Letter pubs.acs.org/NanoLett
Solid-State Semiconductor Optical Cryocooler Based on CdS Nanobelts Dehui Li,† Jun Zhang,† Xinjiang Wang,§ Baoling Huang,§ and Qihua Xiong*,†,‡ †
Division of Physics and Applied Physics, School of Physical and Mathematical Sciences, Nanyang Technological University, Singapore 637371 ‡ NOVITAS, Nanoelectronics Centre of Excellence, School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore 639798 § Department of Mechanical and Aerospace Engineering, School of Engineering, The Hong Kong University of Science and Technology, Kowloon, Hong Kong ABSTRACT: We demonstrate the laser cooling of silicon-oninsulator (SOI) substrate using CdS nanobelts. The local temperature change of the SOI substrate exactly beneath the CdS nanobelts is deduced from the ratio of the Stokes and antiStokes Raman intensities from the Si layer on the top of the SOI substrate. We have achieved a 30 and 20 K net cooling starting from 290 K under a 3.8 mW 514 nm and a 4.4 mW 532 nm pumping, respectively. In contrast, a laser heating effect has been observed pumped by 502 and 488 nm lasers. Theoretical analysis based on the general static heat conduction module in the Ansys program package is conducted, which agrees well with the experimental results. Our investigations demonstrate the laser cooling capability of an external thermal load, suggesting the applications of II−VI semiconductors in all-solid-state optical cryocoolers. KEYWORDS: Optical refrigeration, CdS nanobelts, Raman scattering, photoluminescence up-conversion, electron−phonon interaction
O
achievable cooling temperature. More importantly, the semiconductor can be used as heat sink for many electronic and photonic devices, such as sensors and detectors.13 Although many theoretical5,6,11,12,14−16 and experimental17−19 studies have addressed various aspects of the laser cooling of semiconductors, no net cooling has been achieved until the recent demonstration of 40 K cooling using CdS nanobelts.20 The problems preventing the realization of net laser cooling in semiconductors for a long time include a high nonradiative recombination rate,5 a low extraction efficiency due to the total internal reflection, and the reabsorption effect.6 Compared to GaAs-based quantum wells, CdS nanobelts exhibit strong and enhanced electron-longitudinal optical phonon interactions. Furthermore, the high luminescence extraction efficiency and high external quantum yield play important roles as well.20 In this report, we demonstrate the concept of all-solid-state optical refrigerator based upon a CdS nanobelt, which is in good thermal contact with a patterned silicon-on-insulator (SOI) substrate. The top crystalline Si layer in a SOI substrate provides the local temperature probe based upon the ratio of Si Stokes and anti-Stokes Raman intensities, which exponentially depends on temperature. We demonstrate that the temperature of the Si layer beneath the CdS nanobelt drops 30 and 20 K starting from 290 K under a 3.8 mW, 514 nm and a 4.4 mW,
ptical refrigeration or laser cooling of solids based on the anti-Stokes luminescence was first proposed by Pringsheim in 1929.1,2 Unlike the laser cooling of diluted atoms where the translational energy of atoms is removed via the interaction with the laser field,3 laser cooling of solids is achieved by taking away the lattice vibrational energy during the anti-Stokes emission process, i.e., annihilation of phonons.1,2,4 To realize the laser cooling of solids, the material requires having high crystalline quality with properly spaced energy levels and high external quantum efficiency.5,6 Due to those stringent requirements, laser cooling of solids experienced much greater challenges in materials than laser cooling of atoms. The net laser cooling of solids has not been achieved until 1995 when Epstein and co-authors demonstrated the first laser cooling in ytterbium-doped glasses.4 Since then, a great effort has been put to improve the cooling efficiency and to achieve a large cooling temperature in cooling of a variety of the rare earth (RE) doped crystals and glasses.7−10 To date, the minimum achievable temperature of ∼110 K starting from the room temperature has been realized in a 5% doped Yb:YLF crystal, which surpasses the commercial thermoelectric Peltier coolers, reaching the limit imposed by the Boltzmann’s distribution governning the atomic resonance in the REdoped systems.10 Compared with the atomic resonance involved in the RE doped systems, laser cooling of semiconductors with excitonic resonance dominated6,11,12 exhibits a number of advantages such as more efficient pump light absorption and much lower © 2014 American Chemical Society
