www. biophotonics-journal.orgJournal of

BIOPHOTONICS

REPRINT

FULL ARTICLE

Photothermal bleaching in time-lapsephotoacoustic microscopy

Liang Gao**, Lidai Wang**, Chiye Li, Alejandro Garcia-Uribe, and Lihong V. Wang*

Department of Biomedical Engineering, Washington University in St. Louis, One Brookings Dr., St. Louis, MO, 63130, USA

Received 11 September 2012, revised 23 October 2012, accepted 25 October 2012Published online 26 November 2012

Key words: photoacoustic microscopy, nanoparticle, photothermal bleaching, timelapse imaging

1. Introduction

Photobleaching, a common phenomenon in fluores-cence microscopy, occurs when a fluorophore perma-nently loses its ability to fluoresce due to long expo-sure to intense excitation light [1, 2]. On the onehand, photobleaching undesirably limits the observa-tion time for monitoring a dynamic process; on theother hand, photobleaching can be beneficiallyemployed in molecular diffusion or motion studiesvia techniques such as fluorescence-recovery afterphotobleaching (FRAP) [3, 4] or fluorescence lossin photobleaching (FLIP) [5]. Properly exploitingphotobleaching requires a thorough understanding

of its dependence on the excitation light power. Sucha dependence in fluorescence microscopy has beencarefully characterized [1, 2, 6].

However, photobleaching is not exclusively a fea-ture of fluorescence microscopy: It also exists inother imaging modalities. One example is photoa-coustic microscopy (PAM) [7, 8], in which the sam-ple is irradiated by a short laser pulse (ns duration).The absorbed light energy is converted into heat andproduces an ultrasonic wave via thermoelastic ex-pansion. The ultrasonic wave is detected by an ultra-sound transducer. Contrasts in PAM stem from lightabsorption by endogenous or exogenous absorbers,such as hemoglobin [9] or nanoparticles [10, 11], re-

# 2013 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Journal of

BIOPHOTONICS

We studied the phenomenon of photothermal bleach-ing – a gradual reduction of contrast agent particles dur-ing repeated scans in photoacoustic microscopy. The de-pendence of the photothermal bleaching rate on theexcitation pulse energy, pulse duration, and the absor-ber’s size was determined while the laser focal diameterwas held constant. Our results showed that, the depen-dence of the photothermal bleaching rate on the excita-tion pulse energy differed before and after the absorberswere raised to their melting point by the deposited laserenergy. Based on this finding, we suggested an optimalexcitation pulse energy, which balances the photother-mal bleaching rate and signal amplitude, for time-lapseimaging applications.

Dependence of photothermal bleaching rate k on the de-livered laser pulse energy and temperature rise.

* Corresponding author: e-mail: lhwang@wustl.edu** Authors contributed equally to this work.

J. Biophotonics 6, No. 6–7, 543–548 (2013) / DOI 10.1002/jbio.201200184

spectively. These absorbers can be intentionallyphoto-destructed using strong excitation light, result-ing in a gradual reduction of photoacoustic signalsduring repeated scans.

The photobleaching mechanism in PAM differsfrom that in fluorescence microscopy. Upon absorb-ing a photon, an electron transits from the groundstate to an excited state. The electron’s energy canbe released via primarily two paths: radiative decay,i.e., fluorescence, or non-radiative decay, i.e., thermaldissipation. In fluorescence microscopy, photo-bleaching is the photochemical destruction of afluorophore – during radiative decay the excitedelectrons are trapped in a relatively long-lived tripletstate (lifetime ~1 ms), thus allowing the fluorophorea much longer time to undergo irreversible chemicalreactions with the environment than would be al-lowed by the singlet-singlet transition [12]. In PAM,photobleaching is the photothermal destruction ofan absorber – the heat generated during non-radia-tive decay alters the structure of the absorber, thusreducing sample’s absorption coefficient at the origi-nal excitation wavelength. As an example, Figure 1shows the scanning electron microscopic images of100 nm gold nanoparticles (GNPs) before and afterthey were imaged by PAM. The GNPs were frag-mented by the deposited laser energy during photo-acoustic imaging and thus vanished. To distinguishthe photobleaching in PAM from its fluorescentcounterpart, we term this photothermally inducedabsorption change as “photothermal bleaching”.

