Introduction

Microscale bubbles have found many applications in medical imaging1, biomedical analysis2, drug delivery3 and microfluidic elements4,5,6,7. Various methods have been adopted to generate microbubbles, including ultrasonic cavitation8, laser induced cavitation9,10, microfluidic flow-focusing11 and liquid boiling12. Photothermal heaters produced by laser light combined together with photothermal materials can carry on localized heating to initiate the phase transition of the liquid13,14,15,16,17,18. Previous reports have shown that the microbubbles can be successfully produced by using highly focused laser beams or optical fiber to illuminate photothermal substrates13,14,15, liquids16 and nanoparticles17,18. Compared with other systems8,9,10,11, photothermal system allows for accurately controlling the forming location and the status of microbubbles. The microheater has a dominant influence on the growth and movement of microbubbles. For example, it has demonstrated that the growth rate of microbubbles can be tuned by the power of the microheater15,16. The microbubbles are attracted towards the microheater by thermocapillary forces and Marangoni forces14,19. So far, the reported photothermal microbubbles were generated directly on the microheater13,14,15,16,17,18. Whether the microbubbles can be emerged away from the microheater? It is an interesting topic, which has not been detailedly explored.

Our previous study has been demonstrated a novel microscale photothermal heater based on the combination of a microwire and graphene oxide (GO)20. GO nanosheets (GONs) in N,N-Dimethylformamide (DMF) solvent were successfully trapped and deposited on the surface of the microwire with enhanced evanescent fields21,22,23,24 by injecting near-infrared light into the microwire. The GO-deposition exhibited strong photothermal properties25,26,27,28,29, serving as a line-shaped heater with a length of several hundred micrometers. When the superheat limit of liquid was reached, a lot of spherical microbubbles with diameters from tens to hundreds of micrometers were generated directly on the GO-microheater immersed in DMF, which is a common phenomenon in a traditional photothermal system13,14,15,16,17,18. The present work is devoted to novel microbubbles with approximately ellipsoidal shapes that produced on the smooth microwire rather than on the microheater. Similar to liquid drops on thin wires30,31, each microbubble exhibited a symmetrical conformation with respect to the microwire axis. Bubble behaviors induced by the microheater including constant growth, directional transport and coalescence were investigated experimentally and theoretically, showing the similar behaviors to the reported photothermal microbubble14,19. The existence of the approximately ellipsoidal microbubbles will gain new insight into the mechanism of bubble formation and bubble dynamics at the microscale.

Results

Figure 1a shows schematic illustration of experimental setup to obtain GO-microheater. An inverted optical microscope with a charge-coupled device (CCD) camera was used for real-time monitoring. A liquid cell was mounted on an x-y manual translation stage of the microscope. A 2.2-μm diameter, 2.0-mm long microwire was fabricated by drawing a single mode optical fiber (SMF-28, Corning Inc.) using two stepper motors through the flame-heated technique. The microwire was immersed in the GONs-DMF suspension and fixed by two microadjusters. Figure 1b shows a scanning electron microscopy (SEM) image of 2.2-μm-diameter wire with good surface smoothness. An amplified spontaneous emission broadband light source (ASE, ~20 mW, 1527–1566 nm) was excited by an erbium-doped fiber amplifier (EDFA, 1546–1562 nm), whose output power was 40 mW at 1527–1566 nm. Light from EDFA was coupled into the microwire and output spectra were recorded by an optical spectrum analyzer (OSA, Yokogawa AQ 6370C) with a resolution of 0.02 nm. Transmission electron microscopy (TEM, Fig. 1c) and SEM (see supplementary information, Fig. S1) images show that GONs have scale-like shape with micrometer long wrinkles. In the experiment, the concentration of GONs-DMF suspension was optimized at 0.05 mg/ml. Since a higher concentration could affect imaging due to absorption and a lower concentration would limit the quantity of GO-deposition.