Received: May 17, 2014 Revised: June 30, 2014 Published: July 17, 2014 4724
dx.doi.org/10.1021/nl501831f | Nano Lett. 2014, 14, 4724−4728
Nano Letters
Letter
and 514 nm, and Nd:YAG solid state 532 nm laser) and probe beams (532 nm laser) were collimated and focused through a 50× objective onto a single CdS nanobelt laid on top of the SOI substrate. All spectra were collected by a confocal triple grating spectrometer (Horiba-JY T64000) in a backscattering configuration.24,25 During the laser cooling experiments, each spectrum was taken within 5 min interval when the pumping laser was blocked temporally except for the 532 nm laser pumping. A pump−probe luminescence thermometry (P2LT) technique was used to cool and measure the local temperature variation of CdS nanobelts suspended on the holes as well.20 It should be noted that P2LT method is similar to differential luminescence thermometry,8 both of which give almost the identical results.20 Basically, the P2LT method is based on the change of the band gap energy with temperature. One pump beam (488, 502, 514, and 532 nm) is used to cool the sample, while a probe beam (532 nm with very low intensity) is used to detect the local cooling. For each 5 min, the pump beam is switched off temporarily, and the probe beam is turned on to take the Stokes spectrum, from which the local temperature change can be deduced. Both pump beam and probe beam are focused to the same spot by using a 50× objective. More information about the P2LT method can be found elsewhere.20 The local temperature change of the Si layer beneath the CdS nanobelt is deduced from the intensity ratio of the Stokes and anti-Stokes Raman signals. The phonon population follows the Bose−Einstein statistics, as such the intensity ratio which exponentially depends on a dimensionless factor ℏω0/kBT under nonresonant conditions26,27
532 nm laser pumping, respectively. In contrast, the heating effect has been observed pumped by 502 and 488 nm lasers. The theoretical study based on the general static heat conduction module in Ansys program package gives consistent results with experimental data. The CdS nanobelts were grown in a home-built vapor transport chemical vapor deposition (CVD) system. The detailed growth conditions, morphology and crystalline characterizations can be found elsewhere.21,22 The as-grown nanobelts are several micrometers in width, tens of micrometers in length, and 30−200 nm in thickness, respectively. The cartoon presentation of the concept and the actual device image are shown in Figure 1a and b. To achieve the largest
IStokes Ianti‐Stokes
⎛ ℏω ⎞ = A exp⎜ 0 ⎟ ⎝ kBT ⎠
(1)
where IStokes and Ianti‑Stokes are the Stokes and anti-Stokes Raman intensities, ℏω0 is the phonon energy, A is a parameter depending on the materials and excitation wavelength, and kBT is the thermal energy. This exponential dependence can be used as a very sensitive local thermometry. For Si in this experiment, we found that anti-Stokes Raman signal becomes very weak below 140 K. Therefore, our method is no longer applicable to determine the temperature below 140 K. The Stokes and anti-Stokes evolution with temperature from the Si layer upon a 4.4 mW 532 nm laser irradiation is given in Figure 1c. As the temperature decreases, the ratio of the Stokes and anti-Stokes Raman signal continually decreases, as plotted in Figure 1d. By applying eq 1, the fitting curve and parameters are given in Figure 1d as well, giving a phenomenological calibration equation as
Figure 1. Semiconductor solid-state optical cryocooler. (a) Cartoon presentation of the optical cryocooler. (b) An SEM image of an actual device consisting of a single CdS nanobelt supported on a patterned SOI substrate. (c) The temperature-dependent intensities of the Stokes and anti-Stokes Raman from Si top layer in the SOI substrate excited by a 532 nm laser with a power of 4.4 mW. The spectra are normalized by the Stokes intensity and vertically offset for clarity. (d) The intensity ratio of the Stokes and anti-Stokes Raman signal extracted from (c) (blue dots). The red curve is a fitting according to eq 1 with the fitting parameters listed.
