Open Access
Add to BibSonomy
Abstract
We report on a new mechanism capable of inducing the insulator-metal transition (IMT) in VO2 via surface plasmon polaritons (SPP). Our theoretical model predicts that for a bilayer Au-VO2 sample an enhanced electromagnetic energy density at the Au-VO2 interface will occur at 1064nm laser wavelength when SPPs are excited in the Au layer. This effect can assist the IMT in the VO2 layer and at the same time, the SPP absorption can be used to detect it. Changes in the optical properties of the VO2 thin layer in such structure can be observed in the reflected light in Kretschmann configuration, via a shift in the nadir location due to light absorption at resonance. This optical mechanism occurs at 2mW threshold transition energies and fully saturates at 5mW.
© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Vanadium dioxide is a transition metal oxide that has been studied for the past 60 years because it is a typical highly correlated material that undergoes a characteristic phase transition between insulating to metallic states near room temperature [1]. This phase transition is associated with large changes in the VO2 optical properties in the IR spectral range [2]. This makes VO2 an attractive candidate for various metamaterial and optical technologies, such as IR optical or plasmonic switches, VO2 plasmonic modulators, and for infrared camouflage, thermal regulation, and infrared tagging and labeling [3–8]. Previous studies identified several different mechanisms enabling the IMT in VO2: ultrafast optical laser pulses [9] thermally [1], electrically [10], and by applying high pressure [11]. Photo-induced IMT has been so far only demonstrated using strong ultrafast laser pulses [9] although Muskens [12] has shown that plasmonic nano-antennas do have an effect on the transition by lowering the required laser fluence. In this manuscript we investigate an alternative process for photo-inducing the IMT in VO2 by surface plasmon polaritons (SPP) excited at the interface between VO2 and metal layers. This transition mechanism offers several advantages: on one hand, the transition can be induced purely optically with simple optics and low powered lasers, without the need for complex ultrafast laser systems; on the other hand, it does not require additional processing such as adding electrical contacts to the sample; and it is not subject to the slowness of the thermal response.
The approach that we utilize to both induce and detect the IMT in a 5-nm thin VO2 layer is based on surface plasmon resonance (SPR), in which SPPs can be excited at the VO2 - gold interface over a narrow range of incident angles of a laser beam in the Kretschmann configuration [13]. At resonant conditions the electric field at the interface of the Au-VO2 is enhanced by a factor 16 times the incident E-field strength for our sample geometry, and is able to reach threshold values for IMT in VO2 with lower peak input laser powers compare to the direct excitation at room temperature by either standard or ultrafast optical pulses [9] or applied dc electric field [10, 14].At the same time, the angular position of the resonance absorption is extremely sensitive to the optical properties of the interface materials, and thus its shift can be used to indicate the change in the VO2 optical properties associated with the IMT. This detection principle has been used previously [15] to monitor the change in SPR properties due to a thermally-induced IMT in a VO2 layer. In our experiment we observed a clear change in the SPR nadir position similar to predictions based on known VO2 optical constants in the insulating and metallic states, when we increased the laser power above the threshold value of 5mW. While we did not apply an alternative method such as thermally inducing the IMT to independently confirm the occurrence of IMT in the VO2 layer, the observed SPR behavior is fully consistent with this expectation.
2. Sample preparation
The sample used in this study was grown on plain soda lime glass substrate in two stages. A layer of 31nm of Au was deposited by evaporative deposition on the glass substrate. Next, a 5nm VO2 layer was deposited via reactive biased target ion beam deposition onto the Au layer, which has been described by West [16]. Both Au and Ag are adequate metallic layers for our proposed structure because of their excellent dielectric properties for SPP generation at 1064nm [17]. However, Ag quickly oxidized during the growth environment for the VO2 layer. Thus, the more stable Au was chosen for the base metal layer. The thickness of the VO2 layer was chosen such that the SPP electric field was fairly uniform across the full VO2 thickness.
