Photoacoustic technique to measure temperature effects on microbubble viscoelastic properties

Microbubbles are currently being used for a variety of biomedical applications, including contrast-enhanced ultrasound imaging,¹ molecular imaging,² super-resolution imaging,³ drug and gene delivery,⁴ thrombolysis,⁵ and gas transport.⁶ A microbubble typically comprises a phospholipid monolayer shell encapsulating a gas core of 1–10 μm diameter. The lipid shell stabilizes the microbubble by limiting coalescence, dissolution, and ripening.⁷ The lipid monolayer also imposes viscoelastic properties that impact the acoustic response of the microbubble in both the small-amplitude linear and large-amplitude nonlinear regimes.⁸ It is therefore important to characterize the mechanical properties of the shell in order to better predict and control microbubble behavior for a given application.

It is well established that the viscoelastic properties of lipid monolayers show variations with temperature.⁹ This is particularly important for clinical applications of microbubbles because the mechanical properties can differ significantly between ambient and body temperatures. In fact, a number of studies have confirmed that the dynamic response of microbubbles subjected to an ultrasound field is dependent upon temperature. Studies of populations of microbubbles have shown changes in size, acoustic attenuation and scattering, and stability as a function of temperature for both commercial and laboratory synthesized microbubbles.^10–12 While microbubble population studies are important in assessing the response for in vivo applications, it is difficult to accurately determine temperature-dependent shell properties from such measurements. Measurements of the dynamic response of individual microbubbles, though generally more complex experimentally, can be more informative in this regard.

Studies on the temperature-dependent response of individual microbubbles have been limited. Using high-speed imaging to track microbubble oscillations, Vos et al.¹³ showed increased radial excursions at body temperature versus room temperature, and Mulvana et al.¹⁴ observed increased occurrences of non-spherical oscillations, jetting, and gas expulsion at body temperature. Grant et al.¹⁵ used AFM force spectroscopy to show a temperature effect on compression stiffness and creep-displacement. In this letter, we describe the use of a laser-based ultrasonic technique to probe the frequency response of individual lipid-coated microbubbles over a broad temperature range. The frequency spectra were analyzed to determine the temperature-dependent lipid shell elasticity and viscosity, as well as the maximum microbubble radial oscillation. We first considered the steady-state microbubble response, where the temperature was gradually increased. We then examined the transient response to rapid heating from ambient temperature to slightly above body temperature. Our results illustrate the utility of this measurement technique as a platform for the study of temperature-dependent microbubble properties.

Microbubbles were fabricated with phospholipid shells consisting of 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC) and 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)–2000] (DSPE-PEG2000) at a molar ratio of 9:1. Both lipids were suspended in phosphate buffered saline at a lipid concentration of 2 mg/ml, mixed and heated to 45 °C by probe sonication, and stored in a 3-ml glass serum vial sealed with a perfluorobutane gas headspace. Microbubbles were formed by vigorous shaking using a dental amalgamator for 40 s, after which the serum vial was immediately quenched to room temperature in an ice bath. A microbubble washing and separation technique using centrifugation¹⁶ was conducted to remove undesired bubble sizes and residual lipid in solution. Separated microbubbles were kept in concentrated form until measurements began, at which point the microbubbles were diluted with phosphate buffered saline and pipetted onto a microscope glass slide. All microbubbles were tested within 12 h of synthesis.

