Laser refrigeration of hydrothermal nanocrystals inphysiological mediaPaden B. Rodera,1, Bennett E. Smithb,1, Xuezhe Zhoua,1, Matthew J. Cranec, and Peter J. Pauzauskiea,d,2

aDepartment of Materials Science & Engineering, University of Washington, Seattle, WA 98195; bDepartment of Chemistry, University of Washington,Seattle, WA 98195; cDepartment of Chemical Engineering, University of Washington, Seattle, WA 98195; and dFundamental & Computational SciencesDirectorate, Pacific Northwest National Laboratory, Richland, WA 99354

Edited by Qihua Xiong, Nanyang Technological University, Singapore, Singapore, and accepted by the Editorial Board October 13, 2015 (received for reviewJune 9, 2015)

Coherent laser radiation has enabled many scientific and techno-logical breakthroughs including Bose–Einstein condensates, ultra-fast spectroscopy, superresolution optical microscopy, photothermaltherapy, and long-distance telecommunications. However, it hasremained a challenge to refrigerate liquid media (including phys-iological buffers) during laser illumination due to significant back-ground solvent absorption and the rapid (∼ps) nonradiativevibrational relaxation of molecular electronic excited states. Herewe demonstrate that single-beam laser trapping can be used toinduce and quantify the local refrigeration of physiological mediaby >10 °C following the emission of photoluminescence fromupconverting yttrium lithium fluoride (YLF) nanocrystals. A simple,low-cost hydrothermal approach is used to synthesize polycrystallineparticles with sizes ranging from<200 nm to>1 μm. A tunable, near-infrared continuous-wave laser is used to optically trap individual YLFcrystals with an irradiance on the order of 1 MW/cm2. Heat is trans-ported out of the crystal lattice (across the solid–liquid interface) byanti-Stokes (blue-shifted) photons following upconversion of Yb3+

electronic excited states mediated by the absorption of opticalphonons. Temperatures are quantified through analysis of thecold Brownian dynamics of individual nanocrystals in an inhomo-geneous temperature field via forward light scattering in the backfocal plane. The cold Brownian motion (CBM) analysis of individualYLF crystals indicates local cooling by >21 °C below ambient con-ditions in D2O, suggesting a range of potential future applicationsincluding single-molecule biophysics and integrated photonic,electronic, and microfluidic devices.

laser refrigeration | nanocrystal | hydrothermal | physiological | anti-Stokes

Advances in cryogenic sciences have enabled several obser-vations of new low-temperature physical phenomena including

superconductivity (1), superfluidity (2), and Bose–Einstein con-densates (3). Heat transfer is critical in numerous fields of scienceand technology including thermal management within integratedmicroelectronics (4–6), photonic (7, 8) and microfluidic (9) cir-cuits, and the regulation of metabolic processes (9–11). In 1929,Pringsheim (12) proposed that solid-state materials could experi-ence refrigeration if they exhibited biased emission of anti-Stokes(blue-shifted) radiation relative to a fixed optical excitation wave-length. Epstein et al. (13) experimentally demonstrated this conceptfirst in 1995 using rare-earth–doped fluoride glass materials. Opticalrefrigeration of a condensed phase with a rhodamine dye by anti-Stokes radiation has previously been reported (14–16) but theresults remain controversial (17, 18). More recently, it has beenshown that rare-earth–doped yttrium lithium fluoride (Yb3+:YLF) crystals grown in high-temperature Czochralski reactors(19) can be cooled to cryogenic temperatures (∼90 K) (20) invacuo using continuous-wave near-infrared (NIR) laser excita-tion. Furthermore, the laser refrigeration of doped yttrium alu-minum garnet (Yb3+:YAG) materials has recently been reported inair at atmospheric pressure (21). Anti-Stokes photoluminescencehas also been reported (22) to cool cadmium sulfide (CdS) nano-ribbons in vacuo by as much as 40 °C below room temperature. In

contrast with anti-Stokes processes, optomechanical laser refrig-eration has also been demonstrated based on a novel mechanismof angular momentum transfer between a circularly polarized laserand a birefringent crystal (23).To date, laser refrigeration of nanocrystals in aqueous media

has not been reported stemming primarily from the large NIRoptical absorption coefficient of water (αH2Oj975 nm ∼ 0.5/cm) (24).It has remained an open question whether these known coolingmaterials could act to refrigerate aqueous media and undergohypothesized cold Brownian motion (CBM) (25, 26), or whethersolvent heating from the background absorption coefficient ofwater would overwhelm the cooling of individual YLF crystals.Furthermore, it is not obvious a priori that YLF crystals madethrough hydrothermal processing would have sufficiently lowbackground impurity levels to achieve laser cooling (27, 28). Inthis work, we demonstrate the local laser refrigeration of hydro-thermal YLF nanocrystals dispersed within several differentaqueous media including deionized water, heavy water (D2O),and physiological electrolytes. Refrigeration >10 °C below am-bient conditions is observed in PBS following anti-Stokes pho-toluminescence from optically trapped (29), rare-earth–doped YLFnanocrystals undergoing CBM.

