Electro-hydrodynamic atomization of drug solutions for inhalation purposes
http://www.100md.com
《应用生理学杂志》
1 Particle Technology Group, DelftChemTech, Faculty of Applied Sciences, Delft University of Technology, 2628 BL Delft;
2 Department of Pulmonary Diseases, University Hospital Utrecht, 3584 CX Utrecht, The Netherlands
ABSTRACT
Monodisperse aerosols show therapeutic advantages, but they are difficult to generate. A new method (electrohydrodynamic atomization) is described. A high voltage is applied to a nozzle through which a solution, containing dissolved drug, is pumped. At the nozzle tip, a liquid cone is formed and a stream of monodisperse droplets is released. The droplet diameter is governed by the density, conductivity, and the flow rate of the fluid. The droplets are charged and need to be neutralized. Therefore, a corona discharge system is used. Methylparahydroxybenzoate was used as a model drug, and additional data were generated by using beclomethasone dipropionate (BDP). At a flow rate of 1 ml/h and 0.5% methylparahydroxybenzoate, 1.58-μm particles were produced with a geometric SD of 1.18. Increasing the flow rate to 3 ml/h and the concentration to 3% resulted in 4.55-μm particles with a geometric SD of 1.29. The experiments with BDP resulted in similar particle sizes. The mass of BDP was found to range between 1.42 and 6 μg/l air. Aqueous solutions cannot be sprayed by using this setup. This method can be used to deliver antiasthma drugs to patients.
keywords:steroids; electric fields
INTRODUCTION
IN THE TREATMENT OF ASTHMA and other chronic obstructive lung diseases, inhalation therapy is the most frequently applied method to administer drugs. Direct local administration into the lungs leads to an immediate effect, and smaller doses are needed compared with oral administration. However, with the use of conventional inhalation devices, only a fraction of the inhaled drug reaches the lower airways, where it has its therapeutic effect (1). A large part is deposited in the mouth and throat, after which it is swallowed and subsequently may be absorbed in the gastrointestinal tract. The low efficiency of the inhalation equipment is related to the less optimal size distribution of the particles released, although recently developed inhalers can show improved distributions (10, 22). Research has shown that, in adults, monodisperse 2.8-μm bronchodilator particles were optimal in terms of efficacy (18, 19). These experiments point to the fact that the range of optimal aerosol particle sizes might be much smaller (2-3.5 μm) than the currently assumed 1-5 μm. It was also shown that administration of these monodisperse aerosols opened the way to reduce the dose emitted from metered or dry powder inhalers by ~80% without losing any clinical effect (20). Large particles are less efficacious; therefore, they may be eradicated from the emitted dose without reducing the therapeutic effect. However, they still have the potential to elicit side effects. Smaller particles are exhaled directly or deposit in the alveoli, where they are not effective due to the lack of smooth muscle. They are most probably rapidly absorbed into the systemic circulation and might be responsible for fast-emerging side effects such as hypokaliemia (21). As a result of these findings, an interest in (producing) monodisperse aerosols has emerged as a means to improve the therapeutic quality (16).
To apply the bronchodilator data to inhaled corticosteroids would be speculative, but the physics governing deposition of such aerosols are, of course, identical to those for bronchodilators. Therefore, administration of monodisperse steroid aerosols also has the potential to reduce total dose and side effects, but the optimal formulation needs to be defined.
The means to produce monodisperse or narrow size-ranged steroid aerosols are, however, limited. Present systems, like the spinning-top generator, are cumbersome in their use and are, therefore, confined to a laboratory environment (4). Electrohydrodynamic atomization (EHDA) is a new technique able to produce monodisperse droplets of a defined size in the micrometer range. It is based on techniques that are new to the field of medical atomization, and, although the system is still in its infancy, it seems a promising technique because of its potential to be converted into small hand-held devices. The aim of this study was to find out whether EHDA could be used to generate corticosteroid aerosols in a size range between 1 and 5 μm with a low geometric standard deviation (GSD) and in quantities sufficiently high to make administration to patients feasible.
MATERIALS AND METHODS
Background of EHDA. In this section, we give a short overview of EHDA. For further information, we refer to Cloupeau and Prunet-Foch (6), Grace and Marijnissen (7), and Hartman (8). The energy source for EHDA is an electric field (see Fig. 1). A liquid is supplied to a nozzle, and an electric field is generated between the nozzle and a counterelectrode. When the electrical stress overcomes the surface tension of the liquid, a cone is formed, from which a thin jet emerges. The jet breaks up into monodisperse droplets (see Figs. 2 and 3).
The droplet size is dependent on the liquid flow rate and the liquid properties (i.e., conductivity, surface tension, and density), and Eq. 1 predicts the primary droplet size D in the so-called varicose breakup regime (Dvaricose), where the jet breaks up in an axisymmetric way as opposed to an asymmetric breakup
(1)
where c is a constant; is density of the liquid (in kg/m3); 0 is permittivity of vacuum (8.85 × 1012 · C2 · N1 · m2); is applied flow rate of the liquid (in m3/s); K is conductivity of the liquid (in S/m); and is surface tension of the liquid (in N/m).
The constant c was found to be 2.05 (8). The droplet size subsequently reduces because of evaporation of the solvent, leaving much smaller dry particles (size depending on the concentration of the dissolved drug).
When the electrical conductivity of the liquid is (too) low (108-1010 S/m), insufficient current can flow, and the liquid cannot be sprayed. The conductivity needs to be higher than 106 S/m, but too high levels lead to an unstable spray process (polydispersity) (9). Liquids with a high surface tension (pure water) are hard to spray: high nozzle voltages are needed so that the breakdown threshold of the surrounding gas is passed easily (15).
If the liquid flow rate is too low, no stable cone can be formed. At increasing flow rates, more liquid leaves the cone, and the electric charges needed for a particular cone shape become insufficient. The cone shape cannot be maintained, and a new equilibrium will be found with a larger jet diameter (larger droplets) (8).
The droplets produced have a high charge and need to be neutralized. Because of evaporation, droplets shrink, and the charge in the droplets increases, causing uncontrollable disruption. A direct-current corona discharge (using a sharp needle setup) can be used to discharge the droplets. In the corona region, a dense cloud of free electrons is produced, forming negative ions in the air. These discharge the positively charged droplets. The applied potential difference and the distance between needle and nozzle govern the neutralization of the droplets (11).
