- Email: [email protected]
Surface roughness contribution to the adhesion force distribution of salmeterol xinafoate on lactose carriers by atomic force microscopy
Surface Roughness Contribution to the Adhesion Force Distribution of Salmeterol Xinafoate on Lactose Carriers by Atomic Force Microscopy NAZRUL ISLAM,1 PETER STEWART,1 IAN LARSON,1 PATRICK HARTLEY2 1
Department of Pharmaceutics, Victorian College of Pharmacy, Monash University, Parkville, Victoria 3052, Australia
2
Division of Molecular Science, CSIRO, Clayton, Victoria 3168, Australia
Received 30 September 2004; revised 23 January 2005; accepted 24 January 2005 Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.20381
ABSTRACT: Adhesion force distributions of silica spheres (5 and 20 mm) and salmeterol xinafoate (4 mm) particles with inhalation grade lactose surfaces and spin coated lactose films were determined by atomic force microscopy (AFM) to investigate the influence of surface roughness on the force distributions. The roughness of lactose particles and films was determined by both AFM and confocal microscopy (CM); the lactose particles showed RMS Rq values between 0.93 and 2.2 mm. The adhesion force distributions for silica and SX probes were significantly different for the different lactose carriers and broad, e.g., the adhesion force distribution between a 5 mm silica sphere and lactose particles ranged from 5 to 105 nN. This contrasted with distributions on smooth spin coated lactose films (RMS Rq of 0.28 nm) which were not significantly different and were narrow, e.g., the adhesion force distribution between a 5 mm silica sphere and spin coated lactose films was between 42 and 68 nN. In addition, no significant difference in adhesion force distribution occurred with silica probe size on the lactose carrier surface. The use of X-ray photoelectron spectroscopic analysis confirmed that the lactose surfaces were free of impurities that might contribute to variation in adhesion. Although the almost atomically flat films showed some adhesion variability, the surface roughness of the lactose particles was a major contributing factor to the broad distributions seen in this study. ß 2005 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 94:1500–1511, 2005
Keywords: roughness
AFM; adhesion force distribution; salmeterol xinafoate; lactose; surface
INTRODUCTION Inter-particular interactions in dry powder inhaler (DPI) formulations, containing micronised drug particles and lactose carriers, influence the efficiency of drug dispersion. It is likely that both the magnitude of the force of adhesion and the adhesion force distribution will effect the dispersion of drugs from interactive powder mixtures of drug and carrier. The adhesion force distribution will not only be influenced by the intrinsic Correspondence to: Peter Stewart (Telephone: 61 3 99039517; Fax: 61 3 99039583; E-mail: [email protected]) Journal of Pharmaceutical Sciences, Vol. 94, 1500–1511 (2005) ß 2005 Wiley-Liss, Inc. and the American Pharmacists Association
1500
variability in force between the interacting particles due to surface chemistry effects,1,2 but will also be influenced by the physical nature of the surfaces.3,4 The adhesion force will comprise a number of force components including contact potential, Coulombic, capillary, and intermolecular forces.5,6 Contact potential forces involve the interaction between contiguous surfaces with differing work functions that result in the development of a contact potential between the surfaces, resulting in an adhesion force. Coulombic interactions involve the interaction between oppositely charged particles or between a charged particle and a neutral surface. Capillary interaction requires the formation of a liquid between the contacting surfaces with the resultant capillary
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005
ADHESION FORCE DISTRIBUTIONS
attraction related to surface tension of the liquid bridge and its contact angle with the surfaces. Intermolecular interactions involve van der Waals forces and hydrogen bonding between the interacting particles. The inherent variability of interaction between surfaces will originate from the differing chemistry of the surface and will be related to specific interactions of the differing crystal faces due to molecular packing,7,8 differing adsorbed moisture contents,9 and differing distribution of amorphous/crystalline surface domains.8,10 These factors will influence the electrostatic charging potential, intermolecular interactions, and liquid bridge formation of the interacting surfaces and lead to an inherent force distribution even for atomically smooth surfaces. The interactive force distribution will also be affected by extrinsic factors, in particular the surface morphology, contact geometry, and adsorbed impurities. Adhesion forces in various systems have been determined by vibration,11 centrifugation,12,13,15 impact separation,14,16,17 and recently by atomic force microscopy (AFM).18 The AFM colloid probe technique enables the direct measurement of adhesion force between single particles and surfaces, and is becoming the method of choice to characterise particle adhesion in pharmaceutical systems. Recently, average adhesion forces in pharmaceutical powders, such as between a single small lactose particle and other lactose particle surfaces and compacted lactose disks,19,20 between zanamivir drug particles and lactose carriers,21,22 between beclometasone dipropionate (BDP), and a series of untreated and modified lactose surfaces23 have been reported. However, while many replicates were determined, force distributions were not shown in these papers. In other studies, typical log-normal force distributions, characterised by geometric means were observed for adhesion between a silica colloid probe and a lactose particle24 and the cohesive force between a single particle of salbutamol sulphate and compacts of sulbutamol sulphate at different relative humidity25 were measured. However, in these studies the reasons for the distribution of adhesion forces in terms of intrinsic and extrinsic factors were not addressed. A number of studies have addressed the influence of surface roughness on the force distribution. The influence of surface roughness on the force distribution between single particles and both smooth and rough substrates has been reported.26–28 Wide force distributions between spherical polystyrene particles and polished and
1501
PVD-coated aluminium substrates were observed; whereas, a narrow distribution occurred between the same particle and the smooth surface of a silicone wafer.26 Similar results also occurred with force measurement between spherical gold particles and rough substrates (silica and stainless steel) and smooth substrates (polished silica) were investigated.27 The authors of both studies concluded that the distribution of forces (narrow or broad) depended upon the distributions of asperity radius or the effective contact area of the contacting surfaces. Wide distributions of force at different RHs (30%–90%) between a spherical steroid particle, prepared by a crystallization method, and a lactose surface prepared by recrystallising lactose on a plate were observed; however, force distributions between the drug and smooth stainless steel substrates were narrower than those of the former substrate due to the variation in surface roughness of the substrates.28 Normal distribution of the adhesion force between salbutamol sulphate and budesonide against a surface modified lactose substrate at the RHs of 15% and 75% were observed.29 The authors also found log-normal force distributions and high variability in force measurement with increased substrate surface roughness, thought to be due to the variability in contact area between drug and substrates. The wide distributions of adhesion forces between pharmaceutical powders (beclomethasone dipropionate and lactose crystal) and rough polymeric surfaces (polypropylene, polycarbonate and acrylonitrilebutadine-styrene) observed were due to the roughness of interacting surfaces.30 Therefore, while the broad adhesion force distributions between interacting surfaces, determined using AFM, have been attributed to the morphology of surfaces, in particular, the surface roughness, the distinction between the extrinsic adhesion due to surface roughness and the true intrinsic adhesion has not been determined. Thus, surface roughness has been shown to contribute to broad force distributions, but the extent of this contribution is not known; the contribution of the chemistry of the surface to intrinsic force distributions is also not known. Thus, the purpose of this research was to study the influence of surface roughness on the adhesion force distribution using samples of inhalation grade lactose and probes of salmeterol xinafoate and silica. Intrinsic force distributions were estimated by eliminating surface roughness effects through spin coating of the lactose samples to produce almost atomically smooth surfaces. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005
1502
ISLAM ET AL.
EXPERIMENTAL Materials Salmeterol xinafoate (SX), micronized (inhalation grade; <4 mm), was obtained from GlaxoSmithKline, Melbourne, Australia. Aeroflo 95, Aeroflo 65, and Aeroflo 20, inhalation grades of a-lactose monohydrate, were donated by Foremost Farms, Rothschild, WI; Inhalac 70, Inhalac 120, Inhalac 230, and Sorbolac 400 were donated by Meggle GmbH, Germany; Pharmatose 125M and Pharmatose 110M was donated by DMV International, Veghel, The Netherlands, and Lactose Special dense, a non inhalation grade was donated by Lactose New Zealand, Hawera, New Zealand. Ammonium acetate (Analar, BDH, Victoria, Australia), methanol (HPLC grade, Biolab, Victoria, Australia), absolute alcohol (HPLC grade, CSR, Victoria, Australia) were used as supplied. Methodology Decantation and Removal of Fine Lactose About 20 g of lactose was dispersed (4–5 min) in absolute ethanol pre-saturated with lactose to make a homogeneous suspension and then allowed to settle for a predetermined time. The supernatant liquid was then decanted and replaced by fresh saturated ethanol, vigorously mixed and the process repeated until the supernatant liquid was clear. During removing the supernatant, special care was taken to ensure minimum disturbance of the lower part of suspension. The coarse particles remaining in the beaker after 15 cycles of decantation were wet sieved by spraying with saturated ethanol three times. Finally the samples were dried at room temperature, 21 18C. For first three repeats, 10 min interval was used and then 5 min interval was applied until supernatant liquid was clear. Measurement of Adhesion Force The adhesional forces were measured by atomic force microscopy (Nanoscope Multimode IIIa, Digital Instruments, Santa Barbara, CA) using the colloid probe technique18 in air and ambient humidity. Sample Preparation Lactose Particle. Araldite Five-minute Epoxy Resin (Selleys Chemical Company Pty. Ltd., JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005
Melbourne, Australia) was mixed, spread onto a clean silicon wafer ensuring a relatively uniform thin film and allowed to dry for 4 min. Lactose particles were sprinkled onto the resin by blowing the lactose particles with a gentle nitrogen flow. The samples were dried overnight and any unattached particles were removed by blowing with nitrogen. This sample was stuck onto a stainless steel sample stub with double-sided adhesive tape (Scotch, 3M Australia Pty Ltd., Sydney, Australia). Lactose Film (Spin Coating). Several pieces of silicon wafer of 1 1 cm2 size were placed in a beaker containing a washing solution (surfactant and ethanol) and sonicated for 30 min. After washing several times with milli-Q water, these silicon wafers were further sonicated with ethanol for 30 min. These sonicated wafers were again washed with milli-Q water and kept under a UV lamp for 30 min. One piece of silicon wafer was placed on the centre position of the plate of a PhotoResist Spinner (Headway Research, Inc., Garland, TX) ensuring the smooth side of the wafer was up. After adjusting the spinner speed to 5000 rpm, one drop of lactose solution (20% lactose in water) was dropped onto the wafer by a pasture pipette. The wafer was spun until all solvent was evaporated. These samples were stored in a desicator containing silica gel. All coatings were thicker than the X-ray photoelectron spectroscopy (XPS) sampling depth, i.e., at least 10–15 nm thick since no silicone was observed in XPS measurements. Attaching Colloid Probes to the AFM Tips Silica Probe. Using a micromanipulator (World Precision Instruments, Inc., Sarasota, FL, Model M3301R), a piece of tungsten wire was dipped into a small amount of freshly mixed epoxy resin (Araldite Epoxy Resin 24 hours, Selleys Chemical Company Pty. Ltd., Melbourne, Australia). This process was directed through an optical microscope. After withdrawal from the resin, the tip of the wire was brought in contact with a clean microscope slide to remove excess resin from the wire tip. The resin coated wire tip was brought into contact with the apex of the V-shaped silicon nitride cantilever (NanoprobeTM SPM Tips, Type NP-S, Digital Instruments, Santa Barbara, CA). It was important to transfer only a small amount of resin on the cantilever tip to prevent the immersion of the particle in the wet resin. After cutting the wire tip, it was touched onto a 5 mm (or 20 mm)