cooling efffect, we intentionally selected the thickness of the nanobelts used to be around 100 nm.23 The SOI substrate was patterned into grids by e-beam lithography and etched by reactive ion etching, which is meant to decrease the thermal conductive loss and to provide a reference without a thermal load (suspended segment). The thicknesses of device layer Si, buried SiO2 layer, and handle Si layer of the SOI substrate are 2.5, 1.5, and 500 μm, respectively. The optical cryocooler demonstration was conducted from the frame part, where the CdS nanobelt was in good thermal contact with the crystalline Si top layer. The chip was mounted on the coldfinger of a continuous flow microscopy cryostat to control the starting temperature and to keep the surrounding temperature constant. The probe beam of 532 nm was kept as low as 0.8 mW in order to minimize the cooling or heating effect from probe beam except for 532 nm pumping, when the power of the probe beam was the same as the pumping beam. The 532 nm laser is far away from the emission peak of CdS; therefore, we selected it as the probe beam to eliminate the influence of emission from CdS on the Si Raman peaks. Both pump (Ar ion laser 488, 502,
IStokes Ianti‐Stokes
⎛ 739 ⎞ ⎟ = 0.517 exp⎜ ⎝ T ⎠
(2)
The temperature resolution for this method is estimated to be around 1.5 K at room temperature and 0.1 at 180 K. Figure 2a displays the Stokes and anti-Stokes Raman spectrum evolution of the Si layer beneath the CdS nanobelt upon a continuous 4 mW 532 nm laser radiation starting at 290 K. Each spectrum was taken within a 5 min interval. The local temperature change ΔT of the Si beneath the CdS nanobelt extracted from the Raman spectra (Figure 2a) is shown in Figure 2b versus the pumping time. One can observe that the local temperature gradually decreases from 290 K and reaches an equilibrium temperature ∼270 K after ∼30 min. The maximum cooling effect is ∼20 K. The local temperature 4725
dx.doi.org/10.1021/nl501831f | Nano Lett. 2014, 14, 4724−4728
Nano Letters
Letter
with a mean emission wavelength of 507 nm, where P0 is the power of the pumping laser. To compare the results of the laser cooling of SOI substrate with the previous reported results of the laser cooling of CdS itself,20,23 we conducted both the cooling of SOI substrate using the ratio of the intensities of Stokes and anti-Stokes and the cooling of suspended CdS nanobelt itself by P2LT method on the same belt. The normalized temperature change for the laser cooling of CdS nanobelt itself is plotted versus wavelengths as well shown as black squares in Figure 2d. Compared with the cooling of the CdS nanobelt itself, the heating of substrate and the heat dissipation via substrate during the probe time lead to a smaller normalized temperature change for 514 and 532 nm laser pumping and a larger normalized temperature change for 502 and 532 nm laser pumping for the laser cooling of the SOI substrate. Numerical simulations on the laser cooling of the SOI substrate through CdS nanobelts have been performed by using the general static heat conduction module in the Ansys program package. The symmetry treatment is adopted in the geometric modeling to reduce the simulation domain to 1/4 of the original size. As shown in Figure 3a, the laser beam irradiates the CdS nanobelt on a circle spot with a diameter of 2 μm. The belt is placed perpendicular to and at the center of the Si frame, which is 4 μm wide and 2.5 μm thick. The thickness of middle layer SiO2 and the bottom Si substrate are 1.5 and 500 μm, respectively. The Si ridge repeats each other with a gap of 10 μm in two directions on the middle layer SiO2 surface and forms a set of square grid. During the simulation, the in-plane and cross-plane thermal conductivities of Si are taken as 148 W/(m·K) and 117 W/(m· K), respectively, considering the boundary scattering of phonons in Si thin film,28,29 while the bottom silicon substrate is considered isotropic and the thermal