Journal of Materials Science

, Volume 48, Issue 5, pp 2215–2223

A generalized approach for measuring microcapsule permeability with Fluorescence Recovery After Photobleaching


  • Jan Scrimgeour
    • School of PhysicsGeorgia Institute of Technology
  • Adriana San-Miguel
    • School of Chemical and Biomolecular EngineeringGeorgia Institute of Technology
    • School of PhysicsGeorgia Institute of Technology
  • Sven H. Behrens
    • School of Chemical and Biomolecular EngineeringGeorgia Institute of Technology

DOI: 10.1007/s10853-012-6997-7

Cite this article as:
Scrimgeour, J., San-Miguel, A., Curtis, J.E. et al. J Mater Sci (2013) 48: 2215. doi:10.1007/s10853-012-6997-7


Measurement of the permeability of microcapsules with Fluorescence Recovery After Photobleaching (FRAP) enables analysis of individual microcapsules, the study of their batch heterogeneity, and the survey of changes in an individual capsule’s transport properties under different environmental conditions. We present a modified protocol and analysis which expands the parameter space of microcapsule sizes and permeabilities that can be accurately measured with FRAP. Simulations and experiments show that the hybrid analysis, which accounts for both permeation through the shell and free diffusion, precisely captures the kinetics of typical fluorescence recovery curves. The modified approach eliminates the need for painstaking controls to determine the FRAP measurement volume and the spatial distribution of bleached tracers, while accurately extracting the permeability of the microcapsule from FRAP data acquired using a straightforward procedure.


Microencapsulation for controlled or retarded release is a powerful and increasingly common strategy to address the inherent degradability or volatility of many reactive ingredients such as pharmaceutical actives, agrochemicals, fragrances, or enzymes to name but a few [1, 2]. The idea is to shield the active from oxidation or chemical attack by enclosing it inside a protective solid or core-shell capsule from which it is only released as needed. The capsules can primarily serve as storage vessels, delivery vehicles, or as a physical barrier between poorly compatible ingredients in a liquid formulation; in either case the capsule’s permeability for some mobile species is vitally important [3].

There are many ways to assess the average permeability of a bulk sample of microcapsules. One of the few and most commonly used methods to quantify permeability of individual vessels is Fluorescence Recovery After Photobleaching (FRAP) [39]. Here we demonstrate a straightforward modification of this technique that enables accurate quantification of permeability for a much wider range of capsules than possible with the current approach.

FRAP monitors the transport kinetics of a fluorescent tracer species by photobleaching a volume of the species and then recording the increase in fluorescence intensity as unbleached tracers diffuse into the region. Ibarz and colleagues [10] first demonstrated how to measure the permeability of single capsules using a specialized analysis which assumes that fluorescence recovery intensity is dominated by the permeation of unbleached tracers into a photobleached capsule. The approach employs a static, strongly focused laser beam to photobleach fluorescent tracers located inside of a capsule that is embedded in a solution of fluorescent tracers. This analysis has proven to be robust in situations where photobleaching of all the fluorescent tracers is achieved [10, 11] and where it is limited to the inside of the target capsule (see schematic in Fig. 1a).
Fig. 1

Fluorescence recovery curves from simulated FRAP experiments on capsules. Best fits to the data are shown for the Soumpasis, permeation, and hybrid analyses for three different bleaching geometries. a The intensity recovery curve is well fit by the permeation analysis for the ideal case of a uniformly fully bleached capsule (with no exterior bleaching volume). The hybrid analysis is also capable of fitting the result, and both yield similar values for the permeation time scale, τperm = 73 s and τhybrid = τ2 = 81 s, where the shell thickness is h = 0.1R. b In the case that the capsule interior is not fully bleached (but there is no exterior bleaching volume) the recovery curve is only well fit by the hybrid analysis. For this simulation, z/R = 0.5 and h = 0.15R, leading to a fit that yields τ2 = 207.6 s and α = 0.67. c Fluorescence recovery curve for scenario where the capsule interior is fully bleached but some external volume beyond the capsule is bleached (and measured), identified by the purple region. The hybrid analysis fits the curve well and for z/R = 3 and h = 0.15R, a similar permeability time is found τ2 = 201.7 s as in (b) and α = 0.19

