ASPECTOS GEOGRÁFICOS

4.6.1 Experimental

Although there has been wide application of a variety of tensile testing techniques to pharmaceutical powder systems little work has examined the compressive strength of powder compacts. Newton et al (1993) described the measurement of a compressive to tensile strength ratio for cylindrical pharmaceutical specimens which could be used to characterise the mechanical properties of powders. Such a specimen geometry allows the application of both the diametral compression test and a compressive strength test, thus this technique was employed for this study.

Twelve millimetre diameter specimens with equi-dimensional geometries (thickness/diameter ratio range 0.94 to 1.00) were prepared using a specially constructed extra-long die at compaction pressures of 25, 30, 35, 40, 45 and 50MPa in conjunction with the electrical circuit (Chapter 2, Sections 2.3.1 and 2.3.2). Preparation of specimens was limited to this compaction pressure range because of the equi-dimensional requirement which is the minimal prerequisite for compressive strength testing (Chapter 1, Section 1.2.3.6). Twenty replicate specimens were made at each pressure with 10 used for diametral compression tests and 10 used for compression tests. Diametral compression tests were performed and the tensile strength calculated as described in Section 4.2.1 whilst compressive tests were performed as described in Section 2.3.4 with the compressive strength (Oc) calculated using equation 4.3;-

W

(Equation 4.3)

where Wmax represents the maximum force registered under a compressive load and r is the specimen radius.

No attempt was made to reduce platen-specimen fiictional end-efiects as Newton et al (1993) reported that the insertion of a Teflon sheet between the specimen and platen

in their tests did not significantly alter the force-time curve profile or the maximum load detected.

4.6.2 Results

All specimens subjected to diametral compression testing failed in the normal manner with force-time loading profiles similar to that shown in Figure 4.12(a) where the point at which failure occurs is clear. Failure was far less clear with compression testing with a typical force-time profile recorded as shown in Figure 4.12(b). A slightly rising plateau was not evident in these tests as reported by Newton et al (1993) but the force tended to reach a maximum with a very short plateau , lasting for an approximate duration of 2-3 seconds, before the force began to fall relatively rapidly. Figure 4.13 shows a scanning electron micrograph of a specimen subjected to compression testing. Under an increasing compressive force specimens went through a number of stages of deformation and fi’acture leading to their eventual catastrophic failure as follows :-

(a) Specimen assumes a cone or bell-shape as the force increases.

(b) Fracture begins very near to the circumference of the specimen at the ‘unconfined’ wider end.

(c) Fracture occurs around the entire specimen circumference and runs up towards the ‘confined’ end.

(d) Specimen eventually loses a ‘collar’ of material fi*om its circumference.

(e) The remainder of the specimen has a dumbbell shape if loading is continued further. Thus Figure 4.13 reflects stage (d) of the failure and the significance of the sequence of failure is discussed below (Section 4.6.3).

Table 4.5 shows the values of tensile and compressive strength measured over the range of compaction pressures employed (which is also illustrated graphically in Figure 4.14). In each case the value represents the mean of 10 determinations. The compressive to tensile strength ratio is also shown for specimens made at each compaction pressure and can be used to give an indication of the type of mechanical behaviour of the material.

Figure 4.14 shows the specimen tensile and compressive strengths as a function of the lower punch compaction pressure. Linear regression analysis was performed on the data in the same manner described in Section 4.2.2.

