluster needs to conserve energy and momentum. The bond energy between one water cluster and its surrounding water molecules should be weakened, roughly between the normal hydrogen bonds (~0.26 eV) and the van der Waals bond (~0.026 eV). A green photon with an energy of ~2.5 eV can meet the energy conservation requirements via cleaving off many such bonds. Momentum conservation is satisfied because of the rapid change of the electrical field in the perpendicular direction, as is the case for the surface photoelectric effect (26).

Fig. 1.

Proposed photomolecular effect and measured water temperature rise under laser illumination. (A) The photomolecular effect happens when a TM-polarized laser shines on the air–water interface at an angle such that an electrical field component perpendicular to the surface changes rapidly across the interface, leading to a net force acting on polar water molecular clusters. The clusters are driven out of the air–water interface when the photon energy matches the required energy to cleave off the cluster from its surroundings. (B) Temperature rise of the water surface as a function of time after a 532 nm green laser at 45° incident angle is turned on, measured by an IR camera. Time zero is defined when the laser turns on. The surface water temperature rise under the TM-polarized laser is much larger than under the TE-polarized laser. (C and D) Temperature rise as a function of incident angle measured by (C) an IR camera and (D) a thermocouple, shows consistency between the two, with a peak at 45° under the TM-polarized laser. The power of laser is 1.4 W. The 1/e² radius of the laser is 7.5 mm. The radius of the air–water interface is 15 mm.

Sum-frequency spectroscopy is an established method to study light interaction with a liquid interface (28). However, such a nonlinear process requires high laser power. IR spectroscopy and Raman spectroscopy had been used to study water clusters produced in well-controlled conditions (29–32). Our aim is to show that photomolecular effect happens at the air–water interface even under low light intensities comparable with solar radiation, for which there are no established methods. We probe the photomolecular effect by studying the thermal and optical responses in the bulk water and the vapor phases, through measuring their polarization, angle, and wavelength dependence and directly observing the spectra signatures of clusters in the infrared region.

Results

Liquid-Water Surface Temperature Response.

We used an IR camera (Fig. 1 B and C) and a thermocouple (Fig. 1D) to measure the temperature response of the water surface induced by a green laser under different angles of incidence and polarizations (Materials and Methods; SI Appendix, Note S1 and Table S1http://www.pnas.org/lookup/doi/10.1073/pnas.2320844121#supplementary-materials). Fig. 1B illustrates the water surface temperature response with time measured by the IR camera with the incident angle of 45° for both the transverse magnetic (TM) and transverse electrical (TE) polarized lasers. For the TM-polarized laser, the water surface temperature increases by 0.7 °C in the first 30 s after the laser turns on. It reaches a steady state with a rise of 0.95 °C after ~200 s. In contrast, the temperature rise under the TE-polarized laser is less than 0.3 °C.

The water surface temperature rises measured by both IR camera (Fig. 1C) and thermocouple (Fig. 1D) are consistently higher under the TM-polarized laser than under the TE-polarized beam. Furthermore, the temperature rise shows strong incident angle dependence, peaking at ~45° under the TM-polarized laser beam; in contrast, the temperature rise is nearly zero at normal incidence. SI Appendix, Fig. S2 shows power dependence of the temperature rise at 45° incident angle, which is approximately linear.

Vapor-Phase Temperature Distribution Above an Evaporatively Cooled Surface.

We compare in Fig. 2A the vapor-phase temperature distributions with and without laser illumination, measured with a thermocouple (black circles and blue triangles) and an IR camera (black and blue line) (Materials and Methods; SI Appendix, Fig. S3 and Note S2). Without laser illumination, water is cooled down from room temperature (22.8 °C) due to natural evaporation, leading to a temperature gradient from air to water. However, we observe that when the TM-polarized green-laser with an incident angle of 45° is turned on, the temperature of the vapor phase (Fig. 2A and SI Appendix, Fig. S4) at 2 mm above the air–water interface decreases from 21.9 to 21.4 °C, while the water surface temperature increases from 20.7 to 21.3 °C. This phenomenon can be interpreted according to our hypothesized photomolecular effect (SI Appendix, Fig. S5). Visible light can cleave off water clusters from the air–water interface and inject them into the air. Some of these clusters break up, absorbing heat and cooling down the air. Since the air convection is small, most of the clusters will condense back to the water surface, releasing heat to increase the water surface temperature.

Fig. 2.