The exact thickness composition of our sample was designed using a 4 × 4 optical matrix formalism that is describe in [18] and implemented in Mathematica code. Simulations were carried out for various thicknesses of the Au layer to establish the largest absorption and the narrowest full width at half max (FWHM) resonance for the SPP resonant curve. VO2 was added in the simulations to predict the optimum Au and VO2 layer thicknesses for optimum SPR response for the sample. The overall goal was to obtain optimal SPR response [13], which would correspond to the strongest E-field enhancement at the Au surface. It was determined that 31nm Au with 5nm VO2 was optimal for 1064nm light. The optical susceptibility values used for the simulation were for Au: ε′ = −48.480, ε′′ = 3.6006 [17] for insulating bulk VO2: ε′ = 9.56, ε′′ = 2.81, and for metallic bulk VO2: ε′ = −1.71, ε′′ = 5.97 [2].
3. Experimental setup
Figure 1 shows a schematic representation of the sample in an attenuated intensity reflection setup. The laser light is incident into a cylindrical glass prism from the bottom right, then travels through the sample’s glass substrate and reflects off the gold layer. Beyond total internal reflection, SPPs are generated at a critical incident angle at the glass-Au surface interface. Index matching fluid is used between the glass substrate and the glass prism.
Fig. 1 Diagram of the path of the laser light through the glass prism and VO2 sample. Θ represents the incident angle of laser light in the diagram. The diagram is showing the case for internal reflection which is also the case for the generation of SP. The SP are represented by the + and - signs and the black arrows represent the SP electric field. The glass substrate and glass prism are not to scale.
For SPP monitoring, the attenuated total reflection assembly shown in Fig. 1 was mounted at the center of rotation of a goniometer that has an angular resolution of 0.004 degrees. While the sample is accurately located at the center of rotation, the orientation of the sample was set by eye and it was difficult to get the sample perfectly normal to the goniometer stage. This could lead to a small systematic error in the incident angle. To avoid possible repositioning errors, all the measurements were made without changing the sample’s original position, once it was mounted.
The experimental apparatus set-up is shown in Fig. 2. The arms of the goniometer operate in a Θ, -Θ motion. To remove possible backlash errors in the goniometer motion, the system was always operated in the same rotation direction during the data acquisition. One arm of the goniometer held the output of the single mode polarization maintaining fiber followed by the polarizer set up for p-polarized light, a signal pickoff for power reference measurements, and a lens with focal length 75.6mm focusing the laser beam onto the sample. The lens focuses the beam down to an elliptical spot with a vertical minor axis of 45µm and a major axis in the range of 62-64µm (which varies with the incident angle). The second arm of the goniometer held a collimating lens and photo detector. The backside of the sample has a thermal imaging camera to monitor the sample temperature for any laser heating effects.
Fig. 2 Diagram of the experimental setup. The lock-in-amplifier is set up do a differential measurement from the sample signal and the pick off.
A Nd:YAG 1064nm laser and other optics were located on a nearby optical table. A half-wave plate and polarizer in front of the fiber coupler were used to control the laser light power on the sample. A mechanical chopper modulated the laser beam power at 503 Hz, and the read-outs from both the laser pickoff and sample detector were input to a lock-in amplifier. We used a differential measurement process to handle any drifts in the laser’s power by subtracting the pick off power reference signal from the sample signal. The output of the lock-in amplifier was recorded with a computer via a USB data acquisition module.
The nadir position of the SPR for each laser power level was measured by monitoring the reflection during the uniformed angular motion from a lower angle below the total internal reflection point to a higher angle beyond total internal reflection. For each recorded position, the goniometer motion was stopped for 20 seconds to let any vibrations to settle before acquiring the reflection response. At each angle, data was collected for 100 seconds with a data acquisition system operating at 16kHz to extract the average and variance signal. The stage was then moved to the next angle, repeating the data acquisition process until the SPR angular response was mapped.
4. Results
To numerically model the optical response in addition to the use of a 4 × 4 optical matrix method, we also employed a finite difference time domain (FDTD) calculation, EM Explorer software which allowed us to also extract the spatial distribution of the electric fields. Figure 3(a) shows the sharp decrease in p polarized light reflectivity as the incident angle approaches the critical angle of total internal reflection. Both the 4 × 4 matrix method (shown as solid lines) and the FDTD calculations (shown as dotted lines) agree with each other in the estimation of the nadir location of the SPR for both fully insulating VO2 properties (blue data lines) compared with fully metallic VO2 properties (red data lines), They also clearly predict that the IMT in VO2 must cause the nadir location of the SPR to shift to smaller angles. The narrowness of the SPP resonance helps elucidate the shift in the SPR curve when the IMT occurs, which was one of the design parameters for the sample.
