The Use of AFM and Surface Energy Measurements to Investigate Drug-Canister Material Interactions in a Model Pressurized Metered Dose Inhaler Formulation

The aim of the project was to investigate the interactions between micronized salbutamol sulphate, budesonide, and formoterol fumarate dihydrate and different canister surfaces materials (Aluminium, anodized aluminium, perfluoroalkoxy, fluorinated ethylene propylene—polyether sulphone, and polytetrafluoroethylene) used in pressurized metered dose inhalers (pMDIs).

The surface component approach for polar and apolar interfacial interactions was used to predict the adhesion behavior of micronized drugs with the inner surfaces of pMDI canisters. This was achieved using a combination of in situ colloid probe atomic force microscopy (AFM) measurements and theoretical treatment of the surface free energy measurements, via a contact angle–based technique of the interacting surfaces.

All three drugs exhibited similar dispersive surface energy free values. A greater variation was, however, found in the polar component of the surface free energy measurements. These results were also reflected in the dispersive and polar components of the canister materials. Moreover, the linear relationship between the work of adhesion and AFM measured adhesion was shown to be correlated on the polar contributions of the surface free energies of the interacting materials. AFM measurements indicated that salbutamol sulphate was found to have the strongest adhesive forces with respect to the canister surface materials while budesonide and formoterol fumarate dihydrate appeared to have similar adhesive characteristics. The present study suggests that investigations into the design and characterization of pMDI formulations would benefit from considerations of the polar contribution of the surface free energy and relative work of adhesion of the drug and various components of a pMDI system.

INTRODUCTION

The majority of pressurized metered dose inhaler (pMDI) formulations are based on suspensions of micronized drug in propellant, with or without stabilizing excipients. Historically, device materials used with both chlorofluorocarbon (CFC) and hydrofluoroalkane (HFA) propellants have been found to cause suboptimal operation due to elastomer swelling, polymer extraction, poor lubrication, adsorption, and adhesion of the formulation to container walls (Tiwari et al. 1998). Examples of recent patents dealing with these issues include US 6,143,277 (Ashurst et al. 1996) which describes coating the inside of aerosol containers with fluorocarbon and non-fluorocarbon polymers.

Within a suspension-based pMDI system, there is a considerable risk for irreversible adhesion of micronised drug particulates to internal canister wall. For example, an increase in the force of adhesion between drug particles and canister walls may result in a decrease in emitted dose (Michael et al. 2001; Young et al. 2003). When considering low dose medicaments (such as formoterol fumarate dihydrate: approximately 6–12 μ g/dose) the possibility of drug loss to the canister wall becomes critical, since the internal surface area of the canister may theoretically approach the projected surface area of the total suspended drug.

To date, research has focused on the nature and range of the interactions presents in aqueous and non-aqueous systems. However, only a limited amount of research has been conducted on investigating the physical interactions between drug particles and pMDI canister surfaces (Rogueda et al. 2003). Furthermore, such observations have generally been empirical and have studied the subject in a semi-quantitative manner (Young and Buckton 1990; Chiboswki et al. 1992; Parson et al. 1992; Podczeck et al. 1996; Young et al. 2003). Consequently, a more rational approach into the understanding of these interactions would be beneficial in formulating pMDI suspensions.

The primary objective of this quantitative study is to assess the direct contribution of the relative interactions between drug particulates and a series of pMDI canister materials. Furthermore, by applying the Surface Component Approach (SCA) model for colloidal systems (Traini et al. 2005), it may be possible to relate theoretical adhesion measurement to practical measurements of the surface energies of interacting surfaces. More specifically, the study involved an attempt to correlate the interfacial free energy of interaction (thermodynamic work of adhesion), modelled using the van Oss theory for interfacial forces in non-aqueous media and the surface energy of different polymer coated and non-coated aluminium canister walls (measured using conventional contact angle methods) with direct adhesion measurements of micronised drug probes on respective surfaces (using atomic force microscopy, AFM).