conductivity is set as 155 W/(m·K). Similarly, the in-plane and cross-plane thermal conductivities of CdS nanobelts are estimated to be 16 and 8.7 W/(m·K) from the measured bulk value 16.7 W/(m·K).30 The thermal conductivity of SiO2 film is set as 1.5 W/(m·K). The thermal contact conductance of SiO2/Si interface is taken as 26 MW/(m2·K).31 The value for the CdS/Si interface is estimated to be around 10 MW/(m2K) according to the diffusive mismatch model32 and surface roughness. The effective thermal contact conductance between the SOI wafer and the cryostat is obtained by fitting the experimental thermal response relaxation
Figure 2. (a) Time evolution of the Stokes and anti-Stokes from Si layer on the SOI substrate with a 4.4 mW 532 nm laser pumping starting from 290 K. All of the spectra are normalized by the Stokes Raman intensity from the Si and offset vertically for clarity. (b) The time evolution of the temperature change of Si extracted from (a). (c) The time evolution of the temperature change of Si for various pumping wavelengths (532, 514, 502, and 488 nm) starting from 290 K. The red arrows indicate the pumping lasers are switched off. (d) The experimental results of normalized temperature change of suspended part (black squares) and contact part (red dots) for different pumping wavelengths starting from 290 K.
change ΔT versus time for different pump wavelengths (532, 514, 502, and 488 nm) is plotted in Figure 2c. The temperature of sample reaches a stable value after pumping ∼20−30 min. A net laser cooling of 20 and 30 K is achieved for 532 and 514 nm laser pumping, while a heating effect is observed for 502 and 488 nm laser pumping, which is consistent with our previous report.20 After the pump laser is stopped as indicated by red arrows in Figure 2c, the local temperature of Si gradually returns to the starting temperature of 290 K. The red dots in Figure 2d provide the experimental results of the normalized temperature change ΔT/P0 for the contact part of the nanobelt
Figure 3. (a) Temperature distribution of the calculation domain for a 4.4 mW, 532 nm laser pumping. The inset depicts the temperature profile along the z-axis from the laser spot center on the CdS top surface to the Si/SiO2/Si substrate. (b) The simulated time evolution of the local temperature change of Si layer for a 4.4 mW, 532 nm laser pumping along with the experimental results. 4726
dx.doi.org/10.1021/nl501831f | Nano Lett. 2014, 14, 4724−4728
Nano Letters
Letter
We also carried out the theoretical calculation based on the procedures we developed in our previous publication.20,23 The normalized temperature change ΔT/P0 can be expressed as
time. The body heat sink density is extracted from the steadystate temperature distribution. The simulated transient thermal response curve for the case of a 4.4 mW, 532 nm laser is shown in Figure 3b together with the experimental data. The cooling temperature for the Si top surface in the case with a 7.8 mW, 532 nm laser is also well reproduced (26 K) using these parameters. The laser cooling powers in these two cases are then calculated as ∼68 μW and ∼92 μW, respectively, in good agreement with the estimated value from Sheik−Bahae− Epstein theory. The simulated steady-state temperature distribution is also shown in Figure 3a. Apparently, the top Si thin film is quite uniformly cooled in this case. Using the input laser power and the calculated cooling power, 33 the corresponding refrigeration coefficient of performance (COP) is estimated to be around 1.3% for the case of 4.4 mW, 532 nm laser pumping and 2.0% for 4.3 mW, 514 nm laser pumping. Finally, the starting temperature dependence of the laser cooling of SOI substrate has been investigated above 150 K, below which the anti-Stokes Raman signal from Si is too weak to be used as the temperature probe. The time evolution of the Si Raman intensity normalized by the Stokes intensity starting at 270 K is shown in Figure 4a. After turning on the pumping