Vapor phase temperature and refractive index responses. (A) Vapor-phase temperature distribution above the air–water interface before and after green TM-polarized laser illumination. Solid lines (black when laser off, blue when laser on) measured by an IR camera, circles (laser off), and triangles (laser on) measured by a thermocouple. (B) Schematics of the mirage effect experiment: a probe laser beam (635 nm) above the air–water interface deflects due to the refractive index change in air created by green laser illumination as in (A). (C) Effective average refractive index change as a function of the probe beam distance above the air–water interface under the TM-polarized and TE-polarized green pump laser. (D) Schematics of steady-state transmitted beam deflection measurement. (E) Normalized beam deflection as a function of the incident angle of a green pen laser (520 nm) for TM (black squares) and TE (red circles) polarizations. (F) Normalized beam deflection as a function of wavelength for TM and TE polarizations at 45° incident angle. (G and H) Comparison of vapor phase temperature distributions measured away from the hot air–water interface (53 °C) with and without laser. The power of laser is 1.4 W. (I) Explanation for the vapor phase temperature distribution above hot water.

Vapor-Phase Refractive Index Change under Light.

We probe the refractive index change of the vapor phase above the air–water interface through the mirage effect (33, 34) (Fig. 2 B and C; Materials and Methods; SI Appendix, Figs. S6 and S7 and Note S3). A green laser with a power of 1.4 W and an incident angle of 45° works as pump and a red pen laser (635 nm) works as probe (red dash line in Fig. 2B) with a small incident angle of ~0.033 rad relative to the water surface. The probe beam displacement is measured by a beam position sensing detector between pump off and 30 s after pump on, which can be mapped into the refractive index change along the pathlength. The effective average change of the refractive index ($\overline{\mathrm{\Delta }n}$) of the vapor phase across the water surface (~30 mm in diameter) under the TM-polarized pump laser (wavelength 532 nm, power 1.4 W, 1/e² radius 7.5 mm) is −6.7 × 10⁻⁴, −5 × 10⁻⁴, and −2.5 × 10⁻⁴ at 1 mm, 3 mm, and 5 mm above the air–water interface, respectively. In contrast, the $\overline{\mathrm{\Delta }n}$ of the vapor phase is lower than the measurement resolution (10⁻⁴) when the same pump laser is TE-polarized. Since the thermo-optic coefficient ($dn/dT$) of air is −9 × 10⁻⁷ K^-1 (35), the large refractive index change of the air is not due to the air temperature change. We also estimated the refractive index change of air is on the order of 10⁻⁶ when the relative humidity changes from 0 to 100% around room temperature (36), which is again much smaller than the measured change. As a result, we instead attribute $\overline{\mathrm{\Delta }n}$ to the existence of water molecular clusters in air.

We also measured the beam deflection in the transmission configuration (Fig. 2D; Materials and Methods; SI Appendix, Fig. S8 and Note S4). In this measurement, the pump laser beam exciting molecular clusters also serve as a probe to measure the beam deflection. The direction of the transmitted beam changes (SI Appendix, Fig. S9) when the laser power is changed. Although this method is inspired by the past pump–probe studies of thermal conductivity of solid, in which a probe beam is used to detect the refractive index change and surface deformation caused by a pump laser-induced heating (37–40), we found that the calculated beam deflection assuming the same effects (SI Appendix, Note S4.2) is over two orders of magnitude too small. For example, for a TM-polarized red laser at 45° incident angle to a glass container with water (air–water/glass/air) with an assumed absorptance of 1% (SI Appendix, Fig. S9) at the interface, the calculated beam deflection assuming only thermal effects in the liquid is <0.12 µrad/mW (SI Appendix, Figs. S10–S12 and Note S4.2), while the measured value is 49.1 µrad/mW (SI Appendix, Fig. S9). We attribute the measured beam deflection to mainly the air-side refractive index change, consistent with the above-discussed measurements using the surface mirage effect (Fig. 2 B and C).

We use the normalized beam deflection (SI Appendix, Note S4.3) to evaluate the polarization, angle, and wavelength dependence of the incident laser (SI Appendix, Table S2) on the photomolecular effect at the air–water interface. As shown in Fig. 2 E and F and SI Appendix, Figs. S13 and S14 and Tables S3 and S4, the TE-polarized beam (red circles) has a much smaller normalized transmitted beam deflection than the TM-polarized beam (black squares). Fig. 2E illustrates the normalized transmitted beam deflection of a green laser (520 nm), which maximizes at the incident angle of 45°, consistent with the water surface temperature rise shown in Figs. 1 C and D and 2F and SI Appendix, Fig. S14 illustrate the photomolecular effect is strongest for green light (520 nm) at different incident angles, consistent with previous evaporation experiments on hydrogels (13), which show the highest evaporation rate under green light.