Fig. 3 (a) The reflection of the P polarized light (Rpp) from two different simulations methods; the (dotted) FDTD method the (solid) 4 × 4 optical matrix method. The blue lines are for case when the VO2 is in the insulating state and the red lines are for case of metallic VO2. (b) close-up view of the nadir region of the SPR curve showing the nadir shift produced by the SPP-induced IMT in the VO2.
Figure 3(b) zooms into the central region of the SPR curve to highlight the shift in the SPR nadir location. The nadir location shifts approximately 0.1° between the two material states of VO2. The nadir location for the insulating state is at 42.3 degrees while for the metallic state the nadir is at 42.2 degrees. The same shift is predicted by both the 4x4 matrix method and in the FDTD model. Away from the resonant points, the two methods show small discrepancies from each other.
Figure 4 shows the normalized SPR reflectivity of p polarized light over a wide range of angles with the laser at a very low power (definitely below the IMT threshold) and at a high power (above the projected IMT threshold). In this data, the angle step size of 0.04° was used and the signals was normalized to the maximum signal measured at a 44.12 angle (which is not shown). Since the maximum energy is transferred to SPP near the bottom of the reflectivity resonance, we expect to observe the effect of IMT in that region when the laser power exceeds the necessary threshold for IMT behavior.
Fig. 4 SPR data for two power levels. The vertical axis is normalized Rpp. The horizontal axis is the incident angle in degrees. The blue curve is for 10µW power level scan. The red curve is for a 5.4mW scan. The dashed box highlights the region where the IMT differences are detectable between the two power levels.
Indeed, the two curves show a clear separation at the resonant point highlighting the IMT in this sample in good agreement with our theoretical predictions of a ~0.1° change of the resonant point as indicated in Fig. 3(b) for a thin film VO2 sample with our geometry. The model calculations in Fig. 3 show a consistent difference between the two material states of VO2 (either fully insulator or metallic) at all angles of the SPP, while Fig. 4 only shows a difference near the SPP resonant point. Below 44.88° and above 45.5° the two curves converge and show a reflectivity commensurate with an insulating VO2 state, while between those two angles, the reflection better matches the metallic VO2 for the 5.4 mW SPR curve and the insulating VO2 for the 10 µW SPR curve. The 44.88° angle is 0.35° below the resonant angle for the insulator state of VO2. For VO2 irradiated at 5.4 mW laser power, as the angle of the incident beam moves off resonance, the SPP generation will decrease with a corresponding decrease in the electromagnetic energy density. Outside the IMT initiation region shown by the dashed box in Fig. 4, the electric field enhancement at the Au-VO2 interface is weaker and the VO2 only exhibits the behavior of the insulating phase.
Figure 5 zooms into the experimental measurements for the SPR reflectance over a very narrow angle range using a smaller angular step size of 0.016° for better resolution of the nadir. Each data set was normalized by their averaged signal amplitudes. This had the effect of rescaling the data onto the same amplitude range for easier visual comparison of the nadir locations. In this case the vertical scale is denoted as “Amplitude scale Rpp” instead of the “Rpp Normalized” used in Fig. 4. Here one can see a clear angular shift in the SPR nadir point of approximately 0.07°, which is 70% of the model’s prediction. To determine the precise angle of the nadir location for each laser power, a weighted polynomial fit was used which provided a minimum location and a value for the confidence interval for the quality of the data. The short vertical dotted lines in Fig. 5 show where the fit’s nadir location falls for the two laser power levels examined in Fig. 5.
Fig. 5 This graph shows the SPR response curve of two different laser power scans on the sample. The blue diamonds show the data points for the 90µW power scan with uncertainty. The red squares show the data points for the 5.5mW power scan with uncertainty. The dashed blue line is the corresponding weighted polynomial for the 90µW scan. The dashed red line is the corresponding weighted polynomial for the 5.5mW scan. The dotted vertical blue and red lines are visual guides for the respective nadir locations for the polynomial fit.
Figure 6 shows experimental results for the VO2 SPR nadir’s location shifts vs. the laser power input. Despite relatively low signal to noise, the change in the SPR minimum position above a certain laser power is clear. At low laser power the nadir’s location is stable. The threshold for the onset of the transition appears to be around 2mW, and further increase in laser power caused the nadir position to shift toward a lower angle. It is also observable that above approximately the 5mW power level the nadir position seems to reach saturation and the angle is longer changing.
