Pump–Probe Microscopy: Theory, Instrumentation, and Applications

2 Spectroscopy 32(4) April 2017 www.spectroscopyonline.com

Excited-state dynamics provides an intrinsic molecular contrast of samples examined. These dynamics can be monitored by pump–probe spectroscopy, which measures the change in transmission of a probe beam induced by a pump beam. With superior detection sensitivity, chemical specificity, and spatial–temporal resolution, pump–probe microscopy is an emerging tool for functional imaging of nonfluorescent chromophores and nanomaterials. This article reviews the basic principle, instrumentation strategy, data analysis methods, and applications of pump–probe microscopy. A brief outlook is provided.

Pu-Ting Dong and Ji-Xin Cheng

s a pioneer of femtochemistry, Nobel laureate Ahmed copy to measure fluorescence lifetime (13). In 2007, the Warren Hassan Zewail (1–3)recorded the snapshots of chemi- group reported pump–probe imaging with a high-frequency A cal reactions with sub-angstrom resolution through modulation scheme (14). Their work demonstrated the fea- an ultrafast femtosecond transient absorption (TA) technique. sibility of imaging melanin by using two-color two-photon In a transient absorption experiment, a laser pulse pumps a absorption (TPA) or excited state absorption (ESA) processes. molecule into an excited state. The excited state itself exhibits Since then, extensive research has been conducted by har- relaxation dynamics on the femtosecond or picosecond tim- nessing the merits of pump–probe microscopy. A majority of escale. A second laser pulse then probes the population in the the research focused on nonfluorescent chromophores such excited state at different temporal delays with respect to the as hemoglobin and cytochromes, which absorb light but do excitation. This analysis method reveals the dynamics of the not emit fluorescence efficiently (15). Fu and colleagues used excited state and is termed as pump–probe spectroscopy. two-color absorption to measure the degree of oxygenation Pump–probe microscopy, also known as transient absorp- based on the different decay constants of deoxyhemoglobin tion microscopy, is an emerging nonlinear optical imaging and oxyhemoglobin (16). Pump–probe microscopy can effi- technique that probes the excited state dynamics, which is re- ciently discern hemoglobin and melanin, the two major ab- lated to the third-order nonlinearity (3,4). Pump–probe mi- sorbers in a biological tissue. Based on their signatures from croscopy is an attractive spectroscopic imaging technique with the time-resolved curves, hemoglobin shows a purely positive the following advantages: First, it is nondestructive to cells and response because of excited state absorption, whereas melanin tissues and can be performed without tissue removal (5). Thus, (eumelanin and pheomelanin) demonstrate a negative (ground it can be used as a repeatable diagnostic tool. Second, it is a state bleaching) signal when the pump beam and probe beam label-free technique and doesn’t need an exogenous target (4). spatially and temporally overlap (5). In addition, pump–probe Third, as a nonlinear optical technique, pump–probe micros- microscopy enables the discrimination of melanomas by de- copy can image endogenous pigments with three dimensional termining the ratio between eumelanin and pheomelanin. (3-D) spatial resolution (6). Fourth, unlike linear absorption, Melanin play an important role in skin and hair pigmentation which suffers from scattering in a tissue sample, the pump– and melanomas (17). Without external staining, pump–probe probe technique only measures absorption at the focal plane, imaging yielded novel insight into the differentiation of eu- which offers optical sectioning capability (6). Fifth, compared melanin and pheomelanin among thin biopsy slices and has to scattering measurements, this absorption-based method been used to probe the metastatic potential of melanocytic has a weaker dependence on the particle and thus is highly cutaneous melanomas (16). Besides applications to pigments sensitive to nanoscale subjects (8–11). Sixth, pump–probe in biological tissue, pump–probe microscopy has also been microscopy with near-infrared laser pulses permits biological applied to distinguish various kinds of pigments in arts based applications with an enhanced penetration depth and a lower on their decay differences (18–21). level of tissue damage (12). Another significant application of pump–probe microscopy In 1990s, Dong and coworkers used pump–probe micros- is for characterization of single nanostructures including gold www.spectroscopyonline.com April 2017 Spectroscopy 32(4) 3 nanorods (22)and single-wall nanotubes (SWNTs) (23–26). Specifically, Jung and coworkers for the first time deployed the phase of the pump–probe signal as a contrast to distinguish semiconducting Second excited state carbon nanotubes from metallic ones (25). Tong and colleagues further used this contrast for imaging semiconducting and metallic nanotubes in living Pump Probe cells (26). By tuning the excitation wave- First excited state length, which is resonant with the lowest Vibrational state electronic transition in SWNTs, Huang and colleagues exploited the band-edge relaxation dynamics in isolated and bundled SWNTs (23). Through as- Excited state absorption Stimulated emssion Ground state depletion sembling SWNTs with CdS, Robel and Ground state colleagues demonstrated the charge- transfer interaction between photoex- Figure 1: Three major processes in a pump–probe experiment: (a) Excited state absorption, (b) cited CdS nanoparticles and SWNTs by stimulated emission, and (c) ground-state depletion. For ground-state depletion, the number of transient absorption (24). the molecules in the ground state is decreased upon photoexcitation, consequently increasing In this review, we summarize the con- the transmission of the probe pulse. For stimulated emission, photons in its excited state can trast mechanisms and instrumentation be stimulated down to the ground state by an incident light field, thus leading to an increase of strategies of pump–probe microscopy transmitted light intensity on the detector. In the case of excited-state absorption, the probe photons and highlight some of these significant are absorbed by the excited molecules, promoting them to the higher energy levels. applications. Because of space limita- tions, we could not cover the entire lit- erature and would recommend to the readers other excellent articles in this Pump pulse train Motorized delay line field (27–32). tau Pump–Probe Theory

In a typical pump–probe measurement, Probe Pump the pump-induced intensity change of the probe is measured by a lock-in Scanning mirrors AOM amplifier referenced to the modulated Modulation frequency Transferred modulation pump pulse. Then this change is nor- tau Loss Pump pulse train Objective malized by the probe beam intensity Delay stack Sample Δ Gain Condenser to generate Ipr/Ipr (33). To express this Lock-in ampli er Photodiode process at molecular level, we define the Ref Out In absorption coefficient for an electronic Filter transition between level “i” and level “j” as