Results and DiscussionPioneering efforts to cool Yb3+:YLF materials in vacuo haverelied on the growth of high-purity YLF single crystals using an

Significance

Although the laser refrigeration of bulk crystals has recentlyshown to cool below cryogenic temperatures (∼90 K) in vac-uum, to date the laser refrigeration of physiological media hasnot been reported. In this work, a low-cost hydrothermalsynthetic approach is used to prepare nanocrystals that arecapable of locally refrigerating physiological buffers (PBS,DMEM) upon near-infrared illumination. Optical tweezers areused in tandem with cold Brownian motion analysis to observethe refrigeration of individual (Yb3+)-doped nanocrystals >10 °Cbelow ambient conditions. The ability to optically generatelocal refrigeration fields around individual nanocrystals prom-ises to enable precise optical temperature control within in-tegrated electronic/photonic/microfluidic circuits, and alsothermal modulation of basic biomolecular processes, includingthe dynamics of motor proteins.

Author contributions: P.J.P. designed research; P.B.R., B.E.S., X.Z., and M.J.C. performedresearch; P.B.R., X.Z., and P.J.P. contributed new reagents/analytic tools; P.B.R., B.E.S.,X.Z., M.J.C., and P.J.P. analyzed data; and P.B.R., B.E.S., X.Z., and P.J.P. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission. Q.X. is a guest editor invited by the EditorialBoard.1P.B.R., B.E.S., and X.Z. contributed equally to this work.2To whom correspondence should be addressed. Email: peterpz@uw.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1510418112/-/DCSupplemental.

15024–15029 | PNAS | December 8, 2015 | vol. 112 | no. 49 www.pnas.org/cgi/doi/10.1073/pnas.1510418112

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 2,

202

0

air- and moisture-free Czochralski process (30). In the experi-ments reported here, a low-cost modified hydrothermal synthesis(31) of Yb3+:YLF is used to prepare crystals shown in Fig. 1.Scanning electron microscopy (SEM) reveals that YLF crystalsexhibit a truncated tetragonal bipyramidal (TTB) morphology (Fig.1B). X-ray diffraction shows that the YLF crystal has a Scheelitestructure (Fig. 1C). Bright-field/high-angle annular-dark-field(HAADF) transmission electron microscope (TEM) imaging (Fig.1 D and E) and electron diffraction suggest that the TTB materialsare polycrystalline and likely form through an oriented attachment(32) process of nanocrystalline grains (Fig. 1E, Inset).A home-built laser trapping instrument (shown in Fig. 2) was

used to observe the Brownian dynamics of individual Yb3+:YLFnanocrystals. The laser trap setup is outlined in Materials andMethods, and the CBM temperature analysis is described in SIAppendix, section A. Briefly, the single-beam laser trap was usedto extract the surrounding local temperature profile of YLF parti-cles through observations of forward-scattered laser radiation pro-files that are processed to yield both the calibrated power spectraldensity and diffusion coefficient for individual YLF crystals (33).The laser refrigeration of 10% (mol%) Yb3+:YLF (i.e., 10% Yb3+

ions, 90% Y3+ ions) nanocrystals by more than 10 °C in PBS andDulbecco’s modified Eagle medium (DMEM) was observed at atrapping wavelength of λ = 1,020 nm (Table 1). To minimize fluidheating at control NIR trapping wavelengths (λ = 975 nm and1,064 nm), experiments discussed below were performed in D2O,unless explicitly stated otherwise, due to its low absorption com-pared with H2O (34).A bright-field micrograph for a characteristic optically trapped

Yb3+:YLF crystal is shown in Fig. 3A. The dependence of laserrefrigeration on the trapping laser’s pump wavelength is shownin Fig. 3C, where YLF crystals doped with 10% Yb3+ are ob-served to cool from 19 °C at a 5.9-MW/cm2 trapping irradiance to4 °C at a 25.5-MW/cm2 trapping irradiance when trapped withλ = 1,020 nm, which is resonant with ytterbium’s E4–E5 transi-tion shown in Fig. 3B. The same Yb3+:YLF crystals are shown toheat from 40 °C to 47 °C when trapped at the same respectiveirradiances with λ = 1,064 nm, which is energetically insufficient

to pump the E4–E5 resonance and subsequently cannot initiateupconversion-mediated cooling (Fig. 3B).The CBM analysis discussed above is limited to reporting local

solvent temperatures, but it does not provide information on theinternal lattice temperature of optically trapped YLF nanocrystals.It is well known that codoping YLF crystals with both Yb3+ andEr3+ ions leads to a thermalized Boltzmann distribution betweenthe E2 (2H11/2) and E1 (