The total length of the device tested is ~1.2 m, and the height is ~0.4 m (see Fig. 4). The device has a spray section, a neutralization section, an evaporation section, and a sampling section. The aerosol can be inhaled from an exit near the sampling section.
The spray section contains a nozzle-ring configuration. The nozzle and the ring are connected to two high-voltage power supplies, which yield the necessary electric field. The ring "focuses" the spray to prevent immediate heavy loss of the charged aerosol. The corona discharge is generated from a grounded sharp needle, placed perpendicular below the nozzle. The distance between the needle and the nozzle can be varied. While they are transported to the exit of the system by a filtered airflow (high-efficiency particle arrester filter), the droplets evaporate, after which the formed particles are sampled and/or inhaled (for accurate size measurements, particles need to be solid). A part of the airflow is sampled by an aerosizer or APS 3310 (TSI, St. Paul, MN), the latter of which is equipped with a 1:100 diluter to ensure that particle size measurement error, due to coincident particles, is <2% by keeping particle concentrations <200/cm3. These time-of-flight particle sizers measure the aerodynamic diameter of the particles and the mass contained in the aerosol in micrograms per liter of air (5).
Solutions used to produce pharmaceutical particles. EHDA produces droplets from a solution. Ethanol was chosen as a solvent because it has excellent liquid properties for EHDA, it is already present in inhalation preparations, and, lastly, the steroid intended to be used in human experiments, beclomethasone dipropionate (BDP), is soluble in ethanol. As BDP is expensive, the initial experiments were performed with a solution of methylparahydroxybenzoate (MPHB) in ethanol. MPHB was chosen because it is ethanol soluble, and the physicochemical priorities closely resemble those of BDP. Table 1 gives an overview of the MPHB solutions (and their conductivity) used.
The conductivity of the fluids was measured with a YSI 32 conductance meter (Yellow Springs Instruments). The densities and surface tension of the solutions in Table 1 were assumed to be equal to the density (790 kg/m3) and surface tension (22.8 mN/m) of ethanol at room temperature because of the low concentrations of MPHB in ethanol.
Influence of the liquid flow and drug concentration on particle size. To obtain the desired particle range between 1 and 5 μm, the effect of increasing liquid flows and drug concentration on size was performed with solutions of 0.5, 2, and 3% wt/wt MPHB in ethanol. The applied liquid flow rates during the experiments increased from 1 and 4 ml/h. During the experiments, the potential on the nozzle and ring, the distance between the ring and corona needle, and the airflow rate were kept constant (nozzle: 13.75 kV, ring: 10.9 kV, distance: 90 mm, and airflow: 6.5 l/min). Particle size and mass (μg/l) were measured by the APS time-of-flight system (5, 13). The stability of the system was tested by repeating size distribution measurements over 10 consecutive 1-min intervals.
Droplet discharge efficiency by the corona. A solution of 5% wt/wt MPHB in ethanol at a liquid flow rate of 1 ml/h was used. Behind the system, an electrostatic precipitator was placed, which is a channel containing two plates of stainless steel facing each other. An electric field between the two plates is created: charged particles will deposit in the precipitator. An Aerosizer was placed downstream to measure the number of particles that exit the precipitator. The influence of increasing potential differences (2-16 kV) and two distances (9 and 11 cm) between the corona needle and the ring was studied (the distance between ring and nozzle is fixed, whereas it was easier to measure the needle-ring distance). In the electrostatic precipitator, a potential difference of 5 kV was applied, keeping one plate grounded.
Generation of BDP aerosols. Results from the initial MPHB experiments were used to select and confirm the EHDA settings to obtain aerosols with a mass median aerodynamic diameter (MMAD) of 1.5, 2.8, and 5 μm (19). Based on the outcome of the MPHB experiments (see RESULTS), we decided to use a 0.5% beclomethasone solution at 1 ml/h to generate a 1.5-μm aerosol, a 2% solution at 1.5 ml/h for a 2.8-μm aerosol, and a 3% solution at 3 ml/h for a 4.5-μm aerosol. Particle size (MMAD and the GSD) and mass (μg/l air) were measured by the APS system.
RESULTS
Influence of the liquid flow and drug concentration on particle size. For the MPHB solutions, the effect of increasing the liquid flow rate on particle size and width of the distribution (GSD) are shown in Table 2. At increasing flow rates, the size of the particles increases; averaged over all concentrations, the MMAD increased by factors of 1.39 and 1.61 when liquid flow was doubled or tripled, respectively. According to Eq. 1, this should have been and , respectively (P = 0.59 and P = 0.11, respectively, one-sample t-test).
At a low liquid flow rate, particles are produced with a small GSD of 1.18, but, at increasing flow rates, the distribution becomes broader, up to 1.37. The ratio of the largest to the smallest GSD is 1.16.
The stability of the system was tested by repeating size distributions over 10 intervals of 1 min. The coefficient of variation for the MMAD ranged between 3 and 5% and for the GSD between 1 and 2.5% for all flow/concentration combinations.
At equal flow rates, a clear increase in particle size for increasing concentration of 0.5, 2, and 3% MPHB solutions is present. In Table 3, the mass of material (μg BDP/l air) is depicted. Averaged over all concentrations, the mass per liter air increased by factors of 1.65 and 1.43 when liquid flow was doubled or tripled, respectively.
Neutralization efficiency. Figure 5 shows the influence of the distance and potential difference between the ring and the corona needle on the number of particles passing the precipitator and counted by the aerosizer. When the ring-needle potential difference was <3 kV, no discharge was apparent: no particles passed the precipitator. At higher potential differences, sufficient production of corona ions starts and the droplets are discharged better until a maximum number of particles is counted. For a ring-needle distance of 7 cm, this maximum occurs at a potential difference between 7 and 12 kV and for a 9-cm distance at 9 kV. Increasing the potential difference beyond these values causes the number of counted particles to decrease.
Particle size and mass of monodisperse BDP aerosols. In Table 4, the chosen settings and the characteristics of the beclomethasone aerosols obtained are depicted. Compared with the initial MPHB experiments, the resulting MMAD and GSD are in good agreement: for the 0.5% 1 ml/h setting, exactly the same particle size was obtained (P = 1.0), whereas for the 3% 3 ml/h, the BDP size was 0.45 μm larger (P = 0.06, unpaired t-test).
