Multiple Tamping Effects on Drug Dissolution from
Capsules Filled on a Dosing-Disk Type Automatic Capsule
Filling Machine
K. B. SHAH**, L. L. AUGSBURGER“,AND K. MARSHALL§
Received August 29. 1986, from the ‘De artmenf of Phamaceufics, School of Pharmacy, University of Maryland, Baltimore, MD 21201, and
§Smifh Kline and French Laboratories, Pkladelphia, PA 19101. Accepted for publication June 6, 1987.
and Development, Skokie, IL 60077.
Abstract 0The effects of number of tampsand tamping force on drug
dissolution from capsulesfilled on an instrumenteddosing-diskautomat-
ic capsule filling machine (Hofliger-Karg)were studied using hydrochlo-
rothiazide as a model, low dose, poorly soluble drug. Generally, there
was a trend toward slower dissolution rate with increasing numbers of
tamps, the effect being most marked when insoluble dicalcium phos-
phate dihydrate was the filler. Higher compression forces improved drug
release when anhydrous lactose was the filler, but adversely affected
the dicalcium phosphate-based capsules. Inclusion of 4% croscarmel-
lose sodium disintegranttended to nullify the effects of number of tamps
or tamping force with both fillers; however, the disintegrantalso marked-
ly enhanced drug dissolution from the dicalcium phosphate-based
formulation. Hydrochlorothiazidedissolution from the latter formulation
without disintegrant appeared to follow a “diffusion from insoluble
matrix” model regardlessof number of tamps or their intensity. Mercury
intrusion pore size distribution data for some plugs suggested that for
tamps of equal force (100 or 200 N), further powder consolidation after
two tamps does not occur.
-... ~ . _ _
I n dosing-disk type automatic capsule filling machines, the
plugs of powder to be filled i n t o capsules are progressively
built up in dosing-disk cavities through a series o f tamps. In
previous the tamping stations o f a model GKF 330
Hofliger-Karg machine were instrumented by installing a
strain-gauged piston in each station. It was shown t h a t
although there was a provision for five consecutive tamps in
that machine, a m a x i m u m of three would normally be
required to achieve the desired fill weight for a given disk
thickness. However, both the number o f tamps and the
tamping force, as well as formulation variables, could influ-
ence the drug release rate from an encapsulated dosage form,
particularly if the drug had a low dose and was poorly
soluble. It has been shown on the simpler dosator machines,
in which plugs are formed by a single compression, that drug
dissolution can be affected by manipulating plug compac-
t i ~ nThus,
. ~ the experiments described below were designed
to investigate the influence o f compression force, tamping
history, filler type, and presence o f disintegrant o n the
dissolution o f a model, low dose, slightly soluble drug, hydro-
chlorothiazide.
Experimental Section
Formulations and Capsule Manufacture-The general formula-
tion consisted of 6% hydrochlorothiazide, USP (Pro Farmaco, Bolo-
gna, Italy), magnesium stearate (Amend Drug and Chemical Compa-
ny, Irvington, NJ) as lubricant, and filler. Direct tableting-grade
anhydrous lactose (Sheffield Products, Memphis, TN) and unmilled
dicalcium phosphate dihydrate (Ditab; Staufer Chemical Company,
Westport, CT) were used as “soluble” and “insoluble” fillers, respec-
tively. Magnesium stearate was used a t the 1.5%level for the soluble
filler and a t the 0.50%level for the insoluble filler. Two formulations
were designed for each filler, one without and the other with 4%
croscarmellose sodium (AcDiSol; FMC Food and Pharmaceutical
0022-3549/87/0800-0639$01.00/0
0 7987, American Pharmaceufical Association
Present address: SSearle Research
Products Division, Philadelphia, PA) a~ disintegrant. Batches of
3000 g (anhydrous lactose filler) or 4000 g (dicalcium phosphate
filler) were blended in a 15.1-L twin-shell blender (Patterson-Kelly
Co., East Stroudsburg, PA) for 10 min. The weight difference allowed
a charge of approximately equivalent volumes in the twin-shell
blender.
Size #1 capsules were filled on a model GKF 330 Hofliger-Karg
automatic capsule-filling machine (Bosch Packaging Machinery Di-
vision, Piscataway, NJ) which had been instrumented with strain
gauges to monitor tamping forces, a~ previously described.1.2 The
machine speed was held constant at 100 strokes per min, and a 1.58-
cm thick dosing disk was used in all runs. Either one tamp or up to
three consecutive tamps (i.e., stations #5, #5 and #4, and X5, #4,
and #3) a t a given compression force were utilized during encapsula-
