Effect of extensional properties of polymer solutions on the droplet formation via ultrasonic atomization.
Ultrasonic atomization is a useful way to produce fine droplets ranging from a few microns to submicrons depending on the mechanical construction of atomizers and physicochemical properties of the atomized fluids (1), (2). Compared with the atomization through pressurized nozzles, ultrasonic atomizers offer smaller drop sizes with more uniform size distribution. Among various applications from such process, drug nebulization is one of the promising areas to enhance the efficiency of specific drug delivery (3), Since the first report on the ultrasonic atomization by Wood and Loomis (4), numerous works, both theoretical and empirical, have been conducted. A brief historical review on this subject can be found in a literature (5).
To quantitatively analyze the atomization process by ultrasonic vibration, Lang (6) proposed a relationship to predict the size of atomized droplets, where the density and surface tension of liquid were considered as major material parameters. Later, more elaborated models were developed to explain the mechanisms of droplet formation and to correlate the observed droplet sizes with liquid phase properties including shear viscosity of polymer solutions (1), (7). Although the influence of solution viscosity on the droplet size was investigated in these studies, the viscosity was mostly relevant to shear flow. Of course, shear viscosity is an important material parameter to influence the flow rate of the solutions and threshold intensity of ultrasound. However, it should be recognized that the steady shear flow experiments probe only the linear response of the dissolved polymer and do not reflect the flow profile involved in the progress of atomization. As a consequence, the extensional properties of the polymer solutions should be considered to more fully understand the atomization process. This is strongly associated with the underlying mechanism of droplet formation, in which droplets are generated from elongated threads or capillaries via Rayleigh instability driven by the surface tension forces (1). In producing polymeric particles via ultrasonic atomization, the viscoelasticity of polymer solutions would inevitably affect the surface-tension-driven breakup of capillaries. To our knowledge, previous studies on ultrasonic atomization little considered the pertinent issue. Hence, our goal was to investigate the influence of extensional characteristics of polymer solutions on their propensity to form droplets during ultrasonic atomization.
The schematic diagram of the ultrasonic atomization unit is depicted in Fig. 1. A transparent acrylic tube (height = 21.5 cm, diameter = 6 cm) was combined with an atomizer (HM-2412, Honda Electronics Co.) having an operating frequency of 2.4 MHz and atomizing capacity of 250 mL/h of water. The intensity of ultrasonic energy was found to be around 3.5 W/cm (2), which was determined by calorimetric method (8). According to the suggested liquid level specified in the catalogue, the level of the surface in solution was maintained at 2 cm above the vibrating plate. As the atomized droplets were emerged from the surface of reservoir, they were delivered through a flexible PVC tube (30-cm long from the outlet of the cylinder to the inlet of filter) by vacuum pump and collected in a Teflon filter having an average pore size of 0.1 [micro]m. The vacuum pump was operated in the range equivalent to air flow rate of 30 L/min and a suction pressure of 21 in Hg. In the given unit of atomization, it was assumed that the suction pressure loaded at the outlet of the cylinder does not invoke any disturbances over the surface of the solution and is only effective for transportation of droplets once they are formed. Otherwise, fluctuations at the fluid surface would be apparent without sonication. No such phenomenon was observed. For the ease of handling high molecular weight polymers for the given capacity of ultrasonic atomizer, the concentrations of solutions were kept as low as 1 wt%. The atomization was conducted up to 1 h and the collected droplets were dried overnight in an oven at 80[degrees]C.
[FIGURE 1 OMITTED]
Surface tension measurements were performed using Wilhelmy method (Nima Technology) at 25[degrees]C. Shear viscosity measurements were carried out in a stress-controlled rotational rheometer (Physica MCR300). Scanning electron microscopy (SEM, JEOL JSM-5800) was used to observe the particle morphology after coating each sample with gold.