Vapor-Phase Temperature Distribution Above a Heated Surface.

In previous experiments, the water is colder than the ambient due to evaporation, with or without light. In these cases, the air is denser on water surface and there is little air motion due to natural convection. The molecular clusters cleaved off by light recondense back, causing heating of water as shown in Fig. 1 B–D. We artificially heat up the water surface electrically to 53 °C to create an upward vertical draft of air flow, which carries away the water clusters. We compare the vapor-phase temperature distributions with and without green-laser illumination onto the water surface above heated water surface (Fig. 2 G and H). Without the laser illumination, the vapor-phase temperature decreases continuously (Fig. 2G). The TE-polarized laser beam does not cause much change in the vapor-phase temperature distribution (Fig. 2G). However, under a TM-polarized laser (Fig. 2H and SI Appendix, Fig. S15B), the temperature near water surface sharply decreases, followed by a flat temperature region between 4 and 15 mm. Purple curve in Fig. 2H shows the vapor phase temperature distribution 5 min after the TM laser turns off, which is close to the steady state without laser (green curve in Fig. 2G and SI Appendix, Fig. S16).

We interpret such observations qualitatively as follows (Fig. 2I). The clusters cleaved off by the TM-polarized light are carried away by the ascending air plume. The clusters dissociate and absorb heat, making the air cooler near the water surface (SI Appendix, Fig. S16). The flat region is formed when the air becomes saturated with water. Further away from the water surface, the air temperature decreases again as more dry air drafted from the surrounding ambient reduces local relative humidity below 100%. We also observed such a flat temperature region above an air–water interface when it is subjected to green LED illumination (SI Appendix, Fig. S17).

Such sharp temperature drop followed by a flat temperature regions were also observed above the surface of porous polyvinyl alcohol hydrogel (PVA) hydrogels with and without polypyrrole absorbers (13). Although the flat temperature region about hydrogel surface is shorter. We interpreted such difference as follows. In hydrogel samples, clusters must first escape the pores before getting into air above the surface. Heavier clusters are easier to be adsorbed inside the pores. Hence, clusters about hydrogel surfaces are smaller and it takes shorter distance to break these clusters into single water molecules (13).

Vapor-Phase Spectral Signatures.

The transmission spectroscopy of the vapor-phase is performed over a hot water surface (53 °C) (Fig. 3A; Materials and Methods; SI Appendix, Fig. S18 and Note S5). The green laser system is coupled to the sample chamber of a commercial UV–Vis–NIR spectrometer to excite the water molecular clusters. A CaF₂ lens is used to focus the light source of the spectrometer to obtain enough intensity and spatial resolution (diameter of the beam at the focal plane is ~1 mm). The beam waist is ~4 mm above the hot water surface. Fig. 3B shows the vapor phase transmission spectra in the region of 3,100 to 4,200 cm⁻¹ with and without the laser beam illumination (TM-polarized), which differ significantly from each other in both magnitude and peaks. Also plotted in the figure is the absorption spectrum of pure water vapor (blue line), i.e., isolated water molecules (41). Clearly, under the TM-polarized laser illumination, the absorptance increased significantly, especially in regions where previous works had reported the absorption of cluster signatures (29–32). We attribute the additional absorption to the water molecular clusters cleaved off from the water surface.

Fig. 3.

Spectra of vapor phase. (A) Illustration of IR transmission measurement above hot water surface (53 °C) with/without TM-polarized green laser illumination. (B) Absorptance spectrum of the vapor phase with laser on/off. The beam waist of IR probe beam is condensed to ~1 mm in diameter and ~4 mm above air–water interface. The blue line represents the absorption of pure water vapor (41). (C) Illustration of Raman spectrum measurement of vapor phase. When the focal point of the laser beam (532 nm) is above the air–water interface (AS), the divergent laser hits water surface, generating clusters. (D) Raman spectra with focal point ~3 mm above (AS) and on (OS) the hot water surface (53 °C) probed with a 50X lens (26.5 mm working distance). (E) Raman spectra of unheated air–water interface (~20.5 °C) probed with a 100X lens (0.21 mm working distance). The ~0.15 mm AS spectrum contains peaks resembling that of the IR spectrum of 46-water cluster reported in the literature (29). (F) Schematic of custom-built Raman scattering measurement. (G) Fourier series of an ideal square wave (blue) and the actual modulated beam (red). (H) 2$\omega$ Raman signal as a function of the TM and TE-polarized laser beam power.

