Monolithic OLED-Microwire Devices for Ultrastrong Magnetic

Jun 30, 2017 - Thus, when spin conservation applies to transitions due to the weak spin–orbit interaction, the permutation symmetry of the electron-...
1 downloads 9 Views 4MB Size
Download PDF
Letter pubs.acs.org/NanoLett

Monolithic OLED-Microwire Devices for Ultrastrong Magnetic Resonant Excitation Shirin Jamali,† Gajadhar Joshi,† Hans Malissa,† John M. Lupton,*,†,‡ and Christoph Boehme*,† †

Department of Physics and Astronomy, University of Utah, 115 S, 1400 E, Salt Lake City, Utah 84112, United States Institut für Experimentelle und Angewandte Physik, Universität Regensburg, Universitätsstrasse 31, 93040 Regensburg, Germany

‡

S Supporting Information *

ABSTRACT: Organic light-emitting diodes (OLEDs) make highly sensitive probes to test magnetic resonance phenomena under unconventional conditions since spin precession controls singlet−triplet transitions of electron−hole pairs, which in turn give rise to distinct recombination currents in conductivity. Electron paramagnetic resonance can therefore be detected in the absence of spin polarization. We exploit this characteristic to explore the exotic regime of ultrastrong lightmatter coupling, where the Rabi frequency of a charge carrier spin is of the order of the transition frequency of the two-level system. To reach this domain, we have to lower the Zeeman splitting of the spin states, defined by the static magnetic field B0, and raise the strength of the oscillatory driving field of the resonance, B1. This is achieved by shrinking the OLED and bringing the source of resonant radio frequency (RF) radiation as close as possible to the organic semiconductor in a monolithic device structure, which incorporates an OLED fabricated directly on top of an RF microwire within one monolithic thin-film device structure. With an RF driving power in the milliwatt range applied to the microwire, the regime of bleaching and inversion of the magnetic resonance signal is reached due to the onset of the spin-Dicke effect. In this example of ultrastrong light−matter coupling, the individual resonant spin transitions of electron−hole pairs become indistinguishable with respect to the driving field, and superradiance of the magnetic dipole transitions sets in. KEYWORDS: Organic light-emitting diodes, conducting polymers, spin-dependent recombination, electron paramagnetic resonance, spin collectivity

T

as illustrated in Figure 1a. This pair of two electronic states exhibits random spin orientation as would be the case for two adjacent charge carriers which are exposed to different random hyperfine fields. Due to the Pauli exclusion principle, doubly occupied electron states can exist only in the singlet manifold. Thus, when spin conservation applies to transitions due to the weak spin−orbit interaction, the permutation symmetry of the electron-spin pair state before the transition governs the transition probability into the doubly occupied state. In recent years, organic light-emitting diodes (OLEDs) have proven to be unique probes of electronic processes sensitive to spin-permutation symmetry as these devices operate under electrical injection of positive and negative charges. They therefore represent solid-state systems for the study of the interconversion of singlet and triplet recombinant carrier pairs.11 Such recombination transitions can be detected either in electroluminescence or in the device current,11 which has proven sufficiently sensitive to monitor even miniscule perturbations to the resonantly controlled nuclear spin

he underlying concept of many approaches to quantuminformation processing by singlet−triplet qubits relies on initialization, control, and readout of the permutation symmetry of a pair of spins. Examples include electrostatically defined coupled quantum dots in GaAs,1−3 silicon,4 or in carbon nanotubes,5 but also in isolated spins of unpaired electrons in bulk crystals, the most prominent example being the diamond nitrogen-vacancy system.6 In parallel to these developments, evidence has been emerging that the magnetoceptive abilities of some migratory avian species also rely on the coherent interconversion between singlet and triplet states of photoinduced radical pairs formed in retinal pigment−protein complexes.7−9 Such interconversion is driven by local hyperfine fields and modified by the Earth’s static magnetic field. It appears to be sufficiently robust to dephasing influences10 so that it can even be perturbed by the oscillatory fields of ambient anthropogenic electromagnetic noise.7 Common to these processes is that they require the involvement of sufficiently low spin−orbit coupling and usually also very weakly spin−spin coupled pairs of localized paramagnetic electronic states, that is, with small exchange energies and dipolar coupling strengths. These pairs can undergo either spatial, energetic, or both spatial and energetic transitions into doubly occupied electronic states, © 2017 American Chemical Society