Figure 2: Schematic illustration of pump probe microscopy. αij(ω) = σij(ω) (Ni– Nj) [1] where σij(ω) is the cross section from electronic state i to j, and Ni and Nj are beam: observed: When the probe pulse is res- the populations of the initial and final onant with i→j transitions (i ≠ 0), then ΔIpr states, respectively. Conventionally, α is =−∑αij(ω)ΔNj d [2] the probe pulse is absorbed by the mol- I i,j positive for absorption and negative for pr ecule, reducing the transmission of the gain (33). where d is the sample thickness. The probe pulse. This negative ΔIpr/Ipr signal The pump pulse acts on the sample expression is derived from the Lambert- change is therefore called excited state by changing the energy level popula- Beer relation within the small signal ap- absorption (ESA). When the probe pulse tion, N→N + ΔN. As a consequence, the proximation. The “j” term describes all is resonant with 0→j transmission, the population of excited states will increase possible excited states (33). probe transmission is enhanced upon at the expense of that of the ground state. Depending on the probe energy, three pump excitation. This positive ΔIpr/Ipr Such change is measured by the probe effects on the transmitted pulse can be phenomenon is called ground-state de- 4 Spectroscopy 32(4) April 2017 www.spectroscopyonline.com

pletion (GSD). When the lowest excited state is dipole-coupled to the ground (a) Pump-probe AFM image (b) 1.6 1.6 Pump state and the probe pulse is resonant )

-1 1.2 1.2 Probe with the transition, stimulated emission cm -1 0.8 0.8 (SE) occurs. An increased transmission M 5 is observed in a SE process. (10 0.4 0.4 These three major processes are illus- Fluorescence (a.u. ) 0.0 0.0 trated in Figure 1. A detailed description 500 600 700 800 Wavelength (nm) is provided in the following sections. STAM 100 30 Pump-probe STAM (x2.5) )

-6 20 Excited-State Absorption Mean molecule 50 No. in focus Excited-state absorption (ESA) is a 0 1 2 3 4 10 0.4 process where the probe photons are ∂∂P/P (10 0.2 285 nm attenuated by excited states as shown 0 280 nm 0.0

Signal intensity (a.u. ) 0 0.0 0.5 1.0 1.5 0.0 0.1 0.2 in Figure 1. Since the 1970s, picosec- Position (μm) 0 2 4 6 8 10 Concentration (μM) ond laser–based ESA measurements have been extensively used to measure Figure 3: Pump probe microscopy with subdiffraction spatial resolution and single-molecule detection ground and excited-state dynamics sensitivity. (a) Subdiffraction-limited imaging of graphite nanoplatelets. Image from conventional (34,35). Compared to two-photon ab- transient absorption microscopy (top left) and AFM image of graphite nanoplatelets (top right). Image sorption, which goes through a virtual from saturation transient absorption microscopy (bottom left) and intensity profiles along the lines intermediate state, excited-state absorp- indicated by the dashed lines in pump–probe image and STAM image (bottom right). Adapted with tion significantly enhances the detection permission from reference 47. (b) Ground-state depletion microscopy with detection sensitivity of sensitivity by bringing a resonance with single-molecule at room temperature. Ensemble absorption and emission spectra of Atto647N in pH a real intermediate electronic state. The = 7 aqueous solution (top).The wavelengths of pump and probe beams are indicated. Ground-state mechanism for this process (36) can be depletion signal as a function of concentration of aqueous Atto647N solution (bottom). The power is described using the following equation:

350 µW for each beam. The blue frame shows the points at lowest concentrations, indicating single- Δt N0σpu[σp′r−σpr]IpuIprexp (− τ ( molecule sensitivity is reachable. Figures adapted with permission from reference 40. ΔI =− dz [3] pr ∫ ℏνpv

where N0 is the molecular concentration at ground state; σ and σ′ are 0.04 1 pr pr

Hb 0.8 the linear absorption cross sections of Eumel 0.02 Eumel 0.6 75 the ground state and excited states for 75 0.4 50 50 Hb the probe beam, respectively; ν repre- 0 25 0.2 pu 25 τ Pheomel s 0 sents the pump frequency and is the

Intensity (a.u) –0.02 –0.2 Pheomel Ink lifetime of the excited state (assume this –0.4 is a single-exponential decay); and Δt is –0.04 –0.6 Ink –0.8 the time delay between pump beam and ω = π/2 THz –0.06 –1 0 0.5 1 1.5 2 2.5 3 3.5 4 –1 –0.5 0 0.5 1 probe beam. Ipu and Ipr denote the in- t (ps) g tensity of pump beam and probe beam, 1 1 (a) (b) respectively. In the presence of a pump 0.8 Eumel pulse, excited-state population would 0.6 Eumel

0.4 give birth to the transmission changes of the probe. Equation 3 demonstrates that 0.2 Hb Hb

s 0 Pheomel Pheomel only at Δt = 0 when the pump beam and –0.2 probe beam are spatially and temporally –0.4 overlaid can ΔIpr have the biggest value. Ink –0.6 Δ Δ Ink Normalized counts per pixel bin As t becomes longer, Ipr depicts as –0.8 ω = π/2 THz ω = 1.4π THz an exponential decay curve convoluted –1 0 –1 –0.5 0 0.5 1 –1 –0.5 0 0.5 1 g g with an instrumental response function that is a Gaussian function. Figure 4: Phasor analysis to interrogate pump probe signal. Experimental transient absorption spectra of hemoglobin (Hb), sepia eumelanin, syhthetic pheomelanin, and surgical ink (top left). Stimulated Emission Phasor difference of mixtures of eumelanin and pheomelanin (eumelanin fraction of 75%, 50% When interrogating the short-lived and 25%) along with their phasor locations on the s-g coordinate (top right). Cumulative histogram excited states in pump–probe experi- phasor plot of 17 ocular melanoma samples at frequency /2 THz (bottom left) and 1.4 THz (bottom ments, the photons in the excited states right). Adapted with permission from reference 54. are stimulated down to the ground www.spectroscopyonline.com April 2017 Spectroscopy 32(4) 5 state by a time-delayed probe pulse as shown in Figure 1. This process is called 2.0 Singlet stimulated emission (37). The absorption (a) (c) Triplet (e)

1.0 ) coefficient decreases with increasing -3 Delay = 0 ps ∆A (1 0 excitation irradiance. The decrease in 0.0 <100 ps absorption happens due to the annihi- 5 ns –1.0 550 600 650 700 750 850 lation of the number densities of both Wavelength (nm)