Figure 1
figure 1

(a) Schematic of experimental setup. An inverted optical microscope with a CCD camera was used for real-time monitoring. The ASE source excites the EDFA and then light from EDFA is coupled into the microwire. (b) SEM image of 2.2 μm-diameter microwire shows good surface smoothness. (c) TEM image of GONs shows a scale-like structure. (d) A 2D field profile of 2.2 μm-diameter microwire under fundamental modes (HE11) at the wavelength of 1550 nm. (e-f) Optical microscope images of the GO-deposition with light launched for 0, 30 s and 60 s, respectively.

Full size image

Mode characteristics of the microwire were analyzed in the supplementary information, demonstrating a 2.2-μm-diameter wire submerged in DMF produced a tight field confinement and strong evanescent field. Figure 1d shows a 2D field profile of the HE11 mode at wavelength of 1550 nm. Effective diameter of mode field (Deff) and power ratio outside the microwire (η) of a 2.2-μm-diameter wire were 12.23–13.76 μm and 86.59–88.32% at wavelength of 1527–1566 nm (see supplementary information, Fig. S1). When the light propagated along the microwire, the evanescent field outside the microwire was absorbed by GONs, producing a thermal gradient based on effective optical-to-thermal energy conversion of GONs20. And then a weak convective flow was generated, driving the randomly dispersed GONs to swim into the evanescent field near the microwire32,33. Subsequently, the GONs near the microwire were trapped and deposited on the microwire by an optical gradient force. The process of the deposition would go on until the laser was turned off. Figures 1e–h show optical microscope images of the deposition with light launched for 0, 30 s and 60 s. Finally, a GO-microheater with length of 282 μm and maximum width of 75 μm was formed within 3.5 min. Note that it is not necessary to form the GO-microheater in certain location of the microwire. The location of the microheater in the liquid can be adjusted by tuning microwire through two microadjusters.

In the experiment, the DMF liquid surface was convex due to surface tension34. When the microwire was immersed into the liquid, it could be divided into three sections (AE, ED and DF as indicated in Fig. 2a) under capillarity effect34. The two microwire segments AE and DF were located at DMF-air interface while the microwire segment ED including GO-microheater BC was immersed in the GONs-DMF suspension. Based on our previous study, the working GO-microheater was capable of initiating the phase transition of its surrounding DMF. As a result, spherical microbubbles were generated directly on the GO-microheater (see supplementary information, Fig. S2). By tuning microwire through two microadjusters, the microwire segment A′C including the GO-microheater BC and microwire segment D′F′ were located at the DMF-air interface and microwire segment CD′ was immersed in the GONs-DMF suspension (as shown in Fig. 2b). In this case, it was difficult to generate microbubbles on the GO-microheater, which was suspended at the DMF-air interface. Instead, a series of microbubbles were produced along a horizontal microwire axis on right side of the GO-microheater. As shown in Fig. 3, each microbubbles exhibited a symmetrical conformation with respect to the microwire axis and took a shape of approximate ellipsoid. Similar to liquid drops on thin wires30,31, the microbubble conformation could be characterized by two parameters: long axis (L) and short axis (S), which were measured horizontally and vertically with respect to the microwire.

Figure 2
figure 2

(a) Before adjusting, the two microwire segments AE and DF were located at the DMF-air interface and the microwire segment ED including GO-microheater BC was immersed in the GONs-DMF suspension. (b) After adjusting microwire, the microwire segment A′C including the GO-microheater BC and the microwire segment D′F′ were located at the DMF-air interface and microwire segment CD′ was immersed in the GONs-DMF suspension. Red arrows show the direction of light propagation.

Full size image
Figure 3
figure 3

Optical microscopy images for the growth process of microbubble A0 at a magnification of 100× after the light was turned on (a) at t = 08′09″, (b) t = 08′14″, (c) t = 08′32″, (d) t = 09′08″, (e) t = 09′31″, (f) t = 10′00″, (g) t = 10′53″, (h) t = 10′54″, detailed process from t = 08′09″ to 10′54″ was shown in movie (Movie S1, supplementary information). Red and blue arrows represent the directions of light propagation and microbubbles movement, respectively. The scale bar represents 100 μm.

