Cong
Zhai
ab,
Chunguang
Hu
*ab,
Shuai
Li
ab,
Yanhua
Ma
ab,
Yajing
Zhang
ab,
Tong
Guo
ab,
Hongbin
Li
ac and
Xiaotang
Hu
ab
aState Key Laboratory of Precision Measuring Technology and Instruments, Tianjin University, Tianjin 300072, China. E-mail: cghu@tju.edu.cn
bNanchang Institute for Microtechnology of Tianjin University, Tianjin 300072, China
cDepartment of Chemistry, University of British Columbia, 2036 Main Mall, Vancouver BC V6T 1Z1, Canada
First published on 16th November 2020
Utilizing droplets as micro-tools has become a valuable method in biology and chemistry. In previous work, we have demonstrated a novel droplet generation–manipulation method in a conventional optical tweezer system. Here, a further study of the droplet composition and its formation principle is performed. First, it is proved through Raman spectra that the principal component of the droplets is HPO42− solution. Considering that the generated droplet size is at the μm level, we have adopted a variety of methods in experiments to reduce external interference. Second, using a confocal microscopic video camera, the growth process of the droplet is completely recorded in a common glass-based chamber. The finite element simulations help us to further understand that the droplet generation process using optical tweezers can be divided into two stages: “capture” caused by optical force field and “aggregation” induced by a photothermal phenomenon and thermal acceleration. Through these studies, the nature of the optical tweezer-generated droplets is revealed. As a general principle for the droplet generation, this method will provide inspiration and prospects in the fields of microfluidics and biophysics-chemistry.
In fact, this droplet generation method was discovered accidentally in an optical tweezer experiment. Taking advantage of high-precision controllability of optical tweezers,14,15 this method can generate, manipulate, and transport micro-droplets in a mixture of inorganic phosphate buffer saline (PBS) and isopropanol (IPA). Although the necessary conditions of droplet generation have been discussed in previous research, there are still basic principles unknown, such as the droplet composition and its formation mechanism. Revealing these fundamental problems will help us to understand the physical and chemical nature of optical tweezer based droplet generation. But there are several challenges. For example, the electrochemical analysis of micro-droplets is obstructed by the tiny size of the droplet and the space of microfluidic chips. It is also disturbed by the strong noise signals from the surrounding solvent. In addition, an advanced measurement with a precise and flexible operation is desired for the investigation of the droplet formation process.
In this paper, we focused on the nature of the micro-droplets and verified the formation process. First, a confocal Raman microscope was utilized to determine the fingerprint spectra of the generated droplet. Via principal component analysis (PCA) of the spectra of droplets and raw components, the main composition of the droplets was found to be HPO42−, a phosphate solution ionized by Na2HPO4 in PBS. In order to visualize the formation process, a microfluidics design for the sample cell and a video camera were applied, and the finite difference time domain (FDTD) method was applied to simulate the effects of thermodynamic field and optical force field induced by highly focused optical tweezers. The simulation results show that multiple physical fields are the key factor for the generation of droplets. Finally, by combining the experimental phenomena with theoretical analysis, we propose an explanation for the physical mechanism of the droplet formation.
Studies suggest that the deformability of the droplet could be affected by the salinity of the aqueous solution,19 thus the volume ratio of IPA to PBS could directly influence the efficiency of the optical tweezer based droplet generation. To meet the consistency of the experiment, we select the same volume ratio consistent with the previous study (4:1 for IPA:PBS).13 Therefore, the 1 ml of PBS is first diluted with 4 ml of high-purity IPA, and subsequently loaded into the injector as the typical raw aqueous sample in this study. Apart from these, no other ingredients are used.
After turning on the laser, the laser beam is emitted through the objective and focused in the centre of the sample chamber, which is held on a precision displacement platform by using a fixing bracket. As shown in Fig. 1b and c, the droplets will appear under the irradiations of the highly focused lasers when the raw sample is injected into the sample chamber by using a catheter (the detailed structure of the sample chamber is shown in Fig. S1 in the ESI†). The sum of droplets depends on the number of active light traps, and the size of the droplets relies on the amount of salt solution injected into the sample chamber. In particular, the two little droplets can converge into one large droplet easily if they are close enough. This means that the generated droplets entirely consist of liquid, bringing more control possibilities to the novel application of droplets. In order to reveal the droplet formation process simply and clearly, the subsequent experiments and analysis are based on a single optical trap.
Fig. 3 lists the re-projected Raman spectra of the droplets and the re-projected spectra of IPA, Na2HPO4, KH2PO4, NaCl, and KCl, respectively. The last four are the specific components of PBS. As Fig. 3a–d show, there is a high similarity between the droplets, Na2HPO4 and KH2PO4. But the cumulative intensity of NaCl and KCl spectra is not obvious as shown in Fig. 3e and f. This is because for a molecule to exhibit a Raman effect, its electric dipole–electric dipole polarizability must change relative to the vibrational coordinate corresponding to the rovibronic state, which is almost non-existent in NaCl and KCl as well as metal ions in aqueous solution. In fact, it has been demonstrated in our previous study that the single NaCl or KCl solution cannot converge into droplets under the light trap because it cannot produce an effective relative refractive index with the medium solution. Hence, these two components will not be considered in a following analysis.