To begin to exploit the photothermal bleaching inPAM, here we studied its dependence on the laserpulse energy, the absorber’s diameter, and the laserpulse duration while the laser focal diameter washeld constant. GNPs were chosen as the targets be-cause they were widely used as contrast agents inphotoacoustic imaging applications [10, 11, 13]. Ourresults revealed that, within the linear excitationrange, photothermal bleaching behaved differentlybefore and after the GNPs were raised to their melt-

ing point. Below this critical point, the photothermalbleaching rate had weak dependence on the laserpulse energy; while above the melting point, thephotothermal bleaching rate increased rapidly.Based on this finding, we suggest a method to deter-mine the optimal excitation laser pulse energy fortime-lapse photoacoustic imaging.

2. Results and discussion

To confine our studies within the linear photoacous-tic imaging range, where the photoacoustic signalamplitude is proportional to the laser pulse energy,we first identified the pulse energy range for linearexcitation. GNPs of 400 nm in diameter were sus-pended in water solution and imaged by a PAM sys-tem [14] using different pulse energies. Since GNPsdiffuse freely in the water solution, the photother-mally bleached particles were rapidly replaced, andthus negligible. The measured photoacoustic ampli-tude versus laser pulse energy is shown in Figure 2.Below ~120 nJ (fluence: 0.33 J/cm2), photoacousticamplitude is linearly related to the pulse energy, solaser pulse energies within this linear range wereused in the following experiments. The focal di-ameter of the laser beam was set to 6.8 mm for allexperiments, and the laser wavelength was 532 nm.

Next, the 400 nm-diameter GNPs were fixed on acover glass to eliminate the effect of diffusion on themeasurement of photothermal bleaching rate (seeMaterials and methods). The photothermal bleach-ing dynamics were measured under different laserpulse energies, and the results are shown in Figure 3.The photothermal bleaching rate k was calculated byfitting a double exponential curve to the acquireddata (see Materials and methods).

Figure 1 Photothermal destruction of contrast agents inPAM. Scanning electron microscopic images of a samplewith 100 nm GNPs fixed on a cover glass (a) before and(b) after it was imaged by PAM at an intentionally highfluence of 82.6 mJ/cm2.

Figure 2 Photoacoustic amplitude vs. delivered excitationlaser pulse energy. The 400-nm-diameter GNPs were sus-pended in water solution and imaged by PAM. The coeffi-cient of determination, R2, was 0.990.

L. Gao et al.: Photothermal bleaching544

Journal of

BIOPHOTONICS

# 2013 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.biophotonics-journal.org

Figure 4 is a log-log plot of photothermal bleach-ing rate versus laser pulse energy. Since the photo-thermal bleaching in PAM originates from instanta-neous optical heating, we estimated the temperaturerise of a GNP upon a pulse laser excitation and cor-related it to the measured photothermal bleachingrate. The light absorption of a spherical GNP obeysBeer’s law, which gives

DE ¼ð

Fð1� e�malÞ ds

¼ðp=2

0

I

A1� e�maD cos q� �

pD2

4sin 2q dq ð1Þ

where F is the optical fluence (J/cm2), sis the pro-jected cross-section perpendicular to the incidentlight direction, l is the optical path length, ma is theabsorption coefficient (mAu

a ¼ 5:6867� 105 cm�1), Dis the absorber’s diameter, I is the laser pulse energy,A is the focal spot area, and q is the polar angle inthe spherical coordinates. Since the thermal confine-ment time for a 400 nm GNP in water solution is

tth ¼d2

c

ath¼ ð0:4� 10�4 cmÞ2

1:3 cm2=s¼ 1:2 ns ð2Þ

(dc is the characteristic dimension of the heated re-gion, and ath is the thermal diffusivity of gold),which is longer than the laser pulse duration (1 ns),the optical heating process was regarded as ther-mally confined, i.e., thermal conduction was negligi-ble during the pulse laser excitation. Consequently,the temperature rise of a GNP approximates

DT ¼ DE

CVqVð3Þ

where CV is the specific heat (CAuV ¼ 0:13 J=g k), q is

the mass density (qAu ¼ 19:3 g cm�3), and V is thevolume. By substituting DE in Eq. (1) into Eq. (3),the temperature rise DT under the excitation of dif-ferent pulse energies was calculated. The results areshown in Table 1 and labeled next to the corre-sponding laser pulse energies in Figure 4.