DISCUSSION
We have shown that it is possible to generate BDP aerosols of various particle sizes with very narrow size distributions using EHDA. Up until now, generation of monodisperse aerosols (GSD 1.2) was only possible with complicated laboratory equipment, like the spinning top, vibrating orifice, or Sinclair-LaMer generators. These devices all are characterized by moving mechanical parts, high use of compressed air, complicated operating process, and/or high temperatures (2). EHDA is characterized by nonmoving parts, no use of compressed air, and easier operation. The present restriction is the choice of ethanol as solvent. The electric field needed to obtain a stable cone when spraying water must be very high, and electric breakdown is likely to occur (causing sparks) (15). Ethanol is a low-conductive fluid, while the surface tension is low, and hence it is ideal for EHDA. The restriction to use ethanol, however, is linked to the apparatus described here. Aqueous solutions have been used to create monodisperse aerosols by using a sheet of CO2 surrounding the nozzle (15). In CO2, stronger electric fields are possible without electric breakdown in the gas. However, by using this approach, volunteers would inhale unacceptable amounts of CO2 gas. Several alternatives are possible: surface tension reduction with surfactants or spraying a suspension with very fine particles. After evaporation, agglomerates are formed, the size of which is dependent on the number of particles per primary droplet.
The high voltages do not degrade drug in solution: chromatography showed only the original BDP and no breakdown products (data not shown). Safety for the patient is guaranteed by shielding the high voltage by insulators, whereas the currents can be limited to a few nanoamperes because of the low-power consumptions of the EHDA process.
Influence of the liquid flow and drug concentration on particle size. We have used MPHB to carry out the initial experiments. We did so because the EHDA theory predicts that, when surface tension and conductivity of the fluid are similar, the primary droplet size will be similar (see Eq. 1). Also, when the concentrations of the dissolved compounds and density are close, the secondary dry (aerodynamic) particle size cannot differ much. Thus MPHB was an adequate choice for the introductory experiments.
The increase of the liquid flow rate by a factor of 2 or 3 should, according to Eq. 1, result in a particle size increase of or , respectively. The results in Table 2 indicate that tripling the flow rate resulted in a slightly smaller particle size increase than expected (it should be instead of 1.61 or 1.08 times lower), which can be ascribed to experimental variability. Changes in conductivity, according to Eq. 1, will exert a certain influence: when it increases, particle size will lower, and it is clear that the conductivity tends to increase at higher drug concentrations. According to Eq. 1, a doubled conductivity will lower the particle size by 21/6 = 1.12. Experimental error, as well as changes in conductivity, can offer a reasonable explanation for the observed phenomena.
We noted that an increase in the flow rate and drug concentration causes the size distribution of the aerosol to widen slightly. Although the reasons for this phenomenon are not exactly known yet, stronger instabilities in the jet formed are assumed. At higher liquid flow rates, the jet becomes thicker, and it is conceivable that stronger instabilities are more probable, resulting in higher variability of the aerosol size. It is also evident that the GSD increases at higher drug concentrations; as said before, higher conductivities make the cone less stable, and hence the GSD tends to increase. Recently, an application of EHDA was described and a GSD of 1.6 was reported, which points to less optimal settings of the process, most probably too high of a flow rate and/or drug concentration (23). One actuation delivered 75 μg of a drug, and, with the use of a 10% solution (=100 μg/ml) of that drug fed to the nozzle at 4 ml/h, 400 μg/h should be available for atomization. Delivering 75 μg in one actuation of a few seconds needs, therefore, a very high flow rate and/or drug concentration, with a strong widening of the size distribution. The advantage of monodisperse aerosols is that one can obtain clinically relevant bronchodilator effects at much lower dosages (6). A dose reduction of 80%, compared with a metered dose inhaler, is possible, removing the necessity of delivering high dosages of the drug. A considerable part of a metered dose inhaler aerosol consists of large particles, which do not enter the low airways. This part of the aerosol can be removed without any detrimental effect on the efficacy. The more of this nonefficacious part that is removed, the lower the delivered dose can be, which means that an aerosol with a narrow distribution is the better choice.
The mass measured (see Table 3) at the APS inlet at 1-m distance from the nozzle is, of course, a function of many parameters (deposition in the system, discharging efficiency, etc.). It is, therefore, not surprising that there is no apparent straightforward relation between liquid flow and aerosol mass. This is not problematic, however, at this stage, because we continuously monitor the particle size and the mass of aerosol per liter of air. Should one want to design a practical inhaler, this is of course a point of interest.
Neutralization efficiency. The neutralization by means of the corona discharge was shown to be effective, because sufficient particles pass the precipitator. We measured BDP masses per liter of air high enough that it becomes possible to deliver, in planned human studies, 100 μg BDP to the patient within short periods of time. The test setup differed somewhat from that used earlier involving a 5% MPHB solution at 1 ml/h. We can extrapolate the results obtained to other circumstances, because discharging droplets depends on the number of positive ions in the droplets and reaction between them and the negative ions. The composition of the primary droplets is of no influence on the number of positive ions present in the fluid, whereas the efficiency of the discharging process is governed by the availability of negative ions from the corona discharge. It is possible that some residual charge is still present, particles may create so-called image forces in the airway mucosa (assuming that the mucosa is a conductive surface), and a changed deposition may result. Yu and Chandra (17) modeled the deposition of small particles (<1 μm, which deposit by diffusion) with and without electric charges and estimated an increase of 40%. These estimates were not confirmed by the in vivo experiments of Melandri et al. (12). In these experiments, the total deposition of charged monodisperse particles was determined by counting the difference between inhaled and exhaled numbers (the site of deposition remains unknown). The particles were charged by using a 5-kV corona discharge, and 0.3-μm particles showed an increased deposition of 18-28%, whereas it was less for 1.1-μm particles. In this respect, one has to realize that larger particles with the same total charge and lower number of charges will present at the surface, so that the net deposition effect will become less. This observation was both experimentally and mathematically confirmed by Chan and Yu (3). They reported that the additional deposition decreases as particle size increases: charged 5-μm particles do not deposit extra in the upper airways, only in the smaller airways. Charged 10-μm particle deposition in all airway generations does not differ from that of uncharged particles. Small, uncharged particles (±0.5 μm) do not deposit well, because all deposition mechanisms are minimally effective at that size; therefore, any additional force, like charge, will influence deposition strongly. Equally foreseeable, less or no influence will be measurable with those particles that deposit well by sedimentation and/or impaction. Furthermore, calculations by Chan and Yu showed that a charge effect diminishes when inspiratory flows rise from a slow 15 l/min to 60 l/min. The reduced residence time makes deposition less probable. We can, therefore, say, taking into account the low residual charges present, that, for the inspiratory flows of 60 l/min to be used in future human experiments and the (relatively) large particle sizes, any additional deposition due to remaining charge will be limited. In this respect, one must also realize that, between the mouthpiece and the nozzle, a distance of ~1 m is present, and so charged particles will be removed by electrostatic deposition during their travel to the outlet.