Since a Raman spectrometer is used to measure scattered light of samples, the intensity in the liquid solution is relatively lower than crystal. In order to get a cleaner signal, the top coverslip is removed and the deposited droplets are dried in a dust-free environment for improving measurement sensitivity (Fig. S2c in the ESI†). Fig. 3a shows the spectral comparison of droplets before (blue line) and after drying (red line). Although there is a slight difference in the shape of the wave crests, the two spectra still represent the same thing. This is because the interatomic distance of the salt will be affected by the quantity of water when the concentration of the salt solution changes, and the bonding between atoms will change accordingly and reflect on the spectrum.21,22 In addition, it could be found through dividing the spectrum that the seemingly different waveforms are actually synthesized by four wavelets with the same Raman shift (more analyses are shown in Fig. S2†). This proves that the drying of droplets can significantly enhance the Raman intensity and sharpen the wave crests, while not affecting the corresponding Raman shift of the characteristic crests.
To further determine the droplet composition from the similar waves (Fig. 3b–d), PCA is implemented and the correlation between the spectra of phosphates, IPA, and droplets is quantified,23 as shown in Fig. 4 (the specific process is shown in the Appendix). The pairwise comparison method is first adopted to analyse the correlation between the droplets and all known components in PBS. In this process, the Raman spectra of the droplets and other components are compared respectively. For instance, the PCA in Fig. 4a involves the Raman spectra of droplets before and after drying. The abscissa indicates the correlation between all sample points from the spectra with the separated primary component, and the ordinate is another correlation with the secondary component. Because the spectra involved in this comparison represent the same component and the drying of the droplets does not significantly affect its spectral signature, the result shows that all the sample points converge with respect to the vertical axis, and the contribution rate of the primary separated eigenvectors (PC 1) reaches 97.8%. The same phenomenon is also shown in Fig. 4b. When the PCA method only involves the spectra of dried droplets and Na2HPO4, the correlation between them is also as high as 95.9%, which is much larger than that of the droplets between IPA, KH2PO4, respectively. To some extent, the phenomena in Fig. 4a and b can be regarded as the spectra participating in this PCA process representing the same substance.
In addition, the correlation between the dried droplets and KH2PO4 is 80.9%, which is tremendously similar to the correlation between Na2HPO4 and KH2PO4 (81.4%, Fig. S3e in the ESI†). Since there are no other similar components in this experiment, it is concluded that Na2HPO4 is the principal component of the dried droplets.
Then, a routine PCA method for the droplets and all components of PBS is executed to determine the composition of the droplet from another aspect. In this process, an equal number of spectra for each component is chosen to constitute the multivariate arrays. Three effective feature vectors are separated through the dimension reduction of PCA, as shown in Fig. 4e. It is clear that all sample points also have a good aggregation on the plane corresponding to the PC 1 zero axes. The contribution ratios of the primary and secondary components to the entire eigenvector matrix are 81.6% and 15.5%. This significant difference in the contribution rate confirms another aspect that the main component of the dried droplets is Na2HPO4. Although HPO42− and H2PO4− in the solution will be affected by the concentration of hydrogen ions and undergo mutual conversion, an additional procedure shows that only Na2HPO4 solution could generate droplets (Fig. S3 in the ESI†). In this procedure, the pure solutions of NaCl, KCl, KH2PO4 and Na2HPO4 are separately tested under the same conditions without any spare component that could affect the pH value of the solution. The results show that only Na2HPO4 solution can generate droplets under the action of optical traps, further verifying that the main component of droplets is HPO42− solution.
Fig. 5c shows the complete generation process of a droplet (Video 1 in the ESI†). Before the laser on, it is hard to perceive that there are tiny droplets free in the mixed solution. These droplets are probably caused by non-covalent interactions between the part phosphate groups of PBS and the hydroxyl groups of IPA during the mixing process. Such interactions can link molecules together like DNA strands.25 According to similarity–intermiscibility theory, the linked clusters may outline an interface between HPO42− solution and IPA, and shape spherical droplets under the action of liquid pressure (Fig. 5d). The additional dynamic light scattering (DLS, Zetasizer Nano ZS90, Malvern) measurements show that the average diameter of these original tiny droplets is about 900 nm, and the average polymer dispersity index (PDI) is 0.235. Normally, these tiny droplets are evenly distributed in the solution and hard to gather together. When the laser is turned on (frames i and ii), a few free tiny droplets near the light trap are immediately captured under the attraction of optical trapping force, converging into a larger droplet. Then, other tiny droplets from the surrounding solution are continually converged into this optically trapped droplet, as shown in frames iii–v. With the continuous injection of the raw sample, the trapped droplet gradually grows larger and is firmly controlled at the centre of the beam. In less than one minute, the size of the generated droplet has increased to more than 10 μm, showing a good growth rate. During this process, the growth can be stopped by interrupting the injection of raw samples, and the growth rate can also be affected by changing the laser intensity or the concentration of PBS to achieve the size controllability (the relevant research of droplet growth rate can be referred to in our previous research).13
A noticeable phenomenon during the formation process is that some tiny droplets are converged from a long-distance position which is already far beyond the scope of the optical trap. This means that the optical trapping force may be only one of the effects in the formation process of droplets. Therefore, we introduce a series of finite element simulations to reveal the influence of physical fields on the droplet generation process.
