Abstract
Among the factors that affect the glass surface chemical durability, pH and complexing agents present in aqueous solution have the main role. Glass surface attack can be also related to the delamination issue causing glass particles' appearance in the pharmaceutical preparation. A few methods to check for glass containers delamination propensity and some control guidelines have been proposed. The present study emphasizes the possible synergy between a few complexing agents with pH on borosilicate glass chemical durability.
Hydrolytic attack was performed in small-volume 23 mL type I glass containers autoclaved according to the European Pharmacopoeia or United States Pharmacopeia for 1 h at 121 °C, in order to enhance the chemical attack due to time, temperature, and the unfavorable surface/volume ratio. Solutions of 0.048 M or 0.024 M (M/L) of the acids citric, glutaric, acetic, EDTA (ethylenediaminetetraacetic acid), together with sodium phosphate with water for comparison, were used for the trials. The pH was adjusted ±0.05 units at fixed values 5.5, 6.6, 7, 7.4, 8, and 9 by LiOH diluted solution.
Because silicon is the main glass network former, silicon release into the attack solutions was chosen as the main index of the glass surface attack and analysed by inductively coupled plasma atomic emission spectrophotometry. The work was completed by the analysis of the silicon release in the worst attack conditions of molded glass, soda lime type II glass, and tubing borosilicate glass vials to compare different glass compositions and forming technologies. Surface analysis by scanning electron microscopy was finally performed to check for the surface status after the worst chemical attack condition by citric acid.
LAY ABSTRACT: Glass, like every packaging material, can have some usage limits, mainly in basic pH solutions. The issue of glass surface degradation particles that appear in vials (delamination) has forced a number of drug product recalls in recent years. To prevent such situations, pharmaceutical and biopharmaceutical manufacturers need to understand the reasons for accelerate surface glass corrosion mainly in the case of injectables.
Some drugs can contain active components with known ability to corrode glass silica networks. Sometimes these ingredients are dissolved in an alkaline medium that dramatically increases the glass corrosion and potentially causes the issue. As this action is strongly affected by time and temperature, flaking may become visible only after a long storage time. The purpose of this investigation is to verify the borosilicate glass chemical durability during controlled conditions of time and temperature when in contact with testing solutions containing different complexing agents by varying the pH. Si concentration in the extract solution is taken as an index of glass dissolution during constant autoclaving conditions for 1 h at 121 °C, which simulates approximately five years of contact at room temperature.
Acetate, citrate, ethylenediaminetetraacetic acid (EDTA), phosphate, and glutarate 0.048 M or 0.024 M solutions were used at increasing pH from 5.5 to 9.0. The chemical durability of two borosilicate tubing glass vials of different glass compositions were compared with the molded one in the worst attack conditions by citric acid. Even if no delamination issue has been experienced by this study in type I molded and tubing containers, the conclusions developed can provide pharmaceutical manufacturers with useful information to prevent glass delamination risk in their processes.
- Glass chemical durability
- Hydrolytic resistance
- Complexing agents
- Borosilicate type I glass
- Soda lime type II glass
- Silicon release
- pH
Introduction
Among the factors that affect the glass surface chemical durability, pH and complexing agents present in aqueous solution have the main role (1). Glass surface attack can be also related to the delamination issue causing glass particles' appearance in the pharmaceutical preparation. A few methods to check for glass containers delamination propensity and some control guidelines have been proposed (2, 3). The present study emphasizes the possible synergy between a few complexing agents with pH on borosilicate glass chemical durability.
Borosilicate type I glass is generally considered the reference packaging material for injectables due to its very high chemical durability, transparency, gas barrier, thermal resistance, robustness, etc. Like all the packaging materials, glass can show limits under some harsh usage conditions, so it is of extreme importance to have a knowledge of the factors that can affect the glass surface chemical durability in order to better address the stability trials of a pharmaceutical preparation, as suggested by the United States Pharmacopeia (USP) <1660> (3).