A capillary breakup extensional rheometer (Thermo Haake CaBER 1) was used to characterize the extensional properties of the solution (9-11). The CaBER is a filament-stretching device that monitors the diameter at the mid-point of a fluid filament as it thins under the action of capillary force. The uniqueness of the CaBER technique is to impose a premominantly uniaxial extensional step deformation on a fluid sample in order to extract information on its transient extensional properties. Thus we can analyze the deformation process of the given material by tracing the time evolution of capillary diameter and adopting a model equation developed by Rodd et al. (12). Initially, the 4-mm diameter plates were separated by a gap of 2 mm (initial aspect ratio of distance to diameter = 0.5) and the fluid bridge confined between the plates was stretched as the top plate moves linearly within 50 ms to a specified distance of 4.65 mm (final aspect ratio of distance to diameter = 1.16). The subsequent evolution of the filament diameter with time was monitored with the laser micrometer of the CaBER. Details of CaBER technique and its useful applications can be found elsewhere (13), In carrying out the measurements and analysis of CaBER, it should be reminded that the strain rate of the extensional deformation may not be equivalent to the strain rate imposed during atomization process. Unfortunately, we were not able to estimate the strain rate of the atomization process and this may lead to a misleading due to the difference in Deborah numbers (ratio of material relaxation time to process time scale) of the corresponding processes. Accordingly, it is suggested that the data obtained from CaBER are suitable for the relative comparisons only. Nevertheless, it is expected that the data from extensional analysis provide a useful insight to understand the ultrasonic atomization of polymer solutions.
Poly(ethylene oxide) (PEO), Poly(vinyI alcohol) (PVA), and Poly(vinyl pyrrolidone) (PVP) of various molecular weights were purchased from Aldrich Chemicals Co. The viscosity average molecular weights ([M.sub.y]) of the PEO samples were 100 k (100 [CHI] 10(3)), 200 k, 300 k, 600 k, and 900 k. For poly(ethylene glycol) (PEG), the molecular weight was 14 k. The weight average molecular weights ([M.sub.w]) of PVA were 9 k, 13 k, 31 k, 85 k, 124 k. The percent hydrolyzation of most PVA samples was reported to be in the range of 87-89, while that of PVA having [M.sub.w] of 9 k was 80. In case of PVP, the available [M.sub.w] were 29 k and 55 k. To produce aqueous solutions to be atomized, 1 g of polymer granules was mixed with 1 dL of deionized water at 80[degrees]C for 10 min and subsequent stirring was continued for 2 h at 25[degrees]C. As an attempt to increase the elasticity of the solutions, 0.1 g PEO of high molecular weights was mixed with PEG solutions to prepare PEG/PEO (9:1 by weight) solutions, which was effectively practiced by Yu et al. (10) and Dontula et al. (14). To make homogeneous solutions of PEG and PEO, PEG was added to the solution after PEO had completely dissolved. In preparation and utilization of the above-mentioned polymer solutions, it is important to start atomization immediately after completion of mixing. If prolonged intervals of more than few hours exist between solution preparation and atomization, the possibility of aggregate formation or physical gelation should be considered as one of the governing factors to affect atomization of polymer solutions. The related issue was addressed in a study on electrospinning of PVA solutions made by various solvents (14), where it was found that aggregates were dominantly formed in solutions of relatively high concentrations (over 8 wt%) and it took hours to induce the transition from sol to gel in PVA/water system. In terms of concentration regime and processing time, we presumed that such phase inhomogeneity is not pertinent for a series of polymer solutions used in this study. Furthermore, ultrasonic wave can act as an effective homogenizer during atomization. With this, the broad choice of systems allows us to clarify the key factors that govern the droplet formation during ultrasonic atomization.
RESULTS AND DISCUSSION
The process of ultrasonic atomization was successfully implemented and we were able to practice atomization for various solutions prepared in this study. The resulting morphologies after atomization were demonstrated in Fig. 2 for selected cases. From the atomization of PEG solution, we were able to obtain spherical particles of average size around 0.9 pm; however, droplets were not obtained from PEO solutions under consideration (results not shown here). From the atomization of PEG/PEO sys tems, somewhat peculiar results were obtained. When PEO 100 k was added to PEG solution, droplets larger than 1 [micro]m were observed (Fig. 2a); furthermore, as the molecular weight of PEO was increased to 900 k, the number of droplets were greatly reduced and the size of droplets was enlarged up to the size of 2 [micro]m (Fig. 2b and c). From the atomization of PVA solutions, although the morphology and population of droplets were slightly varied, fine droplets less than 1 [micro]m were formed when the [M.sub.w], of PVA was below 31 k; whereas no droplet was gained in samples comprising higher molecular weight species. In the atomization of PVP solutions, numerous droplets comparable with the case of PVA 9 k were successfully generated in both grades (Fig. 2h and i).
