Powder Technology,
`0 Elsevier Sequoia
`
`161 - 167
`14 (1976)
`.%A., Lausanne - Printed in the Netherlands
`
`161
`
`Ordered Mixing in Direct Compression of Tablets
`
`hl. J. CROOKS and R. HO
`Department of Pharmacy. University of Sydney, Sydney. N_S.W_ 2006 (Australia)
`(Received November 27, 1975)
`
`SUM?sIARY
`
`The mixing of 2% sulphaphenazole (mean
`diameter 25 pm) with coarse directly com-
`pressible tablet vehicles has been studied using
`a sampling method and scanning electron mi-
`croscopy. At a certain vehicle particle size,
`sulphaphenazole appears to mix by an ordered
`process. After mixing with a 180 - 250-m
`size
`fraction of a sucrose-based vehicle (Dipac) for
`100 minutes, the standard deviation of sulpha-
`phenazole content of 200-mg samples was
`equivalent to that predicted for a random mix.
`The mix did not appear to segregate during
`mixing or on vibration. Under scanning elec-
`tron microscopy in conjunction with energy
`dispersive analysis of X-rays, sulphaphenazole
`appeared to be distributed quite uniformly on
`the coarse vehicle particles.
`
`INTRODUCTION
`
`The advantages of direct compression of
`tablets over traditional granulation techniques
`[ 1 - 31. In its
`have been well documented
`simplest form the process involves just one
`unit operation, that of mixing of drug and
`vehicle, prior to compression. The technique
`is probably most suitable for microdose tah-
`lets where the drug represents less than 5% of
`the total mix. From a drug dissolution stand-
`point the drug should be present in as fine a
`particle size as possible, whereas to ensure
`flowability the vehicle should be in a coarse
`granular form. This poses a mixing problem.
`Classical random mixing theory states that
`such a system would be difficult to mix.
`Recently, the concept of ordered mixing
`has been introduced [4]. This arose from an
`awareness that many glidants and lubricants
`act by adhering onto larger particles, im-
`
`proving flow [ 5 - 7]_ If by careful selection of
`drug and vehicle particle size a non-segregating
`ordered system could be constructed, direct
`compression could be a useful method of
`presenting small amounts of relatively insolu-
`ble drugs in a homogeneous form. The estra-
`granular position of the drug should ensure
`high dissolution rates.
`In this work the mixing of a fine drug,
`sulphaphenazole, with various size fractions
`of two commercially available direct com-
`pression tablet vehicles was studied- Sulpha-
`phenazole was selected as it has a cohesiveness
`and particle size distribution similar to drugs
`which are used in low dosage, e.g. digoxin,
`steroids.
`
`MATERIALS
`
`AND METHODS
`
`Sulphaphenazole was kindly donated by
`Ciba-Geigy Australia. Using an Alpine air-jet
`sieve, the particle size distribution was found
`to be log-normal. The geometric mean weight
`diameter
`(dpw)
`is 25
`/_un and the geometric
`standard deviation (c~) is 1.50. Celutab is a
`dextrose-maltose vehicle obtained from
`Brown and Dureau. From sieve analysis using
`an Endecott sieve shaker, d,, was estimated
`as 325 ~.rm with a us value of 1.75. Dipac, ob-
`tained from A-mstar Corporation, is a sucrose-
`dextrin vehicle with a d,, of 255 pm and og
`of 1.28. By liquid displacement studies at
`25 “C, particle densities of s-ulphaphenazole,
`Celutab and Dipac were estimated as 1.06,
`1.41 and 1.52 g ml-’ respectively.
`Mixing was carried out in an Erweka stain-
`less steel cube mixer (capacity 8 1) rotating at
`20 rpm. The initial load was 800 g, and at
`various time intervals 20 X 200-mg samples
`were removed using a sample thief. The sam-
`ples were assayed by dissolution in 40 ml of
`
`Cosmo Ex. 2021-p. 1
`Argentum v Cosmo
`IPR2018-00080
`
`