ADHESION FORCE DISTRIBUTIONS
diameter silica sphere (Bangs Laboratories, Inc., Fishers, IN) which stuck to the wire tip due to capillary forces. The sphere was transferred to the cantilever tip where it stuck to the resin. Great care was taken so that epoxy resin was not spread around the sphere. SX Probe. Using the method mentioned above, epoxy resin was applied to a cantilever tip. The epoxy-coated cantilever was mounted in the AFM. It was positioned directly above a single drug particle, or a small agglomerate of drug particles, and lowered microscopically until contact occurred. Retraction of the tip enabled initial confirmation of particle attachment. After drying overnight the SX probe was examined under an optical microscope fitted with a video camera to ensure successful attachment. Excess drug particles were removed by blowing with nitrogen. Colloid probes were examined under an optical microscope to ensure attachment to the correct position on the cantilever. The size of the attached particle was determined through a microscope with a video camera attached. The spring constant of the cantilevers was determined by the attachment of known masses to the cantilever and measuring resultant changes in the cantilever’s resonant frequency.31 Force Measurement Collecting Force Data. All force measurements were performed using a Nanoscope1 Multimode IIIa (Digital Instruments, Santa Barbara, CA). The Nanoscope III has a force mode in which the colloid probe is held stationary and the lactose sample, mounted on the piezoelectric tube, is driven in a controlled manner towards and away from the colloid probe. The speed of the approach and retraction can be set by the user. A diode laser and prism/ mirror optical path allow focusing of the laser onto the cantilever, and subsequent gathering the reflected light. A position sensitive photodiode detects the position of the free end of the cantilever due to deflection of the laser light. The software of the Nanoscope generates a file that contains the output of the photodiode and the displacement of the sample. Analysing AFM Force Raw Data. The raw AFM force data can be easily converted into force versus separation plot. To determine the force, zero force and zero separation points need to be defined. Zero
1503
force was defined where the cantilever deflection remained constant (zero deflection line) with sample displacement. The region of zero force should be linear. Zero separation is defined when a change in the sample position causes an equal change in cantilever deflection. This region (also known as the constant compliance region) occurs when the colloid probe is in contact with the sample surface. In this region of compliance the photodiode output is a linear function of the displacement of the sample. It is important to note that the data defined as zero separation were completely linear, as any non-linearity would indicate that the surfaces were not in true contact. Non-linearity of the compliance line may also be an indication of a contaminant on one or both the surfaces. All raw force measurement data presented in this study were converted to force versus separation data using software written by Dr. Patrick Hartley (CSIRO, Melbourne). This program displays the collected data graphically, so that the constant compliance region (defined as zero separation) and a baseline to the zero force can be selected from the raw data. After selecting the zero separation and zero force, a least square fit is made to these two sections of data. The user then enters the values of the spring constant and the radius of probe. The program then converts the data to force versus separation plot. Force was calculated using Hooke’s law, which may be presented as F ¼ kx where k is cantilever spring constant and x is the vertical deflection of the cantilever. X-Ray Photoelectron Spectroscopic Analysis XPS analysis was conducted to detect contaminants on the surfaces. This analysis was performed using an AXIS-HSi spectrometer (Kratos Analytical Inc., Manchester, UK) with a monochromated Al Ka source at a power of 180 W (12 kV 15 mA), and a hemispherical analyser having standard aperture of 1 0.5 mm operating in the fixed analyser transmission mode. The total pressure in the main vacuum chamber during analysis was between 2 and 5 108 mbar. A magnetic immersion lens, which was located underneath the sample, was used as part of the imaging system to neutralise the charge in the spectrometer. A bias electrode with low kinetic energy electrons (1–4 eV) was used to gain a uniform and consistent neutralisation of charge build up during analysis. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005
1504
ISLAM ET AL.
Each specimen was analysed at an emission angle of 08 with respect to the normal surface. Assuming a value of approximately 3 nm for the electron attenuation length of relevant photoelectrons (e.g., C 1s, N 1s, O 1s) in a polymeric matrix, this translates into an approximate value for the XPS analysis depth (from which 95% of the detected signal originates) of less than 10 nm. A circular area with a diameter of approximately 1.0 mm was analysed on each sample. All elements present were identified from survey spectra (Figure 4A and B). The atomic concentrations of the detected elements were calculated using integral peak intensities and the sensitivity factors supplied by the manufacturer. To gain more detailed information about chemical structure, oxidation states etc., high resolution spectra (data not shown) were recorded from individual peaks at 40 eV pass energy (yielding a typical peak width for polymers of 1.0 eV). Using a minimisation algorithm, these data were quantified in order to calculate curve fits and thus to determine the contributions from specific functional groups. Chemically different species were quantified by fitting model functions to corresponding spectra (curvefits). Binding energy (BE) values were calculated by correcting measured peak positions for charging. The hydrocarbon C 1s signal was used as a reference. Since this signal contains contribution from both aliphatic (BE ¼ 285.0 eV) and aromatic (BE ¼ 284.7 eV) hydrocarbons, an intermediate value of BE ¼ 284.85 eV was used. It is important to note that assignments of specific curvefit components based on their respective binding energy can be ambiguous because photoelectron peaks may have contributions from more than one species or compound. The accuracy associated with quantitative XPS analysis was ca. 10%–15% and precision was usually better than 5%. Determination of Surface Roughness Atomic Force Microscope (AFM) Imaging of Lactose Particles and Films. Lactose particles were sprinkled on silicone wafers as described previously. TappingMode surface scans were performed in air at ambient temperature and humidity using a Nanoscope III atomic force microscope (AFM) (Digital Instruments, CA). TappingMode AFM tips, with force constants of about 40N/m and resonant frequencies of 300– JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005
400 kHz were used for imaging. A scan rate of 1.00 Hz was maintained for all experiments. A defined scan size of 10 10 mm2 with 512 512 points in each image was used. The set point was adjusted in such a way that no damage to the sample surface was caused by the AFM tip. Typical drive amplitude voltages were 99–150 mV, which generated a free cantilever vibration amplitude of 100 nm. Roughness Measurement. After capturing an image, the image was flattened to remove the effect of the piezo motion. The root mean square (RMS) roughness (Rq), and average roughness (Ra) were determined using image analysis software available with the AFM. The software used the following equations: sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðZi ZÞ2 RMSðRq Þ ¼ ð1Þ N where Zi is the height at position i (mm). Z is the average height (mm) of the Z values (Z-axis representing topographical height) and N is the number of data points within the given area. The average roughness Ra, was obtained from the AFM analysis software using following equation 1 Ra ¼ Lx Ly
y ZLZ
0
Lx
jfðx; yÞjdxdy
ð2Þ
0
where f(x,y) is the surface relative to the centre plane and Lx and Ly are the dimensions of the surface. Confocal Microscope. The surface roughness of the lactose particles was measured by confocal microscopy (Leica TCS NT, Leica Microsystems) equipped with ArKr (Argon–Krypton) laser (568 nm excitation was used) and 590 LP (long pass) filter for detection of fluorescent emission employed with a LEICA DMRE optical microscope with a galvanometer driven stage. The image was captured using a 40 lens at a format of 512 512 pixels with a pinhole set to an Airy Disk of 1. The optical step size was set to 0.5 mm for large and 0.3 mm for small particles to satisfy Nyquist sampling (i.e., the pixel should be at least two times smaller than the z-resolution of the objective lens). Five individual particles and five different sites on each particle from each grade of lactose samples were examined. RMS roughness (Rq) and average
ADHESION FORCE DISTRIBUTIONS
1505
roughness Ra, were measured by the image analysis software (Simulator, Leica Microsystem) within an area of interest from these topographical images, using following equations (DIN EN ISO 4287). Arithmetic average roughness (Ra) of the profile ordinates within the measured section (average height) is expressed as: 1 Ra ¼ A
ZA
2 Z ðx; yÞdA
ð3Þ
0
RA with A ¼ Area of interest ¼ 0 dA A, and Z ¼ ðx; yÞ ¼ Zi Z where Zi is the height at position i (mm). Z is the average height (mm) of the Z values. The root mean square value of the roughness (RMS Rq) within the measured area is presented as: rffiffiffiffi ZA 1 2 RMSðRq Þ ¼ Z ðx; yÞdA A
ð4Þ
0
RESULTS AND DISCUSSIONS Imaging and Surface Roughness of Lactose Carriers and Spin Coated Films In order to understand the surface topography of the lactose particles and spin coated lactose films, imaging of the lactose carriers and films were undertaken using TappingMode AFM. A cross section of the topographic image of the lactose surface was obtained and the distribution of peaks and valleys are represented in Figure 1. Even though this part of the surface seemed to be relatively smooth when viewed with SEM, the surface showed significant roughness with peaks and valleys varying over about 0.5 mm. The technique was unable to be used to quantify roughness of the whole lactose carrier surface because of the restricted movement of the cantilever on the very rough sections of the lactose surfaces and thus representative determinations of RMS Rq could not be obtained by AFM. However, it proved useful in determining the roughness of the very smooth spin coated lactose films for which the RMS Rq value was 0.28 0.08 nm. The roughness of the lactose carriers were quantified using confocal microscopy. The surface roughness (RMS Rq) of 25 sample areas, each of a 10 10 mm2, from at least five different particles
Figure 1. Surfaces of lactose particle (A) determined by tapping mode atomic force microscopy. A crosssectional image of lactose surface (along the line) is shown in (B).
for the different grades of lactose carriers ranged from 0.93 to 2.84 mm (Table 1). The number of 10 10 mm2 sample areas on each particle depended on the carrier size. Small particle size carriers utilised only one sample area per particle, while for the larger carriers, up to five sample areas per particle were able to be used. A significant difference (ANOVA, p ¼ 0.001) in the mean RMS Rq among the samples was observed. The ratio of Rq/ Ra of all samples was greater than unity, which indicated a high degree of surface irregularity. Adhesion Force and Distribution The adhesion forces between the SX and silica probes and lactose surfaces were determined by AFM using several different lactose carriers. The lactose carriers were inhalation grade; in addition, the fine lactose particles adhering to Aeroflo 65 and 95 were removed through a decantation process using saturated ethanol. Cumulative force distributions were determined for the interaction between spherical silica particles (5 and 20 mm) attached to the cantilever of the AFM, an AFM tip and an SX particle (4 mm) with the JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005
1506
ISLAM ET AL.