Received: March 16, 2017 Revised: June 3, 2017 Published: June 30, 2017 4648

DOI: 10.1021/acs.nanolett.7b01135 Nano Lett. 2017, 17, 4648−4653

Letter

Nano Letters

ensemble.12 Nuclear magnetic resonance imparts perturbations of electronic transport on energy scales as small as a millionth of kT but is still resolved in femtoampère changes to device current.12 The unique appeal of coherent spin electronics in OLEDs is that experiments can, in principle, be carried out at arbitrary resonance frequencies and temperatures. This fact has led to the realization of ultrastrong light−matter coupling in the driving of singlet−triplet pair transitionsthe emergence of the nonperturbative regime of electron paramagnetic resonance (EPR), where the Rabi frequency approaches the energetic separation13 of spin eigenstates. This regime, which leads to the formation of a new singlet−triplet basis of spin wave functions as illustrated in Figure 1b, has been shown14 to be analogous to the emergence of superradiant collectivity of optical dipole transitions in the Dicke effect.15,16 When the DC magnetic field B0 applied to the spin ensemble exceeds the inhomogeneous broadening of the individual spin states and the oscillating driving field B1 is of comparable magnitude to B0, the individual spins become indistinguishable and interact collectively with the driving field.14 We recently reported experiments close to this threshold by using custom-designed nuclear magnetic resonance (NMR) coils surrounding the OLED.17 This “brute force” approach, however, requires extremely high electrical powers, which lead to excessive heat formation.17 This heat cannot be efficiently dissipated and ultimately destroys the OLED. An alternative approach recently put forward is the use of superconducting microwire resonators to achieve B1 ≈ B0.18 However, this method comes with the requirement of low temperatures, where a range of additional EPR features can arise due to long-

Figure 1. Illustration of spin-dependent electronic transitions controlled by spin-permutation symmetry and the change of transition rates under strong AC magnetic resonant drive. (a) Two adjacent localized singly occupied paramagnetic charge carrier states (an electron and a hole polaron), which are weakly spin−orbit coupled and weakly exchange and dipolar spin−spin coupled, form a pair whose spin-permutation symmetry governs the probability to yield one doubly occupied state (an exciton) within the singlet spin manifold. When the two charge carriers experience exposure to locally different proton hyperfine fields, their electron spin states will be arbitrary, and thus the pair permutation symmetry will be arbitrary, too. (b) Under strong AC magnetic field drive, the two charge carrier states become mutually coherent and form a triplet to pair state. The transition probability into a doubly occupied singlet state is then strongly quenched.

Figure 2. Monolithic OLED-magnetic resonance structures and experimental setup. (a) Vertical stack structure of the device layers (SY-PPV and PEDOT are conducting polymers in the OLED). (b) Photograph of the device structure with the active device area of 57 μm diameter visible as a bright spot. The photograph was taken prior to spin coating of the electroactive polymer layers and thermal deposition of the cathode. The microwire runs vertically across the image and the left-hand horizontal connector links to the anode of the OLED. The green box is a photoresist layer which defines the OLED geometry horizontally. (c) Sketch of the experimental measurement configuration employed, with the resonant driving field B1 and the Zeeman field of the spin states B0 indicated. 4649

