`
`73
`
`Ordered Mixing or Spontaneous Granulation ?
`
`J. N. STANIFORTH
`
`School of Pharmacy and Pharmacology, University of Bath, Claverton Down, Bath, BA2 7AY (U.K.)
`
`(Received April 2, 1985)
`
`SUMMARY
`
`Ordered mixes were produced using a
`fructose-based excipient as a coarse carrier
`component and fine-particle pyridoxine
`hydrochloride (vitamin B,) as the adherent
`component. Prior to mixing,
`the fructose
`agglomerates were conditioned for 48 h at
`either 0% RH or 55% RH at 20 °C. Under
`these conditions, one lot of fructose had a
`moisture content of 0.24 wt.% and the other
`lot had a moisture content of 0.74 wt.%.
`Following mixing, the powders were subjected
`to vibrotion at various frequencies in the
`range 25 to 200 Hz and accelerations in the
`range 9.81 to 39.24 m/s’. It was found that
`whilst ordered mixes produced using fructose
`at 0.24 wt.% moisture content were unstable,
`those produced using 0.74 wt.% moisture
`content fructose were extremely stable and
`segregation resistant.
`The formation of ordered units with
`increased adhesion in carriers with higher
`moisture content suggested that these ordered
`mixes could be considered as spontaneous
`granulations.
`
`INTRODUCTION
`
`Hersey [1] first introduced the concept of
`ordered mixing to explain the behaviour of
`interacting particles in a powder. Fine
`particles adhere to other particles to form
`so-called ordered units. Some potential
`adhesive or cohesive forces responsible for
`the formation of ordered units or interparticle
`bonds were described by Hersey [1] and by
`others [2, 3, 4]: London-van der Waals and
`other dipole forces; electrostatic or Coulomb
`forces; forces due to the presence of water
`such as surface tension forces and capillary
`suction potential forces; mechanical forces
`dueto static frictional contact or interlocking
`
`of rough or re-entrant particle surfaces;
`chemical forces such as occur in chemisorp-
`tion or hydrogen bonding. The Table shows
`some examples of ordered mixing systems
`studied in which one of these interactions
`wasconsidered to be the predominant force
`responsible for binding particles in ordered
`units.
`Comparisonof this group of forces with a
`similar list produced forinterparticle attractive
`forces in granules is, perhaps not surprisingly,
`very similar. The only significant difference
`between thelist above and one compiled by
`Rumpf [9] for granules, is the absence of
`solid bridges as a class of forces responsible
`for formation of ordered units.
`
`TABLE
`
`Some examples of forces responsible for binding
`particles in ordered units
`
`Adhesion
`force
`
`Ordered unit
`
`Reference
`
`London-van Considered to contribute Staniforth
`der Waals
`to most ordered units
`etal. [5]
`forces
`
`Electrostatic Coarse saccharide-based
`force
`carrier particles and
`fine potassium chloride
`particles following tribo-
`electrification
`
` Staniforth
`and Rees
`
`Surface
`tensionalor
`capillary
`forces
`
`Coarse sucrose crystals
`andfinesalicylic acid
`particles mixed at three
`different humidities
`
`Stephenson
`and Thiel
`[7]
`
`Mechanical
`forces
`
`Highly re-entrant,macro- Staniforth
`porous lactose or glucose ef al. [8]
`carrier particles and fine
`potassium chloride
`particles
`
`Chemical
`forces
`
`Nopractical examples
`
`0032-5910/85/$3.30
`
`© Elsevier Sequoia/Printed in The Netherlands
`
`Cosmo Ex. 2019-p. 1
`Argentum v Cosmo
`IPR2018-00080
`
`Cosmo Ex. 2019-p. 1
`Argentum v Cosmo
`IPR2018-00080
`
`
`
`74
`
`THEORY
`
`On the basis of this apparent similarity
`where Fx is the London-van der Waals force
`between granules and ordered units, there
`for extended contact and p is the radius of
`may be somejustification for describing
`interparticle contact.
`ordered mixing as a typeofdirect granulation.
`For a particle located onaslightly textured
`surface,p =1X10°m,i.e., Fag = 2.6 X10°7N.
`However,it is known that ordered units are
`prone to becomedestroyed during processing
`This adhesion force is comparable with a
`and, in general, particles in ordered units are
`value obtained experimentally for relatively
`far less strongly adhered than equivalent
`smoothcarrier particles, with slightly textured
`particles in granules.