The fluorescence recovery is quantified in terms of the fractional intensity, f(t):
$$ f\left( t \right) = \frac{I\left( t \right) - I\left( 0 \right)}{{I_{\infty } - I\left( 0 \right)}} $$
where I(t) is the measured fluorescence intensity and I(0) and \( I_{\infty } \) represent the fluorescence at t = 0 and \( t \to \infty \). If more than a single position is monitored (i.e., with confocal laser scanning microscopy), then I(t) is the sum of the intensity collected in that region.
For the permeability dominated recovery, Ibarz showed that the fractional intensity will increase with time as
$$ f_{\text{perm}} = 1 - e^{{ - {\raise0.7ex\hbox{$t$} \!\mathord{\left/ {\vphantom {t \tau }}\right.\kern-0pt} \!\lower0.7ex\hbox{$\tau $}}}} $$
where the time constant τ can be related to the diffusion coefficient, D, of tracers moving through a spherical wall of thickness h and capsule radius R, so that
$$ \tau = \frac{Rh}{3D} = \frac{R}{3P} $$

Hence, the permeability can be extracted from such FRAP data since it is defined as P = D/h.

In practice, however, complete bleaching throughout exactly the capsule interior is difficult to achieve. If an unbleached fraction of tracers remains inhomogeneously distributed inside of the capsule (Fig. 1b) or if the volume corresponding to the bleaching (and the recovery measurement) extends outside of the capsule (Fig. 1c), then the recovered intensity will include a contribution resulting from the free diffusion of unbleached tracers in the measurement volume. When this diffusion is unaccounted for, it will lead to poor fits of Eq. 2 to the data and difficulty in assessing the capsule permeability (for examples, see Figs. 1b, c, 2, 4).
Fig. 2

Permeability measurements of two large colloidosome capsules using a weakly/beam scanned across the full area of the capsule midsection tobleach and the monitor the recovery of fluorescent tracers, a capsule 1, R1 = 23 μm and b capsule 2, R2 = 28.2 μm. The hybrid analysis provides the best fit, with values aτ1 = 2.11 s, τ2 = 277.5 s, and α = 0.253 which correspond to a permeability of P = 27.7 nm/s; bτ1 = 1.66 s, τ2 = 320.3 s, and α = 0.189 which correspond to a permeability of P = 29.4 nm/s

Here we work to loosen the requirements of the bleaching protocol (and hence the onus on the user to perform checks and optimizations) by adapting the analysis to include contributions of freely diffusing tracers to the intensity recovery curve. The term free diffusion is used to refer to the unbleached fluorescent tracers that contribute to the measured recovered intensity without permeating through the capsule shell. For big capsules such as colloidosomes [12], large volumes and rapid diffusion of small tracers make it difficult to uniformly bleach the capsule contents. This is especially true when using a stationary, highly focused laser beam, i.e., a high numerical aperture (NA) objective, which has a correspondingly small focal volume. The bleaching time must be long enough to allow for all the tracers throughout the capsule to diffuse through the beam and to be bleached. Competition of the flux of new unbleached tracers through the capsule shell with the rate of bleaching can make it impossible to achieve homogeneous photobleaching. This is further complicated by the practical difficulty of confirming the successful execution of the bleach, and the need to do so for samples of heterogeneous capsule size and permeability typical of most formulations. Our simulations demonstrate that if uniform interior bleaching is not achieved (Fig. 1b), the measurement of the permeability will be skewed because the predicted recovery described by Eq. 1 does not adequately describe the measured data.

Uniform bleaching of the tracers in the capsule might be more successful if one photobleaches using a low NA objective whose illumination is approximately collimated throughout the capsule. The extended geometry facilitates bleaching throughout the large volume. More importantly, if one also scans the beam across a disk of radius R equal to that of the microcapsule, one can greatly enhance timely bleaching in the capsule interior. A similar scanning approach might be used with a more tightly focused beam (higher NA), but for reasons (discussed below) related to a simplified analysis of this extended bleaching geometry, we propose working with low NA. One disadvantage, however, of using a low NA beam is that its bleaching and measurement volumes (which in practice are likely different) can extend beyond the boundary of the capsule (Fig. 1c) to an extent that is difficult to predict or measure. Therefore, while reducing the amount of unbleached tracer in the capsule, it inadvertently creates an additional contribution to fluorescence recovery outside the capsule boundaries that is not present in stationary FRAP experiments with high NA. Simulations shown in Fig. 1c verify that this additional bleached volume results in intensity curves which again cannot be fit by an analysis that only accounts for permeation through the shell. The analysis introduced here to handle the intra-capsule free diffusion is also capable of handling this extra-capsule free diffusion, so this should not be problematic.