NW1 1.0 (f) the ground state and the state being ex- (b) (d) 2 μm NW2 1.0 10 2 μm 4 8 2 cited, this process can be portrayed as NW3 6

lS(√)l 4 5 3 Count 0.5 2 μm 2 0.5

∆I 0 equation 4: NW1 (i-Si) 2 0 50 100 10 15 20 25 ∆I/lx10 Frequency (GHz) Period (ps) NW2 (i-Si) 1 Signal (norm. ) Δt 0.0 N σ σ I I exp 0.0 0 0 pu pr pu pr (− τ ( 0 50 100 150 0 50 100 150 dz [4] NW3 (n-Si) ΔIpr =− Delay time (ps) Delay time (ps) ∫ ℏν 0 200 400 600 800 1000 1200 1400 pv Pump-probe delay (ps) From equation (4), we can tell at Δt = 0, strongest signal is achieved. As Δt be- Figure 5: Imaging nanomaterials by pump–probe microscopy. (a) TA imaging of graphene on glass comes longer, the transmission change coverslip. 0 stands for defects, 1 is single layer graphene, 2 is double layer, 3 is triple layer, respectively. of probe also demonstrates an exponen- Pump = 665 nm (1.10 mW) and probe = 820 nm (0.68 mW), respectively. Data adapted from reference tial decay curve convoluted with Gauss- 70. (b) Transient absorption image of DNA-SWNTs internalized by CHO cells after 24 h incubation. ian function. Based on the stimulated Gray, transmission of cells; green, S-SWNTs; red, M-SWNTs. Pump = 707 nm, probe = 885 nm. The emission, Min and colleagues achieved laser power post-objective was 1 mW for the pump beam and 1.6 mW for the probe beam. Adapted nanomolar detection sensitivity of non- with permission from reference 26. (c) Decay-associated spectra of the triplet (red) and singlet (black) fluorescent chromophores (37). The excitons of tetracene obtained by global analysis of the ensemble transient absorption spectra with the integrated intensity attenuation of the probe polarization to maximize triplet absorption. Data adapted from reference 75. (d) Pump–probe excitation beam can also be expressed as microscopy decay kinetics following photoexcitation of a localized region in three different Si nanowires; ΔI N σ NW1 (red) and NW2 (green) are intrinsic, and NW3 (blue) is n-type. Curves are fit to a tri-exponential pu =− 0 01 ∼10–7 I S [5] decay. Inset shows the SEM image of three wires. Adapted with permission from reference 76. (e) pu 3D transient absorption microscopic images of gold nanodiamonds in living cells taken from eight where S ~ 10-9 cm2 denotes the beam successive focal planes with 1-μm step. Scale bar: 20 µm. Data adapted from reference 77. (f) Transient -16 2 absorption trace from a single Ag nanocube from a sample with an average edge length of 35.5 ± waist, and σ01 ~ 10 cm represents the absorption cross section from ground 3.4 nm (left). The inset shows the Fourier transform of the modulated portion of the data. Ensemble state to the first electronic state. The transient absorption trace for the Ag nanocube sample (right). Inset gives a histogram of the measured stimulation beam will experience a periods from the single-particle experiments, red line is the distribution calculated from the size transmission gain after interaction with distribution of the sample. Adapted with permission from reference 78. the molecules: ΔI N σ in 2007 (39). Similar to stimulated emis- time. The order of modulation depth of pr =− 0 10 ∼10–7 I S [6] sion, it presents as an out-phase signal the transmitted probe beam by a single pr (Figure 1). The overall mechanism is molecule (Atto647N) is ~10-7, which N I I σ σ consistent with other transient absorp- means we can still demodulate the sig- ΔI ∝ 0−pu pr 10 1 pr S2 [7] tion mechanisms. If expressed in equa- nal from a lock-in amplifier. Based on tion form, the GSD process has the the mechanism above, ground-state From equation 7, we can conclude same expression as stimulated emission depletion microscopy could reach sin- that the stimulated emission process in equation 4. The only difference lies gle-molecule detection (40). The ground shows overall quadratic power depen- in the probe wavelength. For GSD, the state depletion method could also be ap- dence, allowing three-dimensional op- probe is chosen close to the maximal ab- plied to localize fluorescence emission tical sectioning. In addition, the linear sorption peak, whereas the probe beam from fluorophores bound to the surface dependence upon the concentration of in the case of stimulated emission is se- of a nanowire, thus making it possible to analyte allows for quantitative analysis. lected away from the absorption peak. map out the structure of a nanowire [41]. The detected sensitivity would be down Based on ground-state depletion, Zink and colleagues also showed how to 10-9 M if the incident irradiance of single-molecule detection at room tem- GSD microscopy can be applied to mea- pump beam and probe beam are in the perature has been achieved (40). Under sure tubulin modifications in epithelial range of megawatt cm-2 (37). the condition that both pump and probe cells (42). High sensitivity coupled with beams (continuous-wave lasers) were optical sectioning capability makes Ground-State Depletion chosen close to near saturation inten- ground-state depletion microscopy an Ground-state depletion (GSD) micros- sity levels (350 μW at the focus for each important emerging technique. copy is a form of super-resolution light beam), a shot noise limited sensitivity microscopy suggested almost a decade is achieved. The detected sensitivity for Instrumentation ago (38), and it was first demonstrated Atto647N is 15 nM with 1s integration A typical pump–probe imaging setup is 6 Spectroscopy 32(4) April 2017 www.spectroscopyonline.com

in the preceding sentence?] Therefore,

2.5 this scheme is sensitive to artifacts. The (a) 2.0 (c) l Oxyhemoglobin Oxyhemoglobin 2.0 Deoxyhemoglobin Deoxyhemoglobin 1.5 second type with high repetition rates 1.5 stratum comeum 1.0 1.0 allows for averaging more laser shots

0.5 0.5 per unit time. As a result, the detection +1 +1 Signal intensity (μV ) 0.0 0.0 -6 1 sensitivity of ~10 units of absorbance -2 0 2 4 6 4 -2 0 2 6 0 0 Delay (ps) Delay (ps) can be achieved (28). By employing high- (b) -1 2 2 -1 frequency (that is, megahertz) lock-in