`surfaces: 3.8 X 10-4 N. Adhesion forces of
`this order were found to be characteristic of
`unstable ordered units [5].
`In cases where the pores or clefts in a
`carrier surface are too small to produce an
`increased van der Waals attraction, other
`means of enhancing the overall interparticle
`adhesion have been used. One method used
`to reduce disruption of ordered units by
`vibration of powder mixes was electrostatic
`charging whichsignificantly reduced segrega-
`tion [6].
`Another method of improving interparticle
`adhesionis used in moist granulation and is
`due to adsorbed or condensed liquid layers
`on particle surfaces.
`Several equations have been derived to
`allow calculation of adhesion forces between
`particles due to surface tension and suction
`pressure in liquid films [10, 11]. These may
`be summarised using the following equation:
`
`The force balance which exists at a given
`interparticle contact will be influenced by
`both particle properties and the particle
`environment. For example,it is likely that as
`particle diameter reduces below 100 um, the
`van der Waals componentofparticle adhesion
`will becomesignificant. In a low-humidity
`environment, contact or frictional electrifica-
`tion may lead to significant interparticle
`electrostatic forces, but at elevated humidities,
`the electrostatic component of particle
`adhesion will be less significant than forces
`due to the presence of adsorbed or condensed
`liquid layers.
`The theoretical contribution of London-
`van der Waals forces to the adhesionofa fine
`and a coarse particle in an ordered unit can be
`calculated. For a fine drug particle with a 45
`ym diameterresting on the surface of a coarse
`excipient particle having a diameter of 500 um,
`the interparticle contact can be assumed to
`approximate to that for a sphere resting on a
`flat plate [4]:
`AR
`
`Fyx=
`
`6H?
`
`where F’, is the London-van der Waals force,
`A is the Hamakerconstant, R is the radius of
`the sphere and H is the separation distance
`between particle surfaces, taken to be 4 X
`107!° m, i.e., Fs = 1 X 107° N. However, the
`surface of the coarse excipient particle is
`porous and the drug particle can therefore
`be considered to be in extended contact with
`the carrier particle, so that
`7.- 4k, 4
`AE”
`6H?
`6nH>
`or
`
`F..= 2AR ( : p?
`
`AE
`
`12H?
`
`RH
`
`Frsp = YDf(8, 5)
`
`where y is the surface tension of the liquid
`layer, D is the fine particle diameter and
`f(8, 5) is a function of 6 the contact angle and
`§ the bridge angle.
`Values for Fersp for particles in the ordered
`unit described above vary between 2.33 X
`10-6 and 1.45 X 10-5 N,according to the
`method of calculation. Although these values
`are below those for van der Waals forces for
`fine particles adhered in clefts or pores,it is
`assumed that surface tension or suction
`potential forces will be responsible for (a)
`drawing particle surfaces closer together and
`(b) causing extended interparticle contact,
`and this will producean increase in dispersion
`forces between the two particles.
`The aim of the present study was to
`investigate the effect of the presence of
`surface moisture on the stability of ordered
`units formed between a hygroscopic coarse
`excipient particle and a fine particle model
`drug.
`
`Cosmo Ex. 2019-p. 2
`Argentum v Cosmo
`IPR2018-00080
`
`Cosmo Ex. 2019-p. 2
`Argentum v Cosmo
`IPR2018-00080
`
`
`
`75
`
`
`
`
`Frequency Counter
`
`Acceleration
`Monitor
`
`
`
`
` Vibration
`
`MATERIALS AND METHODS
`
`Tabfine, type F94M (manufactured by
`Finnish Sugar Company, Espoo, Finland,
`supplied by Forum Chemicals, Reigate,
`Surrey, U.K.) is an agglomerated fructose-
`based direct compression tableting excipient
`and was used as the coarse carrier system.
`Fine-particle
`pyridoxine hydrochloride
`(vitamin B,) was used as the adherent drug
`fraction at a concentration of 1 wt.%.
`Tabfine F94M was conditioned at two
`different humidities prior to mixing with
`pyridoxine hydrochloride powder. Onelot of
`Tabfine F94M powderwas dried to constant
`weight and stored at 0% RH in a desiccator
`containing fresh silica gel, for 48 h. A second
`lot of fructose-based 'Tabfine was allowed to
`equilibrate at a relative humidity of 55% at
`20 °C. The dried fructose agglomerates were
`found to have a moisture content of 0.24 wt.%,
`whereas the fructose agglomerates stored at
`55% RH had a moisture content of 0.74 wt.%.
