3.1.

PFC Emulsions

Gold NPs were synthesized using a previously published method based on the reduction of sodium citrate.³⁰ Specifically, gold NPs were made by loading 100 mg of sodium citrate and 150 mL of deionized water into a 250-mL round-bottom flask, heating the mixture to boiling, followed by the injection of 1 mL of 25 mM ${\mathrm{HAuCl}}_4$. The temperature of the mixture was slowly decreased to 90°C over a 15 min time period, and 1 mL of 60-mM sodium citrate was then injected, followed by the injection of 0.9 mL of 25-mM ${\mathrm{HAuCl}}_4$. After 15 min, the injections of sodium citrate and ${\mathrm{HAuCl}}_4$ were repeated, resulting in the formation of gold NPs. To miscibilize gold NPs into PFHs (FC-72, Synquest Labs, FL, USA), the NPs were coated with silica,³¹ followed by surface fluorination.^32,33 Specifically, to coat the NPs with silica, a freshly prepared aqueous solution of 3-aminopropyl trimethoxy silane (APTES, 1.6 mL, 1 mM) was added to 150 mL of the gold NPs solution under vigorous magnetic stirring. The APS and gold dispersion were reacted for 15 min to ensure complete complexation of the amine groups with the gold surface. A solution of active silica was prepared by lowering the pH of a 0.54 wt% sodium silicate solution to 11 by progressive additions of 1.0 M HCl. Active silica (9 mL) was then added to 150 mL of the surface modified gold NPs, again under vigorous magnetic stirring. The resulting dispersion ($\mathrm{pH}=8.5$) was then allowed to stand for a minimum of 24 h, to permit the active silica to polymerize onto the surface of the gold NPs. For miscibilization of gold NPs into PFH, the silica-coated gold NPs were redispersed in 10 mL of methanol,³⁴ and the mixture was reacted with 150 μL of 1H,1H,2H,2H-perfluorodecyltriethoxysilane (5 min), and mixed with 50 μL of 30% ammonia/water and reacted for 24 h. After evaporation of methanol, the NPs were solubilized into 2 mL of PFH. All chemicals for silica-coated gold NP synthesis and their miscibilization into PFCs (${\mathrm{HAuCl}}_4$, sodium citrate, APS, sodium silicate and 1H,1H,2H,2H-perfluorodecyltriehthoxysilane) were purchased from Sigma-Aldrich, ON, Canada. The thickness of the silica coating and the size of the gold NPs were determined by transmission electron microscopy using a Hitachi S-5200 scanning electron microscope (Hitachi Canada Ltd., ON, Canada). For NP characterization in the ultraviolet, visible, and near-IR ranges, a Cary-5000 UV-Vis-NIR spectrophotometer (Agilent Technologies, CA, USA) was used. Gold NP-incorporated PFH droplets were prepared using 5 mL of deionized water (Millipore Milli-Q grade, $18.2\mathrm{M\Omega }$), 0.175 mL of gold NP-FC-72 solution, and 0.025 mL of anionic phosphate fluorosurfactant (Zonyl-FSP, Sigma Aldrich, ON, Canada). Droplets were prepared by coarse emulsification by vortexing, followed by membrane emulsification using 10-μm pore size polycarbonate membranes (Nuclepore Track-Etch Membrane, Whatman, NJ, USA).¹⁴

PFH (56°C boiling point) was used instead of the more commonly used perfluoropentane (PFP, 29°C boiling point) as micron-scale PFH droplets are more stable and are less likely to spontaneously vaporize at the temperatures used in these experiments compared to PFP droplets.^35,36 Moreover, the sound speed of bulk PFP could not be measured at 36°C as it is over the boiling point of 29°C, and it would be difficult to use the theoretical photoacoustic equations. The sound speed of bulk PFH has been previously measured at the frequencies and temperatures used.²⁴

3.2.

Photoacoustic Microscope

All acoustic and photoacoustic measurements were made using the SASAM acoustic microscope (Kibero GmbH, Germany). An IX81 inverted optical microscope (Olympus, Japan) was modified where a transducer was positioned above the optical objective. The sample is positioned between the optical objective and transducer, allowing for optical observation during measurements. Ultrasonic pulse echo measurements were made by scanning the sample under the transducer. For photoacoustic measurements, a 532-nm laser was collimated through the back port of the inverted microscope and focused onto the sample with the same objective used to view the sample. A dichoric mirror (Chroma Technology Corp, USA) was used to reflect optical wavelengths between 450 and 620 nm, but pass all other wavelengths to allow for optical viewing and targeting when using the laser [Fig. 1(a)]. The transducer was focused above the laser spot under optical guidance, and the sample was scanned under the transducer with both the optical objective and transducer remaining stationary [Fig. 1(b)].