The appearance in solution of visible shiny needle-shaped particles (named lamellae or flakes) after some conditioning time is a typical macroscopic phenomenon of a strong chemical attack on the glass surface, triggered by time and temperature. Before being faced with a visible or subvisible delamination phenomenon it is possible to foresee negative results by a periodic check of silicon release trend from the glass. But, as also found in the present work, lamellae formation could not be related to a high silicon release because it could be an independent phenomenon (4). In a borosilicate glass also boron could be considered as a glass chemical durability index but it is less favorable because of its lower concentration, about 1/6–1/7 that of silicon. The other oxides present in the glass composition are approximately released in amounts similar to their ratio with silica. Only the silica/Na2O ratio is lower due to the higher sodium concentration in the glass surface layer (5) and its propensity to be released more easily from the glass network.
Flakes are not the simple result of a strong chemical attack on the glass surface but usually depend on the surface glass composition that could have been locally enriched in silica after the evaporation of borates following a forming-melting process of the container or after a surface silica treatment. In the lamellae forming process both mechanisms of chemical hydration of the silica layer and its physical detachment by mechanical stress, due to the different composition of the underneath glass, are involved (6). The physical relationship to flakes removal can be ascribed to the different expansion coefficients of the silica-rich hydrated layer in comparison with the glass bulk underneath. In addition to or alternatively to the mentioned mechanism of lamellae (flakes) forming, it is also quite probable that harsh attack conditions can promote an important silicon release without any delamination, that is, a considerable glass surface corrosion can happen without flakes development.
USP <1660> (3) provides both information about factors that affect the glass durability and recommended approaches to evaluate the potential propensity for glass particles production or delamination of the inner surface. To define the influence on the glass durability of some of the organic compounds mentioned in Table II of USP <1660>, the present study shows the trend of the hydrolytic attack on a molded borosilicate glass by a few complexing agents as a function of the pH. The influence of different glass chemical durabilities was also considered by a comparison between glasses with different compositions: molded and tubing borosilicate and soda lime type II glass vials. The most harsh attack condition by citrate at pH 8 was chosen for this comparison.
It is known that some complexing agents promote the rate of glass surface attack by complexation of some glass components (7, 8), but the correlation with the pH for the actual compositions of borosilicate glass has not yet been fully investigated, considering that pH is one of the most important factors affecting the glass durability due to the slow dissolution of the silica network in an alkaline environment (1, 9).
Table I shows a simplified summary of glass surface chemical reaction mechanisms by aqueous solutions according to the pH, and Table II shows a list of the main factors affecting the same.
As is well known, time and temperature are among the most important factors affecting the glass chemical durability, therefore the autoclaving according to USP/EP is an advisable choice to perform a little time-consuming trial with constant and reproducible harsh attack conditions, provided that the organic compounds to test are thermally stable. Autoclaving at 121 °C for 1 h corresponds approximately to five years of contact glass-aqueous solution at room temperature (8, 10). In the present work only the factors 1 and 2 of Table II have been considered.
Materials and Methods
An autoclave Asal Vapormatic 770, Programmable Logic Controller (PLC) controlled, was used for the autoclaving trials, according to the procedure described in the USP <660> and European Pharmacopoeia (EP) 3.2.1 Pharmacopoeias (1 h at 121 °C). Attack solutions of the acids citric, glutaric, acetic, and ethylenediaminetetraacetic (EDTA), pure for analysis (Carlo Erba RPE Reagents), 0,048 M or 0.024 M (M/L), were prepared for the trials. In place of phosphoric acid, trisodiumphosphate was used for practical reasons. Citric acid 0.048 M corresponds to 1% (w/v) so the molarity was maintained constant also with the other reagents for adequate comparison of the results.
Deionized water of conductivity lower than 0.2 μS was used from a RD 60 Elettracqua deionizer. The pH of each solution (pHmeter WTW L2 with Hamilton electrodes) was adjusted ± 0.05 units at fixed values 5.6, 6.6, 7, 7.4, 8, and 9 by the addition of a LiOH (Carlo Erba RPE Reagents) diluted solution. Lithium was chosen instead of other alkali hydroxides only for practical reasons to prevent any background noise and memory effects in the inductively coupled plasma (ICP) torch after the analysis.