tion to study the multiple tamp effect. Separate runs were made
using compression forces of 100 and 200 N to assess any possible
compression force effect.
The inclusion of hydrochlorothiazide at a fixed concentration of 6%
resulted in a 7-27-mg range in drug content, since fill weights
varied with the number and intensity of tamps, as well a~ with the
density of the filler. Based on the solubility of hydrochlorothiazide
(0.106% in 1:lOO hydrochloric acid at 37 “(29, even the maximum
drug content resulted in a drug concentration during dissolution that
was well below the 10-152 of solubility limit considered satisfactory6
for approximating sink conditions.
Dissolution-The dissolution rate of hydrochlorothiazide was de-
termined by USP method 11, with paddles.6 A six-head dissolution
apparatus equipped with a multiple-drive stirrer (model 2000 Disso-
lution System; Distek, Inc., Somerset, NJ) was employed. The
dissolution medium was 900 mL of dilute hydrochloric acid (1:lOO)
maintained at 37 “C. The paddles were positioned 2.5 cm above the
bottop of the flasks and rotated a t 50 2 2 rpm. Six randomly selected
capsules of each formulation were tested. The capsules were held in
stainless steel spirals to prevent their floating. Dissolution runs
were carried out for 64 min. One set from each of the four formula-
tions was tested for 100 (? 5)% release by increasing the paddle
speed to 250 rpm. A sequential flow system (Dissograph Flow
System, model 49-200-000; Hanson Research Corporation, North-
ridge, CA) was used. This system provides for serial withdrawal of
filtered samples from each flask a t preset time intervals for spectral
analysis and for return of the sample to the corresponding flask
following absorbance measurement. To avoid disturbance of the
stirring medium during runs, a pneumatically driven automatic
sampler (model 27-700-000; Hanson Research Corporation, North-
ridge, CA), which inserts sampling probes only for the specified
sampling time, was employed. The withdrawn samples were
pumped, single file, through a 0.1-cm flow cell (model 170; Helma
Cells, Inc., Jamaica, NY)that was mounted in a dual-beam, micro-
processor-controlled spectrophotometer (model Lambda 3B; Perkin
Elmer Corporation, Oak Brook, IL). All absorbance8 were read at
272 nm7 against dissolution medium in the reference flow cell. The
reference dissolution medium was replenished between dissolution
runs from a reservoir container, using a peristaltic pump. The
spectrophotometer was interfaced through a serial port with a
desktop computer (IBM PC; IBM Corporation, Armonk, NY). The
sobware calculates percent drug dissolved based on drug content of
individual dosage units and takes into account the actual fill weight
of the test capsules.
The mean percents dissolved and standard errors appear in Tables
I-IV.
Journal of Pharmaceutical Sciences / 639
Vol. 76,No. 8,August 1987
 Porosity Measurement-The effect of compression force and the
number of tamps on powder consolidation was further studied by
determining the porosity and pore size distribution of anhydrous
lactose-based plugs containing hydrochlorothiazide. A mercury in-
trusion porosimeter (model 9305 Pore Sizer; Micromeritics, Norcross,
GA)was used. Data acquisition and reduction was largely automat-
ed through serial interfacing with the desktop computer. Intrusion
readings up to 168 Mpa were taken. Visual examination ofthe plugs
revealed no apparent physical damage as a result of the intrusion
pressure. Moreover, total pore volumes calculated from true densi-
ties and plug dimensions agreed to within 22%with total intrusion
volumes.
Results and Discussion
These studies were designed to identify the influence of the
Table I-Effect of Multiple Tamps and Compression Force on Hydrochlorothiazlde Dissolution from Anhydrous Lactose-Based
Capsules'
number of tamps and the extent of tamping, if any, on in
vitro drug release. The formulation variables considered
were filler type and the presence (or absence) of a distinte-
grant.
Anhydrous L ~ ~ ~ ~~ ~ ~ ~ ~~ effect l ~of ~ - i
number Of tamps (up ''
three tampsp each at the same
compression force) on hydrochlorothiazide dissolution at low
(100 N) and high (200 N)ComPression force may be seen in
Figures 1 and 2, respectively. The pore size distributions for
these capsule plugs appear in Figures 3 and 4. It is apparent
that regardless of compression force (high o r low), multiple
tamping adversely affected the dissolution profile. This ob-
servation is surprising in light of a previous finding2 on
similar placebo formulations that the mechanical strength of