The data in Figure 4 gives two slopes: belowIc ¼ 30 nJ, the photothermal bleaching rate k has aweak dependence on the laser pulse energy; whileabove it, the slope equals 3:6� 0:3, indicating a non-linear relation between photothermal bleaching rateand laser pulse energy –k / I3:6. The correspondingtemperature rise for the critical point Ic ¼ 30 nJ wascalculated by Eq. (1) and Eq. (3), yielding

DTc ¼ 967 K ð4ÞAssuming the room temperature was 23 �C, the tem-perature of the GNP after a pulse laser excitationapproximated

T ¼ Troom þ DTc ¼ 990 �C ð5Þ

Figure 3 Photothermal bleaching of 400-nm-diameterGNPs under different delivered laser pulse energies. Thesamples were continuously imaged by PAM at 0.5 Hz C-scan rate.

Figure 4 Dependence of photothermal bleaching rate k onthe delivered laser pulse energy and temperature rise. I isthe laser pulse energy (nJ), and k is the photothermalbleaching rate (s�1). The slope for the data points withpulse energies >30 nJ is 3:6� 0:3 (95% confidence bounds).The coefficient of determination, R2, is 0.996.

Table 1 Calculated temperature rise of a 400 nm diameterGNP upon excitation by different laser pulse energies.

Laser pulse energy (nJ) Temperature rise (K)

15 48325 80644 141858 186969 222378 251390 290099 3190

J. Biophotonics 6, No. 6–7 (2013) 545

FULLFULLARTICLEARTICLE

# 2013 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheimwww.biophotonics-journal.org

close to the melting point of gold (TAumelting ¼ 1; 064 �C).

This implies that the phase change of the absorber(from solid to liquid state) may contribute to the twodifferent photothermal bleaching behaviors observedwithin the linear PA excitation range.

Since photothermal bleaching behaves differentlyaround an absorber’s melting point under linear ex-citation, this dependence suggests an optimal excita-tion pulse energy for the time-lapse photoacousticimaging: Ic, the critical value that raises the GNPs totheir melting point. Below Ic, raising the excitationpulse energy increases the photoacoustic signal line-arly but does not noticeably speed up the photother-mal bleaching rate. While above Ic, although increas-ing the excitation pulse energy can linearly improve

the signal, the photothermal bleaching acceleratesnonlinearly, significantly shortening the time windowfor imaging a dynamic event.

Additionally, the dependence of the photother-mal bleaching rate on the absorber’s diameter wasalso investigated. GNPs with diameters of 50 nm,70 nm, 100 nm, and 400 nm fixed on a glass slidewere excited under the same laser pulse energy(I ¼ 15 nJ). The sample was continuously scanned at0.5 Hz C-scan rate until it became totally photother-mally bleached. Figure 5 shows the photoacousticamplitude versus time for GNPs of different di-ameters. Because maD > 1 for gold, the energy de-position DE is primarily related to D2. By contrast,the temperature rise is inversely proportional to D3.As a result, smaller particles sustained higher tem-perature rises and they had faster photothermalbleaching rates than GNPs with larger diameters.