Figure 1(a) shows a schematic of the experimental setup. Dilute microbubble solution (approximately 5 × 10⁵ bubbles per ml) with a size distribution from ∼2 to 10 μm in diameter was injected between a microscope slide and a cover slip, with a thin gasket separating the two. The slide was placed at the focal plane of a microscope with a 50× objective (numerical aperture of 0.42), and an individual microbubble was positioned at the center of the field-of-view. The microbubble was excited from the ultrasound generated using an electro-absorption modulated diode laser (EML) operating at a wavelength of 1550 nm coupled to a 1.0 W fiber amplifier. The EML was a continuous wave laser modulated with a sine wave such that the power assumed the form: $P=P0[1+sin (2π f t)]$⁠, where 2 $P0$ is the peak power and $f$ is the modulation frequency. The laser intensity was modulated at the frequency of interest using a function generator. The laser output passed through the objective, and the focused spot was positioned by a gimbal mirror on the image plane of the microscope approximately 100 μm from the microbubble. The spot size at the image plane had a 1/e diameter of 3.0 μm and a power of 5.5 mW. At the infrared wavelength of 1550 nm, light is strongly absorbed in water, leading to local thermal expansion and ultrasound generation through the photoacoustic effect. This localized ultrasonic source produces a sinusoidal pressure variation that drives the microbubble into oscillation.

Microbubble oscillations were detected through forward light scattering using a continuous wave laser operating at a wavelength of 488 nm. The laser was sent through the microscope and centered on the microbubble. At the image plane, the detection laser had a 1/e diameter of 9.5 μm and a power of 2.4 mW. A lens was used to collect light on the distal side of the microbubble and send it to a photodetector. The output of the photodetector was, in turn, sent to a radio frequency (RF) lock-in amplifier, where the magnitude and the phase of microbubble oscillation at the excitation frequency were determined. A glass plate was used to redirect a small portion of the excitation laser power into a second photodetector, the output of which served as the reference signal for the lock-in. In order to measure a resonance curve, the function generator driving the EML was set to a particular frequency, and the magnitude and the phase of the bubble response were recorded. This process was repeated as the excitation frequency was stepped across the frequency range of interest.

The microbubble size was determined through optical microscopy prior to each scan. Temperature was measured using a type-k thermocouple inserted into the microbubble solution between the glass slide and the cover slip and controlled using two 10 W resistive heaters driven by a DC voltage source. We investigated both the steady state and the transient response of the microbubbles. In steady-state experiments, the microbubble solution temperature was increased from 21 to 53 °C in ∼2 °C increments. To allow the lipid shell to equilibrate, measurements did not begin until after the temperature had been held at a constant value for at least 5 min.

Figure 1(b) shows three steady-state resonance curves, giving microbubble radial displacement as a function of frequency, for a 3.25 μm radius microbubble at temperatures of 26, 32, and 37 °C. Here, the excitation laser modulation frequency was swept from 1.0 to 3.0 MHz in 25 kHz steps, and the lock-in amplifier had a time constant of 300 ms with 24 dB/oct roll-off. Each curve was obtained by averaging two frequency sweeps, and the solid lines show a Lorentzian fit to the data. Two effects of increasing the microbubble temperature are immediately evident from Fig. 1(b): the resonance frequency decreased and the radial displacement increased.

In the small-amplitude oscillation regime, the dynamic response of a microbubble can be modeled as a simple harmonic oscillator, where the displacement amplitude and resonance frequency are coupled.¹⁷ Here, we measured the resonance curves as a function of temperature for 22 individual microbubbles ranging from 2.25 to 3.75 μm in radius. For these measurements, the lock-in amplifier had a time constant of 300 ms with an 18 dB/oct roll-off, and the frequency scan was taken over a 2-MHz frequency range in 50 kHz steps. Each resonance curve was acquired within approximately 2 min. From each curve, the maximum displacement (ΔR_max), resonance frequency (f), and full-width-at-half-maximum of the resonance peak (Δf) were recorded. The measurements taken over ∼2 °C temperature increments were combined into a single bin and processed together.

