The effect of storage conditions on the physical stability of tablets

Published on: 

Pharmaceutical Technology Europe

Pharmaceutical Technology Europe, Pharmaceutical Technology Europe-01-01-2007, Volume 19, Issue 1

Water interacts with pharmaceutical solids at virtually all stages of manufacture, from synthesis of raw materials to the storage of the final dosage form. The interactions of water with powders is, therefore, a major factor in the formulation, processing and product performance of solid pharmaceutical dosage forms.

Water interacts with pharmaceutical solids at virtually all stages of manufacture, from synthesis of raw materials to the storage of the final dosage form. The interactions of water with powders is, therefore, a major factor in the formulation, processing and product performance of solid pharmaceutical dosage forms.

The amount of water associated with a solid at a particular relative humidity (RH) and temperature depends on its chemical affinity for the solid and the number of available sites of interaction, the surface area and the nature of material.1 The physicochemical properties of pharmaceutical solids (such as flow, compaction, hardness and dissolution rate) are dependent on the presence of moisture, and several reports have been made in the literature to this effect.2–11

Influence of storage conditions

Because of the importance of polymers in tablet and sustained-release matrices, the influence of storage conditions on the mechanical properties of tablets will be discussed separately. Compaction of powders into coherent specimens requires the formation of interparticulate attractions between the particles during the compression phase.

However, after ejection of the tablet from the die, the tablet strength can change.5,12–21 This means that interparticulate attractions are also formed or disturbed during the postcompaction storage phase. Three main mechanistic explanations for the formation of interparticulate attractions have been proposed:

  • Rearrangement of solid material at the particle surfaces within the tablet.5,13,14

  • Continuing deformation of particles in the compact after compaction.14–19

  • Crystallization of dissolved material between neighbouring particles because of the movement of water within the tablet.5,20

The first of these mechanisms is one of the main reasons for the postcompaction changes in the tensile strength of tablets. The rearrangement in the second mechanism can lead to either the formation of solid bridges between particles or to an increase in the bonding surface area of intermolecular attraction forces.

It seems that this restructuring of solid materials is mediated or facilitated by the presence of adsorbed water and that water can show this function — although the point has not been reached where the state and thickness of the adsorbed water vapour allows a true dissolution of the solid material.

For example, the first mechanism was proposed as an explanation for the postcompaction increase in tensile strength of tablets compacted from fine particulates (20–40 μm) of two water-soluble materials, saccharose and sodium chloride.5,10

There are several factors that can change the effect of storage conditions including RH on the tensile strength of tablets. These factors will be discussed in this article.

Solubility of powders and RH. For tablets of saccharose and sodium chloride, the tablet strength increased generally during the storage.11 Although saccharose and sodium chloride are both water-soluble, larger differences in tensile strength of saccharose tablets were observed in comparison with tablets of sodium chloride after storage for 1 week.

For water-insoluble excipients, such as tablets of calcium hydrogen phosphate dihydrate, the tablet strength did not vary with time or storage conditions.11 Changes in tensile strengths of sodium chloride and saccharose could be correlated to the changes in surface area of tablets.11

It has been shown that there is transport or movement of solid material within the tablet during the storage results in a rearrangement of the material such that new interparticulate attractions are formed. This seems to be facilitated by the uptake of water from the environment. It seems reasonable that the RH of the environment during storage of tablets affects both the rate and the amount of water uptake. Thus, both the rate and the total degree of movement of solid material and the subsequent formation of interparticulate attractions are related to the amount of sorbed water.

This was also verified by a study on the effect of precompression processing and storage conditions on compaction properties of sodium chloride, sucrose, calcium hydrogen phosphate dihydrate and acetylsalicylic acid, which showed that the tablet strength increased with the postcompaction storage time for both sodium chloride and sucrose (water-soluble materials) at intermediate and high RH. However, for water insoluble materials of calcium hydrogen phosphate and acetylsalicylic acid no change was observed (Table 1).21

Table 1 The effect of storage condition on the tablet strength of some pharmaceutical powders.*

Furthermore, for sodium chloride tablets it has been shown that the increase in compact strength occurred at 0, 33 and 57% RH, mainly within the first 4 h of storage.21 However, only a minor change, if any, in the postcompaction tablet strength was noticed when the tablets were stored at 0% RH (Table 1). Although there are various studies on the effect of RH on the mechanical properties of tablet during the storage,10–12,22 it is very difficult to predict the effect of RH on the hardness of tablets as it has a complex effect on the physical ageing of compressed tablets and the variation in hardness.19,22