Next, we use a Raman spectrometer to probe the spectrum of the vapor phase above the hot water surface (53 °C). As shown in Fig. 3C, the 532 nm laser of the Raman system works as both the pump and the probe (mixture of TE and TM). We used a 50X long working distance lens with the laser focal plane 3 mm above the air–water interface (denoted as AS; SI Appendix, Fig. S19 and Note S6). The same laser beam intercepts the air–water interface at a range of angles, cleaving off water molecular clusters (the pump beam), which are probed around the focal point in the vapor phase since the Raman spectrometer is confocal. Compared with the case when the laser beam is focused on the water surface (denoted as OS; SI Appendix, Fig. S19), the Raman spectrum of the AS configuration (Fig. 3D) has a clear peak around ~3,650 cm⁻¹, which represents single water molecule (42), confirming that part of the signals indeed come from the vapor phase, superimposed on unavoidable Raman signal from liquid water that is significantly stronger than the vapor phase. In addition to this peak, multiple broad peaks in the range of 3,000 to 3,900 cm⁻¹ appear, which are qualitatively consistent with the transmission spectrum in Fig. 3D.

Cluster peaks and the single water molecular vapor peak (~3,650 cm⁻¹) are also observed in the Raman spectrum when we focus the laser beam ~0.15 mm above cold-water surface (20.5 °C) under natural evaporation using a 100X lens (working distance = 0.21 mm) (Fig. 3E), since the heated water surface height changes too rapidly to allow Raman data collection using 100X lens. Compared to the OS Raman spectrum obtained using the same lens, which is dominated by water, in the AS configuration, multiple peaks of water cluster between 2,900 and 3,900 cm⁻¹ appear, qualitatively similar to that of absorptance spectra of clusters reported in the literature (29–32). The Raman spectrum with different heights of beam waist above the air–water interface also shows cluster peaks (SI Appendix, Fig. S20).

The commercial Raman system does not allow easy probing of the polarization effect. We study the polarization dependence on a custom-built Raman system (Fig. 3F; Materials and Methods; SI Appendix, Fig. S21, Table S5, and Note S7). The green laser is modulated at f (=200 Hz) by a mechanical chopper to mimic a square wave modulation for which there exists no 2f harmonics (Fig. 3G and SI Appendix, Fig. S22). In actual measurements, the intensity of 2f harmonics is 3% of that of 1f. The laser cleaves off water molecular clusters into the vapor phase. The water molecular clusters then interact with the same incoming laser to generate Raman signals at 2f (2ω Raman signal), which are measured by a photodiode detector connected to a lock-in amplifier. Filters are placed in front of the photodetector to ensure that only signals with wavenumber >2,130 cm⁻¹ are measured. At an incident angle of 45°, the intensity of 2f signal is proportional to the square of power, P², for the TM-polarized laser because both the Raman signal of an individual molecule and the number of clusters are proportional to the power of TM-polarized laser (black squares in Fig. 3H), while the 2f scattering signal is approximately linearly proportional to the power of the TE-polarized laser (red circles in Fig. 3H).

Mass Change under Modulated Light.

To quantify the mass change caused by the photomolecular effect, we measured the deflection of a clamped beam hosting a water container in the middle with a pump–probe configuration as shown in Fig. 4A (Materials and Methods; SI Appendix, Note S8). A modulated green-laser directed at the air–water interface leads to a periodic change of the vibrational amplitude of the clamped beam, which we interpret as being caused by the weight change in the container. The vibration is measured by the deflection of a continuous wave probe laser (635 nm) directed at a location without the pump laser or water (SI Appendix, Fig. S23). Fig. 4B illustrates the theoretical frequency response of the vibration of the used PMMA clamped beam subjected to a periodic force modulation at the center (SI Appendix, Note S8.2). The modulated signal in the low frequency range (region yellow) is constant, which means we can use the beam bending caused by the mass change during natural evaporation (without light) as a calibration (1 g/V) to infer the mass change created by the modulated laser beam. Fig. 4C shows the beam deflection under TM-polarized laser beam (black squares) is over five times larger than that under the TE-polarized laser beam (red circles), while the signal of the dry container (blue triangle) without water is smaller than the measurement uncertainty at the resonance frequency. Using the flat region beam deflection of ~10 µV in the low frequency range under the TM-polarized green light, we estimate that the mass change is ~10 µg (SI Appendix, Note S8.2) under the laser power of ~0.89 W and incident angle of 45°. This mass change corresponds to a water layer ~14 nm. Since the modulation frequency is much higher than the characteristic time needed to reach steady state (Fig. 1B), we cannot relate this mass change to a steady-state evaporation rate.