DOI: 10.1021/acs.nanolett.7b01135 Nano Lett. 2017, 17, 4648−4653

Letter

Nano Letters lived triplet excitons in the OLED.11 Superconductivity also makes pulsing of the B1 field difficult because of the high Qfactors of these resonators. However, pulsing of B1 is desirable since it allows the detection of coherent Rabi nutation in the device current, which provides an absolute measurement of the magnitude of the local B1 field acting on the device.17 Here, we present a new device structure for probing the ultrastrong coupling regime at room temperature by optimizing two aspects of resonantly driving OLEDs: we minimize field inhomogeneity by shrinking the active device area to 57 μm diameter and at the same time optimizing coupling of the B1 field to the OLED in an architecture combining a microwire RF source19 and an OLED in one monolithic thin-film device. Figure 2 shows the device structure employed. Conceptually, this approach makes use of some of the properties of previously utilized EPR stripline resonator geometries.20 However, in contrast to these resonators, the microwire structure discussed here is significantly smaller even though it is used for frequencies which are almost 2 orders of magnitude lower. The microwire is therefore operated nonresonantly which allows broad frequency sweeps at very large values for B1. The layer stack, as illustrated in Figure 2, constitutes a system of successively deposited thin-film materials, with intermediate lateral structuring steps which together create a structure to allow for (i) the application of AC currents in the radio frequency to microwave-frequency range for the generation of very homogeneous, high-amplitude in-plane AC magnetic fields as needed for the magnetic resonance; (ii) electrical insulation and thermal isolation of the microwire from the surface of the stack system and heat sinking through the silicon substrate below the layer stack; and (iii) low surface roughness (in the range of a few nanometers) of the layer stack that is required for the deposition of the organic thin-film layers, that is, the layer stacks needed for bipolar injection as used in the experiments presented here. In essence, the device consists of a thin-film copper wire onto which an OLED structure is deposited. Different metallic, dielectric, and adhesive layers have to be used to ensure sufficient electrical and thermal conductivity of the RF microwire, electrical isolation from the OLED, and smoothing of the OLED bottom contact to prevent vertical shorts. The template structure of the device is fabricated up to the step of the photoresist layer under cleanroom conditions. The OLED active layers, the hole injector poly(styrene sulfonate)-doped poly(3,4-ethylenedioxythiophene) (PEDOT:PSS), and the emitting layer superyellow poly(phenylene vinylene) (SY-PPV) are spincoated in a nitrogen-filled glovebox and subsequently contacted with a cathode layer of calcium and aluminum deposited by thermal evaporation under high vacuum. A photograph of the template structure is shown in Figure 2b. The thin-film wire used to generate the RF radiation runs vertically across the image and is shrunk down to 150 μm width beneath the OLED pixel, which appears as a bright spot. The copper microwire is electrically and thermally isolated from the OLED by a thick layer stack consisting of SiNx, SiO2, and spin-on glass (SOG). The bottom electrode of the OLED is made of gold with an additional 3 nm thick oxidized Ti layer and protrudes horizontally to the left of the microwire. The entire fabrication procedure is described in detail in the Supporting Information. As sketched in Figure 2c, the RF source is connected to the copper thin-film wire, generating an oscillating B1 field in the plane of the OLED. The static magnetic field B0 is applied orthogonally to this field. The direct current flowing through

the OLED is measured using an analogue-to-digital converter and a data acquisition (DAQ) system. The OLEDs are driven in forward bias, and the current change under static (B0) and oscillating (B1) magnetic field is recorded at room temperature. Figure 3a shows the current−voltage characteristics of a monolithic device. The OLED current shows the expected diodic behavior, turning on just above a bias of 2 V. At constant voltage, the current of the device changes under application of

Figure 3. Characterization of the monolithic OLED-magnetic resonance structure. (a) Current−voltage characteristics of a working device. (b) Magnetoconductivity characteristics with no RF field (blue) and with RF fields of different amplitudes. At the highest driving fields, a second-harmonic feature appears in the resonance spectrum due to the nonlinearity of the RF source and is of no further relevance. (c) Linear relation between the central peak of the resonance spectrum Bc0 and the driving frequency, compared to the relation extracted at very low fields for the related polymer material MEH-PPV (red line). (d) Low-field magnetic resonance spectrum extracted from the difference between the zero-field and the finite-field magnetoconductivity curves. The spectrum is accurately described by the sum (red) of two Gaussian functions (purple, green) originating from the two hyperfine-field distributions experienced by the electron and hole spins, respectively. 4650