`Ordered mixes containing Tabfine F94M
`and pyridoxine hydrochloride were produced
`by a two-stage process: initially, geometric
`mixing, or trituration, of the drug and
`excipient powders was carried out so as to
`ensure that the fine drug particles and agglom-
`erates of particles were broken up and
`uniformly dispersed throughout the excipient
`system. Secondly,
`the premixed powders
`were loaded into a cube blender (Erweka
`GmbH, Frankfurt, F.R.G.) and mixed for
`approximately 30 min. The homogeneity of
`the mixed powders was analysed byfilling
`an interlocking stack of small Perspex
`cylinders with powder and removing samples
`at ten different levels. The samples were
`analysed for drug concentration by dissolving
`in volumetric flasks made up to 100 ml with
`distilled water and measuring the absorbance
`of the solutions at 290 nm using a U.V.
`spectrophotometer (Shimadzu Instruments,
`Kyoto, Japan). The standard deviation and
`coefficient of variation of the ten sample
`concentrations were determined.
`The interlocking cylinder stack was then
`re-assembled and filled with the powder mix
`of known homogeneity. The stack was
`clampedin place on a vibration table (Fig. 1)
`and vibrations of known frequency and
`acceleration were applied using a vibration
`conditioning amplifier (Derritron Electronics,
`
`Frequency
`Generator &
`Table
` Power Amplifier
`
`Fig. 1. Schematic diagram of vibration model.
`
`Folkestone, U.K.). Frequencies were moni-
`tored using a frequency counter (Ferranti
`Electronics, Manchester, U.K.) attached to
`the output of the conditioning amplifier.
`Accelerations were monitored using an
`accelerometer mounted on the horizontal
`base plate of the cylinder holder on the
`vibration table, as shown in Fig. 1. The
`output from the accelerometer was fed into a
`signal conditioning amplifier and then to a
`vibration level monitor set to read R.M.S.
`accelerations,
`Following vibration under set conditions
`for 15 min, the homogeneity of the powder
`mix was re-determined using the sampling
`technique and spectrophotometric method
`described above.
`
`RESULTS AND DISCUSSION
`
`The difference in moisture content
`between the twolots of fructose-based Tabfine
`was only 0.5 wt.%, but the physical stability
`of the two powder mixes formed was very
`different. Ordered mixes containing the low
`moisture content Tabfine F94M were found
`to be extremely unstable when subjected to
`vibration under different conditions(Fig. 2).
`The segregation tendency of low moisture
`content mixesas characterised by the response
`surface shownin Fig. 2 was typical of that for
`other ordered mixes where segregation occurs
`[12], i.e., maximal segregation occurred in
`conditionsof low frequency andhigh accelera-
`tion. For these powder mixes, the segregation
`tendency was of a magnitude characteristic
`of that for unstable systems considered to
`behavelike partially ordered random mixes.
`In contrast, mixtures containing the higher
`moisture content Tabfine F94M (0.74 wt.%)
`were found to be the moststable of any of
`the drug-excipient systems studied, there
`
`Cosmo Ex. 2019-p. 3
`Argentum v Cosmo
`IPR2018-00080
`
`Cosmo Ex. 2019-p. 3
`Argentum v Cosmo
`IPR2018-00080
`
`
`
`76
`
`CoefficientofVariation
`
`Fig. 2. Relationship between coefficient of variation
`of spot samples drug content and vibration frequency
`and acceleration for ordered mixes containing 1 wt.%
`pyridoxine hydrochloride and Tabfine F94M with a
`
`CoefficientofVariation,
`
`
`
`PercentPowder
`
`
`
`PercertPowder
`
`being no evidence of segregation under any of
`the test conditions (Fig. 3). This apparent
`difference in behaviour may be partly due to
`occlusion of the powder bed by the more
`cohesive Tabfine F94M with 0.74 wt.%
`moisture content, although even the lower
`moisture content sample immobilized large
`proportions of the powder under several
`different vibration conditions (Fig. 4).
`
`tmmobilized
`moisture content of 0.24 wt.%. (a)
`Immobilized (b)
`
`Fig. 3. Relationship between coefficient of variation
`of spot sample drug content and vibration frequency
`andacceleration for ordered mixes containing 1 wt.%
`pyridoxine hydrochloride and Tabfine F94M with a
`moisture content of 0.47 wt.%.