Fig. 4.

Modulated pump–probe and fog measurements. (A) Schematic of modulated clamped beam (MCB) measurement. The container of water is not displayed in the figure. (B) Theoretical beam deflection signal of PMMA bending cantilever as a function of frequency. (C) Measured beam deflection signal as a function of modulation frequency. Black squares and red circles represent TM-polarized and TE-polarized pump laser on water, respectively. The green triangle represents TM-polarized pump laser on the dry container without water. (D) Weight changes (normalized to area) measured by balance as a function of time initially heated by an IR lamp to 33 °C, followed by TM-polarized and TE-polarized laser illumination (1.0 W). The radius of air–water interface is 15 mm. Power/area of water surface ≈1,400 W/m² (1.4 suns). (E) Photos of the fog chamber with and without fog. (F) The steady state temperature of the clouds in a chamber under different LED illumination (1,000 W/m²). The black squares and red circles represent the temperature of the fog chamber with fog and without fog, respectively.

Absorptance Estimation.

Quantitative determination of the absorptance has been difficult. We estimate the absorptance based on the measured water surface temperature as follows (SI Appendix, Fig. S24 and Note S9). First, using the evaporation rate without light and the measured water temperature, we estimate the effective convective heat transfer coefficient at 7.96 W/m²-K. Under 1.4 W TM-polarized green light illumination at 45°, the water surface temperature increased by 0.54 °C (SI Appendix, Fig. S25). Assuming the same heat transfer coefficient, the absorptance is estimated to be 0.84%.

Superthermal Evaporation and Cluster Size Estimation.

Fig. 1 shows that most clusters recondense because the water surface is cold. To suppress the recondensation of the water clusters, we heat up the water to produce a rising plume (Fig. 4D; Materials and Methods; SI Appendix, Fig. S26) or blow air across the cold-water surface (Materials and Methods; SI Appendix, Figs. S27 and S28) to carry away the cleaved clusters so that the light-induced evaporation can be directly measured. Here, we use an IR lamp to heat up water to 33 °C to avoid the possible absorption of the laser excitation when using an electrical heater. With the IR lamp power-on, a TM-polarized green laser (power/area of water surface ≈ 1,400 W/m²) incident at 45° increases the evaporation rate by 0.07 kg/m²-h while the TE-polarized light does not lead to any measurable change (Fig. 4D). We also measured a similar evaporation rate increase under a TM-polarized green laser by flowing a gentle wind across the cold-water surface (SI Appendix, Fig. S28). Assuming only 0.84% light (11.8 W/m²) is absorbed, the value of 0.07 kg/m²-h is ~4 times of the thermal evaporation limit.

The energy needed to evaporate a water molecule is 0.46 eV. The photon energy at 532 nm is 2.33 eV. Assuming all the photon energy is used to break up water molecules, one green photon at this wavelength can evaporate 5.06 water molecules. Using the evaporation rate of 0.07 kg/m²-h for absorbed green light with power/surface area = 11.8 W/m², we estimate that the average number of water molecules cleaved by a green photon is 20. These estimations on the evaporation rate and cluster size assume that all cleaved water clusters are brought away from the surface. In reality, some of the clusters may still condense back, and hence the actual values can be higher.

Cloud Heating.

For a flat surface, the photomolecular absorption can be maximized by orienting the light to an optimal angle of incidence. For droplets, there are always electrical field directions of the light that are perpendicular to the interface. The increased surface area, the curvature of the droplets, and the multiple scattering effects will further enhance the photomolecular effect. Thus, we expect that photomolecular effect will be important for clouds and fogs. We designed and fabricated a cloud chamber and generated water droplets into the chamber, as shown in Fig. 4E (Materials and Methods; SI Appendix, Fig. S29). LEDs of different wavelengths with intensities of 1,000 W/m² were shone on the cloud chamber, and the temperature rise of an empty chamber with a cloud inside was measured using an IR camera. Fig. 4F shows that the chamber itself is not heated up, but visible LEDs heat the cloud significantly, peaking at 520 nm, consistent with the steady state. Past cloud measurements have suggested significantly higher cloud absorptance than what existing models can predict (43–45), which had been subjected to questions due to difficulties in the experiment (46, 47). The photomolecular effect and heating data we demonstrate here can provide both the theoretical basis and experimental support for increased solar radiation by clouds.