Particle size and mass of monodisperse BDP aerosols. Compared with the experiments with MPHB, it is striking how similar the results for BDP are. It is clear that, as long as the basic conditions for EHDA are similar (conductivity and liquid flow rate), the outcome of the process is independent of the type of drug in solution. Other systems are in use to produce aerosols with small distribution widths, like the Sinclair-LaMer, the spinning top, or vibrating orifice generator, and EHDA will produce aerosols with slightly wider distributions at larger particles sizes. Brain defined monodispersity as a GSD < 1.2, and this requirement is not met for larger sizes (2). However, the Sinclair-LaMer generator will denature biological materials, and the spinning top is mechanically demanding, whereas the vibrating orifice is difficult to operate (2, 4, 14).
In conclusion, we have developed an aerosol generator that is characterized by the use of electric fields instead of compressed air. The main advantages are easy aerosol production and construction and most of all a small width of the particle size distribution (see Tables 2 and 4).
FOOTNOTES
Address for reprint requests and other correspondence: P. Zanen, Dept. of Pulmonary Diseases, Univ. Hospital Utrecht, Heidelberglaan 100, 3584 CX Utrecht, The Netherlands (E-mail: p.zanen@hli.azu.nl).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Received 28 December 2000; accepted in final form 20 August 2001.
REFERENCES
1.Biddiscombe, MF, Melchor R, Mak VHF, Marriot RJ, Taylor AJ, Short MD, and Spiro SG. The lung deposition of salbutamol, directly labelled with technetium-99m, delivered by pressurised metered and dry powder inhalers. Int J Pharm 91: 111-121, 1993.
2.Brain, JD, and Valberg PA. Deposition of aerosol in the respiratory tract. Am Rev Respir Dis 120: 1325-1373, 1979.
3.Chan, TL, and Yu CP. Charge effects on particle deposition in the human tracheobronchial tree. Ann Occup Hyg 26: 65-75, 1982.
4.Cheah, PKP, and Davies CN. The spinning top aerosol generator: improving the performance. J Aerosol Sci 15: 741-751, 1984.
5.Chen, Y, Barber EM, and Zhang Y. Sampling efficiency of the TSI aerodynamic particle sizer. Instr Sci Tech 26: 363-373, 1998.
6.Cloupeau, M, and Prunet-Foch B. Electrohydrodynamic spraying functioning modes: a critical review. J Aersol Sci 25: 1021-1036, 1994.
7.Grace, JM, and Marijnissen JCM A review of liquid atomization by electric means. J Aerosol Sci 25: 1005-1019, 1994.
8.Hartman, R, Brunner DJ, Camelot DMA, Marijnissen JCM, and Scarlett B. Electrohydrodynamic atomization in the cone-jet mode physical modeling of the liquid cone and jet. J Aerosol Sci 30: 823-850, 1999.
9.Kelly, AJ. Charge injection electrostatic atomizer modelling. Aerosol Sci Technol 12: 526-537, 1994.
10.Kunkel, G, Magnussen H, Bergmann K, Juergens UR, de Mey C, Freund E, Hinzmann R, and Beckers B. Respimat (a new soft mist inhaler) delivering fenoterol plus ipratropium bromide provides equivalent bronchodilation at half the cumulative dose compared with a conventional metered dose inhaler in asthmatic patients. Respiration 67: 306-314, 2000.
11.Meesters, GMH, Vercoulen PHW, Maijnissen JCM, and Scarlett B. Generation of micron-sized droplets from the Taylor cone. J Aerosol Sci 23: 37-49, 1992.
12.Melandri, C, Prodi V, Tarroni G, Formignani M, De Zaiacomo T, Bompane GF, and Maestri G. On the deposition of unipolary charged particles in the human respiratory tract. In: Inhaled Particles IV, edited by Walton WH.. Oxford, UK: Pergamon, 1975, p. 193-201.
13.Sioutas, C, Abt E, Wolfson JM, and Koutrakis P. Evaluation of the measurement performance of the scanning mobility particle sizer and aerodynamic particle sizer. Aerosol Sci Tech 3: 84-92, 1999.
14.Smaldone, GC, Itoh H, Swift DL, Kaplan J, Florek R, Wells W, and Wagner HN. Production of pharmacologic monodisperse aerosols. J Appl Physiol 54: 393-399, 1983.
15.Tang, K, and Gomez A. Generation by electrospray of monodisperse water droplets for targeted drug delivery by inhalation. J Aerosol Sci 25: 1237-1249, 1994.
16.Thompson, PJ. Drug delivery to the small airways. Am J Respir Crit Care Med 157: S199-S202, 1998.
17.Yu, CP, and Chandra K. Precipitation of submicron charged particles in the human airways. Bull Math Biol 39: 471-778, 1977.
18.Zanen, P, Go LT, and Lammers JWJ The optimal particle size for parasympathicolytic aerosols in mild asthmatics. Int J Pharm 114: 111-115, 1995.
19.Zanen, P, Go TL, and Lammers JWJ The optimal particle size for 2-mimetic and parasympathicolytic aerosols in patients with severe airway constrictions. Thorax 51: 977-980, 1996.
20.Zanen, P, Go TL, and Lammers JWJ The efficacy of a low dosed, monodisperse parasympathicolytic aerosol compared to a standard aerosol from a metered dose inhaler. Eur J Clin Pharmacol 54: 27-30, 1998.
21.Zanen, P, and Lammers JWJ Reducing adverse effects of inhaled fenoterol through optimization of the aerosol formulation. J Aerosol Med 12: 241-247, 1999.
22.Zierenberg, B. Optimising the in vitro performance of Respimat. J Aerosol Med 12: S19-S24, 1999.