Compression Number Percent Drug Dissolved at Various Times, min

Force, N of Tamps 8 16 24 32 40 48 56 64
100 1 25.4 (1.99) 45.6(2.11) 56.5 (1.69) 63.1 (1.59) 68.4(1.63) 74.1 (1.66) 78.6 (1.72) 82.1 (1.58)
2 21.2(2.24) 41.8(0.79) 52.3 (0.69) 58.9 (0.82) 64.9 (0.74) 69.6(0.84) 73.8(0.92) 77.3 (1.06)
3 15.0(0.88) 35.4(1.28) 46.9 (1.06) 54.8 (1.34) 60.3(1.56) 65.1 (1.73) 69.0(1.90) 72.6(2.02)
200 1 26.0(1.22) 48.9 (0.52) 59.4 (0.78) 67.2(1.00) 73.4(1.13) 78.3 (1.19) 82.4(1.26) 86.0 (1.41)
2 17.8(1.25) 40.1 (1.11) 51.4 (0.83) 59.1 (0.95) 64.8(1.07) 69.9 (1.14) 74.2(1.21) 77.3(1.20)
3 15.4(0.79) 35.7(0.95) 47.8(0.91) 55.5 (1.04) 61.5 (1.10) 66.0 (1.18) 70.2 (1.22) 73.7(1.27)
'Magnesium stearate (1 5%)as lubricant. bMean of 6 determinations; standard error of mean in parentheses.

Table 11-Effect of Multiple Tamps and Compression Force on Hydrochlorothlazide Dissolution from Anhydrous Lactose-Based
Capsules Containing 4% Croscarmellose Sodlum'

Compression Number Percent Drug Dissolved at Various Times, min

Force, N of Tamps 8 16 24 32 40 48 56 64
100 1 38.0(1.16) 51.6(0.51) 58.7(0.29) 64.3(0.38) 69.3(0.50) 73.7(0.54) 77.7(0.56) 80.7(0.66)
2 35.9(1.19) 49.8(0.44) 57.2(0.22) 62.7(0.35) 68.0(0.39) 72.5 (0.48) 75.8(0.62) 79.2(0.75)
3 34.3 (2.10) 48.8(1.67) 56.7(1.53) 62.2(1.71) 67.0(1.76) 71.5 (1.87) 75.0 (1.87) 78.0(1.82)
200 1 38.0(1.27) 53.4 (0.83) 61.7(0.65) 67.9(0.72) 72.9(0.77) 77.2 (0.88) 81.0 (0.90) 84.0(0.96)
2 36.2(1 .lo) 51.3 (0.53) 58.4(0.52) 64.5(0.51) 69.5(0.49) 73.8 (0.49) 77.3 (0.38) 80.4(0.44)
3 31.3 (1.44) 47.5 (0.57) 54.8 (0.31) 60.4(0.32) 65.0(0.32) 68.9 (0.35) 72.4(0.40) 75.3(0.40)
'Magnesium stearate (1.5%)as lubricant. Mean of 6 determinations; standard error of mean in parentheses.

Table Ill-Effect of Multiple Tamps and Compression Force on Hydrochlorothlazlde Dissolution from Dlcalclum Phosphate-Based
Capsules'