Lipid shell elasticity was determined using the linearized form of the modified Rayleigh-Plesset model from the study by Marmottant et al.^18,19 The shell elasticity (χ) for small amplitude oscillations can be found by

where $ρL$ is the density of surrounding liquid, R is the microbubble resting radius, $ζ$ is the damping ratio, $κ$ is the polytropic exponent for the gas core, $P0$ is the ambient pressure, and $σw$ is the surface tension of the gas-liquid interface. The surface tension term can be assumed to be negligible compared to the shell elasticity of the microbubble and for this case of a stable microbubble in a saturated medium.²⁰ The damping ratio is given by $ζ=Δf/2f$⁠. Equation (1) includes a correction for the expected 17% decrease in resonance frequency that a microbubble experiences while resting against the cover slip.²¹ Figure 2(a) shows a plot of the shell elasticity (mean ± standard deviation) as a function of temperature. The shell elasticity of 2.3 N/m at 24 °C is consistent with our previous measurement of 2.4 N/m for DPPC-coated microbubbles at room temperature.²² The elasticity decreased linearly with increasing temperature; a linear regression gave a slope of −0.038 N/m per °C. We note that the steady-state elasticity measured here may depend on the history-dependent microstructure and therefore does not necessarily represent the value at thermodynamic equilibrium of the surface lipids. However, the general trend is consistent with the force spectroscopy study of Grant et al.,¹⁵ which showed a decrease in compressional stiffness with temperature. We note that there was also a decrease in microbubble stability at temperatures exceeding 41 °C, the main phase transition temperature of DPPC,²³ for the steady-state response experiments. Finally, the relatively large standard deviations in Fig. 2(a) are an indication of the bubble to bubble variations of measured shell elasticity.

The absorption of the 1550 nm wavelength light produces a temperature field that has both a DC component and a modulated component. In the MHz frequency range, the modulated component decays rapidly in the vicinity of the source. The DC component, on the other hand, produces a local temperature rise at the microbubble position 100 μm away from the source. This temperature increase was estimated by measuring the shell elasticity as a function of generation laser power, in the absence of resistive heating, in the low power regime where a linear response is expected. A linear fit to the data gave a slope of −0.017 N/m per mW. Under uniform heating, the change in elasticity with temperature was −0.038 N/m per °C. From these values, we estimate that the laser source produces a local temperature rise at the microbubble position of 0.45 °C per mW or approximately 2.5 °C for the 5.5 mW power used in the experiments. This temperature rise is accounted for in all the reported results. We note that this measurement approach assumes that the temperature dependent elasticity is known and thus suffers from uncertainty associated with variations of this value within the microbubble population.

Lipid shell viscosity was derived from the total damping ratio²² and is plotted against temperature in Fig. 2(b). Our measured shell viscosity of (1.26 ± 0.62) × 10⁻⁸kg/s at 26 °C compares well with that in the study by Meer et al.,¹⁹ who reported shell viscosities ranging from 2.0 × 10⁻⁹ to 8.0 × 10⁻⁸kg/s from larger amplitude oscillations tracked using high-speed microscopy at room temperature. As with elasticity, the lipid shell viscosity also showed a linear decrease with increasing temperature; a linear regression gave a slope of −0.22 × 10⁻⁹kg/s per °C. The decreasing viscosity trend with temperature is consistent with the findings from the study by Kim et al.,²⁴ who showed by micropipette aspiration that lipid shells are less viscous when closer to their main phase transition temperature. The trend of decreasing elasticity and viscosity with temperature is consistent with thermal expansion of the monolayer and a corresponding reduction in short-range intermolecular cohesion forces between the lipid constituents within the monolayer plane.²² 

Figure 2(c) shows the maximum radial displacement (ΔR_max) as a function of temperature. The value of ΔR_max was obtained from the maximum amplitude voltage signal from the resonance curves by the use of a calibration value of 0.165 pm/μV, following the procedure of Dove et al.²⁵ The radial displacement shows a monotonic increase with temperature. The increasing radial displacement with temperature arises from two effects: (1) an increase in the driving pressure associated with the temperature dependent Grüneisen parameter and optical absorption of water and (2) a softening of the lipid shell. The relative role of each mechanism can be elucidated using a simple harmonic oscillator model, where the maximum radial displacement is given by