Fig. 6 Shows the SPR nadir locations versus the laser light power level. The dashed lines represent the simulation values for 100% insulating VO2, 50% metallic/50% insulating VO2, and 100% metallic VO2. The red line is a visual guide for the transition behavior. The insert displays the SPR nadir locations for the bare 31nm-thick film of Au. For small scans in the vicinity of the resonant points, the data was fitted with a polynomial function. The weighted polynomial was a 3rd order polynomial and the minimum of the polynomial determined the nadir location for that power level scan. The uncertainty in amplitude of the fit at the nadir location was used to determine the uncertainty in the angular location.
To verify that the observed behavior cannot be attributed to a change in the optical properties of the gold layer, an inset in the Fig. 6 shows measurements of the SPR’s nadir location variation measured in the same manner, but on a plain layer of 31nm Au on glass without VO2. In contrast to the Au/VO2 sample, for the pure Au sample, the SPR nadir value is unchanged over the whole tested power range.
A common question raised in the measurement of the IMT of VO2 is the possibility of the observed effect being caused by an increased temperature due to heating by the laser beam. To address this issue, a thermal FLIR camera was employed in the measurement system as shown in Fig. 2. The FLIR camera could detect temperatures changes with 0.1° accuracy. At the optical power levels used in these tests, no thermal increase was detected either in the VO2 layer or in the polycrystalline Au layer. Indeed, heating of the Au layer can also induce changes to the Au optical properties when the Au is heated [19]. This would cause the resonant nadir point to shift to higher angles at higher temperatures as opposed to the VO2 layer, which shifts the nadir to lower angles when it goes through an IMT. However, the temperature-dependent changes in the Au layer will make it hard to reliably detect changes in VO2 layer optical properties if the IMT is attempted to be induced thermally.
In Fig. 6 the relative magnitude change of the expected shift of the nadir’s location from the simulation is also plotted in the figure as black dashed lines for zero, 50% and 100% of metallic VO2. It can be seen that the data is close to the magnitude of the shift predicted by the simulation, especially when considering that the susceptibility values for bulk VO2 were used in simulations, which might not be identical to those in thin films. The relative nadir shift is evidence of a significant change in the optical properties of the VO2. Its magnitude suggests that this is due to IMT as predicted by the simulations. It should be noted, that independent measurements of the optical properties of this thin film sample of VO2 have not been carried out because of concerns regarding possible sample damage caused by thermal cycling. Under these considerations, the observed discrepancy in the nadir shift between our measurements (0.07°) and our numerical modeling (0.1°) is reasonably small.
To estimate the internal electromagnetic field enhancement in our sample, the FDTD model and published optical properties [2,17] were used to compute the internal electric Ez or magnetic Hy field amplitudes at the wavelength of 1064nm. Figure 7 illustrates the FDTD output for the density of electromagnetic energy across the Au and VO2 layers normalized to the incident EM wave at the nadir location for the insulting VO2 of 42.3° and the metallic VO2 case of 42.2°. The cell size in the FDTD simulation was set to 0.5nm, as a compromise between the sufficient spatial resolution and the computational time. Even though the effect of the cell size is visible during the rapid rises in the field strength, it provides reasonably accurate measurement of the field amplitudes inside both the 31nm Au and 5nm VO2 layers. Figure 7 shows a gradual buildup of the electromagnetic energy density across the Au layer, followed by its exponential fall in the VO2 layer when the latter is in the insulator state. For metallic VO2 the electromagnetic energy density is lower on average, and slowly increases in the VO2. In the metallic case, such behavior may be related to the possibility of SPP excitation/coupling in metallic VO2 demonstrated in previous experiments [20].
Fig. 7 FDTD simulations for |Hy|2 field in the sample for the case of Insulating VO2 state (blue line) and metallic VO2 (red line). |Hy|2 is the density of the relative electromagnetic energy in the sample and shows the field enhancement caused by surface plasmons at the resonant point.