Surface Component Approach for Particle Interaction

The traditional DLVO theory (Derjaguin and Landau 1941; Verwey and Overbeek 1948) is based on the assumption that interactions between two molecules or particles suspended in a liquid consists of the sum of the attractive London–van der Waals (LW) dispersion forces, and the predominantly repulsive electrostatic interaction (electrical double layers). However, such theory has not been fully validated in non-aqueous pMDI media (Vervaet and Byron 1999; Smyth 2003).

In the system under investigation the interactions involved are more complex: London–van der Waals forces, electrostatic double layer interactions and Lewis acid/base interactions (comprise of hydrogen bonding and entropic contribution) (van Oss et al. 1988; van Oss 1994) all come into play when two particle surfaces approach closer than a few nanometers in distance. Due to the overlay of a very diffuse electrical double layer (derived from a combination of a low dielectric constant and low ionic strengths), the electrostatic repulsive forces acting between particles are very small (Kitahara 1974; Pugh et al. 1983; Wyatt and Vincent 1989). It is predicted, therefore, that the LW and surface polarity factors will predominate in such systems. With the aim of understanding adhesion of drug particulates to canister walls in non-aqueous systems, the theory for interactions of this kind of system is reviewed in the following section.

A more complete overview of the surface component approach (SCA) can be found in the literature (van Oss et al. 1988; van Oss 1994; Traini et al. 2005). Briefly, the SCA model for the stability of colloidal particles considers two types of interactions: the apolar Lifshitz–van der Waals component (which comprises the dispersion, as well as the induction and orientation contributions to the van der Waals interactions) and a polar, or Lewis acid-base component.

As the apolar and polar components of the free energies of interfacial interaction is additive (Fowkes 1983; van Oss et al. 1988), the total surface free energy of a solid is determined by the sum of the apolar (γ^LW) and the polar (γ^AB) component.

Therefore, according the Good-Girifalco-Fowkes combining rule (Good and Girifalco 1960; Fowkes 1964), the interfacial energy parameters between dissimilar solid surfaces (1 and 2) in an apolar medium (3), γ ₃ ⁺ and γ₃ ⁻ = 0), the free energy of interaction can be used in the simplified form (Equation [1]):

Hence, with knowledge of the surface energy of a solid (γ₁ ^LW,γ ₁ ⁺ and γ₁ ⁻) and the surface tension of the liquid (γ ₃ ^LW,γ₃ ⁺ and γ₃ ⁻), the free energy of interaction for the LW and AB forces can be calculated.

Accordingly, the force of adhesion can be related to the thermodynamic work of adhesion (W_ad = ΔG₁₃₂) by applying one of the contact mechanics models. The Derjaguin-Müller-Toporov (DMT) model is typically applied to small particles with a high stiffness and a small curvature radius, while the Johnson, Kendall, and Roberts (JKR) model is applicable for systems with large particle radii, high surface energies, and compliant materials (Johnson et al. 1971; Derjaguin et al. 1975). Both models are valid depending on the interacting materials and their geometries.

The relationship between the force of adhesion (F_ad) and the thermodynamic work of adhesion (W _ad) is represented in the following equation:

Where R* is the contact radius of the particle radii and n (n = 3/2 for JKR and n = 2 for the DMT) is a pre-determined constant depending on the selected model.

Theoretical Estimation and Direct Surface Energy Measurements Correlation

This study investigated the correlation of the interfacial free energy of interaction, developed using the theoretical SCA model in non aqueous media, and particle adhesion measurements on different polymer coated and non-coated aluminium canister walls. Theoretical adhesion values were calculated from contact angle measurements, while in situ adhesion measurements of micronized drug probes on the respective surfaces were conducted using atomic force microscopy (AFM) (Binnig and Quate 1986).