720 nm pump Tissue principal 810 nm probe 0.8 components modulation, Hartland and coworkers 0 720 nm pump 810 nm probe 0.4 Tissue region 1 detected signals from isolated single- Signal (au ) -1 Tissue region 2 Signal (au ) Sepia eumelanin 0.0 Synthetic pheomelanin Component 1 walled carbon nanotubes with a sensi- -2 -0.4 Component 2 -7 Component 3 Δ × 50 μm tivity of I/I ~ 5 10 (44). Moreover, -1 0 2 4 6 8 10 -1 0 1 2 3 4 5 Interpulse delay (ps) Interpulse delay (ps) in this scheme, the modulation provided by either an AOM or an electro-optic Figure 6: Imaging microvascular and melanomas by pump–probe microscopy: (a) Pump–probe modulator (EOM) operates at a high microscopy is applied to differentiate oxyhemoglobin and deoxyhemoglobin. ESA signal from frequency in the range of 100 kHz to 10 oxyhemoglobin and deoxyhemoglobin with pump = 810 nm (10 mW) and probe = 740 nm (6.4 MHz, where the noise approaches the mW) (left). ESA signal from oxyhemoglobin and deoxyhemoglobin with pump = 740 nm (2.4 mW) shot noise limit. One possible drawback and probe = 810 nm (10 mW) (right). Adapted with permission from reference 36 (copyright 2008 of such setups is their high probability Society of Photo-Optical Instruction Engineers). (b) Ex vivo imaging of microvasculature network of detecting the accumulation of long- of a mouse ear based on endogenous hemoglobin contrast. Red, blood vessel network; green, lived species, such as triplet or charge- surrounding sebaceous glands. Pump = 830 nm (~20 mW, two-photon excitation of Soret band), separated states (27). probe = 600 nm (~3 mW, one-photon stimulated emission of Q-band of hemoglobin). Adapted with When it comes to the detection of permission from reference 37. (c) Pump-probe image of a compound nevus at 0-fs (left) and 300-fs pump–probe signal, a phase-sensitive (right) interpulse delay (top). Regions containing eumelanin have positive signal (red/orange). Pump– lock-in amplifier is usually indispens- probe time delay traces comparing tissue regions of interest 1 and 2 (white boxes in top) with pure ably used to demodulate the probe solution melanins (bottom). The first three principal components found in tissue pump–probe signals signal. Slipchenko and colleagues re- (loadings plot, right). The first two components account for more than 98% of the variance. Pump = ported a cost-effective tuned amplifier 720 nm, probe = 810 nm. Scale bar = 100 μm. Adapted with permission from reference 6. for frequency-selective amplification of the modulated signal. By choosing shown in Figure 2. An optical paramet- intensity is detected by a photodiode. a pump beam of 830 nm and a probe ric oscillator pumped by a high-intensity A phase-sensitive lock-in amplifier beam of 1050 nm, the tuned amplifier mode-locked laser generates synchro- then demodulates the detected signal. can be used for pump–probe imaging of nous pump and probe pulse trains. The Therefore, pump-induced transmission red blood cells. This lock-in free method Ti:sapphire oscillator is split to separate changes of the sample versus time delay improved the single-to-noise ratio by pump and probe pulse trains. Tempo- can be measured from the focus plane. one order of magnitude compared to ral delay between the pump and probe This change over time delay shows dif- conventional detection based on a lock- pulses is achieved by guiding the pump ferent decay signatures from different in amplifier (45). beam through a computer-controlled chemicals, thus offering the origin of Spatial resolution is designated as delay line. Pump beam intensity is mod- the chemical contrast. the distance between two points of the ulated with an acousto-optic modulator Generally speaking, lasers applied sample that can be resolved individually (AOM), and the intensity of both beams in pump–probe microscopy can be di- according to the Rayleigh criteria. The is adjusted through the combination of vided into two types: systems working lateral (r0) and axial (z0) resolutions are a half-wave plate and polarizer. Sub- with relatively high pulse energy (5–100 defined (46) as sequently, pump and probe beams are nJ) and repetition rate of 1–5 kHz, and r = 0.61 • λ and z = 2 • n • λ collinearly guided into the microscope. systems using a low pulse energy (0.5– 0 NA 0 (NA)2 [8] After the interaction between the pump 10 nJ) and >1 MHz repetition rate (27). beam and the sample, the modula- With appropriate detection schemes where λ is the wavelength, n is the re- tion is transferred to the unmodulated that involve multichannel detection on fractive index of the medium and NA is probe beam. Computer-controlled scan- a shot-to-shot basis, the first type can the numerical aperture. By using spa- ning galvo mirrors are used to scan the achieve the signal detection sensitiv- tially controlled saturation of electronic combined lasers in a raster scanning ity of ~10-5 units of absorbance over a absorption, diffraction limit in far-field manner to create microscopic images. broad wavelength range (27). Neverthe- imaging of nonfluorescent species could The transmitted light is collected by less, the presence of multiple excited be broken as shown in Figure 3a. Wang the oil condenser. Subsequently, the states under high excitation density and colleagues designed a doughnut- pump beam is spectrally filtered by an conditions leads to singlet-singlet an- shaped laser beam to saturate the elec- optical filter, and the transmitted probe nihilation (43). [AUTHORS: Sense OK tronic transition in the periphery of the www.spectroscopyonline.com April 2017 Spectroscopy 32(4) 7 focal volume, thus introducing modulation only at the focal center. By ras- 0 (a) 0.1 (b) 0 0.5 -0.05 ter scanning three collinearly aligned 0.05 Synthetic 1 −0.04 -0.1 0 ultramarine blue 0.2 0 beams, high-speed subdiffraction-lim- in acrylic -0.05 -0.15 −0.08 −1 0 0.2 ited imaging of graphite nano-platelets -0.1 -0.2 0 5 10 15 20 −0.12 0 was achieved (47). 0.4 −0.16 −0.2 -0.5 0 0.2 Lapis lazuli Alternatively, Miyazak and colleagues 0 10 20 30 0 10 20 30 0 -1 in casein 0.2 0.4 Probe transient absorption (arb. units) (48) demonstrated the use of annular -0.2 -1.5 0.2 0 5 10 15 20 0.5 beams in pump–probe microscopy to -0.4 Pump-probe delay (ps) 0 improve spatial resolution in the focal 0.1 0 0.2 −0.5 plane, since the point spread function −0.2 (PSF) in pump–probe microscopy is 0 0 0 10 20 30 0 10 20 30 23% (43%) smaller than the diffraction- 0.8 0.10 0.4 limited spot size of the pump (probe) 1 0.6 0 0 0.06 beam. The authors also used intensity 0.4 −1 −0.4 modulated continuous wave laser di- 0.2 0.02 odes in a balanced detection scheme to 0 0 achieve subdiffraction resolution with 0 10 20 30 0 10 20 30 shot-noise limited sensitivity (49,50). Figure 7: Imaging artistic pigments by pump probe microscopy: (a) Transient absorption images of Regarding the sensitivity, single-mol- synthetic ultramarine in acrylic (golden artist colors GMSA 400, top) and lapis lazuli in casein (Kremer ecule detection can be achieved through pigments 10530, bottom) and the corresponding pump-probe delay traces in the indicated region of pump–probe microscopy. Chong and interest (white rectangle) where the line indicates double-exponential fits. Scalar bar = 100 µm. S/N coworkers conducted ground-state = 100. Adapted with permission from reference 19 (copyright 2012 Optical Society of America). (b) depletion microscopy and achieved a Graphs showing pump probe dynamics in test samples with the pigments lapis lazuli, vermilion, caput detection limit of 15 nM with a 1-s inte- mortuum, quinacridone, phthaloblue, and indigo. Adapted with permission from reference 21. gration time, which corresponds to 0.3 molecules in the probe volume, indicating the detection of a single-molecule that is based on phase information and 2 2 I(t) exp σ t 1 erf σ −t* τ absorption signal as shown in Figure 3b modulation frequency can be used as = 2 − − [12] ( 2τ τ ( ( ( 2*σ*τ ( ( (40). In their work, the sample was il- is discussed below. [AUTHORS: Sense luminated by two tightly focused laser OK in the preceding sentence?] where erf (x) is the error function, a beams where the pump beam and the standard function in most mathematical probe beam have different wavelengths Multiexponential Fitting software packages. For single exponen- but both are within the molecular ab- Multiexponential fitting, as the name tial decay, the mathematical equation for sorption band of the analyzed sample. implies, fits the time-resolved curves the time-resolved decay curve is In this case, the pump beam only excites with an exponential decay model. This σ2 2 t τ σ2 t τ I(t) = I +A exp − * * * 1−erf − * a molecule so that it only stays in its method is easy to conduct and under- 0 * 2 [13] ( 2*τ ( ( ( 2*σ*τ ( ( ground state, and, hence, photons from stand. The time-resolved intensity is the probe beam can’t be absorbed. Fast regarded as the conjugation between where τ is the decay constant and I0 is on-off modulation of a strong, saturat- the instrumental response R(t) and the the signal from background. A similar ing pump beam leads to the modulation response from sample S(t): equation can be used for double expo- of transmitted probe beam at the same nential decay. After fitting with this modulation frequency. I(t)=∫R(t – t′)S(t′)dt′ [9] model, we obtain the real decay constant τ along with the laser pulse width σ. Data Analysis Methods Suppose the time resolution of the de- Through the deconvolution approach, Generally, two methods can be used tector is modeled by a Gaussian function we could resolve the time constant to analyze a decay curve. The easier with a full width half maximum as σ: purely from decay of chemicals without method is multiexponential fitting to the effect of laser response function. t2 get the decay constants. However, a R(t) = A1 exp − 2 [10] However, the drawback of this method ( 2 * σ ( drawback is that its accuracy is relatively is that it is sensitive to the initial input low. The other method is called phasor In this case, pump–probe decay is mod- parameters, and therefore its accuracy analysis, a method that needs neither eled by an exponential decay with decay is relatively low. any assumptions regarding the physical constant τ: model nor integration fitting to deter- Phasor Analysis mine the lifetimes of multiexponential S(t) = A exp − t [11] Phasor analysis is a method that trans- 2 ( τ ( signals (51–53). When dealing with a lates the time-resolved decay curve into long lifetime (~1 ns), another method Then the convolution integral is a single point at a given frequency in the 8 Spectroscopy 32(4) April 2017 www.spectroscopyonline.com