DOI: 10.1021/acs.nanolett.7b01135 Nano Lett. 2017, 17, 4648−4653

Letter

Nano Letters an external magnetic field B0. The blue curve in Figure 3b shows the magnetoconductivity of a device operating at a current of 101 nA. The axis tick marks correspond to a current change of 200 pA. The current initially decreases slightly with increasing B0 within the first 0.2 mT from the originthe ultrasmall magnetic field effect22and subsequently increases again. The black curves show the same measurements, offset on the current axis, with oscillating B1 fields applied at a frequency of 85 MHz and with amplitudes of 18.5 ± 1.4, 165 ± 2, and 587 ± 7 μT with the magnitudes and the errors of B1 determined as described in the Supporting Information. The RF field leads to a dip in the magnetoconductivity curves. This dip results from the fact that hyperfine fields induce mixing between singlet and triplet electron−hole pair states in the conjugated polymer, and the external magnetic field B0 tends to suppress this mixing.14,19 The external RF field reopens the mixing channel at the resonant magnetic field, quenching the current. As B1 increases from 18.5 to 165 μT, the resonance deepens but also broadens. This effect is a consequence of power broadening, arising from the fact that the large B1 amplitude results in a greater number of individual spins within the hyperfine-broadened ensemble coupling to the oscillatory field. At 587 μT, the resonance spectrum is clearly broadened, but the amplitude is also decreased. This reduction in resonance amplitude is a signature of the emergence of ultrastrong coupling, which leads to the formation of a new spin basis set and new selection rules for the resonant spin transitions.14 We note that, in addition to the spin-1/2 resonance at 3.07 mT (85 MHz), at the highest driving fields a second-harmonic feature is also observed at 6.14 mT. This second harmonic likely arises from the slight nonlinearity of the RF amplifier, which will lead to a weak second-harmonic component in the RF radiation generated in the microwire at the nominal frequency of 85 MHz. We stress that all measurements are of the steady-state current, without the need for modulation or lock-in detection. This approach is in contrast to our previous work where a millimeter-thick Cu wire was mounted externally on the OLED.19 It is not possible to achieve sufficient continuous-wave RF power in such nonmonolithic devices to enable the direct-current detection demonstrated here without significant heating effects occurring or violation of the requirements (i) and (ii) discussed above. The resonance feature follows the dependence of position of the resonance peak Bc0 on driving frequency f expected for a free electron, as plotted in Figure 3c. The red line shows the relationship extracted in the low-B1 regime for a related polymer material,19 poly[2-methoxy-5-(2-ethylhexyloxy)-1,4phenylenevinylene] (MEH-PPV). To demonstrate the quality of the data, we plot the differential current recorded on resonance at B1 = 52(1) μT in panel d. As with previous reports of time-integrated17 and time-resolved detection,11,19 the spectrum is accurately described by the sum (red curve) of two Gaussian functions, with the two Gaussians marked in purple and green. The two Gaussians arise from the distributions in hyperfine fields experienced by the electron and hole wave functions in the polymer, respectively. In examining the dependence of resonance spectrum on B1 strength, the biggest challenge lies in accurately determining the magnitude of B1. This parameter can neither be measured directly with an external detector, since the RF microwire does not radiatethe resonance effect is generated in the near-field of the microwire; nor can it be calculated accurately, since the microwire invariably suffers from reflection and coupling losses and the local field is highly inhomogeneous.21 In time-resolved