Engineer et al. investigated the effect of temperature and humidity on the hardness of sustained release diphenhydramine HCl matrices.23 The obtained results are tabulated in Table 2, which shows that the plateau hardness is achieved much more quickly for the tablets stored at 40 °C/75% RH than for the tablets stored at 25 °C/60% RH, which could be attributed to a faster acquisition of equilibrium moisture content at high temperature and RH (e.g., 40 °C/75% RH).

Table 2 Hardness values of diphenhydramine HCl-kollidon tablets.*

Deliquescent point of salts. Most of the changes in the mechanical properties of tablets could be related to the deliquescent point of the salts. As the humidity increases (but remains below the critical deliquescent point), the hardness of the salts increases. At a particular humidity, hardness would probably change until the compact attains its equilibrium moisture content. Thereafter the hardness would remain constant.

Above the critical deliquescent point, the hardness of the salts decreases considerably.24 In the case of polymer tablets, the effect of moisture is not as straightforward as salts. For example, Reir and Shangraw reported adverse effects on the strength of Avicel tablets upon storage under elevated humidity. They attributed this to the absorption of water into the Avicel tablets causing bond disruption and loss in tablet strength.25

Moisture contents and compression. Initial moisture content of pharmaceutical powders is an important factor affecting physicomechanical properties of tablets.21,26–37

Some investigations suggest only a limited effect for the powder storage conditions before preparation of compacts on the postcompaction strength changes even for materials prone to crystallize readily when exposed to humid atmospheres.21 A possible explanation is that the particle surfaces might be activated during compression because of interparticulate friction. However, a major effect has been observed in some cases. For example, Nyqvist and Nicklasson compared the effects of inducing moisture into Avicel powder prior to compression with the storage of 'dry' Avicel tablets at elevated RHs.37 The former produced an increase in tablet strength while the latter treatment produced a marked decrease.

Partial moisture loss is the main factor increasing the hardness of tablets because of recrystallization of water-soluble excipients or a water-soluble drug in the void spaces on compression.

The extent of the increase in hardness depends on the specific combination of drug and excipient and their physical properties (such as aqueous solubility, crystallinity and hygroscopicity), as well as the tabletting method.30–33,38–40 Therefore, recrystallization could be avoided by careful selection of excipients. Tablets prepared by direct compression are more susceptible to changes caused by humidity than tablets prepared by wet granulation. The changes in tablet behaviour upon ageing can be correlated with the moisture content of granules in tablets made by wet granulation.

For tablets made by direct compression, it has been shown that those with high initial moisture content increase in hardness upon storage, with the magnitude of such increases being dependent upon the physical properties of the base and the absolute moisture content.30,31 Among the tablets studied, lactose-based tablets with different initial moisture contents were found to be the most resistant to changes upon storage.41

Effect of storage conditions on mechanical properties

The effects of RH on the mechanical properties of hydroxypropyl methyl cellulose (HPMC) polymers have been extensively studied.42–46 It has been shown that the tensile strength of tablets obtained from HPMC powders equilibrated at 0% RH and then stored at different RHs before testing showed an initial plateau at about 5% wt moisture content and then a decrease that should be related to weakening of the interparticle bonds because of the softening of the HPMC polymers.

But when HPMC powders compressed at the same compression force after equilibration at different RHs were stored at the same RH, the tensile strength of the tablets showed the plateau at 10% wt moisture content and then decreased.

The effect of moisture content on a polymer's compaction properties has been reviewed briefly.43,44 The presence of moisture in varying quantities could both increase or reduce the mechanical strengths depending on the formulation content.43–50 Nokhodchi et al. explained the increase/decrease in tensile strength with moisture according to the type of moisture associated with polymers (Table 3).45

Table 3 The effect of the type of moisture on the tensile strength and elastic recovery of HPMC K4M tablets stored at various relative humidities.*

An increase in RH from 23% to 75% caused 7.5-,4.8-,and 2.3-fold increases in internally absorbed water, externally adsorbed water, and monolayer-adsorbed water, respectively. Similarly, the tensile strengths of HPMC K4M tablets concomitantly increased.