23.Zimlich, WC, Ding JY, Busick DR, Moutvic RR, Placke ME, Hirst PH, Pitcairn GR, Newman SP, Macintyre F, Miller PR, Shepherd MT, and Lukas TM. The development of a novel electrohydrodynamic pulmonary drug delivery device. In: Respiratory Drug Delivery VII, edited by Dalby RN, Byron PR, Farr SJ, and Peart JP.. Raleigh, NC: Serentec, 2000, p. 241-246.(Jeroen C. Ijsebaert, Kees)
2 Department of Pulmonary Diseases, University Hospital Utrecht, 3584 CX Utrecht, The Netherlands
ABSTRACT
Monodisperse aerosols show therapeutic advantages, but they are difficult to generate. A new method (electrohydrodynamic atomization) is described. A high voltage is applied to a nozzle through which a solution, containing dissolved drug, is pumped. At the nozzle tip, a liquid cone is formed and a stream of monodisperse droplets is released. The droplet diameter is governed by the density, conductivity, and the flow rate of the fluid. The droplets are charged and need to be neutralized. Therefore, a corona discharge system is used. Methylparahydroxybenzoate was used as a model drug, and additional data were generated by using beclomethasone dipropionate (BDP). At a flow rate of 1 ml/h and 0.5% methylparahydroxybenzoate, 1.58-μm particles were produced with a geometric SD of 1.18. Increasing the flow rate to 3 ml/h and the concentration to 3% resulted in 4.55-μm particles with a geometric SD of 1.29. The experiments with BDP resulted in similar particle sizes. The mass of BDP was found to range between 1.42 and 6 μg/l air. Aqueous solutions cannot be sprayed by using this setup. This method can be used to deliver antiasthma drugs to patients.
keywords:steroids; electric fields
INTRODUCTION
IN THE TREATMENT OF ASTHMA and other chronic obstructive lung diseases, inhalation therapy is the most frequently applied method to administer drugs. Direct local administration into the lungs leads to an immediate effect, and smaller doses are needed compared with oral administration. However, with the use of conventional inhalation devices, only a fraction of the inhaled drug reaches the lower airways, where it has its therapeutic effect (1). A large part is deposited in the mouth and throat, after which it is swallowed and subsequently may be absorbed in the gastrointestinal tract. The low efficiency of the inhalation equipment is related to the less optimal size distribution of the particles released, although recently developed inhalers can show improved distributions (10, 22). Research has shown that, in adults, monodisperse 2.8-μm bronchodilator particles were optimal in terms of efficacy (18, 19). These experiments point to the fact that the range of optimal aerosol particle sizes might be much smaller (2-3.5 μm) than the currently assumed 1-5 μm. It was also shown that administration of these monodisperse aerosols opened the way to reduce the dose emitted from metered or dry powder inhalers by ~80% without losing any clinical effect (20). Large particles are less efficacious; therefore, they may be eradicated from the emitted dose without reducing the therapeutic effect. However, they still have the potential to elicit side effects. Smaller particles are exhaled directly or deposit in the alveoli, where they are not effective due to the lack of smooth muscle. They are most probably rapidly absorbed into the systemic circulation and might be responsible for fast-emerging side effects such as hypokaliemia (21). As a result of these findings, an interest in (producing) monodisperse aerosols has emerged as a means to improve the therapeutic quality (16).
To apply the bronchodilator data to inhaled corticosteroids would be speculative, but the physics governing deposition of such aerosols are, of course, identical to those for bronchodilators. Therefore, administration of monodisperse steroid aerosols also has the potential to reduce total dose and side effects, but the optimal formulation needs to be defined.
The means to produce monodisperse or narrow size-ranged steroid aerosols are, however, limited. Present systems, like the spinning-top generator, are cumbersome in their use and are, therefore, confined to a laboratory environment (4). Electrohydrodynamic atomization (EHDA) is a new technique able to produce monodisperse droplets of a defined size in the micrometer range. It is based on techniques that are new to the field of medical atomization, and, although the system is still in its infancy, it seems a promising technique because of its potential to be converted into small hand-held devices. The aim of this study was to find out whether EHDA could be used to generate corticosteroid aerosols in a size range between 1 and 5 μm with a low geometric standard deviation (GSD) and in quantities sufficiently high to make administration to patients feasible.
MATERIALS AND METHODS
Background of EHDA. In this section, we give a short overview of EHDA. For further information, we refer to Cloupeau and Prunet-Foch (6), Grace and Marijnissen (7), and Hartman (8). The energy source for EHDA is an electric field (see Fig. 1). A liquid is supplied to a nozzle, and an electric field is generated between the nozzle and a counterelectrode. When the electrical stress overcomes the surface tension of the liquid, a cone is formed, from which a thin jet emerges. The jet breaks up into monodisperse droplets (see Figs. 2 and 3).
The droplet size is dependent on the liquid flow rate and the liquid properties (i.e., conductivity, surface tension, and density), and Eq. 1 predicts the primary droplet size D in the so-called varicose breakup regime (Dvaricose), where the jet breaks up in an axisymmetric way as opposed to an asymmetric breakup
(1)
where c is a constant; is density of the liquid (in kg/m3); 0 is permittivity of vacuum (8.85 × 1012 · C2 · N1 · m2); is applied flow rate of the liquid (in m3/s); K is conductivity of the liquid (in S/m); and is surface tension of the liquid (in N/m).
The constant c was found to be 2.05 (8). The droplet size subsequently reduces because of evaporation of the solvent, leaving much smaller dry particles (size depending on the concentration of the dissolved drug).
When the electrical conductivity of the liquid is (too) low (108-1010 S/m), insufficient current can flow, and the liquid cannot be sprayed. The conductivity needs to be higher than 106 S/m, but too high levels lead to an unstable spray process (polydispersity) (9). Liquids with a high surface tension (pure water) are hard to spray: high nozzle voltages are needed so that the breakdown threshold of the surrounding gas is passed easily (15).
If the liquid flow rate is too low, no stable cone can be formed. At increasing flow rates, more liquid leaves the cone, and the electric charges needed for a particular cone shape become insufficient. The cone shape cannot be maintained, and a new equilibrium will be found with a larger jet diameter (larger droplets) (8).
The droplets produced have a high charge and need to be neutralized. Because of evaporation, droplets shrink, and the charge in the droplets increases, causing uncontrollable disruption. A direct-current corona discharge (using a sharp needle setup) can be used to discharge the droplets. In the corona region, a dense cloud of free electrons is produced, forming negative ions in the air. These discharge the positively charged droplets. The applied potential difference and the distance between needle and nozzle govern the neutralization of the droplets (11).
The total length of the device tested is ~1.2 m, and the height is ~0.4 m (see Fig. 4). The device has a spray section, a neutralization section, an evaporation section, and a sampling section. The aerosol can be inhaled from an exit near the sampling section.