Given the diverse conditions produced by confocal microscopes, their laser intensities and optical point spread functions and the natural variations in capsule size and permeability, it is risky to make assumptions about the spatial distribution of bleached tracers after the photobleaching step. It is difficult to confirm that homogeneous bleaching has been achieved within the capsule, and critical to prove that the measurement volume is confined to the interior of the capsule shell. We therefore propose a modification to the experimental protocol and the corresponding analysis which makes assessment and segregation of this unwanted component manageable, and allows the permeability to be measured accurately and in a manner that is forgiving of non-ideal bleaching.

Materials and methods

Fluorescence Recovery After Photobleaching (FRAP)

Scanning FRAP measurements were performed on a Zeiss LSM 510 Confocal Microscope using a 488 nm argon ion laser line and a 10×/0.3 NA objective lens. An imaging region of 460 × 460 pixels was defined with a pixel spacing of 0.38 μm. The microscope was focused in the center of the capsule, and a region of interest for photobleaching was defined to match the radius on the capsule. The laser was then scanned pixel by pixel through the disk-shaped region. Photobleaching was performed at 100 % of the available laser intensity, where the bleaching region had pixels spaced 0.38 μm apart and the dwell time at each pixel was 1.67 μs. The recovery series was acquired at 0.3 % of the laser power used for bleaching, and it consisted of a prebleach image (to allow for normalization), monitoring of the photobleaching, and a set of recovery images, which were acquired at 2–3 s intervals until intensity saturation was reached. MATLAB ® Software (Mathworks) was used for data extraction, analysis, and fitting to the theoretical descriptions in Eqs. 2, 4, and 5. Fluorescein, the fluorescent tracer used, was dissolved in ethanol at a concentration of 4 mM. This stock was diluted to 50 uM in 0.2 M phosphate buffer solution (pH 6) and mixed with the microcapsules. The samples were placed in glass cells fabricated using a No. 1.5 cover glass spacer between two cover glasses or placed in the microfluidic dialysis chamber described further below.

Control measurements of the fluorescein tracer bulk diffusion coefficient were performed in these samples, far from any microcapsules. In each instance, a disk with a radius identical to that of a microcapsule measured during the same experiment was photobleached, yielding a typical value of Dbulk = 260 μm2/s. This value was acquired using the expression \( \tau = {{R_{\text{disk}}^{2} } \mathord{\left/ {\vphantom {{R_{\text{disk}}^{2} } {4D}}} \right. \kern-0pt} {4D}} \), as motivated in the text.

Capsule preparation

The measured capsules are pH-responsive colloidosomes of different types prepared as described previously [13, 14]. All the capsules were made using pH-responsive nanoparticles as the building blocks of the capsule shell. These pH-responsive nanoparticles were synthesized from Eudragit S-100 ® (Evonik) by a nanoprecipitation method, where the polymer dissolved in ethanol is abruptly added to two parts of deionized water, upon which the polymer precipitates into small nanoparticles.

Capsule 1 presented in Fig. 2a was made by ethanol consolidation of a colloidosome made with the Eudragit particles. The colloidosomes were prepared by extracting the solvent from a water-in-oil-in-water emulsion stabilized by the Eudragit S-100 nanoparticles. The aqueous phases were dispersions of these particles, and the emulsion was prepared in batch by first homogenizing the inner aqueous phase in dichloromethane. This water-in-oil emulsion was then re-emulsified in an aqueous particle dispersion. The emulsions were then left at room temperature under constant stirring to allow the dichloromethane to evaporate, which created colloidosomes with at least two layers of particles. These colloidosomes, dispersed in water were then mixed with a small amount of ethanol (a good solvent for the polymer) in a 2:1 ratio. After 30 min the capsule dispersion was transferred to an oven at 77.5 °C to evaporate the ethanol and consolidate the capsule shell.