where P_a is the acoustic drive pressure. The parameters $ζ$, f, and R were determined experimentally. The ambient driving pressure P_a was estimated to be 1.4 Pa using the mean experimentally measured parameters at 22 °C (ΔR_max = 8.4 pm, $ζ$ = 0.099, f = 2.92 MHz, and R = 2.45 μm). Within the measured temperature range, the Grüneisen parameter is expected to increase by 4.25%/°C,^26–29 and the infrared laser absorption coefficient is expected to decrease by 0.57%/°C.³⁰ Taking into account these two effects, the acoustic drive pressure increases by 0.046 Pa/°C. The dotted line in Fig. 2(c) shows the predicted max radial displacement for a 2.45 μm radius microbubble with a constant resonance frequency and damping ratio (f = 2.92 MHz and $ζ$ = 0.099) and a temperature-dependent P_a. The solid curve in Fig. 2(c) shows the predicted max radial displacement accounting for the temperature dependence of P_a, f, and $ζ$⁠, where f(T) and $ζ$(T) were found through a linear fit to the measured data and have slopes of −17.07 kHz/°C and −4.7 × 10⁻⁴/°C, respectively. Comparing the two curves in Fig. 2(c) with the experimental results, it is clear that both shell softening and increased drive pressure play important roles in amplifying the microbubble response with increasing temperature.

In transient experiments, a rapid increase from ambient temperature (27 °C) to slightly above body temperature (38 °C) was applied, and the response was recorded for 20 individual microbubbles, ranging in radius from 2.25 to 3.75 μm. A typical transient response of a single microbubble, showing the time-dependent radius and shell elasticity, is shown in Fig. 3. The dotted vertical lines indicate when heating began and when 38 °C was reached. Prior to heating, at an ambient temperature of 27 °C, the radius and shell elasticity were relatively constant. Once heating began, however, an immediate increase in the bubble radius and a decrease in shell elasticity were observed. Microbubble growth was likely a combination of thermal expansion of the gas core and an influx of dissolved gas from the surrounding fluid.³¹ However, the ideal gas law predicts an increase of only ∼4% in volume upon increasing temperature from 27 to 38 °C, and thus gas diffusion from the medium is likely the dominant mechanism for this stage of the process. When the microbubble radius was near its maximum, the shell elasticity reached a valley at ∼0.3 N/m. This behavior is indicative of a ruptured lipid shell.^18,31 When the temperature plateaued at 38 °C, the bubble radius steadily decreased back to its starting radius due to gas dissolution driven by an increase in Laplace pressure. As the microbubble decreased in size, the lipid shell elasticity increased and plateaued as the lipid reformed into a complete monolayer.³¹ In this case, the elasticity increased up to ∼1.8 N/m, which was less than the starting elasticity at ambient temperature. The average shell elasticity values at ambient temperature and 38 °C for the transient response experiments (2.22 ± 0.19 N/m and 1.66 ± 0.15 N/m, respectively) matched well with the steady-state shell elasticity values at the same temperatures in Fig. 2(a). We also observed a decrease in microbubble stability when the bubbles were held at 38 °C, where 11 of the 20 microbubbles continued to show a slow decrease in radius with time. Interestingly, the elasticity plateaued at the steady-state value, even as the radius decreased below the initial size and the monolayer film likely had coexisting collapse structures.

In conclusion, we have demonstrated the feasibility of using a photoacoustic technique to probe the mechanical properties of single lipid-coated microbubbles as a function of temperature. Picometer-amplitude bubble oscillations were produced by laser generated ultrasonic waves and detected by forward light scattering. The narrow-bandwidth nature of the technique allows for highly sensitive resonance measurements. We demonstrated that slow heating of a microbubble leads to an increase in maximum radial displacement owing to a decrease in lipid shell viscoelasticity and an increase in photoacoustic driving pressure with temperature. In addition, rapid heating of a microbubble resulted in a transient response where the lipid shell underwent rupture and subsequent reformation. These results will be useful in furthering the understanding of microbubble dynamics for in vivo applications.

This work was supported by the National Institutes of Health Research Project Grant (R01CA195051) and the National Science Foundation under Grant No. CMMI 1335426.