Table I: Applications of pump–probe microscopy parts of the signal after Fourier transform of the time-resolved curve: Authors Topic Application References I(t) cos (ωt) dt Single metal ∫ Muskens et al. Nanomaterial 65 g(ω)= [14] nanoparticle ∫I(t)dt Davydova et al. Nanomaterial PtOEP crystal 66 Hot carrier dynamics ∫I(t) sin (ωt) dt Xia et al. Nanomaterial 67 s(ω)= [15] in HfN and ZrN ∫I(t)dt Cui et al. Nanomaterial WSe 68 2 Any multicomponent signal can be Graphene of different Li et al. Nanomaterial 70 described as layers and defects Hot photon dynamics Gao et al. Nanomaterial 61 I (t)= f I (t) [16] in graphene tot ∑ i i i Lauret et al. Gao et al. where f is the fraction of each indepen- 73, 74, 60, i Koyama et al. Nanomaterial SWNTs 62, 63 dent species contributing to the total Ellingson et al. signal: Kang et al.

Phase of semiconductor- ∫Ii(t)dt Jung et al. g = f • • g [17] tot ∑ i i Nanomaterial SWNTs and metallic- 25, 26 i Tong et al. ∫Itot(t)dt SWNTs Chirality grown Gao et al. Nanomaterial 74 of SWNTs By plotting g(ω) against s(ω) at a given Singlet fission frequency, we can map the distribution Wan et al. Nanomaterial 75 of tetracene of different chromophores with distinct Carrier motion in Gabriel et al. Nanomaterial 76 lifetimes in the semicircle coordinate. silicon nanowires Here ω is a free parameter depending Nonfluorescent Chen et al. Nanomaterial 77 on the separation efficiency. Robles and nanodiamond colleagues demonstrated its capability to Hartland et al. Nanomaterial Silver nanocube 78 discriminate eumelanin, pheomelanin, Lo et al. Nanomaterial Single CdTe nanowire 79 and ink by phasor analysis as shown in Mehl et al. Nanomaterial Single ZnO rods 80 Figure 4 (54). Optoelectronic The phasor representation of lifetime Cabanillas et al. Nanomaterial 33 semiconductor images has become popular because it Wong et al. provides an intuitive graphical view of Polli et al. Polymer Polymer blends 83, 81, 84, the fluorescence lifetime content with- Guo et al. 85 Yan et al. out any prior knowledge. Meanwhile, it Guo et al. significantly improves the overall sig- Semiconducting Perovskite film 82, 69 Simpson et al. materials nal-to-noise ratio when used for global Fu et al. Deep-tissue imaging analysis. Besides that, the region of in- Hemoglobin 36, 37 Min et al. of blood vessels terest selected in the phasor plot can be Differentiation mapped back to its corresponding image Fu et al. Melanin between eumelanin 5, 15, 87 Piletic et al. to realize segmentation (56). and pheomelanin Samineni et al. Historical pigments Lapis lazuli 19 Frequency Domain Approach Quinacridone red and Villafana et al. Historical pigments 20 The frequency domain approach is more ultramarine blue suitable for long-lived excited state. In this method, the lifetime information phasor space. One of the most advanta- the fluorescence spectrum at each pixel is extracted through a phase-sensitive geous features of phasor analysis when (55,56). Fu and colleagues further ap- detection. A simple model tan ϕ = ω * applied to fluorescent-lifetime imaging plied this analysis method to hyperspec- τ is applied to calculate the lifetime on microscopy (FLIM) (52,53) is that it has tral stimulated Raman scattering data. the basis of phase change correspond- the capability to quantitatively resolve It allows the fast and reliable cellular ing to different modulation frequency. a mixture of fluorophores with differ- organelle segmentation of mammalian When a modulated pump beam I1(t) = ent lifetimes. Phasors from those mix- cells, without any a priori knowledge of I1(1+cos ωt) is incident on the sample, tures display linearly across the phasor their composition or basis spectra (57). the excited state population is given by plot (54). For the first time, Fereidouni The basic mechanism for this method Miyazaki and colleagues (58) in the fol- and colleagues proved spectral phasor is described through mapping the real lowing equation: analysis was powerful for the analysis of parts of the signal against the imaginary www.spectroscopyonline.com April 2017 Spectroscopy 32(4) 9