electrically detected magnetic resonance experiments, B1 can be determined from the Rabi precession frequency.17 Here, we closely examine the effect of power broadening on the resonance spectrum to determine the magnitude of B1. Power broadening contributes to the spectrum the shape of a Lorentzian curve, whereas the underlying statistical distribution of local hyperfine fields experienced by each spin follows a Gaussian shape, with each carrier species being described by a distinct Gaussian curve. At low driving powers, the effect of this disorder dominates the spectrum so that a pure doubleGaussian shape emerges as shown in Figure 3d. As the power increases, the spectral shape changes gradually from doubleGaussian to Lorentzian, with intermediate fields showing a spectrum consisting of a convolution of the two, a double-Voigt line. Figure S1 of the Supporting Information shows additional examples of resonance spectra recorded at different RF excitation powers, with the power increasing from top to bottom. The spectra clearly broaden with increasing power. Using a global fitting procedure for all these data sets with only the two hyperfine distributions of electron and hole spin along with B1 as the three fitting parameters, we can assign a precise value of B1 to each resonance spectrum, and we can also obtain error margins for the B1 measurements. Details of this fitting procedure are given in the Supporting Information. With accurate values of B1 extracted for a given power, we can now investigate the effect of driving field strength on the resonance spectrum. As described in detail in ref 14, the ensemble of spin states susceptible to EPR changes as the driving field is raised from the weak-coupling low-field (perturbative) to the ultrastrong-coupling high-field regime. Figure 4a summarizes the two limits. In the perturbative regime, the driving field B1 is much smaller than the average difference in the distribution of hyperfine fields experienced by the two charge carriers, δBhyp. This field lifts the degeneracy between the antisymmetric pair states of triplet and singlet symmetry, T0 and S0. EPR transitions then occur to the pure triplet states T+ and T−. In contrast, when B1 is of the order of or larger than δBhyp, the approximation that it is pure singlespin states which are driven under resonance breaks down. Three new triplet superposition states emerge as stated in Figure 4a, besides the pure singlet state.14 As shown in ref 14, and probed experimentally in refs 17, 18, the selection rules for allowed EPR transitions change. Transitions between singlet and triplet are blocked and only occur between the three new superposition triplet levels. This change of basis set is a consequence of indistinguishability of the spin pair with respect to the driving electromagnetic field: once the driving field strength exceeds the level of inhomogeneous broadening, which arises predominantly from the local hyperfine fields, the individual spins of the pairs become quantum mechanically indistinguishable, necessitating a description in terms of a new common wave function. This phenomenon is a manifestation of the Dicke effect,15 which is best known from superradiant luminescence of atomic gases,16 and has a profound effect on the EPR amplitude and spectrum.14,17 Figure 4b plots the normalized device current change ΔI as a function of B0 and B1 on a false-color scale. The maximal resonance amplitude is shown in panel c. Initially, the overall amplitude, which is negative since it measures current quenching, rises with increasing B1 but then saturates around B1 = 0.2 mT. Subsequently, the resonance amplitude decreases again, until the resonance vanishes completely at B1 = 1.1 mT. At even higher B1, the resonance inverts and shows the same effect of 4651

DOI: 10.1021/acs.nanolett.7b01135 Nano Lett. 2017, 17, 4648−4653

Letter

Nano Letters

demonstrates the accuracy of the B1 calibration carried out here. The coils used in the earlier experiment could be pulsed, making it possible to measure Rabi oscillations directly in the time domain to provide an absolute measure of B1. Finally, we note that our measurement technique is quite sensitive to potential heating effects arising from the microwire, since a temperature change would lead to a modification of the current−voltage characteristics, which are recorded for every B1 field. We see no change of the I−V characteristics with B1 off resonance, implying that, in the field range studied here, heating effects are negligible. Besides constituting micron-scale versatile magnetometers, which provide an absolute measure of the magnetic field by measuring the resonance frequency,19 the monolithic OLEDmicrowire structures introduced here offer a versatile approach to easily reach a state of collective spin precession in OLEDs, in which all spins precess in phase coherently. Since it is the electromagnetic field which induces this coherence, the demonstration of on-chip pixel-size collectivity in spin precession provides a means to position multiple individual devices to be driven by one and the same field. As with other coherent quantum systems such as atomic gases in cavities, it may be possible to entangle the collective Dicke state of one pixelmeasured through the device currentwith that of an adjacent pixel, driven by the same field. Such a manifestation would provide a unique experimental approach to roomtemperature macroscopic coherence phenomena with all of the benefits associated with the tunability of both spin−orbit coupling23 and hyperfine fields12 of organic semiconductors.