The other measured capsules (Figs. 2b, 4a, b) were composite capsules of PLGA and the pH-responsive nanoparticles. In this case, the same method to generate colloidosomes made with the pH-responsive nanoparticles was utilized, except that the oil phase was a solution of PLGA in dichloromethane (at 7, 14 or 28 mg/mL). Water-in-oil-in-water emulsions were prepared as explained before, and the dichloromethane was extracted by stirring at room temperature in an open container. Upon extraction of the dichloromethane at room temperature, the PLGA precipitated generating a hybrid shell of PLGA with embedded pH-responsive nanoparticles.

Microfluidic dialysis cell for FRAP measurement on single capsules before and after dissolution

For the PLGA composite colloidosomes, a slight pH increase dissolves the embedded Eudragit particles, leaving holes in the PLGA shell. FRAP experiments in these capsules before and after particle dissolution were performed in a microfluidic dialysis cell, as previously described [15, 16]. This cell allows monitoring a single capsule upon a pH change, so that the permeability can be determined before and after dissolution. The cell, developed by Sato and Breedveld [15], is made of a PDMS chamber with a nanoporous membrane (Whatman Anodisc, 0.02 μm) dividing this chamber into two compartments: the top compartment is used for the fluid exchange by flowing solutions at different conditions. The bottom compartment does not experience bulk flow, but allows for fluid exchange through the 20 nm pore membrane. The bottom chamber is bonded to a glass coverslip, with a no 1.5 coverslip spacer, where the capsules are loaded. The top chamber is connected to syringe pumps to allow fluid exchange. The FRAP experiments before and after dissolution were performed by perfusing phosphate buffer with dissolved fluorescein at pH 6, taking FRAP data, and then exchanging the medium with one at pH 8 and repeating the measurements. Measurements were taken after the intensity stabilized indicating the final pH was reached, since fluorescein is pH-sensitive [16].


FRAP experiments were simulated using finite element analysis in the COMSOL MULTIPHYSICS environment. Simulations of tracer diffusion in response to photobleaching were performed in a three-dimensional volume defined by a cylinder with radius 10 units and height 3 units. Spherical capsules were defined with radius R = 0.5 units and h = 0.15 units, the bottom of the capsule was positioned so that it sat on the lower surface of the simulation volume to match the geometry of the FRAP experiments. The ratio of the simulation to capsule volumes is 1884, supporting the assumption that the capsule is sitting in an infinite reservoir of tracers. Bleach and measurement volumes were defined differently, as the bleach volume is expected to extend further in z than the measurement volume. Due to restrictions imposed by the COMSOL environment the bleach and measurement intensity was assumed to be constant. A cylinder of radius R, with height zbleach = 6R, defining the bleach volume, was centered in the simulation volume and completely contained the capsule. A second cylinder with variable height z defined the measurement volume. The measurement cylinder was centered on the capsule when its height was <2R, as its height increased above 2R the additional height penetrated asymmetrically into the simulation volume. Tracer diffusion constants were defined separately for bulk fluid and the capsule wall and Dbulk = 0.01 units2/s and Dwall = 10−4 units2/s. At the beginning of the simulation, t = 0 s, the tracer concentration within the designated bleach region was set to 0 to simulate photobleaching, while the concentration outside the bleach region was set to 1 to represent unbleached tracer. Simulations were run for 1000 s when permeation of unbleached tracer into the capsule was approximately complete.