Compression Number Percent Drug Dissolved at Various Times, min

Force, N of Tamps 8 16 24 32 40 48 56 64
100 1 12.1 (0.60) 20.5 (0.91) 26.9 (1.19) 32.0 (1.43) 37.0(1.57) 41.3 (1.84) 45.2(1.93) 48.8(2.07)
2 9.2(0.29) 16.0(0.45) 21.1 (0.55) 25.4 (0.74) 29.4 (0.92) 32.9 (1.06) 36.3 (1.12) 39.7(1.25)
3 8.9(0.23) 15.6(0.42) 20.5(0.53) 24.5(0.64) 28.2 (0.71) 31.3 (0.87) 34.6 (0.92) 37.5(1.00)
200 1 10.6(0.55) 17.8(0.68) 22.7(0.72) 27.0(0.75) 30.5 (0.86) 33.8 (1.00) 37.2 (1.12) 40.2(1.26)
2 7.6(0.10) 13.0(0.23) 17.3(0.24) 20.7(0.29) 24.0 (0.31) 27.0(0.35) 29.6(0.39) 32.4 (0.44)
3 7.7(0.32) 13.4(0.34) 17.4(0.41) 20.7(0.42) 23.8 (0.54) 26.6(0.57) 29.5 (0.61) 32.1 (0.71)
a Magnesium stearate (0.5%)as lubricant. bMean of 6 determinations; standard error of mean in parentheses.

Table IV-Effect of Multiple Tamps and Compression Force on Hydrochlorothlazlde Dlssolutlon from Dlcalclum Phosphate-Based
Capsules Containing 4% Croscarmellose Sodium'

Compression Number Percent Drug Dissolved at Various Times, min

Force, N of Tamps 8 16 24 32 40 48 56 64
100 1 28.8 (1.74) 41.9(1.76) 51.9 (1.56) 59.4 (1.63) 66.0(1.54) 70.6(1.57) 75.0(1.50) 78.4(1.50)
2 28.4 (0.94) 40.8(1.47) 50.0 (1.19) 57.3 (0.94) 63.9(0.80) 68.9(0.39) 74.2(0.46) 78.8(0.30)
3 27.7 (1.58) 41.5 (1.82) 51.1 (1.72) 59.3 (1.58) 66.2(132) 72.2 (1.06) 77.2(0.83) 81.3(0.67)
200 1 31.0 (1.72) 44.2(1.83) 53.6 (1.54) 61.7(1.39) 68.4(1.23) 73.6 (1.07) 78.0(1.03) 81.0(1.06)
2 28.7 (1.24) 40.3(1.41) 49.6(1.30) 56.1 (1.04) 62.0(0.94) 67.5(0.85) 72.5(0.73) 76.3(0.73)
3 28.4(1.72) 41.3(1.89) 51.1 (1.97) 59.6(1.96) 66.5(1.92) 72.0(1.86) 76.6(1.70) 79.9(1.60)
a Magnesium stearate (0.50%)as lubricant. Mean of 6 determinations; standard error of mean in parentheses.

640 /Journal of Pharmaceutical Sciences

Vol. 76, No. 8, August 1987
the large, initially formed plug segment did not change after
a second or third tamp a t the same compression force. If plug
mechanical strength is a good indicator of powder consolida-
tion, that finding would suggest that multiple tamping does
not cause further compaction. However, the results of the
present study reveal a uniform decrease in pore size distribu-
tion at both force levels from the first tamp to the second
tamp which does, indeed, point to further consolidation.
Evidently, the previously reported mechanical strength mea-
surements were not sensitive enough to detect these relative-
ly small changes in porosity. At the low compression force
(Figure 3), it may be seen that the median pore diameter
decreases from 12.8 pm with one tamp to 10.8 Fm with two
tamps, although no change was apparent between two tamps
and three tamps. In considering the dissolution data (Figure
11, it is interesting to note that the differences in percent
dissolved at any time between the curves representing one,
two, and three tamps were constant. Although small, these
differences were statistically significant (p < 0.05) at 32 min,
100
but not a t 64 min, based on a two-sided t-test for the two
extremes (one tamp versus three tamps).
The trend toward decreasing dissolution rate with in-
creases in the number of tamps is even more clear in the high
compression force data (Figure 2). While the differences in
percent drug dissolved a t various times still remained more
or less constant for the three tamping conditions, a signifi-
cantly (p < 0.05) higher percent of hydrochlorothiazide
dissolved at the 32- and 64-min marks from the plugs
obtained with one tamp than from plugs obtained with two or
three tamps. The pore size distribution (Figure 41, as before,
shifted toward a smaller size when the number of tamps
increased from one to two, but was not greatly affected by the
third tamp.
The consistent differences a t all times in percent release a t
both low and high compression forces among the three
tamping groups, with maximum release occurring when only
a single tamp was employed, points to an initial “burst” effect
in drug release. This burst effect probably results from the
loose powder that fills the disk cavity after the last tamp and
prior to ejection, and is likely to be most significant in the