Fig. 1

(a) The acoustic microscope is an inverted optical microscope where the laser has been focused through the pack port onto the sample. The mirror reflects 500 to 650 nm (the laser wavelengths) toward the sample, but allows other wavelengths to pass for optical viewing. The transducer measures the signals from above the sample. (b) A close-up view of a measurement of a single droplet. The laser is focused onto the sample using the optical objective with the transducer positioned directly above the droplet.

Transducers with center frequencies of 375 MHz (60-deg aperture, 42% bandwidth) and 750 MHz (100-deg aperture, 37% bandwidth) were used for all acoustic and photoacoustic measurements. For acoustic measurements, a $10{\mathrm{V}}_{\mathrm{pp}}$ monocycle signal was generated at 300 MHz (for the 375 MHz transducer) and at 1000 MHz (for the 750 MHz transducer) at a pulse repetition frequency (PRF) of up to 500 kHz. One hundred signals were averaged at 375 MHz, while 1000 signals were averaged at 750 MHz. For photoacoustic measurements, the laser was focused onto the sample using a $10\times$ optical objective (0.3 numerical aperture). The laser triggered the data acquisition of the signals the transducer recorded. The laser had a pulse width of 330 ps and PRF of 4 kHz. All signals were amplified by a 40-dB amplifier (Miteq, USA) and digitized at 8 GHz. All measurements were made at 36°C to simulate human physiological conditions. Further details on the system can be found elsewhere.^14,37

3.3.

Acoustic and Photoacoustic Measurements

Droplets were embedded into gelatin phantoms to ensure they were immobilized during the measurements. Droplets were added to liquid gelatin at a temperature of 37°C, and then formaldehyde was added to increase phantom stiffness. Droplets close to the phantom surface were measured to minimize the attenuation losses through gelatin. All acoustic and photoacoustic measurements at 375 MHz were made with the droplets embedded within the gelatin. No signal from embedded droplets was detected at 750 MHz due to attenuation through the gelatin, therefore droplets were deposited onto the phantom surface for the 750-MHz photoacoustic measurements. Subsequent measurements using the 375-MHz transducer were not possible, as the droplets would move from the field of view while switching transducers. The acoustic properties of gelatin and water are very similar,³⁸ and the spectral features only depend on the sound speed and density of the droplet and coupling fluid in Eq. (2). Therefore it is unlikely the change in medium between similar liquids would significantly affect the spectral shape.

To record the acoustic and photoacoustic signals, a droplet was positioned over the laser spot and then the transducer was focused above the droplet. The sample stage was scanned over a $10\times 10$-μm area using a 0.5-μm step size, and the signal from the center of the droplet was used for analysis. The optical objective and transducer remained stationary and co-aligned during all scans. The transducer response was removed by normalizing the signal as described elsewhere.³⁹ Briefly, the normalized signal was calculated by dividing the measured spectrum of the droplet by the reference spectrum of the transducer. The bandwidth of a transducer is typically found by measuring the reflected echoes from a glass or quartz substrate that reflects the ultrasound back to the transducer. We have found that this is not an accurate representation of the transducer bandwidth used in photoacoustics since the transducer is used in passive mode. The transducers appear to have better sensitivity when receiving (photoacoustic) than during transmit and receive (ultrasound). A 200-nm gold layer theoretically has a flat photoacoustic spectrum to within 3 dB from 1 to 1000 MHz,⁴⁰ and will give an accurate representation of the transducer spectral response in passive mode; therefore, we used the photoacoustic signal from a gold film as a reference. A Hamming window was applied to all signals, along with a bandpass filter of 100 to 500 MHz for the 375-MHz transducer and 300 to 1000 MHz for the 750-MHz transducer.

3.4.

Droplet Diameter Measurements

The droplet diameter was calculated using three independent methods: 1) the photoacoustic time domain signal; 2) the time domain ultrasound signal; and 3) fitting the measured photoacoustic spectrum to the model. Only the signals measured using the 375-MHz transducer were used in the diameter calculations. These calculations were compared to images recorded from a CCD camera using an optical microscope with a $10\times$ objective ($100\times \text{total magnification}$).