other techniques such as high-resolution exhibit transient absorption signals σ I P(t) = 1 1 {A(ω) cos [ωt+ϕ(ω)]+1} hν s [18] transmission electron microscopy, scan- with opposite phases. Figure 5b shows 1 ning electron microscopy, and scanning the transient absorption image of DNA- where tunneling microscopy proved to be cum- SWNTs internalized by CHO cells, bersome in sample preparation (70). In where gray represents the transmission A(ω) = 1 2 [19] their work, they achieved high speed (2 image. The different colors in the image 1+(ω * τ) μs/pixel) imaging of graphene on vari- result from different phases represent- ous substrates under ambient condition ing two different kinds of SWNTs: green ϕ(ω) = tan–1(ω * τ) [20] and even in living cells and animals. In- represents semiconducting SWNTs and terestingly, the intensity of the transient red represents metallic SWNTs. Gao and

Here, σ1 is the absorption cross sec- absorption images is found to linearly colleagues reported transient absorption tion, ν1is the frequency of the pump, h is increase with the number of layers of microscopy experiments on individual the Planck constant, s is the beam waist graphene. In Figure 5a, in the TA image semiconducting SWNTs with known area at the focal point, ϕ is the phase of graphene, we can clearly observe the chirality grown by chemical vapor de- calculated from the x and y channel sig- location of graphene defects and that of position (CVD) with diffraction-limited nals, and τ is the excited-state lifetime. different layers. It only takes a few sec- spatial resolution and subpicosecond In the case of a long excited-state life- onds to acquire a TA image of graphene. temporal resolution (74)]. time, equation 20 suggests an efficient In addition, with polyethylene glycol Pump–probe microscopy has also been method: tan ϕ = ω * τ. This equation used to functionalize graphene oxide, extensively applied to study nanopar- demonstrates the linear relationship be- these well-dispersed particles have ticles and nanowires. Figure 5c presents tween tan ϕ and modulation frequency shown the capability for in vitro and ex visualization of singlet fission by observ- ω and the corresponding phase images. vivo imaging in Chinese hamster ovary ing the decay-associated spectra of the The slope of this equation yields the (CHO) cells (70). triplet (red) and singlet (green) excitons lifetime of the excited state. It is worth Pump–probe microscopy is also ex- of tetracene. The curves in Figure 5c noting that because of the relatively ploited to study SWNTs. Carbon nano- were obtained by global analysis of the larger shot noise at lower modulation tubes, especially single-wall carbon ensemble transient absorption spectra frequency, the standard deviation is very nanotubes, have attracted much at- (75). As shown in Figure 5d, pump– high (59). tention in the last two decades (71,72). probe microscopy has been used to The excellent properties of SWNTs demonstrate the spatial kinetics of sili- Applications of in thermal conductivity, electronics, con nanowires (76). In addition, nano- Pump–Probe Microscopy optics, and mechanics make them ap- diamonds and nanocubes are of great With its superior detection sensitivity, pealing. Pump–probe microscopy has interest to researchers. Figure 5e dem- chemical specificity and spatial-tempo- proved to be a powerful tool to explore onstrates 3D transient absorption mi- ral resolution, pump–probe microscopy the intrinsic photochemical properties croscopic images of gold nanodimonds has been used to study pigmentation of single-wall carbon nanotubes. Ac- in living cells (77). Figure 5f shows a fast (14), microvasculature (14), ultrafast curate detection of carrier dynamics decay of silver nanocubes resulting from relaxation in SWNTs (23–26,60–63), in these nanostructures is essential for electron-phonon coupling and subse- single semiconductor and metal nano- understanding and developing their op- quent modulations from the coherently structures (64,65), and other nanomate- toelectronic properties. Lauret and col- excited breathing mode (78). rials (66–69). Table I summarizes repre- leagues reported for the first time the Pump–probe microscopy examines sentative applications in various areas. time-resolved study of carrier dynam- intrinsic excited state dynamics of These applications are reviewed in more ics in single-wall carbon nanotubes by semiconductors. Lo and colleagues detail in the following sections. means of two-color pump–probe ex- demonstrated transient absorption periments under resonant excitation measurements on single CdTe nanow- Semiconducting with a selective injection of energy in ires, and they showed for the first time Nanomaterials and Graphene the semiconducting nanotubes (73). that acoustic phonon modes were fast Pump–probe microscopy provides a Jung and colleagues for the first time because of the efficient charge carrier vivid image of graphene with high sen- exploited the phase of the pump–probe trapping at a lower excitation intensity sitivity. Muskens and colleagues have signal as a contrast to study SWNTs (25). (79). Mehl and coworkers (80) reported demonstrated the study of a single Later Tong and coworkers showed that pump–probe microscopy of the indi- metal nanoparticle by combining a transient absorption microscopy offers a vidual behaviors of single ZnO rods at high-sensitivity femtosecond pump– label-free approach to image both semi- different spatial locations. Dramatically probe setup with a spatial modulation conducting and metallic SWNTs in vitro different recombination dynamics were microscope (65). Besides metal nanopar- and in vivo in real time with submi- observed in the narrow tips compared ticles, Zhang and colleagues also imaged crometer resolution, by choosing appro- with dynamics in the interior. Cabanil- graphene with single-layer sensitivity priate near-infrared wavelengths (26). las-Gonzalez and colleagues (33) high- through transient absorption, whereas Semiconducting and metallic SWNTs lighted the contribution of pump–probe 10 Spectroscopy 32(4) April 2017 www.spectroscopyonline.com spectroscopy to the understanding of the 650 nm, and successfully harvested two- microscopy, since