■

ASSOCIATED CONTENT

* Supporting Information S

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.7b01135. Details of the fabrication steps of device structure templates for the monolithic polymer devices used in this study and details of the global least-squares fit analysis of the power-dependent electron paramagnetic resonance spectra for the determination of the field amplitude B1 (PDF)

Figure 4. Spin-Dicke effect in monolithic OLED-microwire magneticresonance structures. (a) Change of the spin basis set probed with EPR from the low B1 field regime to the onset of the spin-Dicke effect at high resonant driving fields. (b) Raw data of magnetoconductivity curves as a function of driving power at 85 MHz, plotted on a falsecolor scale, showing power broadening of the resonance, bleaching, and a subsequent inversion of resonance sign at high driving fields. (c) Plot of the amplitude of the resonance maximum at 3.07 mT as a function of B1 compared to results obtained with a different conjugated polymer material, MEH-PPV, in an OLED driven by a RF coil rather than a microwire (data taken from ref 17).

■

AUTHOR INFORMATION

Corresponding Authors

power broadening as seen for very low B1 fields, but with a reversed sign. Our initial exploration of this exotic regime of magnetic resonance was probed in a similar conjugated polymer, MEHPPV, with devices mounted in an external RF coil to generate the high B1 fields.17 This approach requires very high driving powers for the coils and leads to heating of the device and, ultimately, catastrophic breakdown: OLED overheats and the coils burn. Crucially, for B1 fields generated with coils around devices of regular hydrogenated MEH-PPV, which has a comparable hyperfine coupling strength to SY-PPV, we were not able to demonstrate the inversion of the resonance sign17 to signify the emergence of the Dicke regime.14 This transition is now clearly revealed in the new data in Figure 4. Nevertheless, up to a field of B1 = 1.2 mT, the MEH-PPV data17 closely follow those acquired here for SY-PPV. The comparison between the two measurements is important, since it

*E-mail: [email protected]. *E-mail: [email protected]. ORCID

Christoph Boehme: 0000-0001-7323-5757 Notes

The authors declare the following competing financial interest(s): The University of Utah Research Foundation has filed a provisional patent application based on the device concept discussed in this article. Authors Jamali and Boehme are named as inventors on this pending patent.

■

ACKNOWLEDGMENTS This work was supported by the US Department of Energy, Office of Basic Energy Sciences, Division of Materials Sciences and Engineering under Award no. DE-SC0000909. 4652