MATERIALS AND METHOD

Micronized salbutamol sulphate (salbutamol), budesonide, and formoterol fumarate dihydrate (formoterol) were supplied by AstraZeneca (R & D Charnwood, Loughborough, UK). Water was produced by reverse osmosis (Millipore, Molsheim, France). All organic solvents were supplied by BDH (Poole, UK) and were of analytical grade. The model propellant, 2H, 3H decafluoropentane (mHFA), was supplied by Apollo Scientific (Derbyshire, UK) and was purified according method described elsewhere (Goebel and Lunkenheimer 1997). The subsequent purity of the mHFA was in excess of 99.9%, with moisture content less than 9 ppm. Water diffusion into the system could have profound impact on the propellant polarity and consequently particle adhesion (Blondino and Byron 1998; Miller 1990). Subsequently, fresh samples of mHFA were used for each experiment. The canister materials were: Aluminium (ALU), anodise aluminium (AN-ALU), and polymer coated pMDIs (perfluoroalkoxy (PFA), fluorinated ethylene propylene–polyether sulphone (FEP–PES) and polytetrafluoroethylene (PTFE)) (Presspart, Lancashire, UK). All were used as supplied.

Scanning Electron Microscopy of pMDI Canister Materials

Scanning electron microscopy (SEM) was used to characterize the morphology of the canister surfaces. Samples were fixed on adhesive black carbon tabs, which were pre-mounted on aluminium stubs and coated with a thin gold film using a sputter coater (model S150B, Edwards High Vacuum, Sussex, UK). Samples were imaged using a JEOL 6310 SEM (Jeol, Tokyo, Japan) at 10 KeV.

Surface Morphology of the Canister Materials

Previous investigations have suggested that canister morphology significantly influences the drug-surface adhesion (Young et al. 2003). The morphology of the substrates was therefore checked by AFM. Surface materials were cut from commercially available pMDI canisters, using metal snips, to produce ∼ 1 cm² substrates suitable for analysis. Each substrate was mounted on an AFM sample stub with epoxy resin, washed with isopropyl alcohol (commonly used to clean surfaces in the pharmaceutical industry), and dried with nitrogen prior to analysis.

Detailed topographical information of the three drug crystal surfaces was acquired by AFM. Imaging was conducted in air, using Tapping Mode^TM with a high-aspect-ratio silicon probe (OTESP, Digital Instruments, UK), at a scan rate of 0.7 Hz.

All AFM studies were performed using a commercially available Multi Mode AFM with a Nanoscope III controller (Digital Instruments (DI), Cambridge, UK). The root mean squared surface roughness (R_rms) of the deviations was then calculated from the AFM height data over a 5 μ m × 5 μ m area via the following Equation (3):

where, n is the number of points in topography profile and y_i is the distance of asperities (i) from the center line.

Model Drugs Compact Preparation

Drugs surface energy measurements were obtained from compacts. Model surfaces from the micronized drug materials were prepared by direct compression using a servo-hydraulic press (Specac, Model 25010, Specac Ltd., Kent, UK). Approximately 250 mg of micronised drug material was weighed into a 10 mm stainless steel die, spread evenly in an attempt to ensure constant porosity throughout the disc, and compacted at a compression force of 10 kN. Prepared drug compact surfaces were stored in tightly sealed containers in a controlled environment (25°C, 44% RH) for at least 24 hours prior to use.

Contact Angle Measurements for Model Drugs Compacts and Canister Surfaces

Surface free energy of the canister surfaces and drug materials were investigated using the sessile drop method (Good and Girifalco 1960; Good and Stromberg 1979; van Oss et al. 1988; Good 1992; Buckton et al. 1995). The canister materials were washed with methanol and dried using filtered nitrogen prior to analysis. However, the drug materials were formed into model compacts using a method described above.