The distribution of moisture in a material; the range and magnitude of the van der Waals' forces between the particles; and the development of additional bonds by plastic deformation and/or melting of powder particles should control the tensile strength of the tablets. Water molecules initially adsorbed on the surfaces may form a monomolecular layer and increase the van der Waals forces, thereby smoothing out the surface microirregularities and reducing interparticle separation.15 This monolayer-bound water can be regarded as part of the particles' surface molecular structure.51,52


These effects would increase the tensile strength of HPMC K4M tablets with increasing moisture content. As more water molecules adhere to the surface, moisture may transfer into the material.9,53,54 This effect may soften the particles' surfaces; under high pressure, the area of contact between the particles will increase with plastic deformation and more solid bonds may form.51 These effects could account for the tablets' increased tensile strength.

It has been suggested that water adsorption reduced tablet tensile strength because of condensation and multilayer adsorption.45 The extent of multilayer adsorption in Table 3 can be estimated by subtracting the monolayer adsorption from the externally adsorbed moisture values. Although the amount of multilayer adsorption increased, tensile strength did not decrease. Therefore, the effect of moisture on tablet tensile strength is the result of the balance between the amount of monolayer-adsorbed moisture, internally absorbed moisture and externally adsorbed moisture.

Maltodextrines have been extensively studied as excipients.46 They display a completely different type of profile compared with emcompress or Fast-Flo lactose, with significant decreases in crushing strength occurring from the time immediately after ejection until 1 day after ejection. Emcompress and Fast-Flo lactose are primarily brittle materials so the tablet porosity would not be expected to change much over time, whereas the plastically-deforming viscoelastic expands after ejection from the die.

The maltodextrins had an initial rapid increase in tablet porosity, from the time immediately after ejection until 1 day after ejection, after which only small changes in porosity were seen. This initial change was probably a result of viscoelastic expansion of the material and can explain the large loss in crushing force that was seen.

The porosity changes observed after 1day were slight and not sufficient to explain how the consistent maltodextrin tablet-crushing force decreases with time. Bonds weakened within the tablet without concurrent changes in tablet size or other gross changes, such as increased tablet porosity.

The influence of storage conditions on the disintegration time and dissolution rate

Following the administration of a drug through a solid dosage form, a sequence of steps is required before the drug reaches the systemic circulation. An orally administered solid dosage form undergoes disintegration and deaggregation, followed by the dissolution of the drug. The dissolved drug molecules must penetrate the gastrointestinal membrane and be picked up by the blood. Each of the steps involved may limit how fast the drug molecules reach the general circulation. Therefore, it is important to review the effect of storage conditions on the disintegration time of tablets and dissolution rate of drugs.

Disintegration time. Several studies have been reported on the changes in behaviour of disintegration of tablets upon storage.31–33,35,36,39,40,55–59

Such changes were found to be responsible for some bioavailability problems.60,61 The water sorption and its effect on disintegration of tablets made of various excipients have shown that the physical properties of the tablets were influenced by water uptake or loss during storage. At low RHs, the disintegration times increased and at high RHs decreases were observed.36

It has been shown that tablets prepared by direct compression were more susceptible to changes caused by humidity than tablets prepared by wet-granulation.38 There are many studies that examine the effect of ageing on tablet disintegration. Large differences are apparent among different formulations and tabletting methods.31–33,35 39,40

The changes in tablet behaviour upon ageing were correlated with the moisture content of granules in tablets made by wet granulation. Tablets made from granules with high initial moisture content(4%), on storage, showed increases in hardness induced by partial loss of moisture.30

It has been shown that tablets made from granules with low-moisture content (2%) gained hardness during storage as a consequence of the partial loss of moisture. Such increases in hardness, which were not related to initial tablet hardness, showed minimum effect on tablet disintegration.32,33

Molokhia et al. investigated the ageing of tablets prepared by direct compression of bases of different moisture contents.41 The storage-induced changes in disintegration were evaluated for tablets made by direct compression of three different bases with different initial moisture contents.

The increases in hardness may increase the disintegration time. The tablets with low initial moisture content were minimally affected by storage.41 The gain of moisture by some of these tablets led to reduction of the disintegration times. Among the tablets studied, lactose-based tablets with different initial moisture content were found to be the most resistant to changes upon storage.