For the scenario where free diffusion and tracer permeation through the capsule shell are both relevant, the measured FRAP intensity curve will arise from two processes with different time scales. Free diffusion into the bleached areas outside of the capsule will happen almost instantaneously. Similarly, any unbleached fraction inside of the capsule will rapidly diffuse into the measured volume (or redistribute throughout if the entire capsule is included in the measurement). Yet ultimately, full recovery of the original intensity also depends upon the slower process of permeation through the capsule wall. This separation of time scales and the finite total intensity motivates our ansatz of expressing the recovered fluorescence as a convex combination of the two contributions stemming from unhindered diffusion and capsule permeation. Combining the expression developed to account for FRAP recovery by free diffusion from Soumpasis [17] and the expression for permeation introduced by Ibarz (Eq. 1), we hypothesize that the fractional intensity scales as:
$$ f(t) = \alpha f_{\text{free}} + (1 - \alpha )f_{\text{perm}} $$
$$ f_{\text{free}} = \exp \left( { - \frac{{2\tau_{1} }}{t}} \right) \times \left\{ {I_{0} \left( {\frac{{2\tau_{1} }}{t}} \right) + I_{1} \left( {\frac{{2\tau_{1} }}{t}} \right)} \right\} $$
accounts for fluorescence recovery arising from the free diffusion of tracers, with a time constant τ1 and I0 and I1 are the zero and the first order modified Bessel functions. fperm is the expression in Eq. 2, which describes the kinetics of the tracer permeation through the capsule wall via the associated time constant, τ2. The mixing factor, α, weighs the two contributions according to which process dominates the intensity recovery. Smaller α values correspond to a dominance of shell permeation to the final recovered intensity, while larger α values correspond to a sizable component of the recovered intensity originating from free diffusion. The total possible contribution made by free diffusers is determined by the bleaching (and measurement) conditions.

One important subtlety of the proposed analysis and experimental execution is the intentional use of a low NA objective, scanning FRAP as opposed to point FRAP, and the Soumpasis result for ffree. Soumpasis showed that the recovery time of fluorescent tracers in an infinite solution is \( \tau = {{w^{2} } \mathord{\left/ {\vphantom {{w^{2} } {4D}}} \right. \kern-0pt} {4D}} \) for a two-dimensional bleaching area with radius w defined by the waist of a stationary beam. Braeckmans and colleagues [18] extended fluorescence recovery analysis to scenarios where arbitrary bleaching geometries could be scanned with a confocal scanning laser microscope, taking into account the finite axial extent of the beam that naturally occurs in any confocal setup. They showed that for a uniformly bleached disk of radius R and any bleaching depth z (related to the point spread function of the scanned beam), an analytical solution is possible. Importantly, they demonstrated that when the bleaching volume is approximately cylindrical, the 3D recovery is well described by the Soumpasis result where the intensity recovery time is \( \tau = {{R_{\text{disk}}^{2} } \mathord{\left/ {\vphantom {{R_{\text{disk}}^{2} } {4D}}} \right. \kern-0pt} {4D}} \) and Rdisk is the radius of the bleached disk [18]. To ensure that our new hybrid analysis for microcapsules has a simple analytical form, we designed our FRAP experiments to employ a scanned uniformly bleached disk realized using a low NA objective, providing the necessary near-cylindrical intensity distribution. As discussed above, scanning the beam and using a low NA also provide the additional advantage of more efficiently photobleaching tracers in the microcapsule. In principle, more complex expressions accounting for recovery in a highly focused beam (stationary or scanned) might be employed in a similar fashion. In such a case, the source of the freely diffusing tracers would be primarily confined to within the capsule shell as in Fig. 1b. Nevertheless, the hybrid analysis introduced above in Eqn. 4 would still be necessary to insure proper quantification of any unbleached tracers, with an appropriately selected ffree to the match the experimental design.

Results and discussion

Experiments and simulations were performed to test the proposed hybrid analysis in Eq. 4. Large colloidosome capsules were prepared as described elsewhere previously [14] and briefly in the Materials and Methods. In the FRAP measurements, bleaching was achieved using a confocal laser scanning microscope with a weakly focused laser beam (×10 magnification, 0.3 NA, 100 % intensity), which was rapidly scanned across the entire area of the capsule’s midsection. After the bleaching scan, the recovery of fluorescein tracers into the bleached volume was monitored with the same beam in the same scanned area but at a lower intensity (0.3 %). The recovery curves obtained for two capsules are shown in Fig. 2a, b. As expected for such large capsules (R1 = 23 μm and R2 = 28.2 μm) photobleached using a low NA objective, the recovery curves are not fit well by the permeation description (Eq. 2), which only accounts for tracer diffusion through the capsule shell (dotted line). The hybrid analysis, however, visibly fits the data from both capsule 1 and 2 very well (solid line). For comparison, fits to the Soumpasis expression for free diffusion is shown (dashed lines). The analysis reveals that the capsules are difficult for fluorescein to penetrate, with permeation times, τ2 of 277.5 and 320 s, respectively. These correspond to relatively low permeabilities of 27.7 and 29.4 nm/s. The mixing parameter values are closer to 0, α = 0.253 and α = 0.189 indicating that 74.7 and 81.1 % of the recovered intensities originate from permeation of tracers through the shells. Table 1 summarizes the extracted parameters from the fits to experimentally measured capsules.
Table 1