Full size image

Growth, directional transport, disappearance, emergence and coalescence were the types of dynamic microbubble behaviors, visually observed along the microwire. Figure 3 illustrates a growth period of microbubble A0 closest to the GO-microheater at a magnification of 100× (with frames taken from movie S1). Microbubble A0 emerged at t = 08′09″ in Fig. 3a and kept growing until it reached a maximum size at t = 10′53″ (Fig. 3g) and then exploded. Afterwards, a new small microbubble A0′ with size of L = 24 μm, S = 14 μm was generated at t = 10′54″ (Fig. 3h). Table 1 shows that the sizes (L, S) of microbubble A0 are (22 μm, 14 μm), (38 μm, 26 μm), (61 μm, 47 μm), (81 μm, 71 μm), (97 μm, 83 μm), (109 μm, 95 μm) and (143 μm, 128 μm), respectively, at t = 08′09″, 08′14″, 08′32″, 09′08″, 09′31″, 10′00″ and 10′53″. At the same time, other small microbubbles grew large and moved towards GO-microheater which was opposite the direction of the light propagation. Meanwhile, the GO-microheater expanded towards the left along the microwire. This is because the optical trapping occurred at the left side of GO-microheater where the microwire was not covered by GONs and remained a strong optical gradient force. Table 1 shows that the lengths of the GO-microheater are 300, 312, 313, 319, 322, 339, 341 and 351 μm, respectively, at t = 08′09″, 08′14″, 08′32″, 09′08″, 09′31″, 10′00″, 10′53″ and 10′54″, respectively, while the width did not noticeably change.

Table 1 The sizes (L, S) of microbuble A0 and the lengths of the GO-microheater at t = 08′09″, 08′14″, 08′32″, 09′08″, 09′31″, 10′00″ and 10′53″, respectively
Full size table

Other dynamic behaviors such as coalescence and transport were observed. As shown in Fig. 4 (which contains frames from movie S2 at magnification of 250×), microbubble A1 gradually grew up, staying motionless. Figure 4a–4d illustrate the orderly movement of the microbubbles B2, B3 and B4. It can be seen that there was a vacancy between microbubble B1 and B2 at t = 04′24″ (Fig. 4a). At t = 04′25″, microbubble B2 moved to the left by 23 μm, occupied the vacancy and left another vacancy (Fig. 4b). Following, microbubbles B3 and B4 in turn moved to the left by 20 μm and 22 μm at t = 04′26″ and t = 04′27″, respectively (Fig. 4c and 4d). Figure 4e and 4f describe collective disappearance and emergence of the microbubbles. At t = 04′28″, microbubbles B2, B3 and B4 disappeared (Fig. 4e). At t = 04′29″, new microbubbles B6, B7, B8 and B9 emerged at the right side of microbubble B1 (Fig. 4f). Figure 4g–4l show coalescence of the microbubbles. At t = 04′39″, five microbubbles C1, C2, C3, C4 and C5 were closely arranged along the microwire (Fig. 4g). At t = 04′40″, the C5 with size of (19 μm, 14 μm) moved towards C4 (16 μm, 10 μm) and coalesced into a larger microbubble C6 (21 μm, 16 μm) in Fig. 4h. And then, all microbubbles stayed at their positions at t = 04′41″ and t = 04′42″ (Fig. 4i and 4j). After that, microbubble C3 (24 μm, 16 μm) moved towards C2 (24 μm, 18 μm) and then coalesced into a larger microbubble C7 (31 μm, 21 μm) at t = 04′43″, meanwhile, microbubble C6 moved left by 23 μm (Fig. 4k). Following, microbubble C7 moved towards C1 (28 μm, 17 μm) and then coalesced into a larger microbubble C8 (36 μm, 26 μm), while microbubble C6 remained motionless (Fig. 4l). This shows that coalescence of microbubbles is one of the reasons for the growth of the microbubbles.

Figure 4
figure 4

Optical microscope images of dynamic processes of microbubbles at a magnification of 250×, detailed process from t = 02′29″ to t = 04′46″ is shown in movie (Movie S2, supplementary information).