Equilibrium advancing contact angle (CA) of each sample was measured with a NRL goniometer (Rame'-Hart Inc., Mountain Lakes, New Jersey, USA) equipped with a 2.3X objective lens and a 10X Ramsden type eyepiece. Measurements were conducted under ambient conditions (25°C, 44% RH) using three liquids (water, diiodomethane and ethylene glycol) with known non-polar, acid (electron acceptor), and base (electron donor) surface tension components. The total surface free energy (γ^Tot) of each material was then calculated from the sum of non-polar and polar components (van Oss et al. 1998) using the Young-Duprè equation (4):

where γ_L ^LW and γ_s ^LW are the dispersive contribution of the surface free energy for the liquid and solid, respectively; γ_s ^+/− and γ _L ^+/− are the polar contribution of the surface free energy for the liquid and solid, respectively.

In Situ Atomic Force Microscopy Measurements

The interactions of drug particles with canister surfaces, were evaluated using AFM equipped with an in situ cell (DI, Cambridge, UK), using mHFA. The mHFA contained a saturated solution of the respective drugs under investigation (Traini et al. 2005). It is essential to note that conventional AFM systems are currently limited to studies in air and/or low vapour pressure liquid environments. Thus, mHFA was chosen for its similarities to the physico-chemical properties of HFA227, used in pMDIs (Rogueda 2003, Table 1).

TABLE 1 Comparative table of some properties of propellants mHFA and HFA227

	mHFA	HFA227
MW (g mol^−1)	252.051	170.03
nd	1.26089 ± 0.0008	1.2207
d (g cm^−3)	1.583	1.388
b.p. (°C)	53.6	−16.5
m.p. (°C)	−80	−131
ϵ	15.05 at 18.66 kHz	4.071 (3.9 bar)
γ (mN m^−1)	13.59	7.55
η (mPa s)	0.537	0.266
Water solubility (ppm)	390 ± 40	610
(Modified from Rogueda 2003).

The use of mHFA as a model system for studying the behavior of pMDIs, when characterization methods cannot be adapted to the pressure regimen, has been extensively studied by Rogueda (2003).

The AFM is an ideal tool for the single particle studies because it can provide quantitative information on both the separation energy (via force–distance measurements) and surface geometry (via topographic images). Individual micronised drug particles were mounted onto micro-fabricated AFM cantilevers (nominal spring constant 0.38 N · min^{− 1}) using a quick setting epoxy resin. Extreme care was taken during drug probe preparation to limit the amount of drug-glue contact. Detailed methodology for drug probe preparation and measurements are described elsewhere (Ducker et al. 1991; Young et al. 2004).

The integrity of all three drug probes were investigated prior to and post measurement using a high-magnification 500× long-working-distance reflective microscope. Drug probes appeared visibly proud of the cantilever surface, with no observable differences between the start and end of the experimental procedure (indicating no macroscopic change in drug probe morphology).

The interaction (adhesion) between each drug probe and canister surface material was measured using the colloid probe approach under in situ conditions with mHFA (Ducker et al. 1991; Young et al. 2003) by ramping the drug probe, toward, in contact, and away from the surface. This produced a force-distance curve which could be examined to produce the force of adhesion and energy of separation.

A total of 4096 force–distance curves were determined for each individual drug probe and three substrate surfaces (scan area: 10 × 10 μ m) under in situ conditions with the following settings: approach retraction cycle 500 nm, cycle rate 4.07 Hz, and constant compliance region 60 nm and a compressive loading of 23 nN. The area under each force–distance curve between each drug probe and canister surface was integrated to obtain separation energy values. The adhesion energy between each canister surface and drug was repeated in triplicate (n = 3).

A drawback in the use of the colloid probe technique in adhesion studies is the fact that the actual size and geometry of the AFM probe is unknown and, therefore cannot be normalized. Consequently, variability in probe contact radius geometry is to be expected. Accordingly, to avoid significant variations in contact area between each probe and the model canister surface, the same drug probes (3 for each drug) were used for all the measurements and a great deal of care and attention were taken to maintain the integrity of the colloid probe throughout the study. This allows the inter-relationship between each probe and the various substrate surfaces to be elucidated.