80

B
3
2 00
v,
0
Q
2 40
0
s
20
A

0 , . . I . . .
0 13 20 38 61 66 Lo 10 4 I 0.2
TIME (mid
PORE SIZE (urn)
Figure 1--Effect of multiple tamps at low compression force (100 N) on
hydrochlorothiazide dissolution from anhydrous lactose-basedcapsules Flgure 3-Hfect of multiple tamps at low cornpression force (100 N) on
(7.5% magnesium stearate). Key: (0) tamping at station #5; (0) pore size distributions of plugs (anhydrous lactose-based formulation of
tamping at station #5,4; (A)tamping at station 15, 4, 3. hydrochlorothiazide, 7.5% magnesium stearate). Key: ( 0 )tamping at
station #5; (a)tamping at station #5, 4; (A)tamping at station #5, 4, 3.
100 7

80 -
e3
g 80-
td 60-
CL
0
0
a
2 40- w
I
0 I-
be Q s-
5
20 -
u
f
1
4a 10 4 1 0.2
01 I 1
0 13 28 38 62 86
TIME (mid PORE SIZE (urn)

Flgure 2-€ffect of multiple tamps at high compression force (200 N) on Flgure 4-Effect of multiple tamps at high compression force (200 N) on
hydrochlorothiazidedissolution from anhydrous lactose-based capsules pore size distributions of plugs (anhydrous lactose-based formulation of
( 1 5% magnesium stearate). Key: see Figure 1. hydrochlorothiazide, 7.5% magnesium stearate). Key: see Figure 3.

Journal of Pharmaceutical Sciences I641

Vol. 76, No. 8, August 7987
case where only one tamp is employed. Previous work2
showed that the full plug length is not realized with just a
single tamp, and that a void volume remains which is filled
volumetrically from the powder head over the dosing disk
prior to the scrape-offwhich precedes ejection. Since complete
plug length is realized when two tamps are employed, cap-
sules filled when employing two tamps will contain no loose
powder. Assuming the intact plug to release drug a t the same
rate in all cases, this initial burst could account for the
consistently higher dissolution profiles of these capsules. The
amount of loose fill would be less at the low compression force
(i.e., less consolidation of the intact plug, thereby leaving less
room for the loose fill) than it would be at the high compres-
sion force. This latter point thus could account for the greater
(but consistent with time) differences in percent dissolved
between the single-tamp and two-tamp data at the higher
(Figure 1) and lower compression forces (Figure 2).
The effect of single-tamp compression force on hydrochloro-
thiazide dissolution is shown in Figure 5, and the pore size
distribution data for those capsule plugs are given in Figure
6. An examination of these pore size distribution data indi-
100
cates that a greater consolidation of the powder mass oc-
curred with an increase in compression force. Curiously,
however, the increased densification resulted in an improved
dissolution rate for this soluble system. A similar observa-
tion has been reported by Botzolakis et aL8 for a similar
hydroch1orothiadize:anhydrous lactose formulation filled on
a dosator machine, and by Greenberg regarding an unspeci-
fied formulation, also filled on a dosator machine. Exactly
why dissolution rate increased with increased compression
force is not clear, but the effect in part may be related to
greater particle fracture or frictional shearing a t higher
compression force resulting in more clean surfaces uncon-
taminated with hydrophobic lubricant. For these lactose
formulations, both the burst effect and greater consolidation,
together, yield overall improved dissolution profiles a t high-
er compression forces.
That the curves in Figure 5 do not show equal spacing at
all time points is not inconsistent with the discussion of
100
80

80

e3
0 60
v)
v,
0
c7
2 40
n
z
20
01 I 1 , 1
0 13 26 3e 62 65
TIME (mid
0
0 13 26 38 62 66
Figure 7-€ffect of multiple tamps at low compression force (100 N) on
TIME (mid
hydrochlorothiazidedissolution from anhydrous lactose-based capsules
containing 4% croscannellose sodium disintegrant (1.5% magnesium
Figure 5-€ffect of compression force on hydrochlorothiazide dissolu-
stearate). Key: see Figure 7.
tion from anhydrous lactose-based capsules (7.5% magnesium stea-
700 N; (0)
rate) utilizing only a single tamp. Key: (0) 200 N.