[FIGURE 2 OMITTED]
In discussing the results provided in Fig. 2, one can simply consider the influence of molecular weight and its relevance to chain overlapping as a critical factor to affect the ultrasonic atomization of polymer solutions. Of course, it would be difficult to produce droplets once the shear viscosity and the degree of chain overlap in the given solution become too high. To determine the critical overlap concentration ([c.sub.*]) of the polymer solutions studied, we have used an equation proposed by Graessley (16): [c.sub.*] = 0.77/[eta], where [[eta]] is the intrinsic viscosity of the polymer solution. For the estimation of intrinsic viscosities of polymers used in this study, the reported values of molecular weight were employed in the equations: [[eta]] = 3.3 [CHI] 10(-4) [M.sub.y] (0.72) for PEO (17), [[eta]] = 6.51 [CHI] 10(-4) [M.sub.w] (0.628) for PVA (18), and [[eta]] = 3.0 [CHI] 10(-4) [M.sub.w] (0.65) for PVP (19) with [[eta]] in unit of dL/g, respectively. As seen in Table 1, it is regarded that the polymer solutions based on PVA 85 k and PVA 124 k, which were not atomized at all, form indeed chain overlap at a total concentration of 1 wt%. However, even when we increased the polymer concentration to 5% which is well above c* for all polymers used here, atomization was still possible for polymers such as PVA 9 k, PEG and the two PVP's. Hence, it is inferred that the existence of chain overlapping, despite of its apparent importance, does not provide a definite guideline for the droplet formation itself. Moreover, it was revealed from Fig. 3 that shear viscosities of PVA 31 k solutions are close to those of PEG/PEO 600 k solution and only slightly lower than those of PEG/PEO 900 k. The similarity in shear viscosities can also be found in the comparison between PVA 9 k and PEG/PEO 100 k. Despite of the comparable level of shear viscosities between them, the resulting atomization behavior in terms of droplet population and size was quite different as clearly shown in Fig. 2. Besides the influence of shear viscosities on atomization. [text incomplete in original source] consider the effect of surface tension on the sizes of droplets produced from different polymer solutions. The size of atomized droplet (d) can be estimated by using Lang's equation (6): d = 0.34(8[pi][sigma]/[rho]f(2)) (1/3), where [sigma] is surface tension of the solution and f is frequency of ultrasound (2.4 MHz). Based on the measured values of surface tensions at a concentration of 1 wt% (74.5 mN/m for PEG/PEO 600 k; 58.3 mN/m for PVA 31 k), the size ratio of the two appears to be only 1.09, which underestimated the observed value. Thus, it was noticed that the differences in shear viscosities and surface tensions do not discrimi nate the tendency encountered in ultrasonic atomization of polymer solutions.
TABLE 1. Critical concentration ([c.sup.*]) for chain overlap in polymers used in this study. Polymer Molecular weight (a) [c.sup.*] (wt%) PEG 14 k 2.41 PEO 100 k 0.58 600 k 0.16 900 k 0.12 PVA 9 k 3.89 13 k 3.09 31 k 1.79 85 k 0.95 124 k 0.75 PVP 29 k 3.21 55 k 2.14 (a.) [M.sup.y] was used for PEO, whereas [M.sup.w] was used for PVA and PVP.
[FIGURE 3 OMITTED]
In the electrospinning of polymer solutions, it has been well known that uniform fibers are not formed unless the conditions for concentrations or viscosities (either shear or extensional) are satisfied. Under such circumstances, electrospraying is dominant over electrospinning, leading to droplets formation rather than fibrous morphology. More recently, the importance of elasticity for generating the critical stress required to suppress the Rayleigh instability has been rigorously investigated by Yu et al. (10) in a study on the electrospinning of PEG/PEO solutions. A similar argument is pertinent here and a clue for the observed atomization behavior was found from the assessment of extensional behavior of the polymer solutions. The related data were obtained from CaBER and were displayed in Figs. 4 and 5. As seen in Fig. 4, the overall breakup timescale for each sample was very short (less than 6 ms) compared with the results reported in other studies (10), (11). This implies that the polymer solutions used here were not sufficiently viscous to be successfully tested in CaBER, mainly because of the relatively low molecular weight and concentration of polymers. However, it is worthy to note that we were able to distinguish the extensional behaviors of various polymer solutions by comparing the initial capillary diameters and time to reach the breakup. As the molecular weight of the polymer increases for each series of the solutions, the initial capillary diameter also increases and so does the breakup time, which is indicative of greater resistance of the sample to the axial step-strain and subsequent thinning process. Although we can derive, at least in a qualitative way, a meaningful interrelationship between extensional property and atomization behavior, we performed additional measurements for the samples having increased fluid viscosity as an attempt to capture more reliable thinning behavior, which enabled quantitative analysis. In a study on the CaBER of low viscosity fluids (12), it was suggested that the operability for CaBER can be improved by increasing molecular weight and concentration of polymers or solvent viscosity. In this study, change of concentration or solvent viscosity is a viable choice for the given range of polymer molecular weights. However, increase of concentration would lead to a different degree of chain overlap and thus may not be suitable for the representation of atomization properties at a concentration of 1 wt%. Hence, we increased the solvent viscosity by adding glycerol (50 wt%) to water, which was effectively practiced to provide a reference Newtonian fluid in studying breakup of capillary jets composed of PEO (20). Of course, the variation of chain conformation in the mixed solvent of glycerol and water could be problematic; it was presumed, however, that a relative comparison of data would be allowed and the results of CaBER measurements from polymer/water/glycerol system are displayed in Fig. 5 As expected, the increased viscosity of the fluid delayed the breakup event and various thinning profiles were more evident depending on the kind of polymers. The most striking effect of changing matrix viscos ity is that the extended thinning patterns were revealed only in solutions based on PVA 85 k, PVA 124 k, PEG/PEO 600 k, and PEG/PEO 900 k. These polymers were esteemed to be more elastic than the others; but it seems that the elasto-capillary thinning behavior was hidden in the water matrix due to the low Ohnesorge number (Oh, ratio of viscous breakup timescale to Rayleigh timescale) (12).