`

`162
`
`0.5% sodium carbonate and absorbance mea-
`surement at 250 nm using a Varian 635 u-v.
`spectrophotometer.
`After 100 minutes some systems were es-
`amined under a JS1M-U3 scanning electron
`microscope
`(JEOL Co. Ltd.) fitted with a
`device for energy dispersive analysis of X-rays
`(EDXX
`- Nuclear-Diodes
`Inc.).
`
`DEGREE OF 3IISEDXESS
`
`In the selection of a miser and the required
`mising time,
`it is necessary to compare a mea-
`sure of the variation of drug content
`in the
`mis (normally
`the standard deviation) with
`various parameters_
`The most common parameter used is the
`standard deviation
`for the fully randomised
`showed that the standard
`mix, o:; _ Lzxey
`[S]
`deviation of a fully randomised
`two-compo-
`nent system of identicd densities and particle
`size was gir-en by
`
`<I c = _Y_Vi,l’
`
`(1)
`where s and y are the proportions of the two
`components
`and N is the number of particles
`in the sample taken_ For a mix of two corlpo-
`nents, N is given by [ 33
`
`(2)
`
`where if’ is the weight of the sample and d,,
`and p are the volume-number mean diameter
`and density respectively of each component.
`The value of uR in a directly compressible
`system where components
`are of differing size
`is probably of limited absolute significance,
`particularly
`if crdered rather than random
`mising occurs_ However,
`it can be of value in
`selection of particle size of drug.
`less than 2%
`Where on is small, representing
`of the mean, the standard deviation of the
`sampiing and anaiyticai procedures
`(us) cannot
`be ignored. Thus the lowest standard deviation
`that could be achieved
`for the idealised system
`as described by Lacey wouId be uz, where
`
`U E = OR + Us
`
`(3)
`A more useful value in the practical situa-
`tion is the standard deviation necessary to
`or
`comply with pharmacopoeial
`specifications
`uA [9] _
`a manufacturer’s
`own specification,
`In this work, a, was estimated
`for 95% of
`
`samples falling within 110% of the mean, x,
`then
`
`+1.96u,
`
`= tO.10~
`
`(4)
`or ux = 0.05x. Thus if the mean drclg content
`of a 200-mg
`sample is 4 mg, (I_~ would be equal
`to 0.20 mg.
`
`RESULTS XSD DISCUSSIOS
`
`Figure 1 shows the change in standard de-
`viation with time of a mix of 2% sulpha-
`phenazole
`in Celutab. Studies were carried
`out with various size fractions of Celutab
`in
`addition
`to the unsifted material- Values of
`included on the ordinate-
`are
`GR~
`OE
`and
`0.4
`As the number of sulphaphenazole
`particles
`per 200-mg
`sample greatly exceeds
`the num-
`ber of Celutab particles, uR is relatively
`inde-
`pendent of Celutab particle size. LMising with
`the unsifted Celutab
`is very poor. After rapid
`initial mixing the system segregates, and after
`30 min the standard deviation of sulpha-
`phenazole content
`is greater than 1 mg. Seg-
`regation also occurs in the 250
`- 355~pm and
`the 355
`- 500-pm Celutab, and to a lesser
`extent with the 180
`- 250~pm
`fraction_ Using
`these size fractions of Celutab
`it was not pos-
`sible to mis within a manufacturer’s
`specifica-
`tion of u_, equal to 0.2 mg in 100 min_
`
`=E
`“3
`
`0
`
`20
`
`40
`
`60
`
`100 Y
`80
`tlme.mins
`
`Fig_ l_ Plot of standard deviation of sulphaphenazole
`content of 20 X 200-mg samples us. time for mixing
`of 2% sulphaphenazole with Celutab - Unsifted, 0;
`180 - 250-pm. 0; 250 - 355pm. A; 355
`- 500~Pm. 0;
`0; 710 - 1000 pm. a_ UR
`500 - ilO+m.
`is the thee-
`retical standard deviation of the fully randomised mix
`neglecting standard deviation of the sampling and ana-
`lytical procedure, US_ UE
`is equal to the sum of UR
`and US_ U,
`is a typical manufacturers mixing specifi-
`cation as given by eqn. (4).
`
`Cosmo Ex. 2021-p. 2
`Argentum v Cosmo
`IPR2018-00080
`
`