4S3/2) manifolds of Er3+ and an intensegreen upconversion emission that is visible to the unaided eye(shown in Fig. 4A). This upconversion process is enabled by thelong (ms) photoluminescence lifetimes from the rare-earth pointdefects (35). It has also been shown that this upconvertedphotoluminescence from Er3+ may also be used to infer tem-perature changes through ratiometric thermometry by analysisof the photoluminescence emission from different Boltzmannthermal populations (36) given by the equation

I2I1∝ exp

�−ðE2 −E1Þ

kbT

�. [1]

In brief, changes in the ratio of the integrated emission bands I2and I1 that stem from transitions between energy states E2 andE1, respectively, and a common ground state are directly corre-lated to a change in the particle’s temperature. Furthermore, ithas been recently reported that strong visible upconversion inrare-earth codoped nanocrystals can be used for efficient biologicalimaging and labeling (37).Photoluminescence spectroscopy of optically trapped YLF

nanocrystals provides a unique capability of observing particle-to-particle variability within an ensemble (38). For the codoped2%Er3+, 10%Yb3+:YLF particles reported, substantial fluctu-ations in upconversion photoluminescence were observed, indi-cating that ensemble calibrations are inapplicable to quantitativeratiometric temperature measurements of individual nanocrys-tals (SI Appendix, section B). However, ratiometric thermometrycan still be used during laser trapping experiments to makequalitative observations of temperature changes as the trappingirradiance is increased, as shown in Fig. 4 B and C. The

Fig. 1. Synthesis and characterization of YLF crystals. (A) Schematic of Scheelite crystal structure of YLF with I41/a space group symmetry. (B) SEM image ofa faceted (Yb3+)0.1(Y

3+)0.9 LiF4 particle exhibiting TTB morphology. (Scale bar, 1 μm.) (C ) Powder XRD pattern of YLF crystals following hydrothermalsynthesis indicating a pure Scheelite crystal phase. (Inset) Schematic of TTB morphology relative to YLF’s unit cell. (D) Bright-field TEM image of an in-dividual Yb3+:YLF grain. (Scale bar, 200 nm.) (Inset) High-resolution TEM image taken from the indicated region. (Scale bar, 2 nm.) (E ) HAADF image ofthe YLF grain in B showing regions of high contrast suggesting the presence of polycrystalline domains. (Inset) SAED from the indicated region. (F) X-rayfluorescence compositional-analysis spectrum of an individual YLF crystal taken within the TEM confirming the elemental crystalline composition in-cluding Y, Yb, and F species.

Roder et al. PNAS | December 8, 2015 | vol. 112 | no. 49 | 15025

APP

LIED

PHYS

ICAL

SCIENCE

S

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 2,

202

0

decrease (increase) in the logarithmic ratio of I2 to I1 (Fig. 4 Band C) with increasing irradiance reflects a decrease (increase)in the internal lattice temperature (39), which agrees well withthe observed temperature changes measured via laser trapping lightscattering temperature analysis (Fig. 4D). Specifically, laser trappinganalysis of the particles’ CBM indicates that codoped 2%Er3+,10%Yb3+:YLF undergoes laser refrigeration (ΔT = −4.9 ± 2.8 °C)when trapped at λ = 1,020 nm and heating (ΔT = 21.8 ± 10.11 °C)when trapped at λ = 975 nm. Furthermore, it has been proposedrecently that codoping YLF crystals with other upconverting rare-earth ions can enhance cooling through energy transfer enhancedcooling (40).These results illustrate the potential of using singly- and

codoped YLF nanocrystals as a platform for precision circuitcooling, physiological refrigeration, biological imaging, and in situratiometric thermometry. Potential applications for these mate-rials include precision temperature control in integrated electronic(4–6), photonic (7, 8), and microfluidic (9) circuits, as well astriggering and probing fundamental metabolic processes (9–11).In particular, the ability to measure and to modulate temper-ature could enable the investigation of the kinetics and tem-perature sensitivity of cellular processes, including ion channelactuation (41), conformational folding dynamics of RNA (42),and dynamic stepping motion of molecular myosin (V) motorproteins (43).Analyzing the CBM of a nanocrystal dispersed in a liquid

phase to measure the nanocrystal’s temperature also providesthe unique capability to predict the local temperature gradient inthe medium surrounding the trapped nanocrystal. Because theaspect ratio of the truncated tetragonal bipyramid morphologyencountered for YLF crystals is near unity, we approximate theradius R of the particles using an equivalent-sphere model andcan extract the local temperature field a distance r from theparticles’ surface [at temperature Tp, excluding the temperature

discontinuity at the particle’s surface from the Kapitza resistance(44)], which is given by (45)