Statistical Analysis

The statistical analysis of the energy of interaction between individual colloid probes and the respective substrates were compared using one-way analysis of variance. Levels of significance were based on 95% probability values (ANOVA, Fisher's pairwise p < 0.05).

RESULTS AND DISCUSSION

Characterization of Surface Morphology

To fully understand the relationship between surface free energy (γ), work of adhesion (W) and drug–surface interactions in a mHFA system, canister surfaces were first investigated using tapping mode AFM and SEM. Representative AFM surface plots of the ALU, AN-ALU, PFA, FEP-PES and PTFE and relative scanning electron micrographs of the different canister surfaces are shown in Figure 1.

FIG. 1 Taping mode height images (right) and SEM microphotographs (left) of ALU, AN-ALU, PFA, FEP-PES, and PTFE canister surfaces.

In general, SEM inspection of the canisters surface morphology suggested ALU and AN-ALU to have relatively planar surfaces containing linear striations similar to that reported previously (Young et al. 2003). In comparison, PTFE had a homogeneous surface, while both PFA and FEP-PES presented creased, lined surfaces, probably due to the non-uniformity of the applied coating. In addition, the height data obtained by AFM was processed to produce root mean squared roughness values. Analysis of the R_RMS values over multiple 5 μ m × 5 μ m areas of each surface suggested statistical differences (ANOVA p < 0.05) (n = 25). The surface roughness measurements are shown in Table 2.

TABLE 2 Analysis of the root mean square surface roughness (R_RMS) values of each canister surface materials (n = 25, ± Std)

Materials	RRMS(nm) 5 μ m × 5 μ m	Standard deviation (±)
ALU	37.36	8.66
AN-ALU	20.93	4.43
PFA	53.01	16.23
FEP-PES	20.48	7.24
PTFE	44.93	16.05

The rank order of the surface roughness for the canister surfaces were as follows: PFA > PTFE > ALU > AN-ALU > FEP-PES.

Variations in surface roughness between canister surfaces were expected. It has been previously reported that a roughness value of less than 100 nm will not have a significant influence on surface energy measurements (Buckton et al. 1995). The critical effects of surface rugosity and more specifically surface asperities on particle adhesion is still an issue of debate.

Determination of Contact Angle and Relative Surface Energy Values for Drug Compacts and Canister Surface Surfaces

The surface energy components (γ^LW, dispersive and γ⁺, γ⁻ polar) and total surface free energy (γ^TOT) of the canister surfaces and drug compacts were determined using the contact angle method, and are summarized in Table 3.

TABLE 3 Surface energy values for canister surface material and drug compacts calculated by contact angle (mJ · m⁻²) (n = 3, ± STD; Surface energy values of drug compacts previously published in Traini et al. 2005)

	Surface energy (mJ · m^−2)
	γ^LW	γ^+	γ^−	γ^TOT
AN-ALU	43.36	0.54	24.68	50.51
	(0.76)	(0.24)	(2.23)	(0.52)
ALU	40.44	0.79	16.05	47.67
	(0.24)	(0.24)	(1.72)	(1.23)
PFA	17.33	0.42	10.36	21.49
	(0.78)	(0.11)	(0.91)	(0.16)
FEP-PES	24.19	0.02	6.64	24.68
	(0.87)	(0.02)	(0.23)	(0.47)
PTFE	18.88	3.14	2.04	23.9
	(0.8)	(0.57)	(0.27)	(0.59)
Salbutamol	46.49	8.25	18.48	71.12
	(0.69)	(1.46)	(1.16)	(2.97)
Budesonide	49.07	0.34	22.47	53.66
	(0.39)	(0.43)	(3.78)	(2.75)
Formoterol	48.51	0.11	35.04	51.69
	(0.40)	(0.15)	(3.18)	(3.24)

The dispersive γ^LW surface energy parameter for the three drug materials suggested no significant difference.