The spray section contains a nozzle-ring configuration. The nozzle and the ring are connected to two high-voltage power supplies, which yield the necessary electric field. The ring "focuses" the spray to prevent immediate heavy loss of the charged aerosol. The corona discharge is generated from a grounded sharp needle, placed perpendicular below the nozzle. The distance between the needle and the nozzle can be varied. While they are transported to the exit of the system by a filtered airflow (high-efficiency particle arrester filter), the droplets evaporate, after which the formed particles are sampled and/or inhaled (for accurate size measurements, particles need to be solid). A part of the airflow is sampled by an aerosizer or APS 3310 (TSI, St. Paul, MN), the latter of which is equipped with a 1:100 diluter to ensure that particle size measurement error, due to coincident particles, is <2% by keeping particle concentrations <200/cm3. These time-of-flight particle sizers measure the aerodynamic diameter of the particles and the mass contained in the aerosol in micrograms per liter of air (5).
Solutions used to produce pharmaceutical particles. EHDA produces droplets from a solution. Ethanol was chosen as a solvent because it has excellent liquid properties for EHDA, it is already present in inhalation preparations, and, lastly, the steroid intended to be used in human experiments, beclomethasone dipropionate (BDP), is soluble in ethanol. As BDP is expensive, the initial experiments were performed with a solution of methylparahydroxybenzoate (MPHB) in ethanol. MPHB was chosen because it is ethanol soluble, and the physicochemical priorities closely resemble those of BDP. Table 1 gives an overview of the MPHB solutions (and their conductivity) used.
The conductivity of the fluids was measured with a YSI 32 conductance meter (Yellow Springs Instruments). The densities and surface tension of the solutions in Table 1 were assumed to be equal to the density (790 kg/m3) and surface tension (22.8 mN/m) of ethanol at room temperature because of the low concentrations of MPHB in ethanol.
Influence of the liquid flow and drug concentration on particle size. To obtain the desired particle range between 1 and 5 μm, the effect of increasing liquid flows and drug concentration on size was performed with solutions of 0.5, 2, and 3% wt/wt MPHB in ethanol. The applied liquid flow rates during the experiments increased from 1 and 4 ml/h. During the experiments, the potential on the nozzle and ring, the distance between the ring and corona needle, and the airflow rate were kept constant (nozzle: 13.75 kV, ring: 10.9 kV, distance: 90 mm, and airflow: 6.5 l/min). Particle size and mass (μg/l) were measured by the APS time-of-flight system (5, 13). The stability of the system was tested by repeating size distribution measurements over 10 consecutive 1-min intervals.
Droplet discharge efficiency by the corona. A solution of 5% wt/wt MPHB in ethanol at a liquid flow rate of 1 ml/h was used. Behind the system, an electrostatic precipitator was placed, which is a channel containing two plates of stainless steel facing each other. An electric field between the two plates is created: charged particles will deposit in the precipitator. An Aerosizer was placed downstream to measure the number of particles that exit the precipitator. The influence of increasing potential differences (2-16 kV) and two distances (9 and 11 cm) between the corona needle and the ring was studied (the distance between ring and nozzle is fixed, whereas it was easier to measure the needle-ring distance). In the electrostatic precipitator, a potential difference of 5 kV was applied, keeping one plate grounded.
Generation of BDP aerosols. Results from the initial MPHB experiments were used to select and confirm the EHDA settings to obtain aerosols with a mass median aerodynamic diameter (MMAD) of 1.5, 2.8, and 5 μm (19). Based on the outcome of the MPHB experiments (see RESULTS), we decided to use a 0.5% beclomethasone solution at 1 ml/h to generate a 1.5-μm aerosol, a 2% solution at 1.5 ml/h for a 2.8-μm aerosol, and a 3% solution at 3 ml/h for a 4.5-μm aerosol. Particle size (MMAD and the GSD) and mass (μg/l air) were measured by the APS system.
RESULTS
Influence of the liquid flow and drug concentration on particle size. For the MPHB solutions, the effect of increasing the liquid flow rate on particle size and width of the distribution (GSD) are shown in Table 2. At increasing flow rates, the size of the particles increases; averaged over all concentrations, the MMAD increased by factors of 1.39 and 1.61 when liquid flow was doubled or tripled, respectively. According to Eq. 1, this should have been and , respectively (P = 0.59 and P = 0.11, respectively, one-sample t-test).
At a low liquid flow rate, particles are produced with a small GSD of 1.18, but, at increasing flow rates, the distribution becomes broader, up to 1.37. The ratio of the largest to the smallest GSD is 1.16.
The stability of the system was tested by repeating size distributions over 10 intervals of 1 min. The coefficient of variation for the MMAD ranged between 3 and 5% and for the GSD between 1 and 2.5% for all flow/concentration combinations.
At equal flow rates, a clear increase in particle size for increasing concentration of 0.5, 2, and 3% MPHB solutions is present. In Table 3, the mass of material (μg BDP/l air) is depicted. Averaged over all concentrations, the mass per liter air increased by factors of 1.65 and 1.43 when liquid flow was doubled or tripled, respectively.
Neutralization efficiency. Figure 5 shows the influence of the distance and potential difference between the ring and the corona needle on the number of particles passing the precipitator and counted by the aerosizer. When the ring-needle potential difference was <3 kV, no discharge was apparent: no particles passed the precipitator. At higher potential differences, sufficient production of corona ions starts and the droplets are discharged better until a maximum number of particles is counted. For a ring-needle distance of 7 cm, this maximum occurs at a potential difference between 7 and 12 kV and for a 9-cm distance at 9 kV. Increasing the potential difference beyond these values causes the number of counted particles to decrease.
Particle size and mass of monodisperse BDP aerosols. In Table 4, the chosen settings and the characteristics of the beclomethasone aerosols obtained are depicted. Compared with the initial MPHB experiments, the resulting MMAD and GSD are in good agreement: for the 0.5% 1 ml/h setting, exactly the same particle size was obtained (P = 1.0), whereas for the 3% 3 ml/h, the BDP size was 0.45 μm larger (P = 0.06, unpaired t-test).