Dissolution rate. In the development of tablets, it is generally recognized that ageing may cause chemical and physical changes that may modify the dissolution of drugs from compressed tablets. These changes in dissolution may alter the bioavailability of the active drug substance.

Although tablet disintegration is often a necessary precursor for drug dissolution, it does not ensure that the drug substance will dissolve, and hence have the potential to be systemically absorbed. Therefore, it is crucial to discuss the effect of storage conditions (temperature, RH or moisture content) on the dissolution rate of drugs from tablets.

It has been reported that the hardness of tablets was increased because of partial moisture loss in compressed tablets. No significant decreases in dissolution occurred as a result of the hardness increase induced by the moisture content. However, Chowhan and Palagyi showed that in the absence of any moisture-induced phenomenon, when moisture contents of the granulation were low, hardness produced by the use of higher compression induced a considerable decrease in dissolution.30

In other studies, they showed that tablets made from granules with low moisture content gained hardness during storage and this increase in hardness had no appreciable effect on the dissolution rate of the drug.30–32

It has also been shown that the exposure of sustained-release diphenhydramin tablets to moisture could affect the release rate of drug from kollidon matrices.24 The summary of the results is listed in Table 4. This table shows that pronounced changes in both the rate and extent of drug release have been found, even as early as 1 h following exposure at 40 °C/75% RH. Dissolution rate shows that the rate of release appeared to trend down even after a 4-week exposure time (see k values in Table 4). The researchers have tried to fit the dissolution data to the following equation to see the effect of moisture on the mechanism of release. Q=ktn where Q is the percentage of drug release at time t; k is the release rate constant; and n is the release exponent. An n value of 0.5 or 1 indicates diffusion or erosion-controlled mechanism, respectively.

Table 4 Kinetic parameters based on equation Q5ktn.

It is clear from Table 4 that the exposure at 40 °C/75% RH did not affect the n values very much, but did affect the T50% (the time required for 50% of drug release).

Extensive research has been performed on the effect of moisture content on the release rate of theophylline from amylodextrin tablets by Steendam et al.7,62 They found that small variations in moisture content resulted in large changes of the release rate. Release of theophylline slowed down when the moisture content increased. Tablets with moisture content of 4.9% and 9.6% disintegrated rapidly after immersion in the aqueous buffer, resulting in fast release of the drug.

A unique relationship between porosity and release rate, which was independent of moisture content and compaction pressure, was observed. Above a critical porosity of 7.5%, crack formation was followed by disintegration and fast release.

Below this critical porosity, tablets stayed intact despite the formation of cracks, and sustained release was observed. To describe more clearly the effect of moisture content on the drug release, T50% could be plotted against moisture content. Below 7.5%, tablets disintegrated rapidly, resulting in fast dissolution of the drug. For higher moisture content, T50% increased steeply as a result of decreasing porosities of the tablets and the absence of disintegration.

There was a significant reduction in T50% when moisture was 23.4%. This can be explained by the rubbery state of amylodextrin with this moisture content (glass transition temperature of about -8.50°C).

During the decompression stage of the compaction cycle, extensive elastic relaxation of the rubbery polymer will occur, leading to the formation of relatively porous tablets.

Furthermore, the bonding properties of these tablets are relatively low because of a low elastic modulus and multilayer water adsorption. The latter weakens interparticle bonding in the tablet because water now acts as a kind of lubricant. Moreover, these factors lower the tablet strength and facilitate the occurrence of erosion, resulting in faster release.


It can be concluded that control over moisture content is essential for the production of amylodextrin tablets with reproducible release characteristics. In the case of swellable hydrophilic polymers (such as HPMC) it has been shown that variation in HPMC moisture content over the range 2.25–10.85% had no significant effects on release profile of hydrochlorothiazide, despite the fact that HPMC moisture content correlated with tensile strength of tablets.8,45,50 The different behaviours in the effect of moisture content on the release rate of drugs from amylodextrin or HPMC is because of their behaviour against moisture. It has been shown that release from swellable systems such as HPMC is hardly affected by the initial tablet porosity.63,66

Figure 1 Schematic representation of drug release from a leaching-based drug delivery system.

The release of drug is governed by diffusion through the swollen gel-layer instead of diffusion through a porous network.67 As amylodextrin tablets do not form a gel layer, which acts as a barrier to control diffusion of the drug out of the tablet. Instead, release will occur through leaching mechanism (see Figures 1 and 2 for leaching and diffusion mechanisms, respectively).