The photoacoustic method used the time of flight of the positive and negative pressure waves from the droplet and Eq. (3) to determine the droplet diameter. The pressure wave emitted from a droplet during optical irradiation has two significant features: an initial positive sharp pressure peak that gradually decreases to a negative pressure peak. This waveform has a characteristic “$N$-shape” in the time domain, analogous to pressure waves generated from a bursting balloon.^41,42 The theoretical time domain signal for a 5-μm PFH droplet is shown in Fig. 2(a) (black dotted line), where the $N$-shape waveform is clearly visible. However, the finite bandwidth of a transducer can alter the signal. For example, the same signal subjected to a Hamming window and a 100- to 500-MHz bandpass filter (to emulate the limited transducer bandwidth) is shown in Fig. 2(a) (red solid line). Instead of an N-shape, two peaks are visible, which align with the positive and negative peaks of the $N$-shape waveform. These two peaks can be used to determine the droplet diameter using Eq. (3), where $t_1$ is the positive peak on the first signal and $t_2$ is the negative peak on the second signal.

Fig. 2

(a) The theoretical photoacoustic time domain signal from a 5-μm droplet using Eq. (2) (black dotted line), and the same signal after applying a Hamming window and 100- to 500-MHz bandpass filter (red solid line). (b) The measured ultrasound echoes from a 7.2-μm PFH droplet embedded in gelatin. The signals labeled $t_1$ and $t_2$ were used to determine the droplet diameter using Eq. (3) (photoacoustic, left) and Eq. (4) (ultrasound, right).

The ultrasound pulse echo method used the time of flight of the echoes from the top and bottom of the droplet [$t_1$ and $t_2$ in Fig. 2(b)] in Eq. (4) to determine the droplet diameter. This approach required that the droplet was sufficiently large so that the two echoes were separated in time and could be resolved. This method could not be used for droplets smaller than approximately 5 μm, as the signals would overlap.

The photoacoustic spectral method was used to calculate the diameter by comparing the spectrum from the measured photoacoustic signal to the model [Eq. (2)]. The model was calculated using a constant density of $1650\mathrm{kg}/{\mathrm{m}}^3$, and the sound speed was varied from 480 to $486\mathrm{m}/\mathrm{s}$ with increasing frequency from 100 to 1000 MHz [Eq. (5)]. The diameter was adjusted until the minima and maxima between experiment and model matched. The diameter found from these signal processing methods were then compared to a direct optical measurement from the microscope.

4.1.

Droplet Configuration

Silica-coated gold NPs were used as the optical absorbing NP in this study due to their strong optical absorption peak around 540 nm. Transmission electron microscopy images demonstrated these NPs were of uniform size, with a 20-nm-diameter core and 5-nm-thick silica shell [Fig. 3(a)]. The optical absorption of the PFH solution containing NPs peaked between 530 to 540 nm [Fig. 3(b)], an optimum range for the 532-nm pulsed laser used in this study. The laser energy for all photoacoustic measurements was $30\mathrm{nJ}/\mathrm{pulse}$, focused to a spot size approximately 10 μm in diameter, resulting in a laser fluence of $38\mathrm{mJ}/{\mathrm{cm}}^2$. Although damage to gold NPs has been reported for laser fluences as low as $25\mathrm{mJ}/{\mathrm{cm}}^2$ at 532 nm,⁴³ damage thresholds can depend on the gold NP surface and coating, and other laser parameters such as pulse length. In this study, the photoacoustic signals from the NP-loaded droplets (ranging in size from 2 to 15 μm) were stable over repeated measurements using laser fluences of $38\mathrm{mJ}/{\mathrm{cm}}^2$, indicating damage to NPs was negligible. At higher fluence levels (over $50\mathrm{mJ}/{\mathrm{cm}}^2$), the measured photoacoustic signal decreased rapidly over repeated pulses, which could be attributed to NP damage. The vaporization threshold for these droplets was between 50 and $120\mathrm{mJ}/{\mathrm{cm}}^2$, considerably less than 1.5 to $3\mathrm{J}/{\mathrm{cm}}^2$ required to vaporize PFH droplets containing lead sulphide NPs reported previously using a 1064-nm laser.¹⁴ This is likely due to the much higher absorption coefficient of the gold NPs ($110{\mathrm{cm}}^{-1}$ at 532 nm) compared with lead sulphide NPs ($4.3{\mathrm{cm}}^{-1}$ at 1064 nm).