Table 1. Average Root Mean Square Roughness (RMS Rq) of Different Lactose Particles and Spin Coated Lactose Films, and Average Adhesion Forces between Probes of Salmeterol Xinafoate, Silica Probes, and Cantilever Tip and Lactose Particles and Spin Coated Lactose Films (Standard Deviations Are Within Brackets) Roughness Parameters
Samples Particles Aeroflo 65 Aeroflo 95 Aeroflo 20 Inhalac 120 Inhalac 230 Sorbolac 400 Aeroflo 65 decanted Aeroflo 95 decanted Films Aeroflo 65 Aeroflo 95
RMS Rq (mm)
Rq/Ra
1.05 (0.16) 2.21 (1.16) 1.16 (0.22) 1.38 (0.54) 1.49 (0.38) 0.93 (0.43) 1.17 (0.47) 2.84 (1.22)
1.24 (0.02) 1.25 (0.02) 1.25 (0.04) 1.26 (0.02) 1.20 (0.01) 1.21 (0.03) 1.22 (0.03) 1.24 (0.03)
0.26 (0.08)
—
surface of several inhalation grade lactose samples (Figure 2). Cumulative adhesion force distributions between the 5 mm silica probe and individual lactose carriers showed a broad distribution of adhesion forces over the range of 5–105 nN (Figure 2A). Significant differences in the adhesion distribution (p < 0.001) were observed among the samples using all the adhesion data; however, all pair wise multiple comparison procedure (Student– Newman–Keuls methods) showed significant (p < 0.05) difference between Sorbolac 400 and all grades of other lactose, whereas no differences (p < 0.05) occurred between Inhalac 230, Aeroflo 65 and Aeroflo 20; Aeroflo 20, Aeroflo 65 and Inhalac 120; and between Aeroflo 65 and Inhalac 120. Furthermore, Sorbolac 400 produced the highest adhesion force and Aeroflo 95 showed the lowest adhesion. The mean adhesion forces were between 29.2 and 59.2 nN and the standard deviation ranged from 14.6 to 28.5 nN (Table 1). A wide force distribution, ranging from 5 to 100 nN was also observed between the 20 mm silica probe and different lactose surfaces (Figure 2B). Significant differences (p < 0.001) occurred among the samples and all pair wise multiple comparison procedure (Student–Newman–Keuls methods) showed significant (p < 0.05) differences between Aeroflo 95, Aeroflo 65, Inhalac 230 and Sorbolac 400; Inhalac 120, Aeroflo 20 and Inhalac 230; Sorbolac 400 and Aeroflo 20; and Aeroflo 65 and Aeroflo 20. No differences (p < 0.05) observed between Inhalac 120, Aeroflo 95 and Sorbolac 400; and Inhalac 230, Aeroflo 65 and Aeroflo 20. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005
Adhesion Forces (nN) Silica-Probe (5 mm)
Silica-Probe (20 mm)
AFM Tip
SX-Probe (4 mm)
38.76 29.19 39.11 36.19 44.93 59.23 — —
(14.6) (17.5) (18.9) (25.1) (28.5) (25.0)
39.80 (25.8) 52.42 (36.8) 30.16 (12.5) 46.67 (26.4) 36.67 (14.9) 40.79 (11.7) — —
49.57 (7.8) 26.82 (8.6) 33.77 (21.8) 31.56 (10.1) 24.24 (7.8) 70.60 (13.9) — —
55.38 (33.0) 49.33 (27.8) 51.25 (24.3) 64.15 (39.8) 52.51 (34.9) 60.56 (33.9) — —
63.39 (4.5) 62.94 (1.4)
93.24 (6.5) 93.70 (5.3)
41.38 (1.0) 41.36 (1.5)
51.83 (8.6) 50.89 (5.9)
The mean adhesion forces were between 30.2 and 52.4 nN and the standard deviation ranged from 11.7 to 36.8 nN (Table 1). Interestingly, the adhesion force distribution measured between the AFM cantilever tip and the lactose surfaces were quite different from those measured by the silica probes (Figure 2C). The adhesion forces between the AFM tip and lactose surfaces ranged from 5 to 90 nN. The average force values were between 24.2 and 70.6 nN reflecting a greater discrimination between the lactose samples. Significant differences (p < 0.001) were observed among all lactose samples except between Aeroflo 95 and Inhalac 230 (p < 0.05). The highest adhesion force occurred between Sorbolac 400 and AFM tip. The standard deviation of the samples ranged between 7.8 and 21.8 nN and were significantly less than that of the silica probes. This effect is clearly seen in the cumulative force distribution where the distributions are much narrower. In general, the force distributions of the 20 mm silica probe were not significantly different (p ¼ 0.982) to those of the force distributions occurring between the silica probes (5 mm) and lactose surfaces. While it might be expected that the larger probe would produce a greater adhesion force, the similarity in the average forces and force distribution between the 5 and 20 mm probe is probably related to the geometry of the two probes in contact with the rough lactose surface producing a contact area similar for both probes. The larger silica probe is likely to be adhered across the top of several peaks (multiple points of
ADHESION FORCE DISTRIBUTIONS
B
A
100
80
A ero flo 20 A ero flo 65 A ero flo 95 Inhalac 120 Inhalac 230 So rbo lac 400
60 40 20
Cumulative % < than
% Cumulative % < than
100
80
A ero flo 20 A ero flo 65 A ero flo 95 Inhalac 120 Inhalac 230 So rbo lac 400
60 40 20 0
0 1
10
100
1
1000
Adhesion force (nN)
10
100
1000
Adhesion force (nN)
D 100
A ero flo 20 A ero flo 65 A ero flo 95 Inhalac 120 Inhalac 230 So rbo lac 400
80 60 40 20 0
100 Cumulative % < than
C Cumulative % < than
1507
80 Aeroflo 95 original Aeroflo 65 original Aeroflo 20 Inhalac 120 Inhalac 230 Sorbolac 400
60 40 20 0
1
10
100
1000
Adhesion forec(nN)
1
10
100
1000
Adhesion force (nN)
Figure 2. Adhesion force distributions (A) between lactose particles and silica probe (5 mm); (B) between lactose particles and 20 mm silica; (C) between atomic force microscopy (AFM) tip and lactose surface; and (D) between lactose particles and salmeterol xinafoate (SX) probe (4 mm). Each of the force distributions resulted from over a 100 force plots from at least 5 particles and 20 different measurements on each particle.
contact), on the surface and thereby reducing the contact area and experiencing a lower adhesion force, while the smaller probe may fit into the depressions and valleys on the lactose surface. The decreased adhesion force and the narrower cumulative distributions seen in the force measurements with the cantilever tip were most likely due to decreased contact area of the tip and less variation in the contact area as it is likely that the small contact area of the tip of the cantilever of the AFM will be less influenced by the size of the depressions and valleys present on the lactose surface. The effect of the adhering probe size can be seen in Figure 3 where the probes (drawn to scale) are placed over a real lactose surface profile. This hypothetical example demonstrated that greater variability in surface contact can be pro-
duced by the larger silica probes (5 and 20 mm) than with contact by the fine cantilever tip. The adhesion force distribution of the SX probes on the lactose surfaces demonstrated a broad range of adhesion forces over the range of 5.5– 175.7 nN (Figure 2D). The average adhesion forces were from 49.3 to 64.2 nN while the standard deviation of the adhesion force ranged from 24.3 to 39.8 nN and were greater than those of the silica spheres and AFM cantilever tip. The increased variance of the distribution curves may be related to the contact geometry between a non-spherical SX particle and the surface and can be visualised in Figure 3D. Therefore, in order to eliminate the effect of surface roughness, an attempt has been taken to determine the adhesion force between those JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005
1508
ISLAM ET AL.
A
B
C
D
Figure 3. Particle surface interaction and the geometry of contact: interaction between, 20 mm sphere (large round), 5 mm sphere (small round), a model drug probe attached with AFM tip and lactose carriers.
probes and spin coated lactose films which were almost atomically flat and smooth. Although the state of the crystallinity of lactose films was unclear, the average roughness of this film measured by AFM TappingMode was 0.28 nm. Whether the lactose film is crystalline, amorphous or some combination of crystalline and amorphous will have an influence on the magnitude of the adhesion force; however, the purpose of producing the films was to minimise roughness and to observe the effect on the broadness of the distribution. The force distribution between the SX probe and spin coated lactose films was relatively narrow with adhesion forces ranging from 35 to 70 nN and each lactose sample was almost superimposable (Figure 4A). The average adhesion forces between SX probe and spin coated lactose films (Aeroflo 65 and Aeroflo 95) were 52 and 51 nN, respectively. Force distribution between silica probes (5 and 20 mm) and lactose films also showed a similar pattern with narrow force distributions, i.e., 45– 70 nN for the 5 mm and 78–105 nN for 20 mm silica probes, respectively (Figure 4B and C). While the adhesion force distribution for all probes on the spin coated surface is relatively narrow in comparison with real surfaces, there is still a distribution of forces. This could be associated with JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005
intrinsically different adhesion of the spin coated surface or may be associated with error due the AFM methodology. The experimental outcomes further supports the hypothesis that the roughness of contact partners as well as the geometry of contact area are a dominant factors in controlling adhesion forces. As discussed previously there was no significant difference between the distributions of the 5 and 20 mm silica probe and the surface of the lactose particles. However, when the adhesions of these probes were determined on a smooth spin coated film, the adhesion forces were significantly (p ¼ 0.001) different (Table 1). Increased adhesion force was observed with increased the probe size. This finding further suggests that the true area of contact was mainly responsible in controlling adhesion forces. Comparison between SX Probe and Silica Probe A two way analysis of variance test showed that the adhesion forces between SX and individual lactose particles and silica and lactose particles were significantly different ( p < 0.014). Except for Sorbolac 400, the adhesion force between SX and lactose surfaces were significantly higher
ADHESION FORCE DISTRIBUTIONS
Cumulative % < than
A 100 80
Aeroflo 65 Aeroflo 95
60 40 20 0 1
10
100
1000
Adhesion force (nN/m)
B Cumulative % < than
100 A ero flo 20 A ero flo 65 A ero flo 95 Inhalac 120 Inhalac 230 So rbo lac 400
80 60 40 20 0 1
10
100
1000
Adhesion force (nN)
C Cumulative % < than
100 Aeroflo 65 Aeroflo 95
80 60 40 20 0 1
10
100
1000
Adhesion force (nN) Figure 4. Adhesion force distribution on spin coated films of Aeroflo 65 and Aeroflo 95; force between (A) SX particle and lactose film; (B) 5 mm silica sphere and lactose film; and (C) 20 mm silica sphere and lactose film. Each of the force distributions resulted from over a 100 force.