DISCUSSION
We have shown that it is possible to generate BDP aerosols of various particle sizes with very narrow size distributions using EHDA. Up until now, generation of monodisperse aerosols (GSD 1.2) was only possible with complicated laboratory equipment, like the spinning top, vibrating orifice, or Sinclair-LaMer generators. These devices all are characterized by moving mechanical parts, high use of compressed air, complicated operating process, and/or high temperatures (2). EHDA is characterized by nonmoving parts, no use of compressed air, and easier operation. The present restriction is the choice of ethanol as solvent. The electric field needed to obtain a stable cone when spraying water must be very high, and electric breakdown is likely to occur (causing sparks) (15). Ethanol is a low-conductive fluid, while the surface tension is low, and hence it is ideal for EHDA. The restriction to use ethanol, however, is linked to the apparatus described here. Aqueous solutions have been used to create monodisperse aerosols by using a sheet of CO2 surrounding the nozzle (15). In CO2, stronger electric fields are possible without electric breakdown in the gas. However, by using this approach, volunteers would inhale unacceptable amounts of CO2 gas. Several alternatives are possible: surface tension reduction with surfactants or spraying a suspension with very fine particles. After evaporation, agglomerates are formed, the size of which is dependent on the number of particles per primary droplet.
The high voltages do not degrade drug in solution: chromatography showed only the original BDP and no breakdown products (data not shown). Safety for the patient is guaranteed by shielding the high voltage by insulators, whereas the currents can be limited to a few nanoamperes because of the low-power consumptions of the EHDA process.
Influence of the liquid flow and drug concentration on particle size. We have used MPHB to carry out the initial experiments. We did so because the EHDA theory predicts that, when surface tension and conductivity of the fluid are similar, the primary droplet size will be similar (see Eq. 1). Also, when the concentrations of the dissolved compounds and density are close, the secondary dry (aerodynamic) particle size cannot differ much. Thus MPHB was an adequate choice for the introductory experiments.
The increase of the liquid flow rate by a factor of 2 or 3 should, according to Eq. 1, result in a particle size increase of or , respectively. The results in Table 2 indicate that tripling the flow rate resulted in a slightly smaller particle size increase than expected (it should be instead of 1.61 or 1.08 times lower), which can be ascribed to experimental variability. Changes in conductivity, according to Eq. 1, will exert a certain influence: when it increases, particle size will lower, and it is clear that the conductivity tends to increase at higher drug concentrations. According to Eq. 1, a doubled conductivity will lower the particle size by 21/6 = 1.12. Experimental error, as well as changes in conductivity, can offer a reasonable explanation for the observed phenomena.
We noted that an increase in the flow rate and drug concentration causes the size distribution of the aerosol to widen slightly. Although the reasons for this phenomenon are not exactly known yet, stronger instabilities in the jet formed are assumed. At higher liquid flow rates, the jet becomes thicker, and it is conceivable that stronger instabilities are more probable, resulting in higher variability of the aerosol size. It is also evident that the GSD increases at higher drug concentrations; as said before, higher conductivities make the cone less stable, and hence the GSD tends to increase. Recently, an application of EHDA was described and a GSD of 1.6 was reported, which points to less optimal settings of the process, most probably too high of a flow rate and/or drug concentration (23). One actuation delivered 75 μg of a drug, and, with the use of a 10% solution (=100 μg/ml) of that drug fed to the nozzle at 4 ml/h, 400 μg/h should be available for atomization. Delivering 75 μg in one actuation of a few seconds needs, therefore, a very high flow rate and/or drug concentration, with a strong widening of the size distribution. The advantage of monodisperse aerosols is that one can obtain clinically relevant bronchodilator effects at much lower dosages (6). A dose reduction of 80%, compared with a metered dose inhaler, is possible, removing the necessity of delivering high dosages of the drug. A considerable part of a metered dose inhaler aerosol consists of large particles, which do not enter the low airways. This part of the aerosol can be removed without any detrimental effect on the efficacy. The more of this nonefficacious part that is removed, the lower the delivered dose can be, which means that an aerosol with a narrow distribution is the better choice.
The mass measured (see Table 3) at the APS inlet at 1-m distance from the nozzle is, of course, a function of many parameters (deposition in the system, discharging efficiency, etc.). It is, therefore, not surprising that there is no apparent straightforward relation between liquid flow and aerosol mass. This is not problematic, however, at this stage, because we continuously monitor the particle size and the mass of aerosol per liter of air. Should one want to design a practical inhaler, this is of course a point of interest.
Neutralization efficiency. The neutralization by means of the corona discharge was shown to be effective, because sufficient particles pass the precipitator. We measured BDP masses per liter of air high enough that it becomes possible to deliver, in planned human studies, 100 μg BDP to the patient within short periods of time. The test setup differed somewhat from that used earlier involving a 5% MPHB solution at 1 ml/h. We can extrapolate the results obtained to other circumstances, because discharging droplets depends on the number of positive ions in the droplets and reaction between them and the negative ions. The composition of the primary droplets is of no influence on the number of positive ions present in the fluid, whereas the efficiency of the discharging process is governed by the availability of negative ions from the corona discharge. It is possible that some residual charge is still present, particles may create so-called image forces in the airway mucosa (assuming that the mucosa is a conductive surface), and a changed deposition may result. Yu and Chandra (17) modeled the deposition of small particles (<1 μm, which deposit by diffusion) with and without electric charges and estimated an increase of 40%. These estimates were not confirmed by the in vivo experiments of Melandri et al. (12). In these experiments, the total deposition of charged monodisperse particles was determined by counting the difference between inhaled and exhaled numbers (the site of deposition remains unknown). The particles were charged by using a 5-kV corona discharge, and 0.3-μm particles showed an increased deposition of 18-28%, whereas it was less for 1.1-μm particles. In this respect, one has to realize that larger particles with the same total charge and lower number of charges will present at the surface, so that the net deposition effect will become less. This observation was both experimentally and mathematically confirmed by Chan and Yu (3). They reported that the additional deposition decreases as particle size increases: charged 5-μm particles do not deposit extra in the upper airways, only in the smaller airways. Charged 10-μm particle deposition in all airway generations does not differ from that of uncharged particles. Small, uncharged particles (±0.5 μm) do not deposit well, because all deposition mechanisms are minimally effective at that size; therefore, any additional force, like charge, will influence deposition strongly. Equally foreseeable, less or no influence will be measurable with those particles that deposit well by sedimentation and/or impaction. Furthermore, calculations by Chan and Yu showed that a charge effect diminishes when inspiratory flows rise from a slow 15 l/min to 60 l/min. The reduced residence time makes deposition less probable. We can, therefore, say, taking into account the low residual charges present, that, for the inspiratory flows of 60 l/min to be used in future human experiments and the (relatively) large particle sizes, any additional deposition due to remaining charge will be limited. In this respect, one must also realize that, between the mouthpiece and the nozzle, a distance of ~1 m is present, and so charged particles will be removed by electrostatic deposition during their travel to the outlet.