The FDTD computation predicts that with the threshold transition laser power level of 2.0mW (which equals 1.76 kV/m electric field strength in air) will result in an enhanced electric field of 29.6 kV/m in the VO2 at 42.3°, which is the nadir of simulation for the insulating case of VO2. From the data in Fig. 4, it was noted for the high power level scan that the IMT initiation occurred 0.35° below the resonance angle of the insulator state of VO2. Utilizing that 0.35° shift below the VO2 insulator’s nadir angle in the FDTD simulation, a calculation of the enhance electric field strength was also performed at that angle for 5 mW laser power level (which is equal to 4.40kV/m in air). The simulation computed electric field enhancement was 30.9 kV/m in the VO2. These separately computed IMT threshold values show the consistency of the approximate onset of the IMT measured in this sample.
A brief comparison of electric fields from alternative IMT mechanisms in VO2 helps to highlight the benefits of this new approach. For a DC electric field method, researchers required a field on the order of 34 kV/m [14] with a four lead resistance measurement in a 200µm long VO2 sample, while in a second report, a field estimated at 65 MV/m at room temperature [10] was estimated via an atomic force microscope measuring through the thickness of a VO2 thin film on a Si doped substrate. The E-field from the four lead resistance measurements is ~10% higher than our calculated E-field induced by the SPP at the transition thresholds. Another transition process is the photo-induced IMT in VO2 via a pump probe ultrafast measurement, which has been reported to occur using an amplified Ti:Sapphire laser at 800nm [9,21]. Assuming an average pulse power for transition threshold of 7mJ/cm2 with a pulse duration of 0.5ps [9] gives an instantaneous peak intensity of 1.23 × 1010 W/cm2, which correlates to a 304 MV/m peak electrical field strength in air. From [12] another short pulse process for inducing the IMT in VO2 was used along with plasmonic antennas to lower the transition energy. It used an instantaneous peak intensity of 2.63 × 108 W/cm2. The electric field amplitude for their beam would be at 44.5MV/m. Both of these photo induced methods involved very high electric fields.
5. Conclusion
From the FDTD model it can be seen that SPP excitation concentrates significant electromagnetic energy density into the VO2 layer. The observed changes of the reflectivity upon SPP excitation correlate with the changes in the optical properties expected for the IMT in the VO2 sample. The optical model suggests that our sample would have a 0.1° shift in the resonance’s nadir location. Experimentally, a resonance’s nadir shift of 0.07° is observed. It is also documented that the optical properties of VO2 appear to start to exhibit the shift in the resonance’s nadir location at approximately 2mW of laser power. At about 5 mW of power, changes in the shift in the resonance’s nadir location have saturated, suggesting that the sample has fully transitioned. Finally, wider angle scans around the resonance performed at 10 µW and at 5.4 mW show that the shift in the SPP resonance curves only appears near the resonance’s nadir. There is an unobservable difference between the two resonant curves at high and low laser powers away from the SPP resonant point. Near the resonant point for the higher power measurements the VO2 appear to behave like a metallic material while the lower power measurements indicate the behavior of the insulating state of VO2.
Funding
Defense Threat Reduction Agency grant number HDTRA 1-16-1-0056
Acknowledgments
NASA Langley Research Center Loan/agreement number 0160035
References and links
1. F. J. Morin, “Oxides which show a metal-to-insulator transition at the neel temperature,” Phys. Rev. Lett. 3(1), 34–36 (1959). [CrossRef]
2. H. W. Verleur, A. S. Barker, and C. N. Berglund, “Optical properties of VO2 between 0.25 and 5 eV,” Phys. Rev. 172(3), 788–798 (1968). [CrossRef]