Greater differences were found in the electron-donor (γ⁻) and electron-acceptor (γ⁺) surface energy parameters. Analysis of the polar components indicates a difference in rank order for γ⁺: salbutamol > budesonide > formoterol and for γ⁻: formoterol > budesonide > salbutamol. Generally, the total specific surface energy values γ^Tot showed salbutamol to have the highest value, while budesonide and formoterol presented lower values that were not significantly different.

When comparing canister surfaces, significant differences in the dispersive γ^LW surface energy were observed. Analysis of the data indicated a rank order for dispersive γ^LW surface energy parameter of: AN-ALU > ALU > FEP-PES > PTFE > PFA, In addition, ALU and AN-ALU had γ^LW values were approximately double those of the polymer coated canisters. A reasonable explanation for such difference in surface energy values may be related to the chemical nature of the coating material used. It is generally agreed that surface free energy decreases with increased fluorine content (Zisman 1963).

Analysis of the polar components of the surface free energy indicated a difference in rank order for γ ⁺: PTFE > ALU > AN-ALU > PFA > FEP-PES, and for γ⁻: AN-ALU > ALU > PFA > FEP-PES > PTFE. Overall, the total specific surface energy values γ^Tot showed a ranking decrease of: AN-ALU > ALU > FEP-PEF > PTFE > PFA. As expected, greater values of γ^Tot were observed for the non-polymer coated canister surfaces.

Determination of the Theoretical Work of Adhesion of Different Polymer Coated and Non-Coated Aluminium Canister Walls

The experimentally calculated surface energies of the drugs and can surfaces were modelled using the SCA in non aqueous media to give theoretical values for the work of adhesion between each drug and canister surface (Traini et al. 2005). The dispersive component of the thermodynamic work of adhesion (of the three drug particulates against the respective canister surfaces) together with the acid/base, and total component, in the presence of mHFA are presented in Table 4.

TABLE 4 Theoretical thermodynamic work of adhesion (mJ · m⁻²) (Δ G₁₂₃) calculated using the SCA model between the drug materials and canister surfaces

	Theoretical thermodynamic work of adhesion (mJ.m^−2)
	Salbutamol			Budesonide			Formoterol
Materials	γs ^LW	γ^AB	γ^TOT	γs ^LW	γ^AB	γ^TOT	γs ^LW	γ^AB	γ^TOT
AN-ALU	18.16	34.85	53.00	19.24	12.78	32.02	19.01	11.94	30.95
ALU	16.74	30.66	47.40	17.74	13.13	30.87	17.52	13.06	30.68
PFA	2.98	24.08	27.06	3.16	9.94	13.10	3.12	9.79	12.92
FEP-PES	7.72	15.88	23.59	8.18	4.22	12.38	8.08	3.18	11.26
PTFE	4.14	23.42	27.55	4.38	18.46	22.84	4.32	21.89	26.22

Surface tension parameters for mHFA were: γ^LW = 13.59, γ⁺ = 0 and γ⁻ = 0 mJ/m². Comparison of the LW and AB work of adhesion, for each drug with canister surfaces, suggested adhesive forces are controlled by the combination of dispersive and polar interactions. This is exemplified when comparing the rank order in predicted adhesion using dispersive or polar and dispersive calculations. For example, calculation of the work of adhesion for salbutamol with can surfaces shows a rank order for dispersive forces as: AN-ALU > ALU > FEP-PES > PTFE > PFA. However, taking into consideration the acidic and basic component of the free energy, the rank order is: AN-ALU > ALU > PTFE > PFA > FEP-PES. Clearly, the potential for such modification in total surface free energy may significantly influence the particle adhesion in and subsequent formulation performance.

Analysis of budesonide and formoterol theoretical work of adhesion (Table 4), suggested similar changes in rank order, when combining dispersive and polar components. With the aim of quantifying the relative contribution of the various surface free energy components (of both substrate surfaces and drugs) to the thermodynamic work of adhesion of the interacting materials, via the SCA theoretical approach, the values were directly compared with separation energy measurements using in situ AFM.