(a)-(d) Microbubbles B2, B3 and B4 orderly moved to the left from t = 04′24″ to 04′27″. (e) Microbubbles B2, B3, B4 and B5 disappeared at t = 04′28″. (f) New microbubbles B6, B7, B8 and B9 were generated at t = 04′29″. (g) At t = 04′39″, the microbubbles were at their original positions. (h) At t = 04′40″, C4 and C5 mixed together to form C6. (i), (j) At t = 04′41″ and 04′42″, all microbubbles remained the same with those of at t = 04′40″. (k) At t = 04′43″, C2 and C3 mixed together to form C7. (l) At t = 04′44″, C1 and C7 mixed together to form C8. The scale bar represents 50 μm.

Full size image

Discussion

As described in section 3, the GO-deposition as a line-shaped microheater was located at the DMF-air interface. After a period of light illumination within 60 s, the temperature on the GO-microheater was estimated to be 425.95 K, the boiling point of the DMF. When the GO-microheater operated, the temperature distribution in the surrounding DMF was governed by the thermal transfer equation

where k is thermal conductivity (W/(m·K)), T is temperature (K), ∇T is temperature gradient (K), Cp is heat capacity (J/(kg·K)) and Q (W/m3) represents a source term such as GO-microheater35. The temperature field in the GONs-DMF suspension was simulated using finite-element modeling software in COMSOL Multiphysics 4.3 (supplementary information, Fig. S3). Figure 5a shows the partial temperature distribution around the GO-microheater in a rectangular area (in Fig. S3b) circled by lines y = ±300 μm, x = 0 and x = 1000 μm. The temperature was from 364.53 K to 425.95 K. Figure 5b shows the vertical temperature distribution for x = 400, 450, 500, 600 and 800 μm. It can be seen that the vertical temperature distribution is symmetric with respect to the axis of the microwire (y = 0). Figure 5c shows the horizontal temperature distribution for y = 0, ±10, ±30, ±50, ±70 and ±100 μm. It is clear that temperature decreases with the distance from the GO-microheater. The absolute value of slope decreases with the distance from the GO-microheater in Fig. 5c, producing a decreased temperature gradient along the microwire axis from the left to the right.

Figure 5
figure 5

(a) Simulation of temperature distribution driven by the GO-microheater. Black arrows show the direction of temperature gradient. The isotherms are indicated by gray lines. (b) The vertical temperature distribution on lines x = 400, 450, 500, 600 and 800 μm. (c) The horizontal temperature distribution on lines y = 0, ±10, ±30, ±50, ±70 and ±100 μm. The inset of Fig. 5b shows the amplified temperature distribution from x = 450 to 480 μm.

Full size image

The approximately ellipsoidal microbubbles were generated on the microwire away from the microheater. Here the physical mechanism of the microbubble generation can be explained by heterogeneous nucleation. As one of the most important processes in bubble nucleation, heterogeneous nucleation normally occurs at phase boundaries. At these sites, the effective surface energy is lower, which diminishes the free energy barrier and finally facilitates nucleation36,37. There were DMF vapors in the liquid under the GO-microheater. The microwire surface was the phase boundary in this system, so it could provide preferential sites to facilitate nucleation. Meanwhile, the solubility of the DMF vapors decreased with the increasing temperature in the suspension. The DMF vapors were persistently separated and finally integrated into the nucleus at the microwire segment CG in Fig. 2b. The overall result is that the heterogeneous nucleation and vapor integration act cooperatively to facilitate the generation and growth of microbubbles. In addition, we observed that there were other sites on the microwires segment GD′ (in Fig. 2b) which can facilitate heterogeneous nucleation. However, these nuclei failed to grow up because they were farther away from the GO-microheater (supplementary information, Fig. S4). Therefore, the microbubble growth is linked closely with the temperature. Microbubbles would sometimes translate along the wire once growth stopped. After the microbubble vanished from its location induced by explosion, movement, or coalescence, a new one was formed and circulated the process mentioned above.