A preliminary comparison work of the chemical attack on glass demonstrated a substantial equivalence between LiOH and NaOH in the considered pH range of the trials. Small-capacity 23 mL molded type I borosilicate glass containers were chosen to enhance the glass surface attack due to the high surface/volume ratio. To compare the rate of chemical attack on glasses with different glass compositions, borosilicate tubing vials of brimful capacity 28 mL (sample A), borosilicate tubing vials at 10 mL (samples B1 and B2), and soda lime type II 24 mL (sample SL) were used. In particular, the 10 mL tubing vial B2 was formed at higher temperature than B1.
Table III shows the glass chemical composition of the tested vials and the HCl titration values after the hydrolytic resistance test with water at 121 °C for 1 h.
Silicon analysis was performed in the attack solution by ICP Thermo Fisher Dual 6300 just after the autoclaving. Each trial was replicated at least four times to increase accuracy.
A scanning electron microscope Jeol JSM-5900 equipped with an energy dispersive X-ray spectrometry (EDS) system Oxford Link ISIS300 was used for the inner surfaces examination.
Results and Discussion
Autoclaving trials are plotted in Figure 1 and the silicon-extracted results are detailed in Table IV. Considering the attack by water as the comparison test, it is interesting to note that at pH lower than 6 most of the complexing agents used show even a lower extracting power than water. Only citric acid can be compared with water. The overlapping between the curves of acetic acid and water at pH higher than 7 should mean that acetic acid has substantially no complexing properties in the testing conditions. Acetate complex formation with Pb and Cd is important for the crystal glass or glasses containing considerable percentages of them (7).
At pH between 6.0 and 6.5, citric acid 0.048 M begins to show some complexing propensity depending on its concentration considering that citric acid 0.024 M is still next to pure water. Up to pH 7, EDTA and glutaric acid do not show any relevant complexing activity as the silicon release is quite similar to that extracted by water. Excluding acetic acid, the silicon release becomes more and more enhanced at pH 7.4. Citric acid 0.024 M shows an increasing extraction rate with the pH, reaching at pH 9 about one half of the silicon extracted by the more concentrated 0.048 M. In particular, citric acid 0.048 M shows already at pH 7.4 a limit extraction rate for silicon. Glutaric acid 0.048 M shows about one-fourth of the citric acid 0.048 M extraction rate at pH 9.
EDTA 0.024 M shows a completely different behavior due to an almost exponential curve reaching the same silicon extraction value of citric acid 0.048 M at pH 9. At pH higher than 9 (not investigated) it is foreseeable to find also for EDTA the same S-shaped curve as citric acid 0.048 M and the reaching of a higher limit extraction rate for silicon. The exponential curve in the pH range investigated together with the lower concentration of 0.024 M suggests that EDTA is the most aggressive at pH higher than 9.
The sodium phosphate 0.048 M curve shows a shoulder at pH 7, reaching the same extraction power of 0.024 M citric acid at pH 9. Probably the same consideration done for EDTA at pH higher than 9 can be applied here too.
It was not the aim of the present work to study the reaction mechanism of the considered complexing agents, but the behavior difference in function of the pH lets us suggest a few hypotheses. The complexing agent intervention as reaction rate accelerators can be roughly explained considering the OH– ion mechanism of reaction roughly described in Table I: during the hydrolysis of the bridging oxygen between two silicon atoms (Si-O-Si bonds) the other elements present in the glass network (Al, Ca, Na, etc.) became available for complexation leaving a gap in the silica network where the OH– ion can find other Si-O-Si bonds available for further hydrolytic attack.
At low pH, complexing agents are not yet or are poorly dissociated so their chelating power for structural reasons is negligible and they can even work as glass surface protectors as confirmed by the lower Si release found in comparison with water (Table IV). The chelating power becomes more and more effective due to both the acid dissociation by increasing the pH according to the pKa (Table V) and to the molecular structure. Chelates formed by couples of anions are also possible. Acid concentration was also experienced to be effective as demonstrated by the comparison of 0.048 M citric acid that reaches a limit extraction rate for silica and one-half 0.024 M concentration that shows a significant lower but increasing extraction rate in the same pH range.