`
`2007
`Fig. 4. Relationship between vibration frequency and
`acceleration and the percentage of the powder bed
`immobilized through changes in packing geometry
`for Tabfine F94M with moisture content of (a) 0.24
`wt.% and (b) 0.74 wt.%.
`
`Cosmo Ex. 2019-p. 4
`Argentum v Cosmo
`IPR2018-00080
`
`Cosmo Ex. 2019-p. 4
`Argentum v Cosmo
`IPR2018-00080
`
`
`
`77
`
`found to be extremely sensitive to the
`moisture contentof carrier particles.
`Low moisture content carrier particles
`formed unstable ordered mixes, whereas at a
`slightly higher moisture content, the fructose
`agglomerates formed ordered mixes which
`were extremely stable.
`The segregation resistance of these fructose
`systems and the.apparentrole of moisture in
`formation of strong ordered units suggested
`that ordered mixing could be considered to be
`a spontaneousordirect granulation.
`
`REFERENCES
`
`1 J. A. Hersey, Powder Technol., 11 (1975) 41.
`2 M. C. Coelho and N, Harnby, Powder Technoi.,
`20 (1978) 203.
`¢
`3 J. A. Cross and A. Cetronio, Dept. and Filtr. of
`Particles from Gases and Liquids Symp., Soc.
`Chem,Ind. (1978) 227.
`J. Visser, Dep. and Filtr. of Particles from Gases
`and Liquids Symp., Soc. Chem. Ind. (1978) 121.
`J. N. Staniforth, J. E. Rees, F. K. Lai and J. A.
`Hersey, J. Pharm. Pharmacol., 33 (1981) 485.
`6 J. N. Staniforth and J. E. Rees, J. Pharm.
`Pharmacol., 34 (1982) 69.
`P. L. Stephenson and W. J. Thiel, Powder Technol.,
`25 (1980) 115.
`8 J. N. Staniforth, Proc. 23rd Int. Colloq. on Ind.
`Pharm, Rijksuniversiteit, Gent, Belgium (1984) 1.
`H. Rumpf, Agglomeration ’77, Proc, 2nd Int.
`Symp, (1977) 97.
`10 W. Pietsch and H. Rumpf, Chem.Ing. Tech., 39
`(1967) 885.
`11 H. Schubert, Powder Technol., 37 (1984) 105.
`12 J. N. Staniforth and J. E. Rees, J. Pharm.
`Pharmacol., 34 (1982) 700.
`13 M. C. Coetho and N. Harnby, Powder Technol,
`20 (1978) 197.
`14 G. A. Turner and M, Balasubramanian, Powder
`Technol., 10 (1974) 121.
`15 S. J. Gregg and K. S. W. Sing, Adsorption, Surface
`Area and Porosity, Academic Press, London, 2nd
`edn., 1982.
`
`> a
`
`ae
`
`©
`
`Cosmo Ex. 2019-p. 5
`Argentum v Cosmo
`IPR2018-00080
`
`It is considered to be morelikely that the
`higher moisture content sample of Tabfine
`F94M held the increased mass of water as a
`surface film which was condensed at inter-
`particle contacts to form pendular bonds of
`the type commonly found in moist granules.
`The critical humidity above which water
`could be expected to condense to form a
`capillary system between particle contacts
`can be calculated [13]. It is suggested [14]
`that below an RH of 65% capillary forces
`play no part in particle adhesion. However,
`according to the value taken for the B.E.T.
`coefficient for air—wateror air—water saturated
`with fructose, the prediction of the point of
`condensation changes markedly. In addition,
`water exhibits special behaviour at surfaces,
`where polarity has a marked influence on
`adsorption because ofthe ability of the water
`molecule to form H* bonds and water may
`undergo chemisorption as well as physisorp-
`tion [15]. Water may also be associated non-
`stoichiometrically within the fructose lattice
`and it might be expected that although most
`of the water would be present as a thin
`adsorbed film at 55% RH, there may beat
`least some parts of the particle surface where
`capillary condensation could occur and
`contribute to particle bonding.
`In either case, the striking result of the
`presence of elevated surface moisture levels
`appears to be a spontaneous granulation of
`drug and excipient particles through forma-
`tion of extremely stable ordered units.
`
`CONCLUSION
`
`The stability of ordered mixes formed
`between fine pyridoxine hydrochloride
`particles and coarse fructose agglomerates was
`
`Cosmo Ex. 2019-p. 5
`Argentum v Cosmo
`IPR2018-00080
`
`