The movement of the microbubbles was dominated by driving forces and resistance forces, as shown in Fig. 6. One of possible driving mechanisms comes from thermocapillary effect38,39. As described in Fig. 5a, there is a temperature gradient along the interface of the microbubble. In DMF liquid, surface tension decreases linearly with the increasing temperature40,41. So it is accompanied by a corresponding surface tension gradient at the microbubble interface, which induces the resulting interfacial shear stress τS. The interfacial shear stress is proportional to the temperature gradient ∇TS and described by τS = −σT · ∇TS, where σT is the temperature coefficient of surface tension42,43. As illustrated in Fig. 6, the difference in interfacial shear stress drives thermocapillary flow along the microbubble interface (indicated by red arrow), which is oriented from low to high surface tension and opposite to the temperature gradient. The reverse force of the momentum transfer induced by the thermocapillary flow causes the microbubble to move toward high temperature regions, such as the GO-microheater. The thermocapillary force can be expressed by the surface integral of shear stress on the microbubble interface38,39:

where is the temperature gradient along the microbubble interface, dS is the surface element on the microbubble interface.

Figure 6
figure 6

Mechanism of a microbubble movement on the microwire.

There are driving forces (indicated by the black arrow) and resistance forces (indicated by the green arrow) on the microbubble. The driving force comes from thetemperature gradient along the interface of microbubble, which can be thought of as a thermocapillary force arising from thermocapillary flow indicated by red arrows and a non-equilibrium Laplace pressure difference.

Full size image

Another possible driving mechanism arises from the non-equilibrium Laplace pressure difference44. According to the Young-Laplace equation, the Laplace pressure difference on the two sides of the microbubble along microwire can be expressed analytically as34

where ΔPL and ΔPR are the Laplace pressure differences and σ(TL) and σ(TR) are the surface tensions on the left and right side of the microbubble, respectively. R1 and R2 are the radii of the curvature of the surface. As presented in Fig. 6, the Laplace pressure difference on the right side (ΔPR) is larger than that on the left side (ΔPL) because the surface tension on the right side σ(TR) is larger than that on the left side σ(TL). The resultant non-equilibrium Laplace pressure difference can propel the microbubble to move toward the region with lower Laplace pressure difference44.

Meanwhile, there are several resistance forces such as viscous force from DMF liquid and friction force from microwire. Consequently, various motion states were as a result of the interaction of these forces. Several microbubbles remained stationary on the microwire when the driving forces couldn't overcome the resistance forces (such as microbubble A1, A2, B1 and B5 shown in Fig. 4a–f). Others kept on moving towards the microheater when the driving forces dominated over the resisting forces (such as microbubbles B2, B3 and B4 in Fig. 4a–d). Two adjacent microbubbles could have differing translation rates, causing a collision with each other and finally coalescence into a bigger bubble. As an example, microbubble C2 and C3 mixed together to form C7 in Fig. 4k–l. According to the statistics from movie S2, there are altogether 66 microbubbles generated from t = 02′29″ to t = 04′46″, among which, 46 ones coalesced or transported to the microheater, 2 ones transported away from the microheater and 18 ones remain stationary. Figure S5 (supplementary information) shows the details on transport and coalescence of microbubbles generated on another microwire with diameter of 7.0 μm, demonstrating that the phenomena can be repeatable.