According to the low pKa1 and pKa2 (Table V), even considering the higher dissociation rate at pH 7 and 121 °C with the loss of two protons, EDTA still shows a low chelating effectiveness. Data suggest that the loss of the EDTA third proton is necessary to acquire a good chelating effectiveness. The shoulder experienced at pH 7 by sodium phosphate could probably be explained considering that phosphates develop their complexing activity by couples or aggregates of ions and that high temperatures shift toward the weak acid dissociation. For example, at 25 °C, pH 7, about 70% of phosphate is present as H2PO4– ion and the rest is in the form HPO42– ion. So the complexing effectiveness could be related to the ratio between the type of anions in equilibrium at pH 7 and 121 °C during the autoclaving (11).
Now a query arises: What is the behavior of vials with different glass compositions and/or different tube converting temperatures? This topic was approached only in a preliminary way by comparing in the worst attack conditions by citric acid 0.048 M at different pH values the Si release of three borosilicate tubing vials in addition to the molded vial and a soda lime type II vial. Because the glass surface release is also a function of the surface/volume ratio, mainly considering the smallest 10 mL tubing vial, the results were mathematically corrected to 23 mL volume (that is 21.7 mL, 90% of the brimful capacity) by the following exponential equation: Equation 1 was obtained considering the exponential relationship between the average capacity ranges with the glass titration limits, from the EP 1.2.3 or USP <660> surface hydrolytic resistance tables. Equation 1 was found to give a good approximation whenever the chemical durability of different capacities vials have to be compared or extrapolated.
Corrected data of silicon release according to eq 1 are shown in Figure 3 and in Table VI.
The plot of Figure 3 shows that all the glasses give more or less the same Si release at pH 5.5, excluding sample B2, which is significantly higher. Si release of sample B2 approaches and overlaps the other tubing glasses at pH higher than approx. 7.4. Considering that both samples B1 and B2 are made by the same glass, the difference at pH lower than approximately 7.4 should be ascribed to the higher converting temperature of B2 with consequent higher borates evaporation and condensation on the side wall cold surface, as also confirmed by the HCl titration values of Table III.
As soon as the attack by alkaline pH prevails, B1 and B2 Si release equalizes. So tube forming temperature seems to affect the silica availability for exchange and release mainly up to approximately neutral pH. The slightly better behavior of the sample soda lime at pH lower than 7 is in agreement with its alkali-depleted surface as confirmed by its lowest titration HCl value. But as soon as the pH exceeds the neutrality the silicon release increases so much that it became the lowest durable glass vial due to the lack of boron in the glass network even if the difference with the molded borosilicate vial is small. In fact, soda lime type II glass is not for use with alkaline pH solutions because as soon as the alkali-depleted thin layer is corroded the underneath soda lime glass increases more and more its release of components. The low level of silicon release experienced for the sample soda lime should mean that the alkali-depleted layer was not fully destroyed during the autoclaving.
In the case of the tubing vial samples A, B1, and B2, the glass composition looks likely to improve the chemical durability mainly at alkaline pH, while at acid pH the tube-converting temperatures seem to have a negative role, and is probably related to the less chemically durable area involved in the phase separations. Sample A shows the lowest silicon release, that is, the highest chemical durability at pH higher than 7, which is in agreement with the lowest HCl titration values and by its glass composition with the least alkali content. Samples B1 and B2, which are less durable than sample A on the basis of their glass composition, show clearly how the chemical durability can be affected by the converting temperatures at acid pH up to neutral; in fact the higher silicon release at acid pH of sample B2 converted at higher temperature and approaches samples B1 and A at alkaline pH as shown in Figure 3.
On the basis of the glass composition, that is, the higher alkali content among the borosilicate glasses tested, the molded vial should be the least chemically durable, but due to the different forming technology it shows the same silicon release as the most durable borosilicate sample A at acid pH. Data show that at pH ≥7 the higher the glass chemical durability, the lower is the silicon release trend, even if the difference is restricted due probably to the reach of the limiting extraction rate of the glass components.