Extracted parameters from measurements of colloidosomes evaluated using a scanned low NA objective for bleaching and measurement of recovery. The hybrid analysis was used to find τ1, τ2, α, and Phybrid. Capsules 3A and 4A are pH-sensitive and known to become highly become porous (capsules 3B and 4B) above pH 7


τ1 (s)

τ2 (s)


Phybrid (nm/s)

Radius (μm)

Capsule 1






Capsule 2






Capsule 3A






Capsule 3B






Capsule 4A






Capsule 4B






To validate the hybrid approach and the assumed convex relationship introduced in Eq. 4, simulations were performed where the volume associated with the free diffusion was systematically increased (see inset in Fig. 3b). The bleach volume was set to be a cylinder with a radius R and a height zbleach = 6R while the measured volume was assumed to be a cylinder of a radius R and a variable height z. By increasing the ratio of z/R, the bleached volume outside of the shell contributing to the measured recovery intensity is increased. Variations in intensity along the axis of the cylinder are neglected in the simulations. Tracer diffusion constants were defined separately for bulk fluid and the capsule wall and Dbulk = 0.01 units2/s and Dwall = 10−4 units2/s, where 1 unit = 2R. We expect that the time scale τ2 should be independent of the experimental configuration. Further, the value of the mixing parameter is expected to increase as the contribution of the free diffusion to the recovered intensity increases, as will τ1 since it takes longer to reach full intensity recovery in a larger volume. The values of α, τ1, and τ2 were extracted by fitting the resultant recovery curves from the simulated FRAP data. The capsule was modeled as a shell of a thickness h and radius R.
Fig. 3

Data from simulations demonstrating how changing the external measured volume influences τ1, τ2, and α, for z/R = [1, 1.5, 2, 2.5, 3, 5]. aτ2 (squares) and the permeability remain fixed near a value of 200 s and 82 nm/s. τ1 (circles) increases because of the longer time required to fill larger external volumes. b The fraction of the total measured intensity from the exterior bleached and measured volume increases with z/R, which as verified by these simulations, increases α

Figure 3 shows data from simulated experiments where the internal region of the capsule was set as fully bleached for all values of z/R. The extracted recovery curves are fit well by the hybrid analysis but not Eq. 1, with one example corresponding to z/R = 3, shown in Fig. 1c. The experiment was performed in the range z/R = [1, 1.5, 2, 2.5, 3, 5]. Importantly, the permeation time, τ2 was found to be constant (〈τ2〉 = 201.3 ± 3 nm/s) and independent of measurement volume, indicating that the measured permeability of the shell does not vary when experimental conditions change. Further, the extracted τ2 is consistent with the inherent value embedded in the model, since the measured permeability is P = R/(3τ2) = 82 nm/s and the theoretical permeability is P = D/h = 67 nm/s where D = 10−4 μm2/s is the diffusion coefficient associated with permeation through the shell. This demonstrates the robustness of the approach, which conveniently does not demand explicit knowledge of the exact bleaching conditions to extract an accurate permeability.

The data also show that as the exterior bleached volume increases, the mixing parameter α increases from ~0.3 to 0.8 (Fig. 3b). This dependence is expected since the fraction of the total intensity recovery due to free diffusion increases with the exterior bleached region. Figure 3a illustrates the increase of τ1 as the exterior measured volume increases. This is expected since a larger volume requires more time to fill via diffusion.

In another experiment, we simulated the inhomogeneous bleaching of the capsule interior (schematic in Fig. 1b). In this case, the cylinder was set to have z = 0.5R and the region extending beyond the cylinder in the capsule was designated to have unbleached fluorescent tracers. The intensity recovery curve, shown in Fig. 1b, is fit well by the hybrid analysis, yielding the same permeability, τ2 = 207.6 s and P = 80 nm/s, as the other simulated measurements presented above of the same model capsule under different bleaching conditions. This demonstrates that the hybrid analysis also works for experimental conditions where the internal bleaching is insufficient, allowing for accurate quantification of the permeability.