TðrÞ=T0 +Rr

�Tp −T0

�, [2]

where T0 is the bath temperature of the medium. Given that theaverage radius of the Yb3+:YLF particles trapped at λ = 1,020 nmin Fig. 3C is Ravg = 764 ± 293 nm, T0 = 25 °C, and Tp,avg = 3.4 °C at25.5-MW/cm2 irradiance, the distance away from the particlewhere the temperature increases to within 1% of T0 is 6.9 μm(Fig. 2). However, this treatment assumes that the local temper-ature profile around the cold particle behaves according to Eq. 2.Furthermore, it is also conceivable that the region around thecold particle is surrounded by a hot corona that slowly diminishesto the base temperature of the solvent.In the future it can be envisioned that the refrigeration of

particle ensembles and local mapping of the surrounding solventtemperature profile can be achieved through the generation ofmultiple laser traps, via either holographic phase masks (46) orgalvo-steering mirrors (47), to bring a temperature-sensing par-ticle into close proximity to a cooling YLF particle. Futuresynthetic efforts with YLF host crystals will be directed at con-trolling the grain size and morphology in pursuit of morphology-dependent cavity resonances (48) that can increase the optical

Fig. 2. Schematic of laser trapping instrument. An optically trapped YLF crystal in an aqueous fluid chamber. A piezostage driven at 32 Hz produces a peak inthe quadrant photodiode (QPD) power spectrum which is used to extract a calibrated diffusion constant. The particle’s temperature (Tp) and local tem-perature profile is then extracted using CBM analysis.

Table 1. Local cooling of Yb3+-doped YLF crystals in variousmedia at an irradiance of 25.5 MW/cm2, where T0 = 25 °C

Solvent ΔT = (Tp − T0), °C ΔT SD, °C

D2O −15.0 4.1DI water −14.7 3.8PBS −14.9 4.3DMEM −11.2 6.3

15026 | www.pnas.org/cgi/doi/10.1073/pnas.1510418112 Roder et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 2,

202

0

absorption of Yb3+ and reduce the irradiance required to ob-serve the laser refrigeration of physiological media. Furthermore,the low-cost hydrothermal approach reported here could be usedto synthesize novel phases for host crystals (such as β-NaYF4) thatcannot be grown through single-crystal Czochralski methods.

Materials and MethodsYLF Synthesis. The following synthesis was performed following modificationsto Lu et al. (31). Yttrium oxide (Y2O3), ytterbium oxide (Yb2O3), and erbiumoxide (Er2O3) are of 99.99% purity and used as purchased from Sigma-Aldrich.Yttrium nitrate (Y(NO3)3), ytterbium nitrate (Yb(NO3)3), and erbium ni-trate (Er(NO3)3) are obtained by dissolving the oxide in concentrated nitricacid at 60 °C while stirring for several hours until excess nitric acid is removed.The residual solid is then dissolved in Millipore deionized (DI) water to achievea stock concentration of the respective nitrate. Lithium fluoride (LiF), nitric acid(HNO3), ammonium bifluoride (NH4HF2), and EDTA are analytical grade andused directly in the synthesis without any purification. The following prepa-ration uses the synthesis of 2%Er3+,10%Yb3+:LiYF4 as an example. For thissynthesis, 7.04 mL of 0.5M Y(NO3)3, 0.8 mL of 0.5M Yb(NO3)3, and 0.16 mL of0.5M Er(NO3)3 are mixed with 1.17g EDTA in 5 mL Millipore DI water at 80 °Cwhile stirring for 1h. This is solution A. Subsequently, 0.21g of LiF and 0.68g ofNH4HF2 are dissolved in 7 mLMillipore DI water at 70 °C while stirring for 1h toform solution B. Solutions A and B are mixed together while stirring for 20 minto form a homogeneous white suspension which is then transferred to a 23-mLTeflon-lined autoclave and heated to 220 °C for 72 h. After the autoclave coolsto room temperature, the 2%Er3+,10%Yb3+:LiYF4 particles can be recovered bycentrifuging and washing with ethanol and Millipore DI water three times.The final white powder is obtained by calcining at 300 °C for 2 h. Using thesame method, 10%Yb3+:LiYF4 particles are achieved.

TEM Characterization. Bright-field and scanning TEM HAADF images weretaken on an FEI Tecnai G2 F20 at an accelerating voltage of 200 keV. Selectarea electron diffraction (SAED) images were taken with a camera length of490 mm. Energy-dispersive X-ray spectroscopy spectra were obtained with a 60-sacquisition time. The spectra were then processed by subtracting the back-ground and smoothing the peaks.