In Situ Atomic Force Adhesion Measurements of Polymer Coated and Non-Coated Aluminium Canister Walls

Measurement of adhesive forces and separation energies between individual particles and substrate surfaces by in situ AFM, and the influence of various interactive forces on particle adhesion, have been reported previously (Ashayer et al. 2003; Young et al. 2003; Hooton et al. 2004; Young et al. 2004). However, a direct link between the macroscopic measurements of surface energy, theoretical predictions of the thermodynamic interfacial interactions and particle adhesion have not been investigated to such an extent (Begat et al. 2004; Traini et al. 2005).

An obvious issue to consider when comparing the adhesion of a drug particle on a series of can substrates is the variation in roughness. Previous studies, have suggested an increase in surface roughness gives rise to a log-normal distribution in adhesion measurements (Price et al. 2002; Young et al. 2004) in comparison to normally distributed measurements for highly smooth substrate surfaces (Price et al. 2002; Buckton et al. 1995).

Frequency distribution histogram measurements of the separation energy between the colloid drug probes and canister surfaces each exhibited a log-normal distribution. These data suggest therefore that the surface roughness of the canister material does directly influence the adhesion measurements, and its distribution, under in situ conditions (Price et al. 2002; Young et al. 2004). As previously stated, 3 drug probes were utilised to measure the interactions with all substrates to allow rank correlation of the AFM measurements. It is interesting to note, however, that surface energy dominated the system and the effects of roughness did not play such a significant role.

Median separation energy values (e_0.5) were calculated for each individual specific log-normal distribution. A summary table of the AFM measurements for salbutamol, budesonide, and formoterol are presented in Table 5, 6, and 7, respectively.

TABLE 5 AFM median separation energy values for salbutamol drug probes on canister surfaces (Tips 1–3)

	Separation energy (× 10^−18 J)
Salbutamol	Tip 1	Tip 2	Tip 3
AN-ALU	86.40	43.20	80.10
ALU	79.10	28.60	52.80
PFA	1.63	7.46	5.38
FEP-PES	2.84	9.11	6.41
PTFE	4.15	13.10	12.80

TABLE 6 AFM median separation energy values for budesonide drug probes on canister surfaces (Tips 1–3)

	Separation energy (× 10^−18 J)
Budesonide	Tip 1	Tip 2	Tip 3
AN-ALU	254	285	56.10
ALU	139	52.80	22.50
PFA	11.80	7.58	2.77
FEP-PES	4.60	5.85	0.58
PTFE	13.90	13.80	4.53

TABLE 7 AFM median separation energy values for formoterol drug probes on canister surfaces (Tips 1–3)

	Separation energy (× 10^−18 J)
Formoterol	Tip 1	Tip 2	Tip 3
AN-ALU	39.70	26.10	48.00
ALU	1.66	12.00	18.00
PFA	0.76	10.30	3.58
FEP-PES	0.59	7.29	1.93
PTFE	2.07	22.80	31.60

The AFM data for salbutamol drug probes suggested that the force of adhesion followed the following decreasing rank order of: AN-ALU > ALU > PTFE > FEP-PES > PFA for all three probes. For budesonide drug probe, the rank order was: AN-ALU > ALU > PTFE > PFA > FEP-PES. Finally, for formoterol drug probe, the rank order was: AN-ALU > PTFE > ALU > PFA > FEP-PES.

Comparisons Between Thermodynamic Work of Adhesion and AFM Measurements

From Equation (2), it is intuitive to derive that the thermodynamic work of adhesion is directly proportional to the force of adhesion. Thus by plotting the measured separation energy against the theoretical surface energy/work of adhesion, it may, therefore, become possible to correlate theoretical measurements of the interfacial free energy of interaction with experimental measurements of the adhesive interaction.

For comparison, the relationship between the surface free energy calculated by CA (LW-CA), the theoretical work of adhesion (using the dispersive surface free energy only) and the total SCA theoretical work of adhesion (using dispersive and polar contributions of the surface free energy) with AFM separation energy measurements were plotted for salbutamol, budesonide, and formoterol in Figures 2 (A, B, and C), 3 (A, B, and C), and 4 (A, B, and C), respectively.