Figure 2 Schematic representation of drug release from a swelling-based drug delivery system.

Ali Nokhodchi is senior lecturer in pharmaceutical technology at the Medway School of Pharmacy, Universities of Kent and Greenwich (UK).

Yousef Javadzadeh is a a lecturer in pharmaceutics at the Drug Applied Research Centre and School of Pharmacy, Tabriz University of Medical Sciences (Tabriz, Iran).


1. S. Dawoodbhai and C.T. Rhodes, Drug Dev. Ind. Pharm., 15(10), 1577–1600 (1989).

2. A.J. Shukla and J.C. Price, Drug Dev. Ind. Pharm., 17(15), 2067–2081 (1991).

3. G.S. Pande and R.F. Shangraw, Int. J. Pharm., 124(2), 231–239 (1995).

4. F. Khan and N. Pilpel, Powder Technol.,48(1),145–150 (1986).

5. A.R. Fassihi, J. Pharm. Pharmacol., 40 (supplement), 76P (1988).

6. C. Ahlneck and G. Alderborn, Int. J. Pharm., 56(1), 143–150 (1989).

7. R. Steendam et al., Int. J. Pharm., 204(1), 23–33 (2000).

8. M.J. Mosqueras, Int. J. Pharm., 13(2), 147–149 (1996).

9. A. Nokhodchi and M.H. Rubinstein, Pharma Sci., 8(6), 349–356 (1998).

10 A. Nokhodchi and M.H. Rubinstein, Pharma Sci., 11(3), 195–202 (2001).

11. G. Alderborn and C. Ahlneck, Int. J. Pharm., 73(2), 249–258 (1991).

12. J.E. Rees and E. Shotton, J. Pharm. Sci., 60(11), 1704–1708 (1971).

13. G.R.B. Down and J.N. McMullen, Powder Technol., 42(2), 169–174 (1985).

14. C. Nystrom and P.G. Karehil, Powder Technol., 47(3), 201–209 (1986).

15. C. Ahlneck and G. Alderborn, Int. J. Pharm., 5(2), 131–141 (1989).

16. E. Shotton and J.E. Rees, J. Pharm. Pharmacol., 18 (supplement), 160–167S (1966).

17. P.J. Rue and P.M.R. Barkworth, Int. J. Pharm. Technol. Prod. Manuf., 1 (May), 2–3 (1980).

18. N.A. ElGindy and M.W. Samaha, Int. J. Pharm., 13(1), 35–46 (1982).

19. P.G. Karehi and C. Nystrom, Int. J. Pharm., 64(1), 27–34 (1990).

20. J.E. Rees and J.A. Hersey, Pharm. Acta Helv., 47(7), 235–243 (1972)

21. A.A. Elamin, G. Alderborn and C. Ahlneck , Int. J. Pharm., 108(2), 213–224 (1994).

22. R.P. Bhtia and N.G. Lord, J. Pharm. Sci., 68(7), 896–899 (1979).

23 S. Engineer, Z.J. Shao and N.A Khagani, Drug Dev. Ind. Pharm., 30(10), 1089–1094 (2004).

24. N. Lord and P. Shiroman, Drug Dev. Ind. Pharm., 10(5), 729–752 (1984).

25. G.E. Reier and R.F. Shangraw, J. Pharm. Sci., 55(5), 510–514 (1966).

26. W.A. Strickland, J. Amer. Pharm. Ass. Sci.Ed., 45(1), 51–55 (1956).

27. E. Shatton and N. Harb, J. Pharm. Pharmacol., 17(5), 504–508 (1965).

28. M.W. Scott, H.A. Liberman and F.S. Chows, J. Pharm. Sci., 52(9), 994–998 (1963).

29. S. Esezobo and N. Pilpel, J. Pharm. Pharmacol., 26 (Supplement), 47P–56P (1974).

30. Z.T. Chowhan and L. Palagyi, J. Pharm. Sci., 67(10), 1385–1389 (1978).

31. Z.T. Chowhan, Drug Dev. Ind. Pharm., 5(1), 41–62 (1979).

32. Z.T. Chowhan, J. Pharm. Sci., 6(1), 1–4 (1980).

33. Z.T. Chowhan, J. Pharm. Pharmacol., 32(1), 10–14 (1980).