[FIGURE 4 OMITTED]
[FIGURE 5 OMITTED]
From the data obtained from CaBER measurements, the viscoelastic properties of the solutions, such as the apparent extensional viscosity ([eta]E, here "apparent" stands for the deformation history-dependent quantity and henceforth called the extensional viscosity) and the longest extensional relaxation time ([lambda]E, henceforth called the relaxation time), can be estimated by analyzing the corresponding thinning profile, respectively. Here, the extensional viscosities were obtained from the relationship: [eta]E = --[sigma]/(dD(t)/dt) by calculating the initial slopes in evolution of the filament diameter (D(t)) (9). On the other hand, the relaxation time can be evaluated in a elasto-capillary regime which is typically observed at late times close to the breaking point. From the slope of the terminal region, we can get the relaxation time based on the relationship: In D(r) ~ --t/3[lambda]E (11), (12). As mentioned above, due to the lack of elasto-capillary regime in polymer solutions except PVA 85 k, PVA 124 k, PEG/PEO 600 k, and PEG/PEO 900 k, only the available values of AE were summarized with [eta]E in Table 2. The order of extensional viscosities and relaxation time was turned out to be comparable with the results of other studies dealing with the extensional behaviors of PVA and PEO solutions (12), (17). It is noticed from the table that [eta]E is increasing with increasing molecular weight in PVA and PEG/PEO solutions; while of PVP 29 k is slightly higher than that of PVP 55 k, which reflects the higher surface tension of PVP 29 k. Among the samples tested, PVA 124 k has the greatest extensional viscosity and relaxation time. It is worthy to note that the analysis on extensional characteristics of polymer solutions is useful to differentiate the rheological responses of PVA 9 k versus PEG/PEO 100 k and PVA 31 k versus PEG/PEO 600 k. Although shear viscosities of the respective pair were similar in Fig. 3, the extensional viscosities are always higher for PEG/ PEO cases, along with the notable difference in the thinning behaviors.
TABLE 2. Extensional viscosities and relaxation times of polymer solutions (in glycerol/water). Polymer Molecular [sigma] [eta]E,app [lambda]E weight (mN/m) (Pa.s) (ms) PVA 9k 56.7 0.82 -- 13k 57.3 0.84 -- 31k 58.3 0.86 -- 85k 61.5 1.28 1.32 124k 62.0 1.92 2.41 PVP 29k 83.7 1.22 -- 55k 79.2 0.99 -- PEG 14k 75.4 1.21 -- PEG/PEO(9:l) 100k 74.5 1.20 -- 600k 75.8 1.40 1.70 900k 74.2 1.48 1.91
An important and consistent feature captured from Fig. 5 and Table 2 is that numerous droplets around 1 [micro]m could be generated by ultrasonic vibration, provided that the extensional viscosity of the polymer solution was relatively low and the breakup was completed earlier without extended thinning behavior. Otherwise, droplets were not formed as were the cases for PVA 85 k, PVA 124 k, and a series of PEO (results not shown); or only few droplets of enlarged size were obtained in PEG/PEO 600 k and PEG/PEO 900 k. If such polymer molecules capable of elasto-capillary thinning are stretched during the atomiza-tion process, a localized elastic stress grows to balance the capillary pressure. Therefore, under this condition, the droplet formation is inhibited by suppression of capillary breakup (as in PVA 85 k and PVA 124 k) or larger droplets with reduced population are produced owing to the delayed growth of capillary instability (as in PEG/PEO 600 k and PEG/PEO 900 k). In fact, a similar result was obtained in a study on the breakup of laminar capillary jets, where fewer and larger drops were formed for fluids with long breakup lengths (20), Although the importance of fluid elasticity was emphasized above, this argument has been made under the premise of capillary formation. Obviously, however, the capillary formation itself may not be allowed during ultrasonic atomization owing to the high viscosity of the solutions, which also hinders droplet formation.