the variety of pigments elementary processes taking place in or- color TPA images of microvasculature in the artist’s palate is enormous com- ganic based optoelectronic devices. They at different depths with a penetration pared with the biological pigments pres- further illustrated three fundamental depth of ~70 µm. In their following-up ent in skin (20). This is a great approach processes (optical gain, charge photo- study, they chose a longer probe beam of to extract microscopic information for a generation and charge transport). This 810 nm to differentiate oxyhemoglobin broad range of cultural heritage applica- work opens new perspectives for assess- and deoxyhemoglobin as shown in Fig- tions. Samineni and colleagues (19) were ing the role of short-lived excited states ure 6a. Beyond two-photon absorption, the first to conduct a pump–probe study on organic device operation. Polli and other procedures can also be applied to of lapis lazuli, a semi-precious rock, by coworkers developed a new instrument observe microvessels. Min and colleagues analyzing the multiexponential decay approach by combining broadband fem- conducted stimulated emission imaging behavior as shown in Figures 7a and 7b. tosecond pump–probe spectroscopy and of microvasculature network in a mouse The ratio of amplitude for short decay confocal microscopy, enabling simulta- ear based on the endogenous hemoglobin constant to that of long decay constant for neously high temporal and spatial reso- contrast by choosing the pump beam as the synthetic ultramarine pigment is 6.6 lution (81). Guo and colleagues (82) re- 830 nm (two-photon excitation of Soret ± 0.35, while that for natural lapis lazuli ported spatially and temporally resolved band) and probe beam as 600 nm (one- is 2.5 ± 0.05. Thus, they readily could be measurements of perovskite by ultrafast photon stimulated emission of Q-band) distinguished. microscopy. This work underscores the (see Figure 6b) (37). importance of the local morphology Pump–probe microscopy could also Outlook and establishes an important first step be used to differentiate different mela- Looking into the future, we predict the toward discerning the underlying trans- nins. Melanins generally come in two following advancements of this emerging port properties of perovskite materials. polymeric forms: eumelanin (black) technology. First, compact and low-cost Pump–probe microscopy also provides and pheomelanin (red/brown). Their pump–probe microscopy will be devel- new insight into the properties of poly- biosynthetic pathways involve the oxi- oped and made commercially available mer blends by directly accessing the dation of tyrosine leading to the forma- for broad use of this technique by non- dynamics at the interfacing between dif- tion of indoles and benzothiazines (87). experts. Second, handheld pump–probe ferent materials (83).Guo and colleagues Pheomelanin is reddish yellow, and it imaging system will be developed to as- (84) elucidated the exciton structure, exhibits phototoxic and pro-oxidant be- sist precision surgery in the clinic. Third, the dynamics, and the charge genera- havior (88). Eumelanin is a brown–black important applications of pump–probe tion in the solution phase aggregate of pigment that is substantially increased in microscopy will be identified, in which a low-bandgap donor-acceptor polymer melanoma. Therefore imaging the mi- the decay kinetics are used to study cel- by transient absorption. The technique croscopic distribution of eumelanin and lular development and disease stage. enables important applications in con- pheomelanin could be used to separate These advances will make pump–probe trolling morphology. Using ultrafast mi- melanomas from benign nevi in a highly microscopy an important member of the croscopy, Yan and colleagues proved that sensitive manner (16). The differences of nonlinear optical microscopy family with adding an amorphous content to highly the signals of these two different mela- broad use in biology, medicine, and ma- crystalline polymer nanowire solar cells nins are shown in Figure 6c. Eumelanin terials science. could increase the performance (85). has an abrupt positive absorption corresponding to excited-state absorption References Heme-Containing or two-photon absorption, the same as (1) A.H. Zewail, Science 242(4886), 1645–1653 Proteins and Melanins hemoglobin, whereas pheomelanin gives (1988). Responsible for transporting oxygen, he- a negative bleaching signal in Figure 6c. (2) D. Zhong et al., Proc. Natl. Acad. Sci. U.S.A. moglobin is a metalloprotein in the red Their difference arises from stimulated 98(21), 11873–11878 (2001). blood cells of vertebrates. It is an assembly emission or ground-state bleaching, re- (3) S.K. Pal et al., J. Phys. Chem. B 106(48), of four globular protein subunits. Each spectively (6). 12376–12395 (2002). subunit is composed of a protein tightly (4) T. Ye, D. Fu, and W.S. Warren, Photochem. Pho- associated with a heme group. A heme Historical Pigments tobiol. 85(3), 631–645 (2009). group consists of an iron ion in a porphy- Pump–probe microscopy could be fur- (5) W. Min et al., Ann. Rev. Phys. Chem. 62, 507 rin. It is well known that the heme group ther exploited to identify pigments in (2011). portrays strong absorption yet weak flu- historic artworks. The approach could (6) T.E. Matthews et al., Science Translational orescence. These properties make label- extract molecular information with high Medicine 3(71), 71ra15-71ra15 (2011). free pump–probe microscopy imaging resolution in 3D making it attractive in (7) P.A. Elzinga et al., Appl. Spectrosc. 41(1), 2–4 of hemoglobin an ideal approach. Fu this application, since accurate identifi- (1987). and colleagues (36) demonstrated label- cation is of great value for authentication (8) T. Katayama et al., Langmuir 30(31), 9504– free deep tissue imaging of microvessels and restoration (19–21). Villafana and 9513 (2014). in nude mouse ear. They chose a pump colleagues studied the layer structure of (9) M. Hu and G.V. Hartland, J. Phys. Chem. B beam of 775 nm and a probe beam of a painting by femtosecond pump–probe 106(28), 7029–7033 (2002). www.spectroscopyonline.com April 2017 Spectroscopy 32(4) 11