Figure 4.12 Force-time curves for the fracture of 12mm diameter Avicel PHI 02 specimens during (a) diametral compression and (b) compressive strength testing (a)

I'

!■

D«t WD 8E 34.9 CC.V Spot Magn

00 kV 4.0 9x AVICEL PH102

Figure 4.13 Scanning electron micrograph of an Avicel PH 102 specimen subjected to compression strength testing illustrating the typical mechanism of failure

m m

Table 4.5 Compressive and tensile strength of Avicel PHI 02 compacts prepared at valuing compaction pressures

25 30

Compaction pressure (MPa)

35 40 45 50

Tensile strength (MPa) 1.561 1.977 2.385 2.831 3.226 3.592

(Gt)

Compressive strength (MPa) 21.28 28.11 31.56 36.66 42.74 47.73

(Ge)

Ratio 13.63 14.22 13.23 12.95 13.25 13.29

50.0 - 4 5 .0 - 40 .0 - g 35.0 — 30.0 - 25 .0 - -C % § I 20.0 y a . 5 “ 15.0 H 10.0 - 5.0 - 0 20 ~T~ 25 —I---1---1--- r~ 30 35 40 45 Lower punch com paction pressure (MPa)

50 55

Figure 4.14 Tensile ( • ) and compressive ( ^ ) strength as a function of compaction pressure for 12mm diameter Avicel PH 102 specimens (raw data; n=10)

regression analyses reported in Table 4.6.

Figure 4.15 illustrates the relationship between the natural logarithm of both the tensile strength (ot) and the compressive strength ( a j as a function of porosity (e) for the specimens. Linear regression analysis was performed on the data in the same manner described in Section 4.2.2.

A significant linear relationship (as described by Duckworth, 1951) was demonstrated for both cases yielding the regression analyses reported in Table 4.7. A comparison of the compressive to tensile strength ratio was made at a porosity of 0.30 as this represented a porosity value within the experimental range.

4.6.3 Discussion

The failure mechanism seen with these specimens was very complex but can be partially explained by considering the literature on the compression testing of concrete and other such brittle materials (Chapter 1, Section 1.2.3.6). Hawkes and Mellor (1970) reviewed the uniaxial compression test and considered the problems associated with it. The authors stated that displacement boundary conditions are preferred where rigid non-rotating platens are used to constrain the specimen, allowing uniform loading and symmetrical deformation. However, most of the specimens in this study assumed a bell or barrel-shape during loading which can be associated with specimen geometry and fiictional effects at the platens. The latter can induce shear and tensile stresses within the specimen which cause it to fail by axial cleavage. This seems to be the case for most specimens where upon compression fracture was initiated near to the circumference of the specimen and ran axially along it. As Newton et al (1993) found little difference in the force-time profile or maximum force detected when fiictional effects were reduced at the platens, this would seem to suggest that the specimen dimensions were the more important factor which predisposed the specimen to this mechanism of failure. Hawkes and Mellor (1970) and Newman and Lachance (1974) stated that a length to diameter ratio of less than unity should not be employed because end restraint effects can occur throughout the specimen producing barrel-shaped specimens. However, this problem seems particularly difScult to avoid with pharmaceutical specimens because most of the compaction equipment available only just allows equi-dimensional specimens to be prepared. If longer specimens are to be made this would require much longer dies and very high compaction pressures to form

Table 4.6 Linear regression analyses for the diametral compression and compression tests

(at = b.P + bo) and (Oc = b.P + bo)

Strength test procedure b bo r

Diametral compression 0.0825 -0.5083 -0.998

Compression 1.0376 -4.4211 -0.996

s

I

Figure 4.15 In tensile ( • ) and In com pressive (A) strength as a function o f porosity for 12mm diameter A vicel PH 102 specim ens (raw data; n=10)

(Tensile Strength, MPa) (Compressive Strength, MPa)

Table 4.7 Linear regression analyses for the diametral compression and compression tests

(In at = -b.G + In b o ) and (In ac = -b.G + In b o )

Strength test procedure b In bo r strength at G 0 .3 (MPa)

Diametral compression -6.7374 3.1305 -0.999 3.03

Compression -6.5295 5.6581 -0.998 40.42

b; slope; bo; intercept; r: regression coefficient; strength at G0.3: tensile or compressive

coherent compacts. Such specimens would undoubtedly possess inhomogeneous physical structures which would make the data obtained through compression testing less meaningful.