1509
( p ¼ 0.001) than between the 5 mm silica probe and the lactose surfaces. Since the SX was more hydrophobic than silica, the result was unexpected as the more hydrophobic probe would be likely to have less interaction with the hydrophilic lactose surface. The reason behind this was might be due to the dissimilarity of the size and geometry of the probes and the surface roughness characteristics of the lactose particles. As the SX probe was approximately 4 mm in size and plate like, this probe might possess greater contact with the valleys and peaks of the rough lactose surface due to its orientation compared with that of silica probe, which was 5 mm and spherical in shape. Significant differences were also found between SX and 20 mm silica probes (p ¼ 0.010). XPS Analysis of Lactose and SX Particles The surfaces of the lactose carriers and SX particles were analysed using XPS to determine if any inherent surface contamination existed. In particular, surface contamination of the lactose might contribute to the adhesion force distributions and it was important to eliminate surface contamination as a possible cause of the broadened force distribution. No contamination was detected on drug and lactose particles and films used in this study (Figure 5A and B). Figure 5A showed the chemical composition of SX where a characteristic peak for nitrogen (N1s) at about 401.8 eV and common peaks for carbon (C1s) and oxygen (O1s) were at 284.8 and 532.3eV, respectively. Figure 5B also showed the common peaks for carbon (C1s) and oxygen (O1s) at 284.8 and 532.3 eV respectively for lactose particles. From spin coated lactose films showed similar peaks of carbon and oxygen and no peak for silicone was observed which ensured the presence of a continuous film of lactose and no contaminants on the surface. Characteristics high resolution spectra of both SX and lactose also confirmed their chemical structures and composition. As there were no contaminants on the surfaces of both individual lactose particles and spin coated films, the force distributions observed either on lactose particles or spin-coated films were related to the geometry of the contacting partners and the surface chemistry.
CONCLUSIONS Adhesion forces between a single drug particle (SX), silica probes and single lactose particles and JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005
1510
ISLAM ET AL.
affect the contact area between the contiguous particles. Thus it is easy to see how similar sized particles on a rough surface will have differing adhesion forces due to orientation giving rise to broad force distributions. In addition, probes of differing sizes, but similar chemistry showed similar adhesion force distributions on lactose particles. Thus, surface roughness of the lactose carrier in powder mixtures for inhalation will have a marked effect on drug dispersion from the carrier surface. When a distribution of drug particles is adhered on the carrier’s surface, the orientation geometry is more complex and it is likely that the myriad of interaction geometries will provide a distributed pattern of interaction forces. Ideally, it is possible to imagine a match between the drug particle size and distribution and the surface roughness that might be engineered to achieve optimum dispersion.
ACKNOWLEDGMENTS The authors thank Thomas Gengenbach (CSIRO) for taking and analysing XPS measurements.
REFERENCES
Figure 5. (A) Survey spectra of carbon, nitrogen and oxygen on the SX surface; and (B) survey spectra of carbon and oxygen on the lactose particle surface determined by X-ray photoelectron spectroscopy (XPS).
spin coated films have been measured by atomic force microscopy. The broad distribution of adhesion forces between the single drug or silica probes and single lactose particles were attributed to the roughness of the lactose particles affecting the contact between surfaces since the adhesion force distributions on spin-coated films were much narrower. However, adhesion force distributions did exist on these films indicating that an intrinsic variability in surface chemistry does contribute to the overall force distributions; for these lactose samples, surface impurities did not add to the variability. The study demonstrates the very significant effect that roughness has on the adhesion force distribution. The orientation of the adhered drug particle with the rough lactose surface will be important in determining the adhesion force of that particle since it will JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005
1. Hooton JC, German CS, Allen S, Davies MC, Roberts CJ, Tendler SJB, Williams PM. 2004. An atomic force microscopy study of the effect of nanoscale contact geometry and surface chemistry on the adhesion of pharmaceutical particles. Pharm Res 21:953–961. 2. Noel O, Brogly M, Castelein G, Schultz J. 2004. In situ estimation of the chemical and mechanical contributions in local adhesion force measurement with AFM: The specific case of polymers. Eur Polym J 40:965–974. 3. Podczeck F, Newton JM, James MB. 1996. The influence of physical properties of the materials in contact on the adhesion strength of particles of salmeterol base and salmeterol salts to various substrate materials. J Adhes Sci Technol 10:257– 268. 4. Ibrahim TH, Burk TR, Etzler FM, Neuman RD. 2000. Direct adhesion measurements of pharmaceutical particles to gelatin capsule surfaces. J Adhes Sci Technol 14:1225–1242. 5. Stewart PJ. 1986. Particle interaction in pharmaceutical systems. Pharm Int 7:146–149. 6. Zimon AD. 1982. Fundamental concepts of particle adhesion. Adhesion of dust and powder, 2nd edn. Zimon AD, editor. New York: Consultants Bureau. pp 69–91.
ADHESION FORCE DISTRIBUTIONS
7. Muster TH, Prestidge CA. 2002. Face specific surface properties of pharmaceutical crystals. J Pharm Sci 91:1432–1444. 8. Begat P, Young PM, Edge S, Kaerger JS, Price R. 2003. The effect of mechanical processing on surface stability of pharmaceutical powders: Visualization by atomic force microscopy. J Pharm Sci 92: 611–620. 9. Dey FK, Cleaver JAS, Zhdan PA. 2000. Atomic force microscopy study of adsorbed moisture on lactose particles. Advanced Powder Technology 11:401–413. 10. Danesh A, Chen X, Davis MC, Roberts CJ, Sanders GHW, Tendler S JB, Williams PM, Wilkins MJ. 2000. The discrimination of drug polymorphic forms from single crystals using atomic force microscopy. Pharm Res 17:887–890. 11. Stewart PJ. 1981. Influence of magnesium stearate on the homogeneity of a prednisone-granule ordered mix. Drug Dev Ind Pharm 7:485–495. 12. Okada J, Matsuda Y, Fukumori Y. 1969. Measurement of the adhesive force of pharmaceutical powders by centrifugal method. I Yakugaku Zasshi 89:1539–1544. 13. Staniforth JN, Rees JE, Lai FK, Hersey JA. 1981. Determination of interparticulate forces in ordered powder mixes. J Pharm Pharmacol 33: 485–490. 14. Otsuka A, Iida K, Danjo K, Sunada H. 1988. Measurement of the adhesive force between particles of powdered materials and a glass substrate by means of the impact separation method. III. Effect of particle shape and surface asperity. Chem Pharm Bull 36:741–749. 15. Podczeck F, Newton JM, James MB. 1995. Assessment of adhesion and autoadhesion forces between particles and surfaces: I. The investigation of autoadhesion phenomena of salmeterol xinafoate and lactose monohydrate particles using compacted powder surfaces. J Adhes Sci Technol 8:1459–1472. 16. Iida K, Otsuka A, Danjo K, Sunada H. 1993. Measurement of the adhesive force between particles and a substrate by means of the impact separation method. Effect of the surface roughness and type of material of the substrate. Chem Pharm Bull 41:1621–1625. 17. Concessio NM, Jager-Waldau R, Hickey AJ. 1997. Aerosol delivery from an active emission multisingle dose dry powder inhaler. Particulate Science and Technology 15:51–63. 18. Ducker WA, Senden TJ, Pashley RM. 1991. Direct measurement of colloidal forces using an atomic force microscope. Nature 353:239–241. 19. Sindel U, Zimmcrmann I, Schaefer D, Reifenberger R. 1998. Direct measurement of interparticle forces
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
1511
in particulate solids using an atomic force microscope. World Congress on Particle Technology 3: 676–687. Sindeland U, Zimmermann I. 2001. Measurement of interaction forces between individual powder particles using an atomic force microscope. Powder Technology 117:247–254. Berard V, Lesniewska E, Andres C, Pertuy D, Laroche C, Pourcelot Y. 2002. Affinity scale between a carrier and a drug in DPI studied by atomic force microscopy. Int J Pharm 247:127– 137. Berard V, Lesniewska E, Andres C, Pertuy D, Laroche C, Pourcelot Y. 2002. Dry powder inhaler: Influence of humidity on topology and adhesion studied by AFM. Int J Pharm 232:213–224. Young PM, Cocconi D, Colombo P, Bettini R, Price R, Steele DF, Tobyn MJ. 2002. Characterization of a surface modified dry powder inhalation carrier prepared by ‘‘ particle smoothing’’. J Pharm Pharmacol 54:1339–1344. Louey MD, Mulvaney P, Stewart PJ. 2001. Characterization of adhesional properties of lactose carriers using atomic force microscopy. J Pharm Biomed Anal 25:559–567. Young PM, Price R, Tobyn MJ, Buttrum M, Dey F. 2003. Investigation into the effect of humidity on drug–drug interactions using the atomic force microscope. J Pharm Sci 92:815–822. Zhou H, Gotzinger M, Peukert W. 2003. The influence of particle charge and roughness on particle-substrate adhesion. Powder Technology 135-136:82–91. Goetzingerand M., Peukert W. 2004. Particle adhesion force distributions on rough surfaces. Langmuir 20:5298–5303. Tsukada M, Irie R, Yonemochi Y, Noda R, Kamiya H, Watanabe W, Kauppinen EI. 2004. Adhesion force measurement of a DPI size pharmaceutical particle by colloid probe atomic force microscopy. Powder Technology 141:262–269. Price R, Young PM, Edge S, Staniforth JN. 2002. The influence of relative humidity on particulate interactions in carrier-based dry powder inhaler formulations. Int J Pharm 246:47–59. Beach ER, Tormoen GW, Drelich JW, Han R. 2002. Pull-off forces measurements between rough surfaces by atomic force microscopy. J Colloid Interface Sci 247:84–99. Cleveland JP, Manne S, Bocek D, Hansma PK. 1993. A nondestructive method for determining the spring constant of cantilevers for scanning force microscopy. Review of Scientific Instruments 64: 403–405.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005
Department of Pharmaceutics, Victorian College of Pharmacy, Monash University, Parkville, Victoria 3052, Australia
2
Division of Molecular Science, CSIRO, Clayton, Victoria 3168, Australia
Received 30 September 2004; revised 23 January 2005; accepted 24 January 2005 Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.20381
ABSTRACT: Adhesion force distributions of silica spheres (5 and 20 mm) and salmeterol xinafoate (4 mm) particles with inhalation grade lactose surfaces and spin coated lactose films were determined by atomic force microscopy (AFM) to investigate the influence of surface roughness on the force distributions. The roughness of lactose particles and films was determined by both AFM and confocal microscopy (CM); the lactose particles showed RMS Rq values between 0.93 and 2.2 mm. The adhesion force distributions for silica and SX probes were significantly different for the different lactose carriers and broad, e.g., the adhesion force distribution between a 5 mm silica sphere and lactose particles ranged from 5 to 105 nN. This contrasted with distributions on smooth spin coated lactose films (RMS Rq of 0.28 nm) which were not significantly different and were narrow, e.g., the adhesion force distribution between a 5 mm silica sphere and spin coated lactose films was between 42 and 68 nN. In addition, no significant difference in adhesion force distribution occurred with silica probe size on the lactose carrier surface. The use of X-ray photoelectron spectroscopic analysis confirmed that the lactose surfaces were free of impurities that might contribute to variation in adhesion. Although the almost atomically flat films showed some adhesion variability, the surface roughness of the lactose particles was a major contributing factor to the broad distributions seen in this study. ß 2005 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 94:1500–1511, 2005