(10) G. Seifert et al., Applied Physics B 71(6), 795– (40) S. Chong, W. Min, and X.S. Xie, J. Phys. Chem. Cells 150, 51–56 (2016). 800 (2000). Lett. 1(23), 3316–3322 (2010). (68) Q. Cui et al., ACS Nano 8(3), 2970–2976 (11) G.V. Hartland, Chemical Reviews 111(6), (41) K.L. Blythe et al., Phys. Chem. Chem. Phys. (2014). 3858–3887 (2011). 15(12), 4136–4145 (2013). (69) M.J. Simpson et al., J. Phys. Chem. Lett. 7(9), (12) M.J. Simpson et al., J. Invest. Derm. 133(7), (42) S. Zink et al., Microscopy Today 21(04), 14–18 1725–1731 (2016). 1822–1826 (2013). (2013). (70) J. Li et al., Scientific Reports 5, 12394 (2015). (13) C. Dong, et al., Biophys. J. 69(6), 2234 (1995). (43) R. van Grondelle, Biochim. Biophys. Acta. (71) M. Terrones et al., Nature 388(6637), 52–55 (14) D. Fu et al., J. Biomed. Optics 12(5), 054004 811(2), 147–195 (1985). (1997). (2007). (44) B. Gao, G.V. Hartland, and L. Huang, J. Phys. (72) A. Bachtold et al., Science 294(5545), 1317– (15) L. Wei and W. Min, Anal. Bioanal. Chem. Chem. Lett. 4(18), 3050–3055 (2013). 1320 (2001). 403(8), 2197–2202 (2012). (45) M.N. Slipchenko et al., J. Biophoton. 5(10), (73) J.S. Lauret et al., Phys. Rev. Lett. 90(5), 057404 (16) T.E. Matthews et al., Biomed. Optics Express 801–807 (2012). (2003). 2(6), 1576–1583 (2011). (46) D.B. Murphy and M.W. Davidson, in funda- (74) B. Gao, G.V. Hartland, and L. Huang, ACS Nano (17) F.E. Robles et al., Biomed Optics Express 6(9), mentals of Light Microscopy and Electronic 6(6), 5083–5090 (2012). 3631–3645 (2015). Imaging (John Wiley & Sons, Inc., 2012), pp. (75) Y. Wan et al., Nature Chem. 7(10), 785–792 (18) M.J. Simpson et al., J. Phys. Chem. Letters 1–19. (2015). 4(11), 1924–1927 (2013). (47) P. Wang et al., Nature Photonics 7, 449–453 (76) M.M. Gabriel et al., Nano Lett. 13(3), 1336– (19) P. Samineni et al., Optics Letters 37(8), 1310– (2013). 1340 (2013). 1312 (2012). (48) J. Miyazaki, K. Kawasumi, and T. Kobayashi, (77) T. Chen et al., Nanoscale 5(11), 4701–4705 (20) T.E. Villafana et al., Proc. Natl. Acad. Sci. 111(5), Optics Lett. 39(14), 4219–4222 (2014). (2013). 1708–1713 (2014). (49) E.M. Grumstrup et al., Chem. Phys. 458, 30–40 (78) G.V. Hartland, Chem. Sci. 1(3), 303–309 (21) F. Martin, Physics World 26(12), 19 (2013). (2015). (2010). (22) S. Link et al., Phys. Rev. B 61(9), 6086 (2000). (50) E.S. Massaro, A.H. Hill, and E.M. Grumstrup, (79) S.S. Lo et al., ACS Nano 6(6), 5274–5282 (23) L. Huang, H.N. Pedrosa, and T.D. Krauss, Phys. ACS Photonics 3(4), pp 501–506 (2016). (2012). Rev. Letters 93(1), 017403 (2004). (51) G.I. Redford and R.M. Clegg, J. Fluoresc. 15(5), (80) B.P. Mehl et al., J. Phys. Chem. Lett. 2(14), (24) I. Robel, B.A. Bunker, and P.V. Kamat, Adv. 805–815 (2005). 1777–1781 (2011). Mater. 17(20), 2458–2463 (2005). (52) M.A. Digman et al., Biophys. J. 94(2), L14–L16 (81) D. Polli et al., Advanced Materials 22(28), (25) Y. Jung et al., Phys. Rev. Letters 105(21), (2008). 3048–3051 (2010). 217401 (2010). (53) C. Stringari et al., Proc. Natl. Acad. Sci. 108(33), (82) Z. Guo, et al., Nature Commun. 6 7471(2015). (26) L. Tong et al., Nat. Nanotech. 7(1), 56–61 13582–13587 (2011). (83) C.Y. Wong et al., J. Phys. Chem. C 117(42), (2012). (54) F.E. Robles et al., Optics Express 20(15), 22111–22122 (2013). (27) R. Berera, R. van Grondelle, and J.T. Kennis, 17082–17092 (2012). (84) Z. Guo et al., J. Phys. Chem. B 119(24), 7666– Photosynth. Res. 101(2-3), 105–118 (2009). (55) F. Fereidouni et al., J. Biophoton. 7(8), 589– 7672 (2015). (28) D.Y. Davydova et al., Laser & Photonics Reviews 596 (2014). (85) H. Yan et al., Advanced Materials 27(23), 10(1), 62–81 (2016). (56) F. Fereidouni, A.N. Bader, and H.C. Gerritsen, 3484–3491 (2015). (29) B. Nechay et al., Rev. Scient. Inst. 70(6): p. Optics Express 20(12), 12729-12741 (2012). (86) D. Fu et al., Optics Lett. 32(18), 2641–2643 2758–2764 (1999). (57) D. Fu and X.S. Xie, Anal. Chem. 86(9), 4115– (2007). (30) J. Jahng et al., Appl. Phys. Lett. 106(8), 083113 4119 (2014). (87) I.R. Piletic, T.E. Matthews, and W.S. Warren, J. (2015). (58) J. Miyazaki, K. Kawasumi, and T. Kobayashi, Chem. Phys. 131(18), 181106 (2009). (31) L. Huang and J.-X. Cheng, Ann. Rev. Mat. Res. Rev. Scientif. Inst. 85(9), 093703 (2014). (88) J.D. Simon et al., Pigment Cell & Melanoma 43(1), 213–236 (2013). (59) D. Fu et al., J. Biomed. Optics 13(5), 054036– Research 22(5), 563–579 (2009). (32) L. Tong and J.-X. Cheng, Materials Today 14(6), 054036-7 (2008). 264–273 (2011). (60) T. Koyama et al., J. Luminesc. 169 Part B, Pu-Ting Dong and Ji-Xin Cheng are (33) J. Cabanillas-Gonzalez, G. Grancini, and G. Lan- 645–648 (2016). with Purdue University in Indiana. Please zani, Adv. Mater. 23(46), 5468–5485 (2011). (61) B. Gao et al., Nano Lett. 11(8), 3184–3189 direct correspondence to: jcheng@purdue. (34) H.E. Lessing and A. Von Jena, Chem. Phys. Lett. (2011). edu 42(2), 213–217 (1976). (62) R.J. Ellingson et al., Phys. Rev. B 71(11), 115444 (35) A. Von Jena and H.E. Lessing, Chem. Phys. Lett. (2005). 78(1), 187–193 (1981). (63) S.J. Kang et al., Nature Nanotechnol. 2(4), (36) D. Fu et al., J. Biomed. Opt. 13(4), 040503 230–236 (2007). (2008). (64) M.A. van Dijk, M. Lippitz, and M. Orrit, Phys. (37) W. Min et al., Nature 461(7267), 1105–1109 Rev. Lett. 95(26), 267406 (2005). (2009). (65) O.L. Muskens, N. Del Fatti, and F. Vallée, Nano (38) S.W. Hell and M. Kroug, Appl. Phys. B 60(5), Lett. 6(3), 552–556 (2006). For more information on this topic, 495–497 (1995). (66) D.Y. Davydova et al., Chem. Phys. 464, 69–77 please visit our homepage at: (39) S. Bretschneider, C. Eggeling, and S.W. Hell, (2016). www.spectroscopyonline.com Phys. Rev. Lett. 98(21), 218103 (2007). (67) H. Xia et al., Solar Energy Materials and Solar