It is also possible that the testing system used for compressive strength evaluation was not rigid enough and allowed some movement of the bottom platen during loading as this purely rested upon the load cell. This may have allowed non-uniform loading to occur where displacement boundary conditions were absent. However, as no problems have been encountered when using this test configuration for diametral compression and modified diametral compression tests this seems unlikely.

Table 4.5 reflects the change in the compressive to tensile strength ratio with compaction pressure. The ratio is essentially constant over the range of pressures employed (average ratio = 13.43) and any variation merely reflects experimental variation and the inherent variability associated with brittle specimen strengths. Examination of Figure 4.14 (which graphically represents the change in both the tensile and compressive strengths with lower punch compaction pressure) shows little variance in the data. The slopes of the regression lines are not parallel (Table 4.6) illustrating that the specimen compressive strength is more sensitive to changes in the lower punch compaction pressure than is the tensile strength. The change in both the natural logarithm of the Ot and Oc with specimen porosity is illustrated in Figure 4.15 which reflects their linear relationship. That the two regression curves are parallel reflects the constant nature of the compressive to tensile strength ratio with specimen porosity. Table 4.7 gives a tensile and compressive strength value for the specimens at a porosity of 0.30. These two values give a ratio of 13.34 which agrees closely with the ratios obtained at similar compaction pressures. These ratios seem particularly high and are approximately twice that of 6.70 reported for Avicel PHI 01 by Newton et al (1993). In the context of the study of Newton et at (1993) this indicates that these Avicel PH 102 specimens were predominantly brittle in terms of their mechanical properties. However, in their study specimens were prepared on an EKO eccentric press where the ‘contact time’ is much less than that for specimens made on an Instron Physical Testing Machine (Armstrong, 1989). Thus the specimens in this study will have undergone a considerably greater degree of plastic flow and deformation during their compaction resulting in stronger/harder tablets. This could account for the much higher Oc and corresponding compressive strength/tensile strength ratio obtained in this study. This

explanation seems particularly likely in view of the strain rate sensitivity of Avicel. Roberts and Rowe (1985) introduced the term ‘strain rate sensitivity index’ which assigned Avicel PHlOl and Avicel PHI 02 values of 50.6% and 48.3% respectively (Roberts and Rowe, 1986b), indicating that at faster compaction rates these materials showed increasing resistance to deformation, with an accompanying reduction in plastic flow and an increase in brittle behaviour. Alternatively, the values of compressive strength measured in this study may not have been calculated using force values from the same point in the force­ time curve as used by Newton et al (1993). Newton et al (1993) also used Avicel PHlOl with a smaller particle size which would possibly form weaker compacts because of reduced powder densification similar to that seen with lactose (Roberts and Rowe, 1986b). Larger volume specimens were also used in this study and the compressive strength/tensile strength ratios may have been measured at a different porosity value to that of Newton et al (1993). All of these differences between this study and that of Newton et al (1993) are factors which can be used to explain the disagreement between the compressive strength/tensile strength ratios.

From the above consideration of the problems associated with uniaxial compression testing it is evident that compression testing is of a far more complex nature than that of tensile testing (Gordon, 1976). In most cases the imposed stress state is unknown with shear and tensile stresses more than likely initiating failure of the specimen. It is also not entirely clear where failure is first initiated hence in these tests the compressive strength was derived from the maximum force (Wmax) attained before catastrophic failure although failure may well have occurred before this point. This therefore represents the maximum force which the specimen can sustain before failure. Thus the term ‘bearing capacity’ suggested by Darvell (1990) rather than compressive strength does seem to be appropriate in this case. However, this study has demonstrated that in a similar manner to the specimen tensile strength, the natural logarithm of the compressive strength shows a clear linear relationship with the specimen porosity. Such a trend indicates that whatever ‘compressive strength’ has been measured it has been done so in a consistent manner.