T − TC −Kt ·G(v , TC)ΔE ΔT ≡ S ≡ P0 P0 Ckhv
(3)
where t is the thickness of the nanobelt, ΔE is the energy difference between the energy of the pumping laser and the mean luminescence energy, hv is the energy of the pumping laser, k is the thermal conductivity of Si layer, and G(v,T) is the photoconductivity gain relating the absorption coefficient α(v,T) via α(v,T)t = KG(v,T), where K is a constant related to the thickness of the nanobelt and the applied source-drain voltage. The PC measurement provides an indirect way to evaluate the absorption at the temperature above 180 K, when the strong electron-longitudinal optical phonon interaction leads to the complete ionization of excitons and bound excitons.34 C is a parameter related to the heat distribution of the substrate at the interface between the CdS nanobelts and SOI substrate, which we take as a constant for all temperatures we investigated. As we can see, the normalized temperature change ΔT/P0 is proportional to the photoconductivity gain and the energy difference ΔE if we can take the external quantum efficiency (EQE) as unity. In view of the high EQE of 99.8% at room temperature,20 we can take it as unity for all temperatures below the room temperature based on the fact that the EQE increases with the decrease of the temperature. Furthermore, as the temperature decreases, the emission peak blue shifts leading to a large ΔE for a fixed pumping wavelength, while the absorption band tail moves far away from the fixed pumping wavelength, giving rise to a smaller absorption coefficient at the pumping wavelength.23 The energy difference ΔE can be experimentally obtained from the temperature dependent antiStokes photoluminescence spectrum, while the absorption coefficient can be obtained from the temperature-dependent photoconductivity gain as aforementioned. On the basis of eq 3 together with temperature-dependent measured mean emission wavelength, photoconductivity gain, and thermal conductivity of Si,35 the temperature variation dependence of the pumping wavelength at various starting temperatures was calculated under a 532 nm laser pumping with a power of 4 mW as shown in Figure 4c assuming the Kt in eq 3 maintains constant for all starting temperatures. As can be seen from Figure 4c, the optimal cooling wavelength blue shifts, and the cooling temperature ΔT increases first and then decreases with a maximum value presenting at 270 K as the temperature decreases. Below 210 K, the cooling temperature rises again. On the one hand, the energy difference ΔE increases with the lowering of the temperature which would give a larger cooling temperature ΔT. On the other hand, the absorption coefficient decreases at a fixed pumping wavelength on the band tail as the temperature decreases because the band edge shifts far away from the laser pumping as aforementioned. The competition between those two factors gives rise to the largest cooling temperature at 270 K and the smallest cooling temperature appearing at 210 K. Below 210 K, the presence of the donor−acceptor pair (DAP) emission gives rise to the redshift of the mean emission energy, which leads to the increase of the absorption again.22 Therefore, the cooling temperature gradually increases again as shown in Figure 4c. The theoretical values for a 532 nm laser
Figure 4. (a) Time evolution of the Stokes and anti-Stokes from Si layer on the SOI substrate with a 4 mW, 532 nm laser pumping starting from 270 K. All of the spectra are normalized and offset vertically for clarity. (b) The time evolution of the temperature change of Si starting at 270 K (extracted from a) and at 180 K. (c) The theoretical calculation of the normalized temperature change of SOI at various starting tempearture above 180 K. (d) The experimental (blue discrete points) and theoretical calculated (red discrete points) temperature change of SOI beneath the CdS nanobelt at different starting temperatures under a 4 mW, 532 nm laser pumping.