34. J.M. Lausier, J. Pharm. Sci., 66(11), 1636–1937 (1977).

35. S.T. Horhota et al., J. Pharm. Sci., 65(12), 1746–1749 (1976).

36. H. Nyqvist, M. Nicklasson and P. Lundgren, Acta Pharm. Helv., 18(5), 305–314 (1981).

37. H. Nyqvist and M. Nicklasson, Int. J. Pharm. Prod. Mfr., 4(1), 67–73 (1983).

38. K. Uzunarsalan and J. Akbuga, Pharmazi, 46(Apr), 273–275 (1991).

39. A.M. Molokhia, Int. J. Pharm., 22(1), 127–130 (1984).

40. S.A. Sangekar, M. Sarli and P.R. Sheth, J. Pharm. Sci., 61(6), 939–944 (1972).

41. A.M. Molokhia and H.I. AlSho, Drug Dev. Ind. Pharm., 13(9–11), 1933–1946 (1987).

42. S. Malamataris and T. Karidas, Int. J. Pharm., 104(1), 115–123 (1994).

43. A. Nokhodchi and M.H. Rubinstein, Pharm. Sci.,11(3), 195–202 (2001).

44. A. Rajabi-Siahboomi, A. Nokhodchi and M.H. Rubinstein, Pharm. Sci., 10(Oct), 42–48 (1998).

45. A. Nokhodchi, J.L. Ford and M.H. Rubinstein, J. Pharm. Sci., 86(5), 608–615 (1997).

46. M.J. Mollan and M. Celik, Int. J. Pharm., 114(1), 23–32 (1995).

47. S. Malamataris, K. Tsiri and P. Goidas, Sorption of moisture in different size fractions of some direct compression excipients and tensile strength of corresponding tablets, 6th Int. Conf. Pharm. Technol. Paris, V, 195–204 (1992).

48. A. Nokhodchi, M.H. Rubinstein and H. Larhrib, Int. J. Pharm., 118(2), 191–197 (1995).

49. S. Malamataris, P. Goidas and A. Dimitriou, Int. J. Pharm., 68(5), 51–60 (1991).

50. A. Nokhodchi et al., J. Pharm. Pharmacol., 48(11), 1116–1121 (1996).

51. C. Ahlneck and G. Alderborn, Int. J. Pharm., 54(2), 131–141 (1989).

52. G. Zografi and M.J. Kontny, Pharm. Res., 3(4), 187–194 (1986).

53. J.H. Young and G.H. Nelson, Trans. Am. Soc. Agric. Eng., 10(5), 260–263 (1967).

54. J.H. Young and G.H. Nelson, Trans. Am. Soc. Agric. Eng., 10(5), 756–761 (1967).

55. J.F. Bavitz, N.R. Bohida and F.A. Restainos, Drug Dev. Commun., 1 331–347 (1974–1975).

56. A.M. Guyot-Hermann, D. Leblan and M. Draguet-Brughmans, Drug Dev. Ind. Pharm., (11) 551–564 (1985).

57. H. Muti and S. Othman, Drug Dev. Ind. Pharm., 15 2017–2035 (1989).

58. P. Sheen and S. Kim, Drug Dev. Ind. Pharm., 15, 401–414 (1989).

59. M. Sugimoto et al., Int. J. Pharm., 296 (1–2), 64–72 (2005).

60. M.W. Gouda, M.A. Moustaf and A.M. Molokhia, Int. J. Pharm., 5, 345–347 (1980).

61. P. York, Pharmazie, 32, 101–104 (1977).

62. R. Steendam, H.W. Frijlink and C.F. Lerk. Eur. J. Pharm. Sci., 14, 245–254 (2001).

63. P. Gao, P.R. Nixon and J.W. Skoug, Pharm. Res.,12, 965–971 (1995).

64. H.B. Hopfenberg and K.C. Hsu, Pol. Eng. Sci., 18, 1186–1191 (1996).

65. G.W. Raymond Davidson and N.A. Peppas., J. Control. Rel., 3, 259–271 (1986).

66. N. Visavarungroj, J. Herman and J.P. Remon, Drug Dev. Ind. Pharm., 16, 1091–1108 (1990).

67. S.S. Davis, J. Control. Rel., 2, 27–38 (1985).