Keywords: roughness
AFM; adhesion force distribution; salmeterol xinafoate; lactose; surface
INTRODUCTION Inter-particular interactions in dry powder inhaler (DPI) formulations, containing micronised drug particles and lactose carriers, influence the efficiency of drug dispersion. It is likely that both the magnitude of the force of adhesion and the adhesion force distribution will effect the dispersion of drugs from interactive powder mixtures of drug and carrier. The adhesion force distribution will not only be influenced by the intrinsic Correspondence to: Peter Stewart (Telephone: 61 3 99039517; Fax: 61 3 99039583; E-mail: [email protected]) Journal of Pharmaceutical Sciences, Vol. 94, 1500–1511 (2005) ß 2005 Wiley-Liss, Inc. and the American Pharmacists Association
1500
variability in force between the interacting particles due to surface chemistry effects,1,2 but will also be influenced by the physical nature of the surfaces.3,4 The adhesion force will comprise a number of force components including contact potential, Coulombic, capillary, and intermolecular forces.5,6 Contact potential forces involve the interaction between contiguous surfaces with differing work functions that result in the development of a contact potential between the surfaces, resulting in an adhesion force. Coulombic interactions involve the interaction between oppositely charged particles or between a charged particle and a neutral surface. Capillary interaction requires the formation of a liquid between the contacting surfaces with the resultant capillary
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005
ADHESION FORCE DISTRIBUTIONS
attraction related to surface tension of the liquid bridge and its contact angle with the surfaces. Intermolecular interactions involve van der Waals forces and hydrogen bonding between the interacting particles. The inherent variability of interaction between surfaces will originate from the differing chemistry of the surface and will be related to specific interactions of the differing crystal faces due to molecular packing,7,8 differing adsorbed moisture contents,9 and differing distribution of amorphous/crystalline surface domains.8,10 These factors will influence the electrostatic charging potential, intermolecular interactions, and liquid bridge formation of the interacting surfaces and lead to an inherent force distribution even for atomically smooth surfaces. The interactive force distribution will also be affected by extrinsic factors, in particular the surface morphology, contact geometry, and adsorbed impurities. Adhesion forces in various systems have been determined by vibration,11 centrifugation,12,13,15 impact separation,14,16,17 and recently by atomic force microscopy (AFM).18 The AFM colloid probe technique enables the direct measurement of adhesion force between single particles and surfaces, and is becoming the method of choice to characterise particle adhesion in pharmaceutical systems. Recently, average adhesion forces in pharmaceutical powders, such as between a single small lactose particle and other lactose particle surfaces and compacted lactose disks,19,20 between zanamivir drug particles and lactose carriers,21,22 between beclometasone dipropionate (BDP), and a series of untreated and modified lactose surfaces23 have been reported. However, while many replicates were determined, force distributions were not shown in these papers. In other studies, typical log-normal force distributions, characterised by geometric means were observed for adhesion between a silica colloid probe and a lactose particle24 and the cohesive force between a single particle of salbutamol sulphate and compacts of sulbutamol sulphate at different relative humidity25 were measured. However, in these studies the reasons for the distribution of adhesion forces in terms of intrinsic and extrinsic factors were not addressed. A number of studies have addressed the influence of surface roughness on the force distribution. The influence of surface roughness on the force distribution between single particles and both smooth and rough substrates has been reported.26–28 Wide force distributions between spherical polystyrene particles and polished and
1501
PVD-coated aluminium substrates were observed; whereas, a narrow distribution occurred between the same particle and the smooth surface of a silicone wafer.26 Similar results also occurred with force measurement between spherical gold particles and rough substrates (silica and stainless steel) and smooth substrates (polished silica) were investigated.27 The authors of both studies concluded that the distribution of forces (narrow or broad) depended upon the distributions of asperity radius or the effective contact area of the contacting surfaces. Wide distributions of force at different RHs (30%–90%) between a spherical steroid particle, prepared by a crystallization method, and a lactose surface prepared by recrystallising lactose on a plate were observed; however, force distributions between the drug and smooth stainless steel substrates were narrower than those of the former substrate due to the variation in surface roughness of the substrates.28 Normal distribution of the adhesion force between salbutamol sulphate and budesonide against a surface modified lactose substrate at the RHs of 15% and 75% were observed.29 The authors also found log-normal force distributions and high variability in force measurement with increased substrate surface roughness, thought to be due to the variability in contact area between drug and substrates. The wide distributions of adhesion forces between pharmaceutical powders (beclomethasone dipropionate and lactose crystal) and rough polymeric surfaces (polypropylene, polycarbonate and acrylonitrilebutadine-styrene) observed were due to the roughness of interacting surfaces.30 Therefore, while the broad adhesion force distributions between interacting surfaces, determined using AFM, have been attributed to the morphology of surfaces, in particular, the surface roughness, the distinction between the extrinsic adhesion due to surface roughness and the true intrinsic adhesion has not been determined. Thus, surface roughness has been shown to contribute to broad force distributions, but the extent of this contribution is not known; the contribution of the chemistry of the surface to intrinsic force distributions is also not known. Thus, the purpose of this research was to study the influence of surface roughness on the adhesion force distribution using samples of inhalation grade lactose and probes of salmeterol xinafoate and silica. Intrinsic force distributions were estimated by eliminating surface roughness effects through spin coating of the lactose samples to produce almost atomically smooth surfaces. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005
1502
ISLAM ET AL.
EXPERIMENTAL Materials Salmeterol xinafoate (SX), micronized (inhalation grade; <4 mm), was obtained from GlaxoSmithKline, Melbourne, Australia. Aeroflo 95, Aeroflo 65, and Aeroflo 20, inhalation grades of a-lactose monohydrate, were donated by Foremost Farms, Rothschild, WI; Inhalac 70, Inhalac 120, Inhalac 230, and Sorbolac 400 were donated by Meggle GmbH, Germany; Pharmatose 125M and Pharmatose 110M was donated by DMV International, Veghel, The Netherlands, and Lactose Special dense, a non inhalation grade was donated by Lactose New Zealand, Hawera, New Zealand. Ammonium acetate (Analar, BDH, Victoria, Australia), methanol (HPLC grade, Biolab, Victoria, Australia), absolute alcohol (HPLC grade, CSR, Victoria, Australia) were used as supplied. Methodology Decantation and Removal of Fine Lactose About 20 g of lactose was dispersed (4–5 min) in absolute ethanol pre-saturated with lactose to make a homogeneous suspension and then allowed to settle for a predetermined time. The supernatant liquid was then decanted and replaced by fresh saturated ethanol, vigorously mixed and the process repeated until the supernatant liquid was clear. During removing the supernatant, special care was taken to ensure minimum disturbance of the lower part of suspension. The coarse particles remaining in the beaker after 15 cycles of decantation were wet sieved by spraying with saturated ethanol three times. Finally the samples were dried at room temperature, 21 18C. For first three repeats, 10 min interval was used and then 5 min interval was applied until supernatant liquid was clear. Measurement of Adhesion Force The adhesional forces were measured by atomic force microscopy (Nanoscope Multimode IIIa, Digital Instruments, Santa Barbara, CA) using the colloid probe technique18 in air and ambient humidity. Sample Preparation Lactose Particle. Araldite Five-minute Epoxy Resin (Selleys Chemical Company Pty. Ltd., JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005
Melbourne, Australia) was mixed, spread onto a clean silicon wafer ensuring a relatively uniform thin film and allowed to dry for 4 min. Lactose particles were sprinkled onto the resin by blowing the lactose particles with a gentle nitrogen flow. The samples were dried overnight and any unattached particles were removed by blowing with nitrogen. This sample was stuck onto a stainless steel sample stub with double-sided adhesive tape (Scotch, 3M Australia Pty Ltd., Sydney, Australia). Lactose Film (Spin Coating). Several pieces of silicon wafer of 1 1 cm2 size were placed in a beaker containing a washing solution (surfactant and ethanol) and sonicated for 30 min. After washing several times with milli-Q water, these silicon wafers were further sonicated with ethanol for 30 min. These sonicated wafers were again washed with milli-Q water and kept under a UV lamp for 30 min. One piece of silicon wafer was placed on the centre position of the plate of a PhotoResist Spinner (Headway Research, Inc., Garland, TX) ensuring the smooth side of the wafer was up. After adjusting the spinner speed to 5000 rpm, one drop of lactose solution (20% lactose in water) was dropped onto the wafer by a pasture pipette. The wafer was spun until all solvent was evaporated. These samples were stored in a desicator containing silica gel. All coatings were thicker than the X-ray photoelectron spectroscopy (XPS) sampling depth, i.e., at least 10–15 nm thick since no silicone was observed in XPS measurements. Attaching Colloid Probes to the AFM Tips Silica Probe. Using a micromanipulator (World Precision Instruments, Inc., Sarasota, FL, Model M3301R), a piece of tungsten wire was dipped into a small amount of freshly mixed epoxy resin (Araldite Epoxy Resin 24 hours, Selleys Chemical Company Pty. Ltd., Melbourne, Australia). This process was directed through an optical microscope. After withdrawal from the resin, the tip of the wire was brought in contact with a clean microscope slide to remove excess resin from the wire tip. The resin coated wire tip was brought into contact with the apex of the V-shaped silicon nitride cantilever (NanoprobeTM SPM Tips, Type NP-S, Digital Instruments, Santa Barbara, CA). It was important to transfer only a small amount of resin on the cantilever tip to prevent the immersion of the particle in the wet resin. After cutting the wire tip, it was touched onto a 5 mm (or 20 mm)
ADHESION FORCE DISTRIBUTIONS