W
5

I . O I , 1 I
40 10 4 1 0.2 0 13 20 38 62 05
TIME (mid
PORE SlZE(um)
Figure 8-.€ffect of multiple tamps at high compression force (200 N) on
Figure 6--Effect of compression force on pore size distributions of hydrochlorothiazidedissolution from anhydrous lactose-basedcapsules
plugs (anhydrous lactose formulation of hydrochlorothiazide, 1.5% containing 4 % croscarmellose sodium disintegrant (1.5% magnesium
magnesium stearate). Key: (A)700 N; (A)200 N. stearate). Key: see Figure 1 .

642 /Journal of Pharmaceutical Sciences

Vol. 76, No. 8, August 7987
 Figures 1 and 2. In the early stages, dissolution from both From the previously noted porosimetry data (anhydrous
capsules is primarily due to the dissolution of the faster lactose, Figures 3 and 41, it is clear that consolidation
dissolving loose powder. Thus, one would expect little or no occurred only up to two tamps. With both the anhydrous
difference between the two capsules initially, with the curves lactose and dicalcium phosphate formulations, fill weight
gradually diverging to a fairly constant difference at later also increased with up to only two tamps. Thus, it seems
times, owing to the different amounts of loose powder fill as reasonable to infer that even for the dicalcium phosphate
well as to the effect of the two different compression forces on formulations, consolidation and porosity reduction occurred
the intact plug. This is exactly what the data in Figure 5 with up to only two tamps. Based on these results (Figures 9
show. and lo), it is apparent that in vitro drug release depends on
The effects of disintegrant (4% croscarmellose sodium) and the extent of consolidation for this "insoluble" system. A
multiple tamps a t low and high compression force on the comparison of Figures 9 and 10 also reveals that dissolution
dissolution of hydrochlorothiazide from anhydrous lactose- rate decreased (regardless of the number of tamps) a t the
based capsules are summarized in Figures 7 and 8. The higher compression force. This compression force effect is
presence of disintegrant at low compression force yielded most clearly evident in the single-tamp data (Figure 11).
essentially identical dissolution patterns for the capsules Overall, the differences in dissolution profiles generally
filled using any of one, two, or three tamps. The burst effect, increased with time as a function of both the number of
observed previously in the absence of disintegrant, was not tamps and the extent of tamping.
evident here. It is likely that rapid plug dispersal and initial
dissolution, resulting from the presence of disintegrant, SOUARE ROOT OF TIME (rnin"')
overshadowed any such effect. A comparison between Fig-
ures 1 and 7 clearly indicates that for any tamping combina-
tion, a significantly (p < 0.05) higher percent of hydrochloro-
thiazide dissolved in 8 min when the disintegrant was
included in the formulation than was released in 8 min from
I
the formulation without disintegrant.
Figures 7 and 8 also reveal that an increase in the number
of tamps tends to retard dissolution from the disintegrant-
containing lactose formulations. This effect is most clearly
obvious at the high compression force (Figure 8).
Dicalcium Phosphate-Based Formulations-The effects
of the number of tamps and the extent of compression on
hydrochlorothiazide dissolution from the dicalcium phos-
phate-based formulations may be seen in Figures 9-11.
Increasing the number of tamps slowed drug release at both
the low (Figure 9) and high (Figure 10) compression forces,
but only when up to two tamps were employed. When a third
lo 1
tamp was applied at the low compression force (Figure 91, the 0
dissolution rate appeared to decay further; however, the 0 13 26 39 62 66
differences were not significant (p < 0.05) at 32 and 64 min. TIME (rnin)
The application of a third tamp at the higher compression
Figure lO-€ffect of multiple tamps at high compression force (200 N)
force had no apparent effect on drug dissolution (Figure 10). on hydrochlorothiazide dissolution from dicalcium phosphate-based
capsules (0.50%magnesium sfearate). Key: see Figure 9.
SOUARE ROOT OF TIME (min"')
0 2 4 6 8
60 SOUARE ROOT OF TIME (min"')
0 2 4 6 8

40

B
23 30

0
2 20
n

*I
zp
10
10

0
0 13 26 39 62 66 01 , I I I
TIME (mid 0 13 26 30 62 66
TIME (mid
Figure +Effect of multiple tamps at low compression force (100 N) on
hydrochlorothiazide dissolution from dicalcium phosphate-based cap- Figure 11-Effect of compression force on hydrochlorothiazidedissolu-
sules (0.50%magnesium stearate). Key: (0) tamping at station #5; (0) tion from dicalcium phosphate-based capsules (0.50% magnesium
tamping at station X5, 4; (A) tamping at station #5, 4, 3; open symbols: 100 N;(0)
stearate) utilizing only a single tamp. Key: (0) 200 N;open
x-axis = time; closed symbols: x-axis = time "? symbols: x-axis = time; closed symbols: x-axis = time l'?