SEM Characterization. Secondary electron images were taken on an FEI Sirionat an accelerating voltage of 5 keV.

XRD Characterization. Powder X-ray diffraction (XRD) patterns are obtainedon a Bruker F8 Focus Powder XRD with Cu K (40-kV, 40-mA) irradiation (λ =0.154 nm). The 2θ angle of the XRD spectra is from 10° to 70° and the scanningrate is 0.01° s−1. The one minor unlabeled peak in the XRD spectra at 2θ = 44.9°is attributed to a small amount of unreacted LiF precursor [(200) peak].

Laser Trapping Description. The laser tweezer setup is a modified modularoptical tweezer kit (Thorlabs, OTKB), where the original condenser lens hasbeen replaced with a 10× Mitutoyo condenser (Plan Apo infinity-correctedlong WD objective, stock no. 46–144). The 100× objective focusing lens has anumerical aperture of 1.25 and a focal spot of 1.1 μm. The quadrant pho-todiode (QPD) and piezostage were interfaced to the computer through aDAQ card (PCIe-6361 X series, National Instruments) and controlled throughmodified MATLAB software (Thorlabs). Experimental chambers were preparedas follows. Several microliters of the nanocrystal/aqueous medium dispersionwere transferred by a pipette into a chamber consisting of a glass slide andglass coverslip. The edges of the glass slide and the glass coverslip were thensealed with a 150-μm-thick adhesive spacer (SecureSeal Imaging Spacer, GraceBio-laboratories). Nanocrystals were trapped at the center (∼75 μm from thesurface) of the temperature-controlled perfusion chamber (RC-31, Warner In-struments) and held at T0 = 25 °C while voltage traces were recorded at theQPD for 3 s at a sample rate of 100 kHz. The QPD voltage signal was calibratedby oscillating the piezostage at 32 Hz and an amplitude of 150 nm peak-to-peak during signal acquisition, as outlined in ref. 49. Trapping data were ac-quired using a diode-pumped solid-state Yb3+:YAG thin-disk tunable laser(VersaDisk 1030–10, Sahajanand Laser Technologies) at a wavelength of 1,020nm, a 975-nm pigtailed fiber Bragg grating stabilized single-mode laser diode(PL980P330J, Thorlabs), as well as a solid-state Nd3+:YAG 1,064-nm (BL-106C,Spectra-Physics) at an irradiance of 5.9, 10.7, 14.6, 21.2, and 25.5 MW/cm2. EachYLF cooling data point in Fig. 3C in the article represents an average of sixindividual particles with an average radius of 764 nm with an SD of 293 nm.Magnitudes of cold Brownian temperature changes presented in Table 1 weredetermined using methods outlined in ref. 33. Silica beads (SS04N/9857, BangsLaboratories) were used for their monodisperse size distribution (1,010-nmdiameter), and they have shown to minimally heat when trapped with a lasertweezer at NIR wavelengths (50). Electromagnetic simulations of the in-teraction of the trapping laser with an YLF TTB were also performed to predictthe stable trapping configurations of optically trapped YLF particles, detailed

Fig. 3. Laser refrigeration of optically trapped YLF microcrystals. (A) Opticalmicrograph of an optically trapped YLF crystal. (Scale bar, 3 μm.) (B) Crystal-field energy level configuration of Yb3+ dopant ions and used coolingscheme. (C) Extracted temperature (Tp) of optically trapped particles in D2Oas determined using the outlined CBM analysis. Yb3+-doped YLF particles areshown to cool when trapping wavelength is resonant with the E4–E5 tran-sition (λ = 1,020 nm) but heat when the trapping wavelength is below thetransition (λ = 1,064 nm).

Fig. 4. Upconversion and ratiometric thermometry of codoped YLF. (A) Bright-field optical micrograph showing a codoped 2%Er3+,10%Yb3+:YLF particle inBrownian motion (Top Left) and a dark-field optical micrograph of the crystalwhen trapped with λ = 1,020 nm (Bottom Left). (Scale bar, 4 μm.) Upconvertedphotoluminescence can be seen with the unaided eye (Right). (B) Photo-luminescence spectra of the corresponding dark-field image showing the in-tegration regions I2 and I1, representing emission from Er3+ energy states E2(2H11/2) and E1 (4S3/2) to the ground state Eground (4I15/2), respectively.(C) Natural logarithm of the ratio I2/I1 showing a linear increase (Top) withlaser irradiance at λ = 975 nm and a linear decrease (Bottom) with laser ir-radiance at λ = 1,020 nm. (D) Laser refrigeration of the codoped YLF crystalanalyzed in C measured via CBM analysis at an irradiance of 25.5 MW/cm2.