diameter silica sphere (Bangs Laboratories, Inc., Fishers, IN) which stuck to the wire tip due to capillary forces. The sphere was transferred to the cantilever tip where it stuck to the resin. Great care was taken so that epoxy resin was not spread around the sphere. SX Probe. Using the method mentioned above, epoxy resin was applied to a cantilever tip. The epoxy-coated cantilever was mounted in the AFM. It was positioned directly above a single drug particle, or a small agglomerate of drug particles, and lowered microscopically until contact occurred. Retraction of the tip enabled initial confirmation of particle attachment. After drying overnight the SX probe was examined under an optical microscope fitted with a video camera to ensure successful attachment. Excess drug particles were removed by blowing with nitrogen. Colloid probes were examined under an optical microscope to ensure attachment to the correct position on the cantilever. The size of the attached particle was determined through a microscope with a video camera attached. The spring constant of the cantilevers was determined by the attachment of known masses to the cantilever and measuring resultant changes in the cantilever’s resonant frequency.31 Force Measurement Collecting Force Data. All force measurements were performed using a Nanoscope1 Multimode IIIa (Digital Instruments, Santa Barbara, CA). The Nanoscope III has a force mode in which the colloid probe is held stationary and the lactose sample, mounted on the piezoelectric tube, is driven in a controlled manner towards and away from the colloid probe. The speed of the approach and retraction can be set by the user. A diode laser and prism/ mirror optical path allow focusing of the laser onto the cantilever, and subsequent gathering the reflected light. A position sensitive photodiode detects the position of the free end of the cantilever due to deflection of the laser light. The software of the Nanoscope generates a file that contains the output of the photodiode and the displacement of the sample. Analysing AFM Force Raw Data. The raw AFM force data can be easily converted into force versus separation plot. To determine the force, zero force and zero separation points need to be defined. Zero
1503
force was defined where the cantilever deflection remained constant (zero deflection line) with sample displacement. The region of zero force should be linear. Zero separation is defined when a change in the sample position causes an equal change in cantilever deflection. This region (also known as the constant compliance region) occurs when the colloid probe is in contact with the sample surface. In this region of compliance the photodiode output is a linear function of the displacement of the sample. It is important to note that the data defined as zero separation were completely linear, as any non-linearity would indicate that the surfaces were not in true contact. Non-linearity of the compliance line may also be an indication of a contaminant on one or both the surfaces. All raw force measurement data presented in this study were converted to force versus separation data using software written by Dr. Patrick Hartley (CSIRO, Melbourne). This program displays the collected data graphically, so that the constant compliance region (defined as zero separation) and a baseline to the zero force can be selected from the raw data. After selecting the zero separation and zero force, a least square fit is made to these two sections of data. The user then enters the values of the spring constant and the radius of probe. The program then converts the data to force versus separation plot. Force was calculated using Hooke’s law, which may be presented as F ¼ kx where k is cantilever spring constant and x is the vertical deflection of the cantilever. X-Ray Photoelectron Spectroscopic Analysis XPS analysis was conducted to detect contaminants on the surfaces. This analysis was performed using an AXIS-HSi spectrometer (Kratos Analytical Inc., Manchester, UK) with a monochromated Al Ka source at a power of 180 W (12 kV 15 mA), and a hemispherical analyser having standard aperture of 1 0.5 mm operating in the fixed analyser transmission mode. The total pressure in the main vacuum chamber during analysis was between 2 and 5 108 mbar. A magnetic immersion lens, which was located underneath the sample, was used as part of the imaging system to neutralise the charge in the spectrometer. A bias electrode with low kinetic energy electrons (1–4 eV) was used to gain a uniform and consistent neutralisation of charge build up during analysis. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005
1504
ISLAM ET AL.
Each specimen was analysed at an emission angle of 08 with respect to the normal surface. Assuming a value of approximately 3 nm for the electron attenuation length of relevant photoelectrons (e.g., C 1s, N 1s, O 1s) in a polymeric matrix, this translates into an approximate value for the XPS analysis depth (from which 95% of the detected signal originates) of less than 10 nm. A circular area with a diameter of approximately 1.0 mm was analysed on each sample. All elements present were identified from survey spectra (Figure 4A and B). The atomic concentrations of the detected elements were calculated using integral peak intensities and the sensitivity factors supplied by the manufacturer. To gain more detailed information about chemical structure, oxidation states etc., high resolution spectra (data not shown) were recorded from individual peaks at 40 eV pass energy (yielding a typical peak width for polymers of 1.0 eV). Using a minimisation algorithm, these data were quantified in order to calculate curve fits and thus to determine the contributions from specific functional groups. Chemically different species were quantified by fitting model functions to corresponding spectra (curvefits). Binding energy (BE) values were calculated by correcting measured peak positions for charging. The hydrocarbon C 1s signal was used as a reference. Since this signal contains contribution from both aliphatic (BE ¼ 285.0 eV) and aromatic (BE ¼ 284.7 eV) hydrocarbons, an intermediate value of BE ¼ 284.85 eV was used. It is important to note that assignments of specific curvefit components based on their respective binding energy can be ambiguous because photoelectron peaks may have contributions from more than one species or compound. The accuracy associated with quantitative XPS analysis was ca. 10%–15% and precision was usually better than 5%. Determination of Surface Roughness Atomic Force Microscope (AFM) Imaging of Lactose Particles and Films. Lactose particles were sprinkled on silicone wafers as described previously. TappingMode surface scans were performed in air at ambient temperature and humidity using a Nanoscope III atomic force microscope (AFM) (Digital Instruments, CA). TappingMode AFM tips, with force constants of about 40N/m and resonant frequencies of 300– JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005
400 kHz were used for imaging. A scan rate of 1.00 Hz was maintained for all experiments. A defined scan size of 10 10 mm2 with 512 512 points in each image was used. The set point was adjusted in such a way that no damage to the sample surface was caused by the AFM tip. Typical drive amplitude voltages were 99–150 mV, which generated a free cantilever vibration amplitude of 100 nm. Roughness Measurement. After capturing an image, the image was flattened to remove the effect of the piezo motion. The root mean square (RMS) roughness (Rq), and average roughness (Ra) were determined using image analysis software available with the AFM. The software used the following equations: sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðZi ZÞ2 RMSðRq Þ ¼ ð1Þ N where Zi is the height at position i (mm). Z is the average height (mm) of the Z values (Z-axis representing topographical height) and N is the number of data points within the given area. The average roughness Ra, was obtained from the AFM analysis software using following equation 1 Ra ¼ Lx Ly
y ZLZ
0
Lx
jfðx; yÞjdxdy
ð2Þ
0
where f(x,y) is the surface relative to the centre plane and Lx and Ly are the dimensions of the surface. Confocal Microscope. The surface roughness of the lactose particles was measured by confocal microscopy (Leica TCS NT, Leica Microsystems) equipped with ArKr (Argon–Krypton) laser (568 nm excitation was used) and 590 LP (long pass) filter for detection of fluorescent emission employed with a LEICA DMRE optical microscope with a galvanometer driven stage. The image was captured using a 40 lens at a format of 512 512 pixels with a pinhole set to an Airy Disk of 1. The optical step size was set to 0.5 mm for large and 0.3 mm for small particles to satisfy Nyquist sampling (i.e., the pixel should be at least two times smaller than the z-resolution of the objective lens). Five individual particles and five different sites on each particle from each grade of lactose samples were examined. RMS roughness (Rq) and average
ADHESION FORCE DISTRIBUTIONS
1505
roughness Ra, were measured by the image analysis software (Simulator, Leica Microsystem) within an area of interest from these topographical images, using following equations (DIN EN ISO 4287). Arithmetic average roughness (Ra) of the profile ordinates within the measured section (average height) is expressed as: 1 Ra ¼ A
ZA
2 Z ðx; yÞdA
ð3Þ
0
RA with A ¼ Area of interest ¼ 0 dA A, and Z ¼ ðx; yÞ ¼ Zi Z where Zi is the height at position i (mm). Z is the average height (mm) of the Z values. The root mean square value of the roughness (RMS Rq) within the measured area is presented as: rffiffiffiffi ZA 1 2 RMSðRq Þ ¼ Z ðx; yÞdA A
ð4Þ
0
RESULTS AND DISCUSSIONS Imaging and Surface Roughness of Lactose Carriers and Spin Coated Films In order to understand the surface topography of the lactose particles and spin coated lactose films, imaging of the lactose carriers and films were undertaken using TappingMode AFM. A cross section of the topographic image of the lactose surface was obtained and the distribution of peaks and valleys are represented in Figure 1. Even though this part of the surface seemed to be relatively smooth when viewed with SEM, the surface showed significant roughness with peaks and valleys varying over about 0.5 mm. The technique was unable to be used to quantify roughness of the whole lactose carrier surface because of the restricted movement of the cantilever on the very rough sections of the lactose surfaces and thus representative determinations of RMS Rq could not be obtained by AFM. However, it proved useful in determining the roughness of the very smooth spin coated lactose films for which the RMS Rq value was 0.28 0.08 nm. The roughness of the lactose carriers were quantified using confocal microscopy. The surface roughness (RMS Rq) of 25 sample areas, each of a 10 10 mm2, from at least five different particles
Figure 1. Surfaces of lactose particle (A) determined by tapping mode atomic force microscopy. A crosssectional image of lactose surface (along the line) is shown in (B).
for the different grades of lactose carriers ranged from 0.93 to 2.84 mm (Table 1). The number of 10 10 mm2 sample areas on each particle depended on the carrier size. Small particle size carriers utilised only one sample area per particle, while for the larger carriers, up to five sample areas per particle were able to be used. A significant difference (ANOVA, p ¼ 0.001) in the mean RMS Rq among the samples was observed. The ratio of Rq/ Ra of all samples was greater than unity, which indicated a high degree of surface irregularity. Adhesion Force and Distribution The adhesion forces between the SX and silica probes and lactose surfaces were determined by AFM using several different lactose carriers. The lactose carriers were inhalation grade; in addition, the fine lactose particles adhering to Aeroflo 65 and 95 were removed through a decantation process using saturated ethanol. Cumulative force distributions were determined for the interaction between spherical silica particles (5 and 20 mm) attached to the cantilever of the AFM, an AFM tip and an SX particle (4 mm) with the JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005
1506
ISLAM ET AL.