Our observations show that the temperature distribution induced by GO-microheater plays a significant role in the dynamic behaviors of approximately ellipsoidal microbubbles. After the light source was cut off, the temperature on the microheater was gradually decreasing due to Newtonian heat flow and no more source term. Consequently, it was unable to provide adequate heat to maintain the dynamic behaviors of microbubbles and then microbubbles started to gradually disappear and/or shrink. The temperature was initially lower at further distances from the GO-microheater, which resulted in the faster disappearance of microbubbles. As shown in Fig. 7a–7f, at first, five microbubbles D3, D4, D5, D6 and D7 started to shrink gradually until they disappeared, while D1 and D2 remained unchanged with sizes (L, S) of (72 μm, 47 μm) and (55 μm, 45 μm), respectively, as shown in Fig. 7a and 7b. The sizes of microbubbles varied from D3 (24 μm, 14 μm), D4 (26 μm, 15 μm), D5 (26 μm, 18 μm), D6 (29 μm, 18 μm), D7 (31 μm, 19 μm) to D3 (23 μm, 14 μm), D4 (25 μm, 13 μm), D5 (19 μm, 12 μm), D6 (17 μm, 10 μm), D7 (20 μm, 10 μm), respectively. And then, the left two microbubbles D1 and D2 started to shrink after the other microbubbles disappeared. D2 had the quicker velocity of disappearance due to the faster decreasing temperature of D2. The sizes of D1 were (63 μm, 51 μm) in Fig. 7c, (58 μm, 47 μm) in Fig. 7d, (41 μm, 33 μm) in Fig. 7e, (23 μm, 15 μm) in Fig. 7f, respectively. In the experiment, we found that D2 disappeared at t = 16′31″. The sizes of D2 in Fig. 7c and 7d were (41 μm, 33 μm) and (23 μm, 15 μm), respectively.

Figure 7
figure 7

Optical microscope images for disappearance process of microbubbles at magnifications of 100× (a–d) and 400× (e, f), respectively, when the light was turned off at t = 11′49″ (a), 12′25″ (b), 14′41″ (c), 15′17″ (d), 17′24″ (e) and 20′20″ (f).

Full size image

In conclusion, we have presented that various dynamic behaviors of microbubbles with novel conformation under the GO-microheater system. The surrounding liquid was heated to form a temperature gradient field by GO-microheater based on the photothermal effects. A series of approximately ellipsoidal microbubbles were formed and grew symmetrically along the microwire axis on the right side of the GO-microheater based on heterogeneous nucleation. Meanwhile, several microbubbles were found to continuously move towards the GO-microheater or coalesce together under the temperature gradient field. We contributed the driving force of directional transport to thermocapillary effect or non-equilibrium Laplace pressure along the microbubble. When the light was turned off, microbubbles stopped moving and started to shrink gradually. The results revealed that GO-microheater played a significant role in the dynamic behaviors of approximately ellipsoidal microbubbles. It has been demonstrated a new experimental case on microbubble generation with other conformations, which would be not only helpful for understanding the bubble dynamics but also for development of novel photothermal bubble-based devices. However, the detailed mechanism on the symmetrical conformation of the microbubble is requiring further exploration.

Methods

Fabrication of optical microwire

A microwire was fabricated by drawing a single mode optical fiber (SMF-28, Corning Inc.) using two stepper motors through the flame-heated technique45.

Preparation and characteristics of GO

Graphite powders (spectral pure) were purchased from Sinopharm Chemical Reagent Co., Ltd. The reagents and solvents were obtained from commercial suppliers. All aqueous solutions were prepared with ultrapure water (>18 MΩ) from a Milli-Q Plus system (Millipore). GO was prepared by oxidizing natural graphite powder based on a modified Hummers method46,47. As-prepared GO was dialyzed to remove the residual salts and acids completely and the resulting purified GO powder was collected by centrifugation and then dried in the air. GO powder was suspended in ultrapure water (0.5 mg/ml) and exfoliated through ultrasonication in a water bath (KQ218, 60 W) for 3 hours. As a result, the bulk of GO powder was transformed into GONs.

GONs were characterized using SEM, Fourier transform infrared (FTIR) spectra and UV/vis absorption spectra (see supplementary information, Fig. S6–S9). Transmission electron microscopy (TEM, Fig. 1c) and SEM (see supplementary information, Fig. S6) images show that GONs have scale-like shape with micrometer long wrinkles. The dried GO was transformed into GONs through ultrasonication for 3 hours in a water bath and well dispersed in DMF due to the presence of oxygen-containing groups in GONs48,49. The near-infrared (NIR) absorption spectra are sensitive to the concentration of GONs in the DMF solvent (see supplementary information, Fig. S6). Inset of Fig. S9 shows relationship between the absorbance and the concentration of GONs dispersion at the wavelength of 1550 nm. It can be seen that the absorption grows linearly with the concentration.