Both the experimental measurements and the simulated experiments yield values for the fast component recovery time, τ1, that are about an order of magnitude longer than expected. We measured the bulk diffusion constant for fluorescein (see Materials and Methods) to be Dbulk = 260 μm2/s. Therefore, for a bleached disk of radius R, one would expect a recovery time of \( \tau_{1} = {{R_{\text{disk}}^{2} } \mathord{\left/ {\vphantom {{R_{\text{disk}}^{2} } {4D}}} \right. \kern-0pt} {4D}}_{\text{bulk}} \) = 0.6 s for a typical capsule of radius 25 μm. Yet, as listed in Table 1 extracted recovery times for the free diffusion component, τ1, range from 2 to 8 s, approximately a factor of 10× larger. The longer recovery times likely arise from the boundary conditions of the microcapsule superimposed on the bleaching and measurement area. Consider for example, the intra-capsule free tracer recovery for a bleached disk with a radius matching the capsule and a height z = R. The boundary of the microcapsule prevents any radial recovery at the heights of the bleached region. Instead all recovery from free tracers already located within the capsule arises from the unbleached (or lesser bleached) regions above and below this region. This results in longer recovery times because in an infinite solution with no boundaries recovery would occur both radially and along the axial direction. Likewise, for this same example, now consider recovery in the extra-capsule bleached space. Here, recovery depends on z (each plane in z has a different area which must be filled) and further, the boundary creates an additional asymmetry in the x-y plane, allowing recovery to proceed faster along some in-plane directions than others.

Thus, the fast component of the recovery process for a bleached microcapsule is not the value predicted by a Soumpasis analysis because the inherently cylindrical geometry required by the Braeckmans/Soumpasis analysis is broken by the spherical boundary of the microcapsule. Intra-capsule diffusion is dependent on z, while extra-capsule diffusion has both a dependence on z (breaking down for similar reasons that a highly focused extended beam does not work) and furthermore, an in-plane anisotropy stemming from the confining capsule walls. With full awareness of these limitations, we propose that the present hybrid approach is the best currently available one to describe microcapsule permeability for a vast range of experiments without requiring a full 3D analysis of fluorescence recovery and intimate knowledge of the point spread function of the laser. The rough approximation of the free tracer diffusion by the Soumpasis analysis is flawed. Yet, combining it with the microcapsule permeability in our hybrid approach allows an extra degree of freedom to account for the fast component of recovery, which is temporally distinct from permeation through the capsule. As simulations show (see Fig. 3), the final value of the permeability, determined from τ2 is robust even under changing bleach conditions and the associated changes in τ1. The value of this approach is well illustrated by the greatly improved fitting of typical data, shown in Figs. 1 and 2.

The impact of free diffusion on fluorescence recovery can be increased not only by changing the bleached and monitored volume, but also by opening up large holes in the capsule walls [14]. To demonstrate this, we prepared composite colloidosome capsules using polymer nanoparticles (Eudragit S-100) and poly-(lactic-co-glycolic acid) (PLGA) to fill the gaps between the nanoparticles. The nanoparticles are made from a pH-responsive polymer, Eudragit S-100, which is a (2:1) methyl methacrylate: methacrylic acid random co-polymer. The dissociation of the methacrylic acid groups makes this polymer anionic and soluble in water above pH 7. A slight pH increase from pH 6 to 8 dissolves the embedded particles leaving holes in the PLGA shell. The insets in Fig. 4a, b show typical PLGA capsules before and after particle dissolution. Magnified SEM images in the lower left hand corners illustrate the morphology of the wall for pH 6 and pH 8 (prepared with 14 mg/mL PLGA). This enables a comparison of the hybrid analysis for two extreme scenarios: tightly closed capsules with low permeability versus capsules with large holes randomly positioned throughout wall. Measurements on the same individual colloidosomes were performed before and after the pH change, allowing for direct comparison of the change in permeability.
Fig. 4