The effective “capture” area only exists at the centre of laser focus where there is a huge light intensity28,29 (Fig. 6b). However, the experiment shows that the tiny droplets still can continually flow into the formed droplet when the “capture” area is covered. This is consistent with the speculation in the previous section: the particle nature of light is only one of the elements in the generation of droplets. In addition to the light trap force, there must be other force fields to promote the growth of droplets. Considering the undulatory properties of light,30 the finite difference time domain (FDTD) method and “COMSOL” are operated to simulate the thermal energy of the highly focused beam31,32 and fluidity of liquid samples respectively.33 The reason is that under the laser irradiation, the radiation pressure may heat the liquid solution,34,35 affecting the laminar flow in the chamber.36 Due to the loss of laser in the optical path, the final power of the single optical trap is less than 600 mW in the sample chamber. According to the previous research, the centre temperature of the optical trap is estimated to be 30–40 °C.37
Fig. 6c shows the vertical view of the simulation of liquid flow velocity under focused laser light, and Fig. 6d illustrates the vortex flow trend caused by the radiation pressure at the x–z plane. Obviously, the radiation pressure engenders a series of little vortexes in the microfluidic chip and accelerates the circulation of the liquid to a great extent. That means the flow of tiny droplets from the surrounding solution is also accelerated toward the inside of the light beam with the liquid flow and fused with the generated droplet, which is exactly the same as the formation process in Fig. 5c(iii–vi). This process is defined as “aggregation”. Under the combined action of thermal acceleration and optical trapping force, these free droplets are immediately gathered together and shaped into a single larger droplet due to the liquid surface tension.38 Besides, the grown droplet can change the direction of heat transfer, as Fig. 6e shows, improving the liquid flow efficiency in the sample chamber.
Briefly, in the initial phase of droplet generation, the tiny droplets are mainly attracted and gathering into a larger droplet by the force field of the focused laser. This first gathering method of droplet formation is called “capture”. While the droplet size is larger than the scope of the light trap, the gradient of light intensity is completely wrapped by the liquid and cannot continue to capture the tiny droplet effectively. At this moment, the photothermal effect becomes dominant, heating the mixed solution to achieve thermal acceleration in the chamber. The other tiny droplets could still flow into the formed droplets under the drive of the small vortexes, making the formed droplet continue to grow. As the second stage of the droplet generation, this process is called “aggregation”. Although induced by different multiphysics, both “capture” and “aggregation” are necessary conditions for the droplet generation, completely consistent with the experimental phenomenon in Section 3.2.
From the above research, the essence of optical tweezer based droplet generation is revealed that under the comprehensive action of light traps, the tiny droplets can be gathered together to form the controlled droplets. Relying on the high-precision controllability of optical tweezers, the generated droplets have the capability to replace traditional beads in the transport of large molecules, or control the flow of the microchannels through their own tension. At present, the formation rationale of the tiny droplets, i.e. HPO42− droplets, in the raw sample has not been figured out clearly. But the studies on the interface between IPA and HPO42− solutions and the fusion process at the molecular level will be concentrated in future research.
(1) |
(2) |
After this process, each orthogonal eigenvector is normalized to a unit vector. The leading diagonal elements represent the variance of each axis and the counter-diagonal elements are the covariance matrix of different arrays, each unit eigenvector in the orthogonal can be interpreted as a recombinant axis for the fitted data. At this point, the preparation of the characteristic matrix of components is accomplished.
Next, before acting as the recombinant axis, the eigenvectors need to be sorted in descending order. Then the uniformly quantified data project onto the first recombinant axis which also is the largest, to constitute an eigenvectors matrix. If there are multiple crucial components in the original data arrays, it is necessary to calculate its contribution to the entire eigenvector matrix beginning from the largest eigenvector. When the sum contribution cannot reach the set threshold, choosing the eigenvector in turn and calculating the cumulative contribution until the total contribution reaches the threshold. In this process, the reformed eigenvectors matrix can be expressed as follows:
F(m,k) = U(m,n) × K(n,k) | (3) |
In this formula, m is the number of total arrays n is the number of features in each component array, U(m,n) represents the matrix which unified the quantization standard, k is the number of selected eigenvalues when calculating the contribution, and K(n,k) is a matrix formed during this calculation.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/d0na00705f |
This journal is © The Royal Society of Chemistry 2021 |