In this study, micro-particles of water-soluble polymers were successfully formed via ultrasonic atomization within the limited range of rheological properties. Although the shear viscosity and surface tension of the materials have been the major factors involved in previous studies of the atomization process, the understanding of atomization behavior of polymer solutions was rather limited. We observed no correlation between shear viscosity or surface tension and droplet morphology. Instead, it was found that for a given frequency and intensity of ultrasound, the propensity for droplet formation was strongly affected by extensional properties of polymer solutions. It was difficult to produce fine particles from polymer solutions having extended thinning behavior, mainly due to the buildup of elastic stress to delay or suppress the breakup of capillary. Of course, the formation of capillary itself would be hardly promoted in case of high extensional viscosity, producing no droplets.
It is suggested that extensional property should be investigated to better understand and more properly describe the ultrasonic atomization process of polymer solutions.
The present research was conducted by the research fund of Dankook University in 2010.
ABBREVIATIONS caBER capillary breakup extensional rheometry PEO poly(ethylene oxide) PVA poly(vinyl alcohol) PVP polyvinyl pyrrolidone)
(1.) B. Avvaru, M.N. Patil, P.R. Gogate, and A.B. Pandit, Ultrasonics, 44, 146 (2006).
(2.) W.H. Suh and K.S. Suslick,,J. Am. Chem. Soc, 127, 12007 (2005).
(3.) P.P.H. Le Brun, A.H. de Boer, H.G.M. Heijerman, and H.W. Frijlink, Pharm. World Sci., 22, 75 (2000).
(4.) W.R. Wood and A.L. Loomis, Phil. Mag., 4, 417 (1927).
(5.) F. Barreras, H. Amaveda, and A. Lozano, Exp. Fluids, 33,405 (2002).
(6.) R.J. Lang, J. Acoust. Soc. Am., 34, 6 (1962).
(7.) R. Rajan and A.B. Pandit, Ultrasonics, 39, 235 (2001).
(8.) J.M. Loning, C. Horst, and U. Hoffmann, Ultrasonics Sono-chemistry, 9, 169 (2002).
(9.) N.J. Kim, C.J. Pipe, K.H. Ahn, S.J. Lee, and G.H. Mckinley, Korea-Australia Rheo. J., 22, 31 (2010).
(10.) J.H. Yu, S.V. Fridrikh, and G.C. Rutledge, Polymer, 47, 4789 (2006).
(11.) C. Clasen, J.P. Plog, W.M. Kulicke, M. Owens, C. Macosko, L.E. Scriven, M. Verani, and G.H. McKinley, J.Rheol., 50, 849 (2006).
(12.) L.E. Rodd, T.P. Scott, J.J. Cooper-White, and G.H. Mckinley, Appl. Rheol., 15, 12 (2005).
(13.) G. Mckinley, Rheol. Rev, 1 (2005).
(14.) P. Dontula, C.W. Macosko, and J.E. Scriven, AIChE J., 44, 1247 (1998).
(15.) S.L. Shenoy, W.D. Bates, and G. Wnek, Polymer, 46, 8990 (2005).
(16.) W.W. Graessley, Polymer, 21, 258 (1980).
Correspondence to: Hyungsu Kim; e-mail: email@example.com
Contract grant sponsor: Dankook University, 2010.
Published online in Wiley Online Library (wileyonlinelibniry.com).
[c] 2011 Society of Plastics Engineers
Yubin Kim, Hyungsu Kim
Department of Chemical Engineering, Dankook University, 126 Jukjeon-dong,
Suji-gu, Kyungki-do 448-701, Korea
|Printer friendly Cite/link Email Feedback|
|Author:||Kim, Yubin; Kim, Hyungsu|
|Publication:||Polymer Engineering and Science|
|Date:||Dec 1, 2011|
|Previous Article:||Interpenetrating polymer network of blocked polyurethane and phenolic resin. II. Kinetic formation, swelling ratio, thermal properties, and flame...|
|Next Article:||Preparation and properties of porous poly(sodium acrylate-co-acrylamide) salt-resistant superabsorbent composite.|