Particle size and mass of monodisperse BDP aerosols. Compared with the experiments with MPHB, it is striking how similar the results for BDP are. It is clear that, as long as the basic conditions for EHDA are similar (conductivity and liquid flow rate), the outcome of the process is independent of the type of drug in solution. Other systems are in use to produce aerosols with small distribution widths, like the Sinclair-LaMer, the spinning top, or vibrating orifice generator, and EHDA will produce aerosols with slightly wider distributions at larger particles sizes. Brain defined monodispersity as a GSD < 1.2, and this requirement is not met for larger sizes (2). However, the Sinclair-LaMer generator will denature biological materials, and the spinning top is mechanically demanding, whereas the vibrating orifice is difficult to operate (2, 4, 14).
In conclusion, we have developed an aerosol generator that is characterized by the use of electric fields instead of compressed air. The main advantages are easy aerosol production and construction and most of all a small width of the particle size distribution (see Tables 2 and 4).
FOOTNOTES
Address for reprint requests and other correspondence: P. Zanen, Dept. of Pulmonary Diseases, Univ. Hospital Utrecht, Heidelberglaan 100, 3584 CX Utrecht, The Netherlands (E-mail: p.zanen@hli.azu.nl).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Received 28 December 2000; accepted in final form 20 August 2001.
REFERENCES
1.Biddiscombe, MF, Melchor R, Mak VHF, Marriot RJ, Taylor AJ, Short MD, and Spiro SG. The lung deposition of salbutamol, directly labelled with technetium-99m, delivered by pressurised metered and dry powder inhalers. Int J Pharm 91: 111-121, 1993.
2.Brain, JD, and Valberg PA. Deposition of aerosol in the respiratory tract. Am Rev Respir Dis 120: 1325-1373, 1979.
3.Chan, TL, and Yu CP. Charge effects on particle deposition in the human tracheobronchial tree. Ann Occup Hyg 26: 65-75, 1982.
4.Cheah, PKP, and Davies CN. The spinning top aerosol generator: improving the performance. J Aerosol Sci 15: 741-751, 1984.
5.Chen, Y, Barber EM, and Zhang Y. Sampling efficiency of the TSI aerodynamic particle sizer. Instr Sci Tech 26: 363-373, 1998.
6.Cloupeau, M, and Prunet-Foch B. Electrohydrodynamic spraying functioning modes: a critical review. J Aersol Sci 25: 1021-1036, 1994.
7.Grace, JM, and Marijnissen JCM A review of liquid atomization by electric means. J Aerosol Sci 25: 1005-1019, 1994.
8.Hartman, R, Brunner DJ, Camelot DMA, Marijnissen JCM, and Scarlett B. Electrohydrodynamic atomization in the cone-jet mode physical modeling of the liquid cone and jet. J Aerosol Sci 30: 823-850, 1999.
9.Kelly, AJ. Charge injection electrostatic atomizer modelling. Aerosol Sci Technol 12: 526-537, 1994.
10.Kunkel, G, Magnussen H, Bergmann K, Juergens UR, de Mey C, Freund E, Hinzmann R, and Beckers B. Respimat (a new soft mist inhaler) delivering fenoterol plus ipratropium bromide provides equivalent bronchodilation at half the cumulative dose compared with a conventional metered dose inhaler in asthmatic patients. Respiration 67: 306-314, 2000.
11.Meesters, GMH, Vercoulen PHW, Maijnissen JCM, and Scarlett B. Generation of micron-sized droplets from the Taylor cone. J Aerosol Sci 23: 37-49, 1992.
12.Melandri, C, Prodi V, Tarroni G, Formignani M, De Zaiacomo T, Bompane GF, and Maestri G. On the deposition of unipolary charged particles in the human respiratory tract. In: Inhaled Particles IV, edited by Walton WH.. Oxford, UK: Pergamon, 1975, p. 193-201.
13.Sioutas, C, Abt E, Wolfson JM, and Koutrakis P. Evaluation of the measurement performance of the scanning mobility particle sizer and aerodynamic particle sizer. Aerosol Sci Tech 3: 84-92, 1999.
14.Smaldone, GC, Itoh H, Swift DL, Kaplan J, Florek R, Wells W, and Wagner HN. Production of pharmacologic monodisperse aerosols. J Appl Physiol 54: 393-399, 1983.
15.Tang, K, and Gomez A. Generation by electrospray of monodisperse water droplets for targeted drug delivery by inhalation. J Aerosol Sci 25: 1237-1249, 1994.
16.Thompson, PJ. Drug delivery to the small airways. Am J Respir Crit Care Med 157: S199-S202, 1998.
17.Yu, CP, and Chandra K. Precipitation of submicron charged particles in the human airways. Bull Math Biol 39: 471-778, 1977.
18.Zanen, P, Go LT, and Lammers JWJ The optimal particle size for parasympathicolytic aerosols in mild asthmatics. Int J Pharm 114: 111-115, 1995.
19.Zanen, P, Go TL, and Lammers JWJ The optimal particle size for 2-mimetic and parasympathicolytic aerosols in patients with severe airway constrictions. Thorax 51: 977-980, 1996.
20.Zanen, P, Go TL, and Lammers JWJ The efficacy of a low dosed, monodisperse parasympathicolytic aerosol compared to a standard aerosol from a metered dose inhaler. Eur J Clin Pharmacol 54: 27-30, 1998.
21.Zanen, P, and Lammers JWJ Reducing adverse effects of inhaled fenoterol through optimization of the aerosol formulation. J Aerosol Med 12: 241-247, 1999.
22.Zierenberg, B. Optimising the in vitro performance of Respimat. J Aerosol Med 12: S19-S24, 1999.
23.Zimlich, WC, Ding JY, Busick DR, Moutvic RR, Placke ME, Hirst PH, Pitcairn GR, Newman SP, Macintyre F, Miller PR, Shepherd MT, and Lukas TM. The development of a novel electrohydrodynamic pulmonary drug delivery device. In: Respiratory Drug Delivery VII, edited by Dalby RN, Byron PR, Farr SJ, and Peart JP.. Raleigh, NC: Serentec, 2000, p. 241-246.(Jeroen C. Ijsebaert, Kees)