Different laser pulse durations also affect thephotothermal bleaching rate. We excited 100-nm-di-ameter fixed GNPs with two lasers having the samepulse energy (28 nJ) but delivered over differentpulse durations (1 ns and 10 ns). As shown in Fig-ure 6, the laser pulses of 1 ns duration bleached theGNPs much faster than the 10 ns ones (k ¼ 4.40 s�1

at 1 ns pulse duration; k ¼ 0.0775 s�1 at 10 ns pulseduration), indicating that pulse duration plays an im-portant role in the photothermal bleaching process.Although the total deposited energy DE was thesame for both cases, the heat delivered in 10 ns dura-tion diffuses into a larger volume than that deliveredin 1 ns duration. Thus the ultimate temperature ofGNPs raised by 10 ns laser pulses is lower than thatby 1 ns ones. This fact may explain the acceleratedphotothermal bleaching under illumination by short-er duration laser pulses.

3. Conclusion

In summary, we studied the phenomenon of photo-thermal bleaching in PAM. The dependence of thephotothermal bleaching rate on the excitation laserpulse energy, duration, and absorber’s diameter wascarefully measured and analyzed. Our resultsshowed that, under linear photoacoustic excitation,two different photothermal bleaching behaviors existaround the absorber’s melting point. Based on thisfinding, an optimal excitation laser pulse energy,which balances the strength of the photoacoustic sig-nal and the bleaching rate, was suggested for time-lapse imaging applications. Additionally, increasingthe absorber’s diameter or prolonging the laser pulseduration also beneficially reduces the photothermalbleaching rate.

To the best of our knowledge, this is the first timethat the phenomenon of photothermal bleaching was

Figure 5 Photothermal bleaching of different diameterGNPs under the same delivered excitation pulse energy(15 nJ). The samples were continuously imaged by PAM at0.5 Hz C-scan rate.

Figure 6 Photothermal bleaching of 100 nm diameterGNPs under two different laser pulse durations. The laserpulse energy delivered was 28 nJ for both pulse durations.The sample was imaged by PAM at 0.5 Hz C-scan rate.

L. Gao et al.: Photothermal bleaching546

Journal of

BIOPHOTONICS

# 2013 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.biophotonics-journal.org

characterized in PAM. This knowledge will be valu-able not only for the GNP-based PA imaging, butalso for establishing new photothermal-bleaching-assisted techniques such as photoacoustic-recoveryafter photothermal bleaching or photoacoustic-lossin photothermal bleaching.

4. Materials and methods

4.1 Sample preparation

To eliminate the effect of diffusion during the photo-thermal bleaching rate measurement, the targetGNPs were fixed on top of cover glasses by the fol-lowing operations [15–17]. The cover glasses werefirst cleaned in warm detergent (RBS 35, Sigma-Al-drich, USA), flushed with DI water, and dried withnitrogen. Then the cover glasses were put in aplasma oxidizer (Diener Electronics, Germany) for20 min, followed by another 20 min in a 10% solu-tion of (3-Aminopropyl)-triethoxysilane (Sigma-Al-drich, USA) in anhydrous ethanol. After silaniza-tion, the cover glasses were rinsed in anhydrousethanol and kept at 120 �C for 3 hours. Gold parti-cles with mean diameters of 50 nm, 70 nm, 100 nm(TedPella, USA) and 400 nm (Cytodiagnostics, Ca-nada) were pipetted onto the silanized cover glassesand baked in a 45 �C oven overnight. Before im-aging, the samples were washed with DI water on ashaker for 3 hours to remove the unfixed residualgold particles.

4.2 Time-resolved measurement ofphotothermal bleaching dynamics

Time-resolved photothermal bleaching experimentswere performed using a previously described voice-coil scanning PAM [14] with 0.1 NA optical objec-tive. The PAM uses a raster-scanned focused ultraso-nic transducer coupled with confocal optical illumi-nation to detect light-absorption-induced ultrasoundsignals. Two 532-nm-wavelength pulsed lasers (Elfor-light, Ltd., UK), with 1 ns and 10 ns pulse durations,respectively, were employed in the photothermalbleaching experiments. The excitation laser pulse en-ergy was adjusted by rotating a variable ND filter(Thorlabs Inc., USA) in front of the laser output,and monitored by a custom-made photodiode. Be-fore the experiment, the correspondence betweenthe laser pulse energy at the sample and the photo-diode’s measured value was calibrated.