Journal of Pharmaceutical Sciences / 643

Vol. 76, No. 8, August 1987
 Interestingly, all dicalcium phosphate-based plugs re-
mained intact even after the last dissolution sample was
taken at 64 min. HiguchilO has described the following model
for diffusion-controlled drug release from a heterogeneous
matrix in which diffusion takes place in the intergranular
pores:
(1)
where Q is the amount of drug released per unit surface area
after time t , D is the diffusion coefficient of the drug in the
dissolution medium, c is the porosity of the matrix, r is the
tortuosity of the matrix, A is the concentration of the drug in
the compact, and Cs is the solubility of the drug in the
dissolution medium. This equation holds as long as 2A >>
E c s , and predicts a linear fit for the amount released per unit
area versus t”2. A capsule fill weight range of 0.369 to 0.505
g, containing drug at 6% (w/w) in a disc cavity volume of
0.396cm3, corresponds to a range of A of 5.6-7.6% (w/v).The
solubility of the drug in the dissolution medium has been
reported to be 0.106% (w/v) a t 37 0C.4Furthermore, from the
lactose data and an earlier report,” the porosity of the plugs
could reasonably be expected to be in the range of 0.4 to 0.5.
Thus, it is quite reasonable that the condition 2A >> E c s
holds in this case. However, the surface area is not held
constant in a typical dissolution study. Nevertheless, the fact
that the plug did remain intact throughout the dissolution
run suggests a possible fit to the Higuchi model, with the
limitation noted above. As may be seen in Figures 9-11,
when the dissolution data were regressed and plotted as
percent dissolved against t1’2, an excellent fit of the model
was found in all cases (coefficient of determination 2 0.999).
The squared x-intercept ranged from 0.82-1.6 min (Table V)
which could represent the time taken for the capsule shell to
dissolve in the release medium and for the initial direct
exposure of the plug to the dissolving fluid. Decreasing slopes
with increases in the extent and frequency of tamps could be
a function of resultant greater consolidation and, thus, de-
creasing porosity.
The addition of disintegrant improved drug dissolution
over that of the control dramatically, regardless of the
number of tamps andlor the tamping force. As may be seen in
Figures 12 and 13, the number of tamps did not adversely
affect drug dissolution. Neither was there any specific effect
100 7
80 -
8
3 60-

Table V-Regresslon Parameters Using A “Diffusion from

lsoluble Matrlx” Model for HydrochlorothlazldeDlssolutlon from 0
3
IE 40-
Dlcalclum Phosphate-BasedCapsules’
n
Compression Number Regression Parameterb ae
Forces N Of Tamps Slope, %/min”z (x-intercept)Z,rnin F 20 -
100 1 7.1 1 1.26 0.9999
2 5.86 1.65 0.9997
3 5.49 1.40 0.9998 01 I I 1 1

200 1 5.65 0.82 0.9996 0 13 26 39 62 66

2 4.78 1.62 0.9996 TIME (mid

3 4.66 1.37 0.9994
Figure 13-.€ffect of multiple tamps at high compression force (ZOO N)
aMagnesium stearate (0.5%) as lubricant. bObtained by regressing on hydrochlorothiazide dissolution from dicalciurn phosphate-based
mean (n = 6) percent dissolved against timeo2from 64-min dissolution capsules containing 4% croscarmellose sodium (0.50% magnesium
runs. stearate). Key: see Figure 7.

’””1
80 -
lo0l

ow 601

60 -
v,
0

40 -

0
I

13 ze 39 62 e6
!i 401
20

0 13 26 39 62
I 1
66
TIME (mid TIME (mid
Flgure 12-€ffect of multiple tamps at low compression force ( 1 00 N) on Figure 14-Effect of compression force on hydrochlorothiazidedissolu-
hydrochlorothiazide dissolution from dicalcium phosphate-based cap- tion from dicalciurn phosphate-based capsules containing 4% croscar-
sules containing 4% croscarmellose sodium (0.50% magnesium stea- mellose sodium filled using a single tamp (0.50% magnesium stearate).
rate). Key: see figure l. Key: (0) 100 N; (3) 200 N.