Discussion

In SI Appendix, Table S6, we summarize 14 different experiments that support the existence of the photomolecular effect: the cleavage of water molecular clusters by photons. These experiments cross-check different manifestations of the effect: polarization, angle of incidence, and wavelength dependence. Surprisingly, this effect happens at a spectral region where water is least absorbing. The qualitative pictures presented in Fig. 1A can reasonably explain the dependence on polarization and angle of incidence, but not the wavelength dependence. The wavelength dependence and the low intensity of light used suggest that the photomolecular effect is a single photon process, related to the matching of the photon energy to the water cluster binding energy with surroundings. However, we recognize that a quantitative description of the process will need more effort, especially considering that the states of the hydrogen bond network in bulk water and on the surface are unsettled (22, 28, 48). Past work on laser-induced desorption and surface photoelectric effects can be sources of inspiration for more theoretical studies (27, 49).

We had proposed (13) that the photomolecular effect is the reason behind recent reports of solar-driven interfacial evaporation from different porous materials exceeding the thermal evaporation limit (9–11, 50, 51). We believe that the same mechanism could also explain some past experimental observations of light-induced breathing mode of water molecules on solid surfaces (52) and unexplained cloud absorption of solar radiation (43–45, 47), and may also be related to the surprising chemical reactivity of water microdroplets (53). Our demonstration of the photomolecular effect in fog suggests that this process is ubiquitous: from clouds to fogs, ocean to soil surfaces, and plant transpiration. The fact that the photomolecular effect is strongest at green light, which happens to be the most intensive spectrum of the solar radiation, also leads us to wander if plants evolved to reflect more green light to avoid losing too much water when they migrated from ocean to land. After all, biological organisms emerged way after the earth had water and sunlight. We often hear from weather forecasts “after the sun thinned the fog.” Before, we thought this was caused by thermal evaporation. Our work shows that photomolecular effect could also be at play in this natural phenomenon. The stronger surface absorption of green light adds another mechanism to the beautiful color of sunset in addition to Rayleigh’s explanation based on scattering. While it is well known that bulk materials’ absorption mainly happens via electronic and vibrational transitions, our work suggests that interfacial photomolecular effect can lead to photon absorption in water and possibly in other liquids or even solids. Further exploration of the effect can lead to new scientific insights and technologies in desalination, waste-water treatment, drying, and air conditioning.

Materials and Methods

Materials and Container.

Deionized water with resistivity 0.182 $\mathrm{M}\mathrm{\Omega }\mathrm{m}$ is used. The outer diameter, the inner diameter, and height of the glass container are 33 mm, 30 mm, and 10 mm, respectively. The thickness of the bottom glass of the glass container is ~1 mm. The outer diameter and height of polystyrene (PS) container is 36 mm, and 12 mm, respectively.

Laser and LED.

The parameters of the pen lasers used (from Thorlab) are summarized in SI Appendix, Table S2. 1.5 W high-power green laser (from OPTO Engine LLC) has 1/e² diameter ~2.0 mm. The expansion of the laser is done by two focal lenses with different focal lengths (SI Appendix, Fig. S1). Incident angle is changed by two mirrors (SI Appendix, Fig. S1). LED lamps (from Charon) used have a rated power of 100 W and different nominal wavelengths: purple 390 nm, blue 440 nm, green 520 nm, yellow 590 nm, red 650 nm, and IR 815 nm. The OT301 precision position sensing amplifier was purchased from ON-TRAK.

Thermocouple and IR Cameras.

K-type thermocouple (diameter = ~0.15 mm) from Omega was shaped into a circle with a diameter of ~8 mm. IR cameras used in this work are FLIR C5 and FLIR ETS320. In IR camera measurement, the emissivity of water is calibrated by adjusting the reading of water at the temperature of 20.7 °C (measured by a thermocouple). The emissivity of coverslip and PS are calibrated by adjusting the reading of coverslip and PS at the ambient temperature 22.8 °C (measured by a thermocouple). The humidity and ambient temperature are read by a humidity meter (Mengshen, M86).

IR Image and Thermocouple Measurements of Water Surface Temperature.