During the photothermal bleaching experiments,the sample was continuously scanned by the PAM at

0.5 Hz C-scan rate until the signals vanished fromthe image. The image was acquired with 1000 voicecoil scanning steps and 20 mechanical stage scanningsteps, and rendered as a maximum amplitude pro-jection image. The photoacoustic amplitude of animage was calculated by summing all the signalswithin the field-of-view (400� 100 mm2) after remov-ing the background. The photothermal bleachingrate k was calculated by fitting a double exponentialcurve Aexpð�k1tÞ þ Bexpð�k2tÞ to the acquired dataand averaging over the exponential indicesk ¼ Ak1 þ Bk2 [6, 18].

Acknowledgements This work was sponsored by the Na-tional Institutes of Health (NIH) under grants R01EB000712, R01 EB008085, R01 CA134539, U54 CA136398,R01 CA157277, and R01 CA159959. L. V. Wang has a finan-cial interest in Microphotoacoustics, Inc. and Endra, Inc.;however, neither provided support for this work.

Author biographies Please see Supporting Informationonline.

References

[1] R. I. Ghauharali, J. W. Hofstraat, and G. J. Braken-hoff, J Microsc 192, 99–113 (1998).

[2] R. I. Ghauharali and G. J. Brakenhoff, J Microsc 198,88–100 (2000).

[3] E. A. J. Reits and J. J. Neefjes, Nat Cell Biol 3(6),E145–E147 (2001).

[4] Y. H. Sniekers and C. C. van Donkelaar, Biophys J89(2), 1302–1307 (2005).

[5] J. Lippincott-Schwartz, E. Snapp, and A. Kenworthy,Nat Rev Mol Cell Bio 2(6), 444–456 (2001).

[6] G. H. Patterson and D. W. Piston, Biophys J 78(4),2159–2162 (2000).

[7] L. V. Wang, Nat Photonics 3(9), 503–509 (2009).[8] L. H. V. Wang and S. Hu, Science 335(6075), 1458–

1462 (2012).[9] S. Hu and L. V. Wang, J Biomed Opt 15(1), 011101

(2010).[10] Q. Zhang, N. Iwakuma, P. Sharma, B. M. Moudgil,

C. Wu, J. McNeill, H. Jiang, and S. R. Grobmyer, Na-notechnology 20(39), 395102 (2009).

[11] X. M. Yang, E. W. Stein, S. Ashkenazi, and L. H. V.Wang, Wires Nanomed Nanobi 1(4), 360–368 (2009).

[12] R. Y. Tsien, L. Ernst, and A. Waggoner, Fluoro-phores for Confocal Microscopy: Photophysics andPhotochemistry, in: Handbook of Biological ConfocalMicroscopy, J. B. Pawley (Ed.) (Springer, 2006).

[13] S. Mallidi, T. Larson, J. Tam, P. P. Joshi, A. Karpiouk,K. Sokolov, and S. Emelianov, Nano Lett 9(8), 2825–2831 (2009).

[14] L. D. Wang, K. Maslov, J. J. Yao, B. Rao, and L. H. V.Wang, Optics Letters 36(2), 139–141 (2011).

[15] C. George, D. Ricci, and E. Di Zitti, Superlattice Mi-crost 44(4–5), 608–616 (2008).

J. Biophotonics 6, No. 6–7 (2013) 547

FULLFULLARTICLEARTICLE

# 2013 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheimwww.biophotonics-journal.org

[16] N. Nath and A. Chilkoti, Anal Chem 74(3), 504–509(2002).

[17] G. Arslan, M. �zmen, B. Gunduz, X. Zhang, andM. Ersoz, Turkish Journal of Chemistry 30(2), 203–210 (2006).

[18] S. Gavrilyuk, S. Polyutov, P. C. Jha, Z. Rinkevicius,H. Agren, and F. Gei’mukhanov, J Phys Chem A111(47), 11961–11975 (2007).

L. Gao et al.: Photothermal bleaching548

Journal of

BIOPHOTONICS

# 2013 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.biophotonics-journal.org