Roder et al. PNAS | December 8, 2015 | vol. 112 | no. 49 | 15027

APP

LIED

PHYS

ICAL

SCIENCE

S

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 2,

202

0

in SI Appendix, section E. Lastly, visible emission of Er3+ from Er/Yb codopedtrapped YLF host crystals was detected using an Acton SpectraPro 500i spec-trograph with a Princeton liquid-nitrogen-cooled Si detector.

CBM Temperature. Power spectra from the QPD voltage traces were processedaccording to Berg-Sorensen and Flyvbjerg (51) and used to calibrate the QPDtraces following the method of Toli�c-Norrelykke et al. (49). An experimentaldiffusion coefficient was then extracted by fitting the characteristic functionfor the experimental power spectra derived in Berg-Sorensen and Flyvbjerg(51). Given that the temperature of the trapped particle is significantlydifferent from the temperature sufficiently far from the laser focus, theparticle–trap system is not isothermal and behaves according to non-equilibrium dynamics. Thus, equating the experimental diffusion coefficientto nonisothermal Brownian dynamics necessitates the application of CBM, asderived by Chakraborty et al. (25). The CBM diffusion coefficient is thenrelated to the CBM temperature by

DCBM =kbTCBMγCBMðTÞ, [3]

where DCBM is the CBM diffusion coefficient, kb is Boltzmann’s constant, TCBMis the CBM temperature, and γCBMðTÞ is the CBM Stokes drag. To the leadingorder of the temperature increment or decrement ΔT = (Tp − T0), the tem-perature dependence of the viscosity on TCBM can be neglected, giving theeffective temperature (25)

TCBM = T0 +512

ΔT . [4]

To account for the solvent viscosity temperature dependence, we follow themethods of ref. 52 and use the Vogel–Fulcher–Tammann–Hesse (VFT) lawwith the viscosity functional form ηðTÞ= η∞exp½A=ðT − TVFT Þ�. The CBMStokes drag is given by

γCBMðTÞ= 6πRηCBMðTÞ, [5]

where R is the particle radius, and ηCBMðTÞ is the temperature-dependentCBM viscosity that is related to the viscosity of the solvent at room tem-perature, η0, by

η0ηCBMðTÞ≈ 1+

193486

�ln

�η0η∞

���ΔT

ðT0 − TVFT Þ�

−�56243

ln�η0η∞

�−

12,563118,098

ln2�η0η∞

���ΔT

ðT0 − TVFT Þ�2. [6]

Eqs. 4–6 are then used in Eq. 3 to obtain DCBM, which is subsequently com-pared with the experimental diffusion coefficient to determine the particle

temperature Tp [excluding the temperature discontinuity at the particle’ssurface from the Kapitza resistance (44)]. An alternative CBM temperature analysisusing a semiphenomenological expression for DCBM that approximately accountsfor higher-order terms in ΔT (equation 15 of the supporting online materials ofChakraborty et al. (25)] yields consistent results, indicating that these higher-ordercorrections are negligible, for our purposes. For the experiments reported here, theVFT viscosity parameters were fit to experimental data and are as follows:

D2O

η∞ = 3.456 ·10−5   Pa · s

A= 478.6  K

TVFT = 160  K,

and

DI water,   PBS,  DMEM

η∞ = 2.664 ·10−5   Pa · s

A= 536.5  K

TVFT = 145.5  K.

VFT viscosity parameters for DI water, PBS (0.01M, pH 7.4; Sigma P5368), andDMEM (1×, high glucose, pyruvate; Life Technologies cat. no. 11995–065)were assumed to be equivalent because it has been reported that waterviscosity can be used for purposes of modeling particle transport in non–serum-containing media (53).

ACKNOWLEDGMENTS. The authors thank Klaus Kroy of Leipzig Universityfor discussion of CBM analysis, John W. Cahn for discussion of YLF crystallog-raphy, and E. James Davis for manuscript comments and providing an opticalspectrometer with LN2-cooled detector. This research was made possible by agrant from the Air Force Office of Scientific Research Young Investigator Pro-gram (Contract FA95501210400), start-up funding from the University ofWashington, as well as a capital equipment donation from the Lawrence Liver-more National Laboratory. P.B.R. thanks the National Science Foundation for aGraduate Research Fellowship under Grant DGE-1256082. M.J.C. was sup-ported by the Department of Defense through the National Defense Scienceand Engineering Graduate Fellowship Program. P.J.P. gratefully acknowledgessupport from both the US Department of Energy’s Pacific Northwest NationalLaboratory (PNNL) and the Materials Synthesis and Simulation Across Scales(MS3) Initiative, a Laboratory Directed Research and Development (LDRD)program at the PNNL. The PNNL is operated by Battelle under Contract DE-AC05-76RL01830.