`

`However, on increasing the Celutab particle
`size fraction
`to 500
`- 710~flrn mixing
`improves,
`and after 50 min the standard deviation of
`to (Jo _
`sulphaphenazole
`content
`is equivalent
`Further mixing does not occur, but the system
`does not appear to segregate. A similar degree
`of misedness can be achieved after 100 min
`using the 710
`- lOOO-pm Celutab. After mixing,
`this material was vibrated
`in the hopper of a
`Manesty SPl single-punch
`tablet machine
`for
`one hour. The standard deviation of samples
`taken from this material,
`indicated by “V” on
`Fig. 1, suggests that the mix is stable to such
`vibration.
`Mixing of 25-pm sulphaphenazole with 710
`lOOO-pm vehicle particles is undoubtedly
`not
`a random process To further investigate
`the
`system, scanning electron microscopy was
`used. Sulphaphenazole
`particles appear to be
`quite angular. esisting mainly as agglomerates
`(Fig. 2)_ 710
`- iOOO-pm Celutab particles ap-
`pear to consist of 10 - 20 smaller spherical
`particles fused together to form an aggregate,
`which is stable to handling (Fig. 3a). On higher
`magnification
`the surface appears relatively
`porous, with few fines adhering (Fig. 3b). At
`even higher magnification
`the fine structure
`of the Cklutab particle appears porous and
`angular (Fig. 3~).
`After mixing for 100 min with 2% sulpha-
`phenazole,
`the Celutab particle adopts a
`“furry” appearance possibly suggesting the
`presence of adsorbed
`fines (Fig. 4a). At mag-
`nifications corresponding
`to those used in
`
`-
`
`163
`
`(b)
`
`Fig. 2. Scanning electron micrograph of sulphaphenazole
`particles.
`
`Fig. 3. Scanning electron micrographs of a Celutab
`particle (710
`- lOOO+m sieve fraction).
`
`Cosmo Ex. 2021-p. 3
`Argentum v Cosmo
`IPR2018-00080
`
`

`

`164
`
`(a)
`
`(b)
`
`(4
`Fig. 4. Scanning electron micrographs of a 710 - lOOO-pm Celutab particle after mixing for 100 min with 2%
`sulpxphenazole.
`
`Fig. 3(b) and (c), adsorbed particles can be
`visualied (Fig. 4b, c). At 2000X magnification
`such a -particle appears to resemble sulpha-
`phenazoIe (Fig. 4~).
`Using conventional scanning electron micro-
`scopy it is difficult to estimate the number of
`adsorbed sulphaphenazole particles per Celutab
`particle. For this purpose the EDAX system
`x-as used_ 1x1 this mode, the X-rays emitted
`x-hen the electron beam strikes the surface of
`the particle can be analysed to yield informa-
`tion as to the chemical composition of the top
`layer of the sample. The detector is sited ciose
`to the specimen, and energy discrimination
`
`takes place within it. The lower end of the
`range of a solid-state counter detector is 1 keV.
`so elements lighter than sodium will not be
`detected. Thus it is possible to scan Celutab
`particles for sulphur with very little back-
`ground from the vehicle, consisting essentially
`of only hydrogen, carbon, nitrogen and ouy-
`gen. The detector was set to pick up X-rays of
`2807 keV, representing the major peak for
`sulphur.
`Figure 5 is an EDAX photograph of a 710 -
`10OOqn-1 Celutab particle after mixing with 2%
`sulphaphenazoie- The white spots represent
`2.307 keV X-rays and the concentrated areas
`
`Cosmo Ex. 2021-p. 4
`Argentum v Cosmo
`IPR2018-00080
`
`