Table 1. Average Root Mean Square Roughness (RMS Rq) of Different Lactose Particles and Spin Coated Lactose Films, and Average Adhesion Forces between Probes of Salmeterol Xinafoate, Silica Probes, and Cantilever Tip and Lactose Particles and Spin Coated Lactose Films (Standard Deviations Are Within Brackets) Roughness Parameters
Samples Particles Aeroflo 65 Aeroflo 95 Aeroflo 20 Inhalac 120 Inhalac 230 Sorbolac 400 Aeroflo 65 decanted Aeroflo 95 decanted Films Aeroflo 65 Aeroflo 95
RMS Rq (mm)
Rq/Ra
1.05 (0.16) 2.21 (1.16) 1.16 (0.22) 1.38 (0.54) 1.49 (0.38) 0.93 (0.43) 1.17 (0.47) 2.84 (1.22)
1.24 (0.02) 1.25 (0.02) 1.25 (0.04) 1.26 (0.02) 1.20 (0.01) 1.21 (0.03) 1.22 (0.03) 1.24 (0.03)
0.26 (0.08)
—
surface of several inhalation grade lactose samples (Figure 2). Cumulative adhesion force distributions between the 5 mm silica probe and individual lactose carriers showed a broad distribution of adhesion forces over the range of 5–105 nN (Figure 2A). Significant differences in the adhesion distribution (p < 0.001) were observed among the samples using all the adhesion data; however, all pair wise multiple comparison procedure (Student– Newman–Keuls methods) showed significant (p < 0.05) difference between Sorbolac 400 and all grades of other lactose, whereas no differences (p < 0.05) occurred between Inhalac 230, Aeroflo 65 and Aeroflo 20; Aeroflo 20, Aeroflo 65 and Inhalac 120; and between Aeroflo 65 and Inhalac 120. Furthermore, Sorbolac 400 produced the highest adhesion force and Aeroflo 95 showed the lowest adhesion. The mean adhesion forces were between 29.2 and 59.2 nN and the standard deviation ranged from 14.6 to 28.5 nN (Table 1). A wide force distribution, ranging from 5 to 100 nN was also observed between the 20 mm silica probe and different lactose surfaces (Figure 2B). Significant differences (p < 0.001) occurred among the samples and all pair wise multiple comparison procedure (Student–Newman–Keuls methods) showed significant (p < 0.05) differences between Aeroflo 95, Aeroflo 65, Inhalac 230 and Sorbolac 400; Inhalac 120, Aeroflo 20 and Inhalac 230; Sorbolac 400 and Aeroflo 20; and Aeroflo 65 and Aeroflo 20. No differences (p < 0.05) observed between Inhalac 120, Aeroflo 95 and Sorbolac 400; and Inhalac 230, Aeroflo 65 and Aeroflo 20. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005
Adhesion Forces (nN) Silica-Probe (5 mm)
Silica-Probe (20 mm)
AFM Tip
SX-Probe (4 mm)
38.76 29.19 39.11 36.19 44.93 59.23 — —
(14.6) (17.5) (18.9) (25.1) (28.5) (25.0)
39.80 (25.8) 52.42 (36.8) 30.16 (12.5) 46.67 (26.4) 36.67 (14.9) 40.79 (11.7) — —
49.57 (7.8) 26.82 (8.6) 33.77 (21.8) 31.56 (10.1) 24.24 (7.8) 70.60 (13.9) — —
55.38 (33.0) 49.33 (27.8) 51.25 (24.3) 64.15 (39.8) 52.51 (34.9) 60.56 (33.9) — —
63.39 (4.5) 62.94 (1.4)
93.24 (6.5) 93.70 (5.3)
41.38 (1.0) 41.36 (1.5)
51.83 (8.6) 50.89 (5.9)
The mean adhesion forces were between 30.2 and 52.4 nN and the standard deviation ranged from 11.7 to 36.8 nN (Table 1). Interestingly, the adhesion force distribution measured between the AFM cantilever tip and the lactose surfaces were quite different from those measured by the silica probes (Figure 2C). The adhesion forces between the AFM tip and lactose surfaces ranged from 5 to 90 nN. The average force values were between 24.2 and 70.6 nN reflecting a greater discrimination between the lactose samples. Significant differences (p < 0.001) were observed among all lactose samples except between Aeroflo 95 and Inhalac 230 (p < 0.05). The highest adhesion force occurred between Sorbolac 400 and AFM tip. The standard deviation of the samples ranged between 7.8 and 21.8 nN and were significantly less than that of the silica probes. This effect is clearly seen in the cumulative force distribution where the distributions are much narrower. In general, the force distributions of the 20 mm silica probe were not significantly different (p ¼ 0.982) to those of the force distributions occurring between the silica probes (5 mm) and lactose surfaces. While it might be expected that the larger probe would produce a greater adhesion force, the similarity in the average forces and force distribution between the 5 and 20 mm probe is probably related to the geometry of the two probes in contact with the rough lactose surface producing a contact area similar for both probes. The larger silica probe is likely to be adhered across the top of several peaks (multiple points of
ADHESION FORCE DISTRIBUTIONS
B
A
100
80
A ero flo 20 A ero flo 65 A ero flo 95 Inhalac 120 Inhalac 230 So rbo lac 400
60 40 20
Cumulative % < than
% Cumulative % < than
100
80
A ero flo 20 A ero flo 65 A ero flo 95 Inhalac 120 Inhalac 230 So rbo lac 400
60 40 20 0
0 1
10
100
1
1000
Adhesion force (nN)
10
100
1000
Adhesion force (nN)
D 100
A ero flo 20 A ero flo 65 A ero flo 95 Inhalac 120 Inhalac 230 So rbo lac 400
80 60 40 20 0
100 Cumulative % < than
C Cumulative % < than
1507
80 Aeroflo 95 original Aeroflo 65 original Aeroflo 20 Inhalac 120 Inhalac 230 Sorbolac 400
60 40 20 0
1
10
100
1000
Adhesion forec(nN)
1
10
100
1000
Adhesion force (nN)
Figure 2. Adhesion force distributions (A) between lactose particles and silica probe (5 mm); (B) between lactose particles and 20 mm silica; (C) between atomic force microscopy (AFM) tip and lactose surface; and (D) between lactose particles and salmeterol xinafoate (SX) probe (4 mm). Each of the force distributions resulted from over a 100 force plots from at least 5 particles and 20 different measurements on each particle.
contact), on the surface and thereby reducing the contact area and experiencing a lower adhesion force, while the smaller probe may fit into the depressions and valleys on the lactose surface. The decreased adhesion force and the narrower cumulative distributions seen in the force measurements with the cantilever tip were most likely due to decreased contact area of the tip and less variation in the contact area as it is likely that the small contact area of the tip of the cantilever of the AFM will be less influenced by the size of the depressions and valleys present on the lactose surface. The effect of the adhering probe size can be seen in Figure 3 where the probes (drawn to scale) are placed over a real lactose surface profile. This hypothetical example demonstrated that greater variability in surface contact can be pro-
duced by the larger silica probes (5 and 20 mm) than with contact by the fine cantilever tip. The adhesion force distribution of the SX probes on the lactose surfaces demonstrated a broad range of adhesion forces over the range of 5.5– 175.7 nN (Figure 2D). The average adhesion forces were from 49.3 to 64.2 nN while the standard deviation of the adhesion force ranged from 24.3 to 39.8 nN and were greater than those of the silica spheres and AFM cantilever tip. The increased variance of the distribution curves may be related to the contact geometry between a non-spherical SX particle and the surface and can be visualised in Figure 3D. Therefore, in order to eliminate the effect of surface roughness, an attempt has been taken to determine the adhesion force between those JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005
1508
ISLAM ET AL.
A
B
C
D
Figure 3. Particle surface interaction and the geometry of contact: interaction between, 20 mm sphere (large round), 5 mm sphere (small round), a model drug probe attached with AFM tip and lactose carriers.
probes and spin coated lactose films which were almost atomically flat and smooth. Although the state of the crystallinity of lactose films was unclear, the average roughness of this film measured by AFM TappingMode was 0.28 nm. Whether the lactose film is crystalline, amorphous or some combination of crystalline and amorphous will have an influence on the magnitude of the adhesion force; however, the purpose of producing the films was to minimise roughness and to observe the effect on the broadness of the distribution. The force distribution between the SX probe and spin coated lactose films was relatively narrow with adhesion forces ranging from 35 to 70 nN and each lactose sample was almost superimposable (Figure 4A). The average adhesion forces between SX probe and spin coated lactose films (Aeroflo 65 and Aeroflo 95) were 52 and 51 nN, respectively. Force distribution between silica probes (5 and 20 mm) and lactose films also showed a similar pattern with narrow force distributions, i.e., 45– 70 nN for the 5 mm and 78–105 nN for 20 mm silica probes, respectively (Figure 4B and C). While the adhesion force distribution for all probes on the spin coated surface is relatively narrow in comparison with real surfaces, there is still a distribution of forces. This could be associated with JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005
intrinsically different adhesion of the spin coated surface or may be associated with error due the AFM methodology. The experimental outcomes further supports the hypothesis that the roughness of contact partners as well as the geometry of contact area are a dominant factors in controlling adhesion forces. As discussed previously there was no significant difference between the distributions of the 5 and 20 mm silica probe and the surface of the lactose particles. However, when the adhesions of these probes were determined on a smooth spin coated film, the adhesion forces were significantly (p ¼ 0.001) different (Table 1). Increased adhesion force was observed with increased the probe size. This finding further suggests that the true area of contact was mainly responsible in controlling adhesion forces. Comparison between SX Probe and Silica Probe A two way analysis of variance test showed that the adhesion forces between SX and individual lactose particles and silica and lactose particles were significantly different ( p < 0.014). Except for Sorbolac 400, the adhesion force between SX and lactose surfaces were significantly higher
ADHESION FORCE DISTRIBUTIONS
Cumulative % < than
A 100 80
Aeroflo 65 Aeroflo 95
60 40 20 0 1
10
100
1000
Adhesion force (nN/m)
B Cumulative % < than
100 A ero flo 20 A ero flo 65 A ero flo 95 Inhalac 120 Inhalac 230 So rbo lac 400
80 60 40 20 0 1
10
100
1000
Adhesion force (nN)
C Cumulative % < than
100 Aeroflo 65 Aeroflo 95
80 60 40 20 0 1
10
100
1000
Adhesion force (nN) Figure 4. Adhesion force distribution on spin coated films of Aeroflo 65 and Aeroflo 95; force between (A) SX particle and lactose film; (B) 5 mm silica sphere and lactose film; and (C) 20 mm silica sphere and lactose film. Each of the force distributions resulted from over a 100 force.