laser, the anti-Stokes Si Raman intensity gradually decreases and finally reaches a minimum value. The extracted temperature variation of the Si layer based on eq 1 for the starting temperature of 270 and 180 K was exhibited in Figure 4b. The starting temperature dependence of the cooling temperature is shown in Figure 4d as the blue dots for a 4 mW, 532 nm laser pumping. 4727
dx.doi.org/10.1021/nl501831f | Nano Lett. 2014, 14, 4724−4728
Nano Letters
Letter
(10) Seletskiy, D. V.; Melgaard, S. D.; Epstein, R. I.; Di Lieto, A.; Tonelli, M.; Sheik-Bahae, M. Opt. Express 2011, 19, 18229−18236. (11) Rupper, G.; Kwong, N. H.; Binder, R. Phys. Rev. Lett. 2006, 97, 117401. (12) Rivlin, L. A.; Zadernovsky, A. A. Opt. Commun. 1997, 139, 219− 222. (13) Epstein, R. I.; Sheik-Bahae, M. Optical Refrigeration; Wiley-VCH Verlag GmbH & Co. KGaA: Weinheim, 2009. (14) Rupper, G.; Kwong, N. H.; Binder, R. Phys. Rev. B 2007, 76, 245203. (15) Khurgin, J. B. J. Appl. Phys. 2006, 100, 113116. (16) Khurgin, J. B. Phys. Rev. B 2008, 77, 235206. (17) Imangholi, B.; Hasselbeck, M. P.; Sheik-Bahae, M.; Epstein, R. I.; Kurtz, S. Appl. Phys. Lett. 2005, 86, 081104. (18) Finkeissen, E.; Potemski, M.; Wyder, P.; Vina, L.; Weimann, G. Appl. Phys. Lett. 1999, 75, 1258−1260. (19) Gauck, H.; Gfroerer, T. H.; Renn, M. J.; Cornell, E. A.; Bertness, K. A. Appl. Phys. A: Mater. Sci. Process. 1997, 64, 143−147. (20) Zhang, J.; Li, D.; Chen, R.; Xiong, Q. Nature 2013, 493, 504− 508. (21) Xu, X.; Zhao, Y.; Sie, E. J.; Lu, Y.; Liu, B.; Ekahana, S. A.; Ju, X.; Jiang, Q.; Wang, J.; Sun, H.; Sum, T. C.; Huan, C. H. A.; Feng, Y. P.; Xiong, Q. ACS Nano 2011, 5, 3660−3669. (22) Liu, B.; Chen, R.; Xu, X.; Li, D.; Zhao, Y.; Shen, Z.; Xiong, Q.; Sun, H. J. Chem. Phys. C 2011, 115, 12826−12830. (23) Li, D.; Zhang, J.; Xiong, Q. Opt. Express 2013, 21, 19302− 19310. (24) Li, D.; Zhang, J.; Xiong, Q. ACS Nano 2012, 6, 5283−5290. (25) Zhang, J.; Peng, Z.; Soni, A.; Zhao, Y.; Xiong, Y.; Peng, B.; Wang, J.; Dresselhaus, M. S.; Xiong, Q. Nano Lett. 2011, 11, 2407− 2414. (26) Jellison, G. E., Jr.; Lowndes, D. H.; Wood, R. F. Phys. Rev. B 1983, 28, 3272−3276. (27) Gupta, R.; Xiong, Q.; Adu, C. K.; Kim, U. J.; Eklund, P. C. Nano Lett. 2003, 3, 627−631. (28) Asheghi, M.; Leung, Y.; Wong, S.; Goodson, K. Appl. Phys. Lett. 1997, 71, 1798−1800. (29) Hopkins, P. E.; Reinke, C. M.; Su, M. F.; Olsson, R. H., III; Shaner, E. A.; Leseman, Z. C.; Serrano, J. R.; Phinney, L. M.; El-Kady, I. Nano Lett. 2010, 11, 107−112. (30) Holland, M. Phys. Rev. 1964, 134, A471. (31) Lyeo, H.-K.; Cahill, D. G. Phys. Rev. B 2006, 73, 144301. (32) Swartz, E. T.; Pohl, R. O. Rev. Mod. Phys. 1989, 61, 605. (33) Kosloff, R.; Levy, A. Annu. Rev. Phys. Chem. 2014, 65, 365−393. (34) Li, D.; Zhang, J.; Zhang, Q.; Xiong, Q. Nano Lett. 2012, 12, 2993−2999. (35) Fulkerson, W.; Moore, J. P.; Williams, R. K.; Graves, R. S.; McElroy, D. L. Phys. Rev. 1968, 167, 765−782.