The optical path is shown in SI Appendix, Fig. S1. The power and 1/e² radius of the expanded high-power green laser (532 nm) are 1.4 W and 7.5 mm, respectively. The glass container was placed ~30 cm above the optical table (SI Appendix, Fig. S30) with transparent support (glass slide and PMMA) to avoid heat transfer with the optical table heated by the transmitted laser. The steady-state temperature rise of a dry glass container is <0.1 °C under the same laser power measured by an IR camera. The experiment was performed in a quiet laboratory environment and a shield was used to decrease the influence of wind on the steady-state evaporation rate. Under the relative humidity (HD) ~40% and environment temperature of 22.8 °C, the steady-state temperature of water surface is ~20.7 °C and natural evaporation rate is ~0.09 kg/m²-h. It often takes 1 h for the water at room temperature to reach the steady state.

Vertical Vapor Phase Temperature Distribution.

Optical path for measuring the temperature profile of the vapor phase above the air–water interface is shown in SI Appendix, Fig. S3. The power and 1/e² radius of the expanded high-power green laser (532 nm) are 1.4 W and 7.5 mm, respectively. The polarization of pump laser beam was changed by a half-wave plate. A coverslip was used as a thermal emitter to indicate the vapor phase temperature since the vapor itself does not emit much thermal radiation (SI Appendix, Fig. S31). The coverslip is thin (0.1 mm thick) which minimizes heat conduction along the height direction and is suspended ~2 mm above water surface (not touch water surface) by a thin string made of cotton to minimize heat conduction distortion of the temperature distribution. An IR camera (FLIR) was used to measure the temperature of the coverslip. For heated water, Joule heating (resistance wire glued to the bottom of the glass container) is used to heat up the water. We estimate that heating wire absorption of the laser power can lead to 1 °C water temperature rise. For the room temperature test, no heater was used with the glass container.

Vapor-Phase Mirage Effect Measurement.

The optical path of mirage effect of vapor phase above air–water interface is shown in SI Appendix, Fig. S6. The power and 1/e² radius of the expanded high-power green laser are 1.4 W and 7.5 mm, respectively. A pen laser (635 nm) was used as the probe beam, which is arranged perpendicular to the plane of incidence of the pump beam. The probe laser was adjusted to have an angle with water surface (~0.033 rad). A position sensing detector connected to OT301 precision position sensing amplifier (ON-TRAK) was used to measure the shift of probe beam position before and 1 min after the pump laser turned.

Steady-State Transmitted Beam Deflection (SS-TBD).

The set-up of the SS-TBD measurement is shown in SI Appendix, Fig. S8. Pen lasers (parameters shown in SI Appendix, Table S2) were used. We rotated the pen lasers to change the polarization of the laser instead of using a half-wave plate since the half-wave plate will lead to the beam deflection when the power of laser is changing. A position sensing detector connected to OT301 precision position sensing amplifier (ON-TRAK) was used to measure the beam deflection.

Transmission of Vapor Phase.

The optical path of the vapor phase transmission measurement is shown in SI Appendix, Fig. S18. The green laser was added into the commercial Cary Series UV–VIS–NIR spectrophotometer. Expanded high-power green laser was used as pump laser. Due to the limited space of the sample compartment of the commercial spectrophotometer, the pump-beam has some divergence. As a result, the incident angle is in the range of 40° to 50° (average is 45°), the polarization is a TM polarization, and the power and 1/e² radius of the expanded high-power green laser close to air–water interface are 1.4 W and ~9 mm, respectively. These parameters are slightly different from other experiments in this work. CaF lens (IR transmitted, focal length = 2 mm) was used to focus the light source of the spectrometer to get enough intensity and spatial resolution (diameter of beam at focal plane is ~1 mm). The air–water interface was adjusted and placed at ~4 mm below the beam waist.

Raman Spectroscopy.

A commercial confocal Raman spectrometer (HORIBA Scientific) was used to measure the spectrum of liquid water and vapor by varying the position of the focal plane (SI Appendix, Fig. S19). The wavelength of the laser in the Raman system is 532 nm.

In a confocal Raman system, the signal detected by the CCD camera is weighed Raman scattering of all components close to the beam waist (SI Appendix, Fig. S19). We catalog the configurations into four groups based on the position of the beam waist/focal plane: on the water surface (OS), above the water surface (AS), below the water surface (BS), and emersion lens (EL). OS and BS are common configurations used in most of the past Raman studies of water. In our Raman system, the 1/e² radius at the back focal plane of the objective lenses is ~1.1 mm. The 1/e² beam waist radius at the focal plane is ~1.0 µm and 0.5 µm using a 50X lens and 100X lens, respectively. The power of laser is 33 mW and 17 mW at beam waist using the 50X lens and 100X lens, respectively. The working distance of the 50X lens and 100X lens is 26.5 mm and 0.21 mm, respectively.