644 /Journal of Pharmaceutical Sciences

Vol. 76, No. 8,August 1987
due to compression force (Figure 14). Four,percent croscar-
mellose sodium may be sufficient for this insoluble system to
provide disintegration and deaggregation sufficiently rapid
to overcome any effects resulting from multiple tamps or
increased tamping force within the range of these experi-
ments.
Conclusions
Formulators are called upon to optimize formulations not
only with respect to the performance characteristics of the
finished dosage form, but also with respect to the manufac-
turing operation, and a study of the interplay of process
variables as related to the formulation performance is essen-
tial to attaining that goal. In our previous study,2 certain
machine operating variables and formulation factors affect-
ing plug formation and fill weight were studied for a dosing-
disk automatic capsule filling machine. The present study
focused on the effects of certain of these variables on drug
release.
The implications of the two important dosing-disk machine
operating variables (i.e., the tamping force and the number of
tamps) on drug release were explored in a dissolution study
using hydrochlorothiazide as a model, low dose, poorly solu-
ble drug. Anhydrous lactose and dicalcium phosphate were
chosen to represent soluble and insoluble fillers, respectively.
Generally, there was a tendency toward slower drug release
with an increase in number of tamps. This effect was more
dramatic for the insoluble system than for the soluble sys-
tem; however, that variable could be eliminated in either
case when 4% croscarmellose sodium, a tablet disintegrant,
was included in the formulation.
Drug release improved slightly from the anhydrous lac-
tose-based capsules when a higher tamping force was ap-
plied, but dissolution was adversely affected by compression
force when dicalcium phosphate was the filler. Hydrochloro-
thiazide dissolution from this latter matrix appeared to fit a
“diffusion from insoluble matrix” model through the dissolu-
tion times studied. The decreasing slopes of the square-root-
of-time dissolution plots based on that model point to de-
creasing porosity andlor increasing tortuosity a t high
compression force or with multiple tamps. This latter obser-
vation could be used advantageously to control drug release
from an insoluble, nondisintegrating matrix since the same
fill weight may be obtainable with varying degrees of consoli-
dation by adjustment of the number of tamps and the
tamping force.
References and Notes
1. Shah, K. B.; Augsburger, L. L.; Small, L. E.; Polli, G. P. Pharm.
Technol. 1983, 7(4), 42.
2. Shah, K.B.; Augsburger, L. L.; Marshall, K. J . Pharm. Sci.
1986. 75. 291.
3. Mehta, A.M.; Augsburger, L. L. Int. J . Pharm. 1981, 7, 327.
4. Augsbur er, L.L.; Shangraw, R. F.; Giannini, R. P.; Shah, V. P.;
f.
F’rasad, K.; Brown, D. J . Pharm. Sci. 1983, 72, 876.
5. Levy, G. In Papers Presented Before the Industrial Pharmacy
Section, A.Ph.A. 113th Annual Meeting, Dallas, TX, A ril 1966;
American Pharmaceutical Association: Washington, 8C,1966;
289-293.
6. VS. Phurmacopeia XXIlNational Formulary XVI; The US.
Pharmacopeial Convention: Rockville, MD, 1985; 1244.
7. US. Phurmacopeiu XXIlNational Formulary X h ; The US.
Pharmacopeial Convention: Rockville, MD, 1985; p 497.
8. Botzolakis, J. E.;Small, L. E.; Augsburger, L. L. Int. J . Pharm.
1982,12, 341.
9. Greenber R. Proc. 88th National Meeting, Am. Inst. Chem.
Engrs.; Pkladel hia, PA, June 8-12, 1980, Fiche 29.
10. Higuchi, T. J . P L r m . Sci. 1963,52, 1145.
11. Botzolakis, J. E.,Ph.D. Thesis, University of Maryland, Balti-
more, MD, 1985.
Acknowledgments
University of Maryland, in partial fulfillment of the Doctor of
Philosophy degree requirements. ’
Journal of Pharmaceutical Sciences / 645
Vol. 76, No. 8, August 1987