FIG. 2 Plot of salbutamol drug probe AFM separation energy against γ_LW dispersive surface free energy calculated by CA (A), γ_LW dispersive work of adhesion value (B), and Total work of adhesion (W_ad ^Tot) (C), calculated using the SCA model the for the 5 canister materials. (n = 3 drug probes, • = Tip 1; ○ = Tip 2; ▾ = Tip 3.)

FIG. 3 Plot of budesonide drug probe AFM separation energy against γ_LW dispersive surface free energy calculated by CA (A), γ_LW dispersive work of adhesion value (B), and Total work of adhesion (W_ad ^Tot) (C), calculated using the SCA modelthe for the 5 canister materials. (n = 3 drug probes, • = Tip 1; ○ = Tip 2; ▾ = Tip 3.)

FIG. 4 Plot of formoterol separation energy against γ_LW dispersive surface free energy calculated by CA (A), γ_LW dispersive work of adhesion value (B), and Total work of adhesion (W_ad ^Tot) (C), calculated using the SCA model the for the 5 canister materials. (n = 3 drug probes, • = Tip 1; ○ = Tip 2; ▾ = Tip 3.)

Comparison of the AFM adhesion measurements with theoretical values for salbutamol indicated a direct relationship with surface free energy measured using the CA method (with a correlation coefficient of the linear relationship equal to R² = 0.7952, R² obtained from the mean value of the three probes). However, when considering the theoretical work of adhesion (calculated using the SCA method), dispersive components resulted in a less significant relationship (R² = 0.7020). In comparison, when assessing the relationship between adhesion measurement and the total theoretical adhesion values a more positive relationship was observed (R² = 0.9260). Such observations suggest the SCA results in an improved relationship between theory and experimental values, when considering interactions between drug particles and different canister materials.

For example, when considering interactions between salbutamol sulphate and energetically similar polymers, FEP-PES presents the lowest adhesion measurements. Such observations, however, are not equivalent to the rank order of polymer surface energy (LW), where PFA < PTFE < FEP-PES. In comparison, when utilising the SCA, the rank order of predicted adhesion and measured separation energy become analogous. Similar results were observed for budesonide and formoterol interactions.

In general, the linear relationship between theoretical and measured adhesion was observed to improve when comparing theoretical surface free energy and SCA values. Subsequently, when comparing these results with the total theoretical SCA work of adhesion (in mHFA), a similar rank order to that of AFM was observed. Such observations highlight the importance of including the dispersive and AB components of the drug polymer and non-aqueous media when considering formulation development. These latest findings were unforeseen and corroborated a still highly debated opinion that although non-aqueous, partially fluorinated liquids are by no means apolar. Even for budesonide and formoterol the correlation coefficient was observed to increase from R² = 0.6947 to 0.7133 for budesonide and from 0.6768 to 0.8012 for Formoterol giving an indication of the improved relationship between theory (derived by SCA) and experimental values (measured by AFM and CA).

CONCLUSIONS

The present study suggests that it may be possible to predict the interactions in pMDIs systems from thermodynamic considerations of the surface free energy measurement of the interacting materials, derived using the theoretical surface component approach (SCA). This investigation provided further experimental evidence for the existence of non-DLVO type of interactions in pMDI suspension based colloidal systems. Whilst the classical DLVO theory could not satisfactorily illustrate the above phenomena, the inclusion of Lewis AB forces in the SCA model, correlated well with the observed drug particulates behavior different substrate interactions. In addition, this investigation demonstrates that partially fluorinated liquids are not non-polar, and their weak polarity plays an important role in dictating interactions in this kind of pMDI system. Although good correlations were observed between SCA and particle adhesion, it is important to note topography will influence direct comparison. The link between roughness SCA and adhesion is currently under investigation by the authors.