`

`165
`
`equal to uE. Thus a mix was achieved which
`is the best possible from random mixing theo-
`ry. No segregation can be observed, even after
`vibration for one hour in the single-punch
`tablet machine hopper. Figure 7(a) and (b)
`shows scanning electron micrographs of a
`180 - 250-pm Dipac particle and Fig. S(a) and
`(b) shows a similar particle after mising with
`sulphaphenazole_ Adsorbed drug can be iden-
`tified. Under EDAX of three different Dipac
`particles (Fig. 9), it is easy to count the num-
`ber of bound particles. Within each area
`scanned, there appear to be approximately
`15 sulphaphenazole particles. indicating a
`homogeneous distribution of the drug over
`the surface of the vehicle. This represents 5’i
`
`(a)
`
`Fig. 5. Scanning electron micrograph of the system
`Fig. 4 under EDAX.
`
`in
`
`parti-
`the sulphaphenazole
`represent
`of spots
`l/110
`cles. The area scanned is approximately
`of the surface area of a Celutab particle. From
`consideration of the size distribution of the
`two components, there should be approsi-
`mately 2140 sulphaphenazole particles per
`Celutab particle in the mix. From Fig. 5 there
`25 sulphaphenazole
`appear
`to be approximately
`particles in the area scanned, representing 2750
`particles per Celutab particle. This good agree-
`ment suggests that the majority of sulpha-
`phenazole particles are adsorbed onto the
`vehicle particles rather than free within the mix.
`Figure 6 shows mixing data of 2% sulpha-
`phenazole in Dipac. Again, mixing with un-
`sifted Dipac is poor, but with a 180 - 250-ctm
`fraction the observed standard deviation was
`
`0
`
`20
`
`40
`
`60
`
`100 v
`80
`time.mtns
`
`Fig_ 6. Plot of standard deviation of sulphaphenazole
`content of 20 X 200-mg samples L’s. time for mking
`of 2% sulphaphenazole with Dipac - Unsifted, 0;
`180 - 250qm. A; 250 - 355~j.fm. u; 355 - 500-,um, a_
`
`(b)
`Fig_ 7. Scanning electron micrographs of a Dipac par-
`ticle (180
`- 250-pm sieve fraction)_
`
`Cosmo Ex. 2021-p. 5
`Argentum v Cosmo
`IPR2018-00080
`
`

`

`166
`
`(b)
`Fig. A. Scanning electron micrographs of a 180 - 250-
`&ml Dipx particle after mixing for 100 min with BE
`sulphaphenazoIe_
`
`adsorbed drug particles per Dipac particle, com-
`pared with an espected
`ratio of 42 to 1 from
`their size distributions.
`appears
`Ordered mising of sulphaphenazole
`to occur at 2 lower Dipac particle
`size com-
`pared with Celutab. This may
`‘Je because the
`Dipac particle has a more irregular surface,
`allowing close approach of sulphaphenazole
`a smeller size_ Jones [ ‘71 has suggested that
`giidants may adsorb at surface irregularities on
`coarser particles. With larger Dipac size frac-
`tions, mising appears to be poorer and the
`unsifted widely distributed Dipac has a com-
`posite mhing profile.
`
`at
`
`(C
`rfer
`Fig. 9. Scanning electron micrographs of three dii
`DA
`Dipac particles from the system in Fig. 8 under E
`
`Cosmo Ex. 2021-p. 6
`Argentum v Cosmo
`IPR2018-00080
`
`

`

`readily available for dissolution and absorption
`in the gastro-intestinal
`tract.
`
`REFERENCES
`
`It appears that for an ordered mising system
`to be constructed,
`the vehicle should be close
`to monodispersed-
`In a widely distributed
`vehicle, drug may be bound to certain particles
`and some may be free to mis with the remain-
`ing vehicle particles_ Thus if vehicle particles
`themselves segregate from each other, the drug
`will segregate with the fraction to which
`it is
`adsorbed. So if a sample taken from the mix
`does not adequately
`represent the size distri-
`bution of the vehicle as a whole, the drug
`content will be smaller or greater than the
`required mean.
`The mechanism of adsorption of fine mate-
`rials onto coarser particles
`is not well under-
`stood.
`It was suggested that van der LVaals‘ and
`electrostatic
`forces were involved
`in the strong
`adsorption of fine magnesium oside and spray-
`dried lactose onto
`larger particles
`[ 6]_ Travers
`and White demonstrated
`the adsorption of
`micronised
`sodium bicarbonate on coarse
`[IO] _ They proposed
`sucrose particles
`that
`adsorption occurred at cqstal
`indentations
`and irregularities and that electrostatic
`forces
`were probably
`too weak. Other possible
`forces
`involved are chemisorption,
`surface tension or
`frictional
`forces [ 4]_
`The present work suggests that it should be
`possible to design a directly compressible
`sys-
`tem such that ordered mixing occurs. The
`degree of homogeneity
`attainable should be
`as good or better than that achieved by granu-
`lation, with the advantage that the drug is
`
`Cosmo Ex. 2021-p. 7
`Argentum v Cosmo
`IPR2018-00080
`
`