DOI: 10.1021/acs.nanolett.7b01135 Nano Lett. 2017, 17, 4648−4653

Letter

Nano Letters

■

REFERENCES

(1) Petta, J. R.; Johnson, A. C.; Taylor, J. M.; Laird, E. A.; Yacoby, A.; Lukin, M. D.; Marcus, C. M.; Hanson, M. P.; Gossard, A. C. Science 2005, 309 (5744), 2180−2184. (2) Johnson, A. C.; Petta, J. R.; Taylor, J. M.; Yacoby, A.; Lukin, M. D.; Marcus, C. M.; Hanson, M. P.; Gossard, A. C. Nature 2005, 435 (7044), 925−928. (3) Koppens, F. H. L.; Folk, J. A.; Elzerman, J. M.; Hanson, R.; van Beveren, L. H. W.; Vink, I. T.; Tranitz, H. P.; Wegscheider, W.; Kouwenhoven, L. P.; Vandersypen, L. M. K. Science 2005, 309 (5739), 1346−1350. (4) Maune, B. M.; Borselli, M. G.; Huang, B.; Ladd, T. D.; Deelman, P. W.; Holabird, K. S.; Kiselev, A. A.; Alvarado-Rodriguez, I.; Ross, R. S.; Schmitz, A. E.; Sokolich, M.; Watson, C. A.; Gyure, M. F.; Hunter, A. T. Nature 2012, 481 (7381), 344−347. (5) Chorley, S. J.; Galpin, M. R.; Jayatilaka, F. W.; Smith, C. G.; Logan, D. E.; Buitelaar, M. R. Phys. Rev. Lett. 2012, 109 (15), 156804. (6) Hanson, R.; Awschalom, D. D. Nature 2008, 453 (7198), 1043− 1049. (7) Engels, S.; Schneider, N. L.; Lefeldt, N.; Hein, C. M.; Zapka, M.; Michalik, A.; Elbers, D.; Kittel, A.; Hore, P. J.; Mouritsen, H. Nature 2014, 509 (7500), 353−357. (8) Gauger, E. M.; Rieper, E.; Morton, J. J. L.; Benjamin, S. C.; Vedral, V. Phys. Rev. Lett. 2011, 106 (4), 040503. (9) Hogben, H. J.; Biskup, T.; Hore, P. J. Phys. Rev. Lett. 2012, 109 (22), 220501. (10) Lambert, N.; Chen, Y. N.; Cheng, Y. C.; Li, C. M.; Chen, G. Y.; Nori, F. Nat. Phys. 2012, 9 (1), 10−18. (11) Kavand, M.; Baird, D.; van Schooten, K.; Malissa, H.; Lupton, J. M.; Boehme, C. Phys. Rev. B: Condens. Matter Mater. Phys. 2016, 94 (7), 075209. (12) Malissa, H.; Kavand, M.; Waters, D. P.; van Schooten, K. J.; Burn, P. L.; Vardeny, Z. V.; Saam, B.; Lupton, J. M.; Boehme, C. Science 2014, 345 (6203), 1487−1490. (13) Gunter, G.; Anappara, A. A.; Hees, J.; Sell, A.; Biasiol, G.; Sorba, L.; De Liberato, S.; Ciuti, C.; Tredicucci, A.; Leitenstorfer, A.; Huber, R. Nature 2009, 458 (7235), 178−181. (14) Roundy, R. C.; Raikh, M. E. Phys. Rev. B: Condens. Matter Mater. Phys. 2013, 88 (12), 125206. (15) Dicke, R. H. Phys. Rev. 1954, 93 (1), 99−110. (16) Gross, M.; Haroche, S. Phys. Rep. 1982, 93 (5), 301−396. (17) Waters, D. P.; Joshi, G.; Kavand, M.; Limes, M. E.; Malissa, H.; Burn, P. L.; Lupton, J. M.; Boehme, C. Nat. Phys. 2015, 11 (11), 910− 914. (18) Bayliss, S. L.; Greenham, N. C.; Friend, R. H.; Bouchiat, H.; Chepelianskii, A. D. Nat. Commun. 2015, 6, 8534. (19) Baker, W. J.; Ambal, K.; Waters, D. P.; Baarda, R.; Morishita, H.; van Schooten, K.; McCamey, D. R.; Lupton, J. M.; Boehme, C. Nat. Commun. 2012, 3, 898. (20) Wallace, W. J.; Silsbee, R. H. Rev. Sci. Instrum. 1991, 62, 1754− 1766. (21) Joshi, G.; Miller, R.; Ogden, L.; Kavand, M.; Jamali, S.; Ambal, K.; Venkatesh, S.; Schurig, D.; Malissa, H.; Lupton, J. M.; Boehme, C. Appl. Phys. Lett. 2016, 109 (10), 191−195. (22) Nguyen, T. D.; Gautam, B. R.; Ehrenfreund, E.; Vardeny, Z. V. Phys. Rev. Lett. 2010, 105 (16), 166804. (23) Sun, D.; van Schooten, K. J.; Kavand, M.; Malissa, H.; Zhang, C.; Groesbeck, M.; Boehme, C.; Vardeny, Z. V. Nat. Mater. 2016, 15 (8), 863−868.

4653

DOI: 10.1021/acs.nanolett.7b01135 Nano Lett. 2017, 17, 4648−4653