Measurements of the permeability of two stimulus responsive colloidosomes before and after the dissolution of nanoparticles embedded in the capsule walls (capsules 3 and 4). Dissolution was achieved by switching the pH from 6 to 8 using a microfluidics dialysis chamber, which also minimizes the flow in the chamber to facilitate accurate FRAP measurements. Data in a show the recovery curves of two tightly closed colloidosomes (no pores) with the nanoparticles still intact (pH 6). The colloidosome with a slower recovery corresponds to a capsule with a larger amount of nondissoluble PLGA in the capsule wall (an initial concentration of 28 vs 14 mg/mL in the later solidified oil phase of the templating water-in-oil-in-water double emulsion). Inset: SEM images of capsules with 14 mg/mL before dissolution. Scale bars are 1 and 20 μm. b After dissolution, the capsules have large holes in the shell walls and intensity recovery happens quickly. Solid lines represent fits of the hybrid analysis to the data. Dashed lines represent the fits of Eq. 2. Inset: SEM images of 14 mg/mL PLGA capsules after dissolution with pH 8. Holes in the capsule wall are clearly visible. Scale bars are 1 and 20 μm as in “a

Figure 4 shows data from two such colloidosomes monitored individually in a microfluidics dialysis cell [15, 16]. At pH 6, when capsule 3 is intact (R = 32.3 μm, 14 mg/mL PLGA), the hybrid analysis works well, yielding a mixing parameter of α = 0.387 and permeability P = 38.6 nm/s. After dissolution of the embedded nanoparticles, the hybrid analysis gives a larger value α = 0.59, but the analysis is extremely insensitive to the value of τ2. We interpret the sizable increase in α even though there has been no change in the measurement parameters (bleaching volume, capsule radius, etc.) as a signal that the capsule likely has cracks or holes, consistent with the SEM images shown in Fig. 4b. The same general trend occurs for capsule 4, which was produced using a more PLGA. As a result, full recovery takes slightly longer (Fig. 4a) and the measured permeability is slightly lower at P = 27.4 nm/s. After dissolution, the two capsules have nearly identical recovery curves and extremely high permeabilities (Fig. 4b).

Our data have shown that the hybrid analysis performs vastly better and more consistently than either the permeation or Soumpasis analysis alone. It is inherently adaptable to different experimental scenarios and requires no additional calibration. The use of the Soumpasis result for ffree is imperfect since “free” diffusion in our system does not occur in an infinite, isotropic medium where the bleached regions can be filled equally well from all sides. The analysis also ignores the slight additional time delay that must occur when tracers first diffuse into bleached but unmeasured volume at the start of the recovery measurement. The effect will be negligible so long as τ2 ≫ τ1 since the two processes are in series such that the resistances add linearly and the permeation barrier will dominate.


In summary, an experimental and analytical approach for measuring capsule permeability using confocal scanning FRAP has been presented. This hybrid approach offers a distinct advantage because of its amenability to a variety of bleaching conditions and capsule properties including volume and permeability. Two sources of usually unconsidered freely diffusing tracers which alter FRAP recovery kinetics have been identified, (i) the inhomogeneous bleaching of the capsule interior and (ii) the bleaching of volume exterior to the capsule. Simulations verify the impact of these components, demonstrating that the basic permeability analysis for capsules is inadequate when one or both sources are present. Experiments supported by simulations have confirmed the merit of describing the fluorescence recovery with a convex combination of a contribution due to capsule permeation alone and a contribution caused by unhindered tracer diffusion in the liquid medium. All data were fit well by the hybrid analysis, yielding reasonable and consistent values for the permeability. A practical requirement of the proposed protocol is the use of a weakly focused, scanned laser beam which facilitates bleaching of a disk in the interior of the capsules, and more importantly enables the simplified analysis of 3D FRAP recovery using the Soumpasis result for 2D. These studies verify that the hybrid analysis is a robust and straightforward approach for measuring capsule permeabilities over a broad range of capsule sizes and shell wall properties without requiring extensive analysis or assumptions regarding the bleaching conditions.


We gratefully acknowledge support from the Camille and Henry Dreyfus New Faculty Awards Program (SHB), and the National Science Foundation (NSF) Career Award DMR-0955811 (JEC, JS) and NSF PHYS-0848797 (JEC).

Copyright information

© Springer Science+Business Media New York 2012