1. Onnes HK (1911) Disappearance of the electrical resistance of mercury at helium

temperatures. Proc K Ned Akad Wet 14:113–115.2. Kapitza PL (1941) The study of heat transfer in helium II. J Phys (USSR) 4(1-6):181–210.3. AndersonMH, Ensher JR, Matthews MR, Wieman CE, Cornell EA (1995) Observation of

Bose-Einstein condensation in a dilute atomic vapor. Science 269(5221):198–201.4. Feshchenko AV, Koski JV, Pekola JP (2014) Experimental realization of a Coulomb

blockade refrigerator. Phys Rev B 90(20):201407.5. Chowdhury I, et al. (2009) On-chip cooling by superlattice-based thin-film thermo-

electrics. Nat Nanotechnol 4(4):235–238.6. Flipse J, Bakker FL, Slachter A, Dejene FK, van Wees BJ (2012) Direct observation of

the spin-dependent Peltier effect. Nat Nanotechnol 7(3):166–168.7. Bogaerts W, Fiers M, Dumon P (2014) Design challenges in silicon photonics. IEEE J Sel

Top Quantum Electron 20(4):1–8.8. Vicario C, et al. (2013) Pump pulse width and temperature effects in lithium niobate

for efficient THz generation. Opt Lett 38(24):5373–5376.9. Lucchetta EM, Lee JH, Fu LA, Patel NH, Ismagilov RF (2005) Dynamics of Drosophila

embryonic patterning network perturbed in space and time using microfluidics.

Nature 434(7037):1134–1138.10. Lee JH, et al. (2013) Regulation of temperature-responsive flowering by MADS-box

transcription factor repressors. Science 342(6158):628–632.11. Kucsko G, et al. (2013) Nanometre-scale thermometry in a living cell. Nature 500(7460):

54–58.12. Pringsheim P (1929) Two observations about the difference between luminescence

and thermal radiation. Z Phys 57:739–746.13. Epstein RI, Buchwald MI, Edwards BC, Gosnell TR, Mungan CE (1995) Observation of

laser-induced fluorescent cooling of a solid. Nature 377(6549):500–503.14. Zander C, Drexhage KH (1995) Cooling of a dye solution by anti-Stokes fluorescence.

Adv Photochem 20:59–78.15. Clark JL, Rumbles G (1996) Laser cooling in the condensed phase by frequency up-

conversion. Phys Rev Lett 76(12):2037–2040.

16. Clark JL, Miller PF, Rumbles G (1998) Red edge photophysics of ethanolic rhodamine

101 and the observation of laser cooling in the condensed phase. J Phys Chem A

102(24):4428–4437.17. Mungan CE, Gosnell TR (1996) Comment on “Laser cooling in the condensed phase by

frequency upconversion.” Phys Rev Lett 77(13):2840.18. Rumbles G, Clark JL (1996) Rumbles and Clark reply. Phys Rev Lett 77(13):2841.19. Coluccelli N, et al. (2008) Diode-pumped passively mode-locked Yb:YLF laser. Opt

Express 16(5):2922–2927.20. Melgaard SD, Seletskiy DV, Di Lieto A, Tonelli M, Sheik-Bahae M (2013) Optical re-

frigeration to 119 K, below National Institute of Standards and Technology cryogenic

temperature. Opt Lett 38(9):1588–1590.21. Soares de Lima Filho E, Nemova G, Loranger S, Kashyap R (2013) Laser-induced cooling

of a Yb:YAG crystal in air at atmospheric pressure. Opt Express 21(21):24711–24720.22. Zhang J, Li D, Chen R, Xiong Q (2013) Laser cooling of a semiconductor by 40 kelvin.

Nature 493(7433):504–508.23. Arita Y, Mazilu M, Dholakia K (2013) Laser-induced rotation and cooling of a trapped

microgyroscope in vacuum. Nat Commun 4(2374):2374.24. Palik ED (1998) Handbook of Optical Constants of Solids (Academic, Boston), Vol 3.25. Chakraborty D, et al. (2011) Generalised Einstein relation for hot Brownian motion.

Eur Phys Lett 96(6):60009.26. Kroy K (2014) Levitating nanoparticles: Non-equilibrium nano-thermometry. Nat

Nanotechnol 9(6):415–417.27. Hehlen MP, Epstein RI, Inoue H (2007) Model of laser cooling in the Yb3+-doped

fluorozirconate glass ZBLAN. Phys Rev B 75(14):144302.28. Seletskiy DV, Hehlen MP, Epstein RI, Sheik-Bahae M (2012) Cryogenic optical re-

frigeration. Adv Opt Photonics 4(1):78–107.29. Ashkin A, Schütze K, Dziedzic JM, Euteneuer U, Schliwa M (1990) Force generation of

organelle transport measured in vivo by an infrared laser trap. Nature 348(6299):346–348.30. Seletskiy DV, et al. (2010) Laser cooling of solids to cryogenic temperatures. Nat

Photonics 4(3):161–164.