pumping are picked out from Figure 4c as shown in Figure 4d as well (blue dots), which agrees well with the experimental data except for 240 K. In conclusion, we have demonstrated a conceptual all-solidstate semiconductor optical cryocooler using a CdS nanobelt on the SOI substrate. A novel local thermometry has been developed using the ratio of the Stokes and anti-Stokes Raman intensities from the crystalline Si layer on the top of the substrate, which is applicable down to ∼140 K. A 30 and 20 K net laser cooling of SOI substrate has been achieved starting from 290 K under a 3.8 mW, 514 nm and a 4.4 mW, 532 nm laser pumping, respectively. The starting temperature dependence of the cooling temperature has been systematically investigated. On the basis of the temperature-dependent antiStokes photoluminescence, photoconductivity gain, and the thermal conductivity of Si, the relation between the cooling temperature and the pumping wavelength at various temperatures above 180 K is obtained, which agrees well with the experimental results and thermal analysis.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail address:
[email protected]. Author Contributions
D.L. and J.Z. contributed to this work equally. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This work is mainly supported by Singapore National Research Foundation through a NRF fellowship grant (NRF-RF2009-06) and Ministry of Education via two AcRF Tier 2 grants (MOE2011-T2-2-051 and MOE2013-T2-1-049). Q.X. would like to also acknowledge the strong support by start-up grant (M58113004) and New Initiative Fund (M58110100) from Nanyang Technological University (NTU). This work was also supported in part by AFOSR through its Asian Office of Aerospace Research and Development (FA2386-13-1-4112). B.H. thanks the support from the Hong Kong General Research Fund (Grant No. 613211) and National Natural Science Foundation of China (Grant No. 51376154). D.L. acknowledges the World Future Foundation (WFF) for awarding him the WFF PhD Prize in Environmental and Sustainability Research (2014) and the financial support to this work.
■
REFERENCES
(1) Pringsheim, P. Z. Phys. A: Hadrons Nuclei 1929, 57, 739−746. (2) Sheik-Bahae, M.; Epstein, R. I. Nature photon. 2007, 1, 693−699. (3) Chu, S.; Cohen-Tannoudji, C.; Philips, W. D. Nobel Prize in Physics, 1997; http://nobelprize.org. (4) Epstein, R. I.; Buchwald, M. I.; Edwards, B. C.; Gosnell, T. R.; Mungan, C. E. Nature 1995, 377, 500−503. (5) Khurgin, J. B. Phys. Rev. Lett. 2007, 98, 177401. (6) Sheik-Bahae, M.; Epstein, R. I. Phys. Rev. Lett. 2004, 92, 247403. (7) Seletskiy, D. V.; Melgaard, S. D.; Sheik-Bahae, M.; Bigotta, S.; DiLieto, A.; Tonelli, M.; Epstein, R. I. Optical refrigeration breaks the Peltier barrier: cooling Yb:YLF to 155 K. In Laser Refrigeration of Solids III; Epstein, R. I., SheikBahae, M., Eds.; Spie-Int Soc Optical Engineering: Bellingham, 2010; Vol. 7614. (8) Seletskiy, D. V.; Melgaard, S. D.; Bigotta, S.; Di Lieto, A.; Tonelli, M.; Sheik-Bahae, M. Nature photon. 2010, 4, 161−164. (9) Seletskiy, D. V.; Melgaard, S. D.; Di Lieto, A.; Tonelli, M.; SheikBahae, M. Opt. Express 2010, 18, 18061−18066. 4728
dx.doi.org/10.1021/nl501831f | Nano Lett. 2014, 14, 4724−4728