Fig. 3

Transmission electron microscopy image (a) and optical absorption coefficient (b) of the silica coated gold NP PFH solution. The gold NPs were approximately 20 nm in diameter with a 5 nm thick silica coating. The peak optical absorption was between 530 and 540 nm. The scale bar is 20 nm.

The number of NPs inside the droplets was estimated based on the initial concentration of NPs in the PFC liquid ($\sim {10}^{14}\mathrm{NPs}/\mathrm{mL}$ of PFC). Assuming the NPs are distributed uniformly within the droplets, there are approximately 6550 NPs in 5-μm droplets ($6.7\times {10}^{-11}\mathrm{cc}$) and 52300 NPs in 10-μm droplets ($5.7\times {10}^{-10}\mathrm{cc}$). The approximate distance from one NP to its neighbor can be calculated by estimating the volume fraction of spheres that can fill a single droplet using a random close-packing model.⁴⁴ Based on this model, the average distance between NPs is $\sim 230\mathrm{nm}$. Although some nonuniform distribution of NPs within the droplets is possible, as an order of magnitude calculation, the calculated values for the NP quantity and NP-spacing seem reasonable.

4.2.

Sound Dispersion Effects

Other studies have demonstrated excellent agreement between the photoacoustic model [Eq. (1)] and experiments of optical absorbing mm-sized spherical droplets at photoacoustic frequencies less than 20 MHz, where sound dispersion has been neglected.^23,25,45 While PFC liquids have negligible dispersion at 22 MHz,⁴⁶ a small but measureable increase of sound dispersion was found for three PFC liquids up to 1000 MHz.²⁴ For PFH, the liquid used in this study, the sound speed increased from 480 to $486\mathrm{m}/\mathrm{s}$, and the ultrasound attenuation increased from 1368 to $16840\mathrm{dB}/\mathrm{cm}$ from 200 to 1000 MHz at 36°C.²⁴ Using a power law fit, the attenuation as a function of frequency was found to be $\alpha (f)=0.352{\mathrm{f}}^{1.56}$, as described in Ref. 24. Attenuation is linked to changes in sound speed via the Kramers-Kronig relations. To obtain a smooth change in sound speed with frequency, the sound speed over the range of 100 to 1000 MHz was calculated from the attenuation power law fit of $0.352{\mathrm{f}}^{1.56}$ using the Kramers-Kronig relations [Eq. (5)].

The effects of this small but measurable amount of dispersion is shown in Fig. 4, which compares the theoretical time and frequency domain solution using Eq. (1) (constant sound speed) with Eq. (2) (increasing sound speed) for a 5-μm diameter PFH droplet. A Hamming window and 300- to 1400-MHz bandpass filter were used to simulate a wide bandwidth transducer. Good agreement in the location of the spectral minima and maxima between both equations were observed up to 300 MHz, while discrepancies were observed over 500 MHz. Only minor differences were observed in the time domain waveform, where the second peak at 362 ns was shifted slightly. Since this study compares the measured photoacoustic spectra from droplets up to 1000 MHz, it is important to account for sound dispersion when comparing measured results to the model.

Fig. 4

The effects of dispersion on the photoacoustic model showing the time domain solution (a) and frequency domain solution (b) for a 5-μm droplet using Eq. (1) (constant sound speed, black solid line) and Eq. (2) (sound speed increasing from 480 to $486\mathrm{m}/\mathrm{s}$, red dashed line). A Hamming window and 300- to 1400-MHz bandpass filter were used for both solutions. Good agreement was found up to 300 MHz, but discrepancies in the location of the spectral minima and maxima were observed over 500 MHz.

4.3.

Droplet Diameter Measurements

The diameters of 18 micron-sized droplets were calculated using the three signal-based methods (photoacoustic waveform, ultrasound pulse echo, and photoacoustic spectrum) and compared with optical microscopy measurements. The results are summarized in Table 1. The diameter was calculated using several methods to verify the photoacoustic spectral fit; in Eqs. (1) and (2), the spectral minima and maxima shift depending on $q$, which itself depends on the droplet radius and sound speed. The same photoacoustic spectrum could be obtained by varying the droplet radius and sound speed, provided the ratio remained constant. Therefore to ensure the parameters found from the spectral fit were accurate, they were compared to other methods.