15028 | www.pnas.org/cgi/doi/10.1073/pnas.1510418112 Roder et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 2,

202

0

31. Lu C, HuangW, Ni Y, Xu Z (2011) Hydrothermal synthesis and luminescence properties

of octahedral LiYb(4):Er(3+) microcrystals. Mater Res Bull 46(2):216–221.32. Li D, et al. (2012) Direction-specific interactions control crystal growth by oriented

attachment. Science 336(6084):1014–1018.33. Roder PB, Smith BE, Davis EJ, Pauzauskie PJ (2014) Photothermal heating of nano-

wires. J Phys Chem C 118(3):1407–1416.34. Haro-González P, et al. (2013) Optical trapping of NaYF4:Er3+,Yb3+ upconverting

fluorescent nanoparticles. Nanoscale 5(24):12192–12199.35. Auzel F (2004) Upconversion and anti-Stokes processes with f and d ions in solids.

Chem Rev 104(1):139–173.36. McLaurin EJ, Bradshaw LR, Gamelin DR (2013) Dual-emitting nanoscale temperature

sensors. Chem Mater 25(8):1283–1292.37. Wang F, Banerjee D, Liu Y, Chen X, Liu X (2010) Upconversion nanoparticles in bi-

ological labeling, imaging, and therapy. Analyst (Lond) 135(8):1839–1854.38. Mor FM, Sienkiewicz A, Forró L, Jeney S (2014) Upconversion particle as a local lumines-

cent Brownian probe: A photonic force microscopy study. ACS Photonics 1(12):1251–1257.39. Vetrone F, et al. (2010) Temperature sensing using fluorescent nanothermometers.

ACS Nano 4(6):3254–3258.40. Dong GZ, Zhang XL, Li L (2013) Energy transfer enhanced laser cooling in Ho 3+ and

Tm 3+-codoped lithium yttrium fluoride. J Opt Soc Am B 30(4):939–944.41. Huang H, Delikanli S, Zeng H, Ferkey DM, Pralle A (2010) Remote control of ion

channels and neurons through magnetic-field heating of nanoparticles. Nat

Nanotechnol 5(8):602–606.

42. Hodak JH, Downey CD, Fiore JL, Pardi A, Nesbitt DJ (2005) Docking kinetics andequilibrium of a GAAA tetraloop-receptor motif probed by single-molecule FRET.Proc Natl Acad Sci USA 102(30):10505–10510.

43. Iwaki M, Iwane AH, Ikezaki K, Yanagida T (2015) Local heat activation of singlemyosins based on optical trapping of gold nanoparticles. Nano Lett 15(4):2456–2461.

44. Cahill DG, et al. (2003) Nanoscale thermal transport. J Appl Phys 93(2):793–818.45. Rings D, Schachoff R, Selmke M, Cichos F, Kroy K (2010) Hot Brownian motion. Phys

Rev Lett 105(9):090604.46. Grier DG (2003) A revolution in optical manipulation. Nature 424(6950):810–816.47. Neuman KC, Block SM (2004) Optical trapping. Rev Sci Instrum 75(9):2787–2809.48. Roder PB, Pauzauskie PJ, Davis EJ (2012) Nanowire heating by optical electromagnetic

irradiation. Langmuir 28(46):16177–16185.49. Toli�c-Norrelykke SF, et al. (2006) Calibration of optical tweezers with positional de-

tection in the back focal plane. Rev Sci Instrum 77(10):103101.50. Peterman EJ, Gittes F, Schmidt CF (2003) Laser-induced heating in optical traps.

Biophys J 84(2 Pt 1):1308–1316.51. Berg-Sorensen K, Flyvbjerg H (2004) Power spectrum analysis for optical tweezers. Rev

Sci Instrum 75(3):594–612.52. Rings D, Selmke M, Cichos F, Kroy K (2011) Theory of hot Brownian motion. Soft

Matter 7(7):3441.53. Hinderliter PM, et al. (2010) ISDD: A computational model of particle sedimentation,

diffusion and target cell dosimetry for in vitro toxicity studies. Part Fibre Toxicol7(1):36.

Roder et al. PNAS | December 8, 2015 | vol. 112 | no. 49 | 15029

APP

LIED

PHYS

ICAL

SCIENCE

S

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 2,

202

0