3. N. I. Zheludev and Y. S. Kivshar, “From metamaterials to metadevices,” Nat. Mater. 11(11), 917–924 (2012). [CrossRef] [PubMed]
4. S. Chen, H. Ma, X. Yi, H. Wang, X. Tao, M. Chen, X. Li, and C. Ke, “Optical switch based on vanadium dioxide thin films,” Infrared Phys. Technol. 45(4), 239–242 (2004). [CrossRef]
5. L. A. Sweatlock and K. Diest, “Vanadium dioxide based plasmonic modulators,” Opt. Express 20(8), 8700–8709 (2012). [CrossRef] [PubMed]
6. A. Joushaghani, B. A. Kruger, S. Paradis, D. Alain, J. S. Aitchison, and J. K. S. Poon, “Sub-volt broadband hybrid plasmonic-vanadium dioxide switches,” Appl. Phys. Lett. 102(6), 061101 (2013). [CrossRef]
7. B. A. Kruger, A. Joushaghani, and J. K. S. Poon, “Design of electrically driven hybrid vanadium dioxide (VO2) plasmonic switches,” Opt. Express 20(21), 23598–23609 (2012). [CrossRef] [PubMed]
8. M. A. Kats, R. Blanchard, S. Zhang, P. Genevet, C. Ko, S. Ramanathan, and F. Capasso, “Vanadium dioxide as a natural disordered metamaterial: perfect thermal emission and large broadband negative differential thermal emittance,” Phys. Rev. X 3(4), 041004 (2013). [CrossRef]
9. A. Cavalleri, C. Tóth, C. W. Siders, J. A. Squier, F. Ráksi, P. Forget, and J. C. Kieffer, “Femtosecond structural dynamics in VO2 during an ultrafast solid-solid phase transition,” Phys. Rev. Lett. 87(23), 237401 (2001). [CrossRef] [PubMed]
10. B. Wu, A. Zimmers, H. Aubin, R. Ghosh, Y. Liu, and R. Lopez, “Electric-field-driven phase transition in vanadium dioxide,” Phys. Rev. B 84(24), 241410 (2011). [CrossRef]
11. L. Bai, Q. Li, S. A. Corr, Y. Meng, C. Park, S. V. Sinogeikin, C. Ko, J. Wu, and G. Shen, “Pressure-induced phase transitions and metallization in VO2,” Phys. Rev. B 91(10), 104110 (2015). [CrossRef]
12. O. L. Muskens, L. Bergamini, Y. Wang, J. M. Gaskell, N. Zabala, C. H. de Groot, D. W. Sheel, and J. Aizpurua, “Antenna-assisted picosecond control of nanoscale phase transition in vanadium dioxide,” Light: Science &Amp. Applications 5(10), e16173 (2016). [CrossRef]
13. H. Raether, Surface Plasmons on Smooth and Rough Surfaces and on Gratings, 1st ed., Springer Tracts in Modern Physics (Springer-Verlag, 1988), p. 136.
14. G. Stefanovich, A. Pergament, and D. Stefanovich, “Electrical switching and mott transition in VO2,” J. Phys. Condens. Matter 12(41), 8837–8845 (2000). [CrossRef]
15. M. Beebe, L. Wang, S. E. Madaras, J. M. Klopf, Z. Li, D. Brantley, M. Heimburger, R. A. Wincheski, S. Kittiwatanakul, J. Lu, S. A. Wolf, and R. A. Lukaszew, “Surface plasmon resonance modulation in nanopatterned Au gratings by the insulator-metal transition in vanadium dioxide films,” Opt. Express 23(10), 13222–13229 (2015). [CrossRef] [PubMed]
16. K. G. West, J. Lu, J. Yu, D. Kirkwood, W. Chen, Y. Pei, J. Claassen, and S. A. Wolf, “Growth and characterization of vanadium dioxide thin films prepared by reactive-biased target ion beam deposition,” J. Vac. Sci. Technol. A 26(1), 133–139 (2008). [CrossRef]
17. P. B. Johnson and R. W. Christy, “Optical constants of the noble metals,” Phys. Rev. B 6(12), 4370–4379 (1972). [CrossRef]
18. D. W. Berreman, “Optics in stratified and anisotropic media: 4×4-matrix formulation,” J. Opt. Soc. Am. 62(4), 502–510 (1972). [CrossRef]
19. H. Reddy, U. Guler, A. V. Kildishev, A. Boltasseva, and V. M. Shalaev, “Temperature-dependent optical properties of gold thin films,” Opt. Mater. Express 6(9), 2776–2802 (2016). [CrossRef]
20. L. Wang, E. Radue, S. Kittiwatanakul, C. Clavero, J. Lu, S. A. Wolf, I. Novikova, and R. A. Lukaszew, “Surface plasmon polaritons in VO2 thin films for tunable low-loss plasmonic applications,” Opt. Lett. 37(20), 4335–4337 (2012). [CrossRef] [PubMed]
21. D. Wegkamp, M. Herzog, L. Xian, M. Gatti, P. Cudazzo, C. L. McGahan, R. E. Marvel, R. F. Haglund Jr, A. Rubio, M. Wolf, and J. Stähler, “Instantaneous band gap collapse in photoexcited monoclinic VO2 due to photocarrier doping,” Phys. Rev. Lett. 113(21), 216401 (2014). [CrossRef] [PubMed]