Table 1

The diameters of 18 droplets calculated using the photoacoustic time domain signal [Eq. (3)], the ultrasound time domain signal [Eq. (4)], spectral comparison between model and experiment [Eq. (2)], and optical microscopy. The uncertainties in the photoacoustic and ultrasound methods were calculated based on the accuracy of determining the locations of the signal peaks in Fig. 1, and the model uncertainty from the diameter variations in fitting the experiment to theory. The optical method uncertainty was large due to the difficulty in determining the droplet edges, and the limited number of pixels used when acquiring the droplet image.

The average difference in diameter between the three signal-based methods was $0.12\pm 0.09\mu \mathrm{m}$. The direct optical measurement was an average $0.68\pm 0.39\mu \mathrm{m}$ greater than the other three methods. While the three calculated methods based on signal measurement were very close to each other, the optical measurements were 8.5% higher. The droplet images recorded were only 10 to 20 pixels across, which significantly reduced the measurement precision. Poor contrast due to difficulties in observing the droplet through the gelatin phantom prevented higher magnification objectives from being used. Due to the number of pixels used and the difficulty in observing the droplet through the phantom, the uncertainty in the droplet diameters estimated from optical methods was determined to be $\pm 0.5\mu \mathrm{m}$. The larger diameter estimation and uncertainty in the optical method is likely due to the difficulty in determining the edges of a small spherical object near the limits of optical resolution, compounded by the difficulty in seeing through the partially translucent gelatin phantom. The agreement between optical and signal processing methods improved as the droplet diameter increased; for example, the optical method was $13\%\pm 8\%$ greater for diameters 5 to 7 μm and was only $4.8\%\pm 3.2\%$ for diameters over 10 μm. Therefore, it can be inferred that the diameter determined through optical microscopy would likely improve if larger diameters or higher magnifications were used. Since it was difficult to determine the droplet diameter visually, using the signal-based measurements were essential to corroborate the diameter as predicted by the model [Eq. (2)].

4.4.

Photoacoustic Spectral Measurements

The model [Eq. (2)] calculates the photoacoustic pressure wave emitted from a spherical homogeneous liquid undergoing thermoelastic expansion due to absorption of optical energy. To determine if the measured photoacoustic spectra from NP-loaded PFH droplets were in agreement with the model, the photoacoustic signals for three droplets of different sizes were measured: a 5.7-μm droplet measured at 375 MHz, a 2.45-μm droplet measured at 750 MHz, and a 6.95-μm droplet measured at 750 MHz (Fig. 5). No photoacoustic signal was detected for unloaded PFH droplets. All signals were normalized to maximum amplitude, and the droplet diameter was determined by matching the measured photoacoustic spectral maxima and minima to the model [Eq. (2)]. The diameter found from the spectral fit agreed well with the time domain photoacoustic and ultrasound methods used to calculate the diameter (see Table 1).

Fig. 5

The measured photoacoustic signal (red dashed line) compared to the model (black solid line, equation 2) for a 5.7-μm droplet measured at 375 MHz (a) and (b), a 2.45-μm droplet measured at 750 MHz (c) and (d) and a 6.95-μm droplet measured at 750 MHz (e) and (f). A bandpass filter of 100 to 500 MHz was used for the 375-MHz measurement, and a bandpass filter of 300 to 1000 MHz was used for the 750-MHz measurements. A Hamming window was applied to all measurements. Good agreement between model and experiment were found for the waveforms (a), (c), and (e), and in the location of the spectral minima and maxima (b), (d), and (f).

The measured spectra agreed well with the model in both the time and frequency domains for all three droplets (Fig. 5). The gold NPs act exclusively as the optical absorbing material within the droplets, yet the photoacoustic pressure wave emitted is based entirely on the PFH liquid properties, and not of the gold NPs. This is consistent with previous photoacoustic measurements using lead sulphide NPs inside PFC droplets, which showed good agreement in the spectral features between the model and experiment.^47,48 These results also agree with a study that found the photoacoustic wave is generated from the liquid surrounding the NP, and not from the NP itself.²¹ This suggests that the NP-loaded droplets can be modeled as a single photoacoustic source taking into account only the liquid PFC properties, irrespective of the optical absorbing NPs contained within, provided they absorb light sufficiently at the wavelength used. As NP-loaded emulsions gain popularity as photoacoustic contrast agents, the photoacoustic signal from droplets can be maximized by optimizing the size, shape, and type of NPs knowing that the emulsions will emit pressure waves as a single photoacoustic source.