2$\mathit{\omega }$ Raman Scattering Measurement.

The optical path of the 2$\omega$ Raman scattering measurement is shown in SI Appendix, Fig. S21. The high-power green laser (1.4 W) was modulated by a mechanical chopper (modulation frequency f = 200 Hz) placed at the focal plane of the two objective lenses to be closer to an ideal square wave. Its polarization was changed by a half-wave plate. The mechanical chopper provides a reference signal (f) to the lock-in amplifier. A 550 nm long-pass filter and a 600 nm long-pass filter are placed between the sample and photodiode detector to filter the elastic scattering of green laser. The photodiode was placed 1 cm away from the beam spot on surface and not directly exposed to the incident and reflected laser. The photodiode was connected to the lock-in amplifier.

Measurement of MCB.

The optical path is shown in SI Appendix, Fig. S23. The expanded high-power green laser works as the pump. The power and 1/e² radius of the expanded modulated high-power green laser are 0.7 W and 7.5 mm, respectively. The polarization of the pump laser was changed by a half-wave plate. The pump beam was modulated by a mechanical chopper, which provides a reference signal to the lock-in amplifier. A pen laser (635 nm) was used as probe, which is perpendicular to the plane of incidence of the pump laser. The beam position sensing detector was connected to a lock-in amplifier to measure modulated beam deflection signal.

Measurement of Evaporation under an IR Lamp or Air Flow.

Optical paths of the evaporation measurement under an IR lamp and air flow are shown in SI Appendix, Figs. S26 and S27, respectively. Air flow is produced by a fan. The power and 1/e² radius of the expanded high-power green laser are 0.7 W and 7.5 mm, respectively. The surface of balance surface is covered with highly reflective mirror film (3M ESR reflector film) to avoid the heating of balance under laser illumination.

Fog Generation and Temperature Measurement.

The fog generation system is illustrated in SI Appendix, Fig. S29. Here the fog generator is made of an ultrasonic Atomization Maker (20 mm in diameter, 113 kHz, from WHDTS@Amazon). To supply water and prevent the generator to heat up, we built a water-cooling kit by using a silicone tube and plastic tube. The water-cooling kit was glued to the backside of the ultrasonic Atomization Maker. The fog size according to the product manual is about 3 μm in diameter. To avoid dropwise condensation on the chamber surface which blocks the light transmission, we coated a thin layer of hydrogel on the inside surface of the chamber (54).

To show the fog can absorb visible light due to the photomolecular effect, we used LEDs lamps (light intensity, light intensity of 1,000 W/m², 1 sun) with various wavelengths as the light sources and an IR camera to monitor the fog chamber’s surface temperature.

The testing sequence of cloud measurements consists of the following steps: a) Leave the dry and empty fog chamber on the test platform for 60 min, and then monitor its outside top surface temperature for 20 min to make sure it gets thermal equilibrium with the surrounding environment; b) turn on the LED lamp and shine onto the fog chamber (empty without fog) for 60 min, and then monitor the outside top surface temperature of the dry and empty fog chamber for 20 min to make sure it gets thermal equilibrium with the surrounding environment. This step makes sure no absorption of the chamber. The chamber does not heat up under LED; c) turn off the LED lamp and turn on the fog generator, after 30 min when the chamber is filling with fog (Fig. 4E), start to monitor the outside top surface temperature of the fog chamber for 20 min to make sure it gets thermal equilibrium with the surrounding environment; d) turn off the fog generator, after 30 min when almost all the inside fog condense onto the wall, turn on the LED lamp for 60 min, and then monitor the outside top surface temperature of the fog chamber for 20 min to make sure it gets thermal equilibrium with the surrounding environment. In this case, we can make sure no obvious temperature rise due to the absorption of the condensate film; e) drain out the condensation inside the fog chamber and turn on the fog generator, after 30 min when the chamber is filling of fog (Fig. 4E), turn on the LED lamp after 30 min start to monitor the outside top surface temperature of the fog chamber for 20 min to make sure it gets thermal equilibrium with the surrounding environment. For all the measurements under various LED lamps, we keep the light intensity of 1,000 W/m² (1 sun) on the container surface and the same distance between the lamp and the fog chamber.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.