1509
( p ¼ 0.001) than between the 5 mm silica probe and the lactose surfaces. Since the SX was more hydrophobic than silica, the result was unexpected as the more hydrophobic probe would be likely to have less interaction with the hydrophilic lactose surface. The reason behind this was might be due to the dissimilarity of the size and geometry of the probes and the surface roughness characteristics of the lactose particles. As the SX probe was approximately 4 mm in size and plate like, this probe might possess greater contact with the valleys and peaks of the rough lactose surface due to its orientation compared with that of silica probe, which was 5 mm and spherical in shape. Significant differences were also found between SX and 20 mm silica probes (p ¼ 0.010). XPS Analysis of Lactose and SX Particles The surfaces of the lactose carriers and SX particles were analysed using XPS to determine if any inherent surface contamination existed. In particular, surface contamination of the lactose might contribute to the adhesion force distributions and it was important to eliminate surface contamination as a possible cause of the broadened force distribution. No contamination was detected on drug and lactose particles and films used in this study (Figure 5A and B). Figure 5A showed the chemical composition of SX where a characteristic peak for nitrogen (N1s) at about 401.8 eV and common peaks for carbon (C1s) and oxygen (O1s) were at 284.8 and 532.3eV, respectively. Figure 5B also showed the common peaks for carbon (C1s) and oxygen (O1s) at 284.8 and 532.3 eV respectively for lactose particles. From spin coated lactose films showed similar peaks of carbon and oxygen and no peak for silicone was observed which ensured the presence of a continuous film of lactose and no contaminants on the surface. Characteristics high resolution spectra of both SX and lactose also confirmed their chemical structures and composition. As there were no contaminants on the surfaces of both individual lactose particles and spin coated films, the force distributions observed either on lactose particles or spin-coated films were related to the geometry of the contacting partners and the surface chemistry.
CONCLUSIONS Adhesion forces between a single drug particle (SX), silica probes and single lactose particles and JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005
1510
ISLAM ET AL.
affect the contact area between the contiguous particles. Thus it is easy to see how similar sized particles on a rough surface will have differing adhesion forces due to orientation giving rise to broad force distributions. In addition, probes of differing sizes, but similar chemistry showed similar adhesion force distributions on lactose particles. Thus, surface roughness of the lactose carrier in powder mixtures for inhalation will have a marked effect on drug dispersion from the carrier surface. When a distribution of drug particles is adhered on the carrier’s surface, the orientation geometry is more complex and it is likely that the myriad of interaction geometries will provide a distributed pattern of interaction forces. Ideally, it is possible to imagine a match between the drug particle size and distribution and the surface roughness that might be engineered to achieve optimum dispersion.
ACKNOWLEDGMENTS The authors thank Thomas Gengenbach (CSIRO) for taking and analysing XPS measurements.
REFERENCES
Figure 5. (A) Survey spectra of carbon, nitrogen and oxygen on the SX surface; and (B) survey spectra of carbon and oxygen on the lactose particle surface determined by X-ray photoelectron spectroscopy (XPS).
spin coated films have been measured by atomic force microscopy. The broad distribution of adhesion forces between the single drug or silica probes and single lactose particles were attributed to the roughness of the lactose particles affecting the contact between surfaces since the adhesion force distributions on spin-coated films were much narrower. However, adhesion force distributions did exist on these films indicating that an intrinsic variability in surface chemistry does contribute to the overall force distributions; for these lactose samples, surface impurities did not add to the variability. The study demonstrates the very significant effect that roughness has on the adhesion force distribution. The orientation of the adhered drug particle with the rough lactose surface will be important in determining the adhesion force of that particle since it will JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005
1. Hooton JC, German CS, Allen S, Davies MC, Roberts CJ, Tendler SJB, Williams PM. 2004. An atomic force microscopy study of the effect of nanoscale contact geometry and surface chemistry on the adhesion of pharmaceutical particles. Pharm Res 21:953–961. 2. Noel O, Brogly M, Castelein G, Schultz J. 2004. In situ estimation of the chemical and mechanical contributions in local adhesion force measurement with AFM: The specific case of polymers. Eur Polym J 40:965–974. 3. Podczeck F, Newton JM, James MB. 1996. The influence of physical properties of the materials in contact on the adhesion strength of particles of salmeterol base and salmeterol salts to various substrate materials. J Adhes Sci Technol 10:257– 268. 4. Ibrahim TH, Burk TR, Etzler FM, Neuman RD. 2000. Direct adhesion measurements of pharmaceutical particles to gelatin capsule surfaces. J Adhes Sci Technol 14:1225–1242. 5. Stewart PJ. 1986. Particle interaction in pharmaceutical systems. Pharm Int 7:146–149. 6. Zimon AD. 1982. Fundamental concepts of particle adhesion. Adhesion of dust and powder, 2nd edn. Zimon AD, editor. New York: Consultants Bureau. pp 69–91.
ADHESION FORCE DISTRIBUTIONS
7. Muster TH, Prestidge CA. 2002. Face specific surface properties of pharmaceutical crystals. J Pharm Sci 91:1432–1444. 8. Begat P, Young PM, Edge S, Kaerger JS, Price R. 2003. The effect of mechanical processing on surface stability of pharmaceutical powders: Visualization by atomic force microscopy. J Pharm Sci 92: 611–620. 9. Dey FK, Cleaver JAS, Zhdan PA. 2000. Atomic force microscopy study of adsorbed moisture on lactose particles. Advanced Powder Technology 11:401–413. 10. Danesh A, Chen X, Davis MC, Roberts CJ, Sanders GHW, Tendler S JB, Williams PM, Wilkins MJ. 2000. The discrimination of drug polymorphic forms from single crystals using atomic force microscopy. Pharm Res 17:887–890. 11. Stewart PJ. 1981. Influence of magnesium stearate on the homogeneity of a prednisone-granule ordered mix. Drug Dev Ind Pharm 7:485–495. 12. Okada J, Matsuda Y, Fukumori Y. 1969. Measurement of the adhesive force of pharmaceutical powders by centrifugal method. I Yakugaku Zasshi 89:1539–1544. 13. Staniforth JN, Rees JE, Lai FK, Hersey JA. 1981. Determination of interparticulate forces in ordered powder mixes. J Pharm Pharmacol 33: 485–490. 14. Otsuka A, Iida K, Danjo K, Sunada H. 1988. Measurement of the adhesive force between particles of powdered materials and a glass substrate by means of the impact separation method. III. Effect of particle shape and surface asperity. Chem Pharm Bull 36:741–749. 15. Podczeck F, Newton JM, James MB. 1995. Assessment of adhesion and autoadhesion forces between particles and surfaces: I. The investigation of autoadhesion phenomena of salmeterol xinafoate and lactose monohydrate particles using compacted powder surfaces. J Adhes Sci Technol 8:1459–1472. 16. Iida K, Otsuka A, Danjo K, Sunada H. 1993. Measurement of the adhesive force between particles and a substrate by means of the impact separation method. Effect of the surface roughness and type of material of the substrate. Chem Pharm Bull 41:1621–1625. 17. Concessio NM, Jager-Waldau R, Hickey AJ. 1997. Aerosol delivery from an active emission multisingle dose dry powder inhaler. Particulate Science and Technology 15:51–63. 18. Ducker WA, Senden TJ, Pashley RM. 1991. Direct measurement of colloidal forces using an atomic force microscope. Nature 353:239–241. 19. Sindel U, Zimmcrmann I, Schaefer D, Reifenberger R. 1998. Direct measurement of interparticle forces
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
1511
in particulate solids using an atomic force microscope. World Congress on Particle Technology 3: 676–687. Sindeland U, Zimmermann I. 2001. Measurement of interaction forces between individual powder particles using an atomic force microscope. Powder Technology 117:247–254. Berard V, Lesniewska E, Andres C, Pertuy D, Laroche C, Pourcelot Y. 2002. Affinity scale between a carrier and a drug in DPI studied by atomic force microscopy. Int J Pharm 247:127– 137. Berard V, Lesniewska E, Andres C, Pertuy D, Laroche C, Pourcelot Y. 2002. Dry powder inhaler: Influence of humidity on topology and adhesion studied by AFM. Int J Pharm 232:213–224. Young PM, Cocconi D, Colombo P, Bettini R, Price R, Steele DF, Tobyn MJ. 2002. Characterization of a surface modified dry powder inhalation carrier prepared by ‘‘ particle smoothing’’. J Pharm Pharmacol 54:1339–1344. Louey MD, Mulvaney P, Stewart PJ. 2001. Characterization of adhesional properties of lactose carriers using atomic force microscopy. J Pharm Biomed Anal 25:559–567. Young PM, Price R, Tobyn MJ, Buttrum M, Dey F. 2003. Investigation into the effect of humidity on drug–drug interactions using the atomic force microscope. J Pharm Sci 92:815–822. Zhou H, Gotzinger M, Peukert W. 2003. The influence of particle charge and roughness on particle-substrate adhesion. Powder Technology 135-136:82–91. Goetzingerand M., Peukert W. 2004. Particle adhesion force distributions on rough surfaces. Langmuir 20:5298–5303. Tsukada M, Irie R, Yonemochi Y, Noda R, Kamiya H, Watanabe W, Kauppinen EI. 2004. Adhesion force measurement of a DPI size pharmaceutical particle by colloid probe atomic force microscopy. Powder Technology 141:262–269. Price R, Young PM, Edge S, Staniforth JN. 2002. The influence of relative humidity on particulate interactions in carrier-based dry powder inhaler formulations. Int J Pharm 246:47–59. Beach ER, Tormoen GW, Drelich JW, Han R. 2002. Pull-off forces measurements between rough surfaces by atomic force microscopy. J Colloid Interface Sci 247:84–99. Cleveland JP, Manne S, Bocek D, Hansma PK. 1993. A nondestructive method for determining the spring constant of cantilevers for scanning force microscopy. Review of Scientific Instruments 64: 403–405.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 7, JULY 2005