Visual inspection by halogen lamp on a black background of every solution did not reveal the presence of floating flakes, that is, no floating particle of estimated dimension equal to or higher than 50 μm was seen as a result of chemical attack. As a final step of the present investigation, scanning electron microscopy (SEM) examinations of the inner glass surface were performed for the vials tested with citrate 0.048 M at pH 8. Figure 4 shows the examined points of the inner surface. Before and after the worst attack case by citrate 0.048 M at pH 8 molded borosilicate vials are shown in Figures 5 and 6: homogeneous surfaces from shoulder to bottom were found, that is, no delamination or pitting was found on the inner surfaces in contact with the attack solution. A very fine roughness development (Figure 6) is compatible with the harsh conditions of hydrolytic attack. Soda lime type II surfaces in contact with the citrate solution were similar to the molded borosilicate ones.
On the contrary, SEM examination of the most-durable borosilicate tubing (vial A) showed a remarkable pitting in the lower area of the side wall near the bottom (Figure 9). This is related to the known phenomenon of phase separation of silicon-rich phase and sodium borate in the form of droplets (halos in Figure 7) and borates condensation on the “cold” side wall surface during the converting of the tubing vial bottom (13, 14). The chemically less-durable phase separation area is also the origin of the surface pitting after strong hydrolytic attack as shown in Figures 7 and 8. The remaining surfaces are quite similar to the molded vial surfaces. In the head-space area (shoulder and neck) of all the vials tested not in contact with the citrate solution but only with steam, no roughness development was observed as expected.
Conclusions
Chemical durability of 23 mL type I borosilicate glass molded vials has been tested by autoclaving for 1 h at 121 °C according to EP, with a few complexing agent solutions at different pH. The purpose of autoclaving was to simulate a long-term contact of approximately five years of storage at room temperature. Hydrolytic attack was also enhanced by the use of small-volume vials with an unfavorable surface/volume ratio. Sodium phosphate solutions of 0.048 M (M/L) were used, as well as citric, glutaric, and acetic acids. Trials with 0.024 M solutions of the citric acid and EDTA were also performed. Glass chemical durability was tested by analysing the silicon release. The results are summarized as follows:
pH <6: substantially no difference between the extracting power of water and citric acid; the other tested solutions show significantly lower extracting power than water;
pH ∼6.6: increases of silicon extraction rate only by citrate and phosphate;
pH ∼7: while citrate and phosphate increase the silicon extraction rate, EDTA and glutarate approximately equalize water;
pH ≥7.4: substantially no difference between water and acetate solutions, that is, acetate ion does not show complexing activity; citrate 0.048 M reaches a silicon extraction rate limit, while the other complexing agents become more and more aggressive on glass surfaces up to the pH range limit tested of 9.
The results can be interpreted on the basis of both molecular structure and deprotonation pKa equilibria that affect the chelating effectiveness of the tested complexing agents and the scaling up of the hydrolysis of the bridging Si-O-Si bonds by OH− ions. Complexing agent concentration was also experienced to be effective. The comparison between glasses with different compositions—borosilicate molded, borosilicate tubing, and soda lime type II vials—in the worst case of surface attack by citrate, show that the silicon release at pH ≥7 is a little lower the higher the glass chemical durability, while at acid pH the tube converting temperature can increase the silicon release regardless of the glass chemical durability. In the case of the molded vial, homogeneous inner surfaces from shoulder to bottom were evidenced by SEM images after the worst attack condition, that is, no delamination or pitting has been evidenced. On the contrary, sample A showed an evident pitting on the side wall zone of about 3 mm height near the bottom regardless of its better chemical durability, but no delamination was evidenced.
Conflict of Interest Declaration
The authors declare that they do not have any financial or nonfinancial competing interests related to the content of the manuscript.
Acknowledgements
The authors want to thank Mr. Alberto Mezzadri of SM Pack SpA, p.le Vecchia Fornace 6/A 43035 Felino, Italy, for valuable support in the present research work.
This work was partly presented at the PDA Europe Parenteral Packaging, March 11–12, 2014 in Brussels, Belgium, and in the poster session (updated version) of the PDA Annual Meeting April 7–9, 2014 in San Antonio, TX.
- © PDA, Inc. 2017
PDA members receive access to all articles published in the current year and previous volume year. Institutional subscribers received access to all content. Log in below to receive access to this article if you are either of these.
If you are neither or you are a PDA member trying to access an article outside of your membership license, then you must purchase access to this article (below). If you do not have a username or password for JPST, you will be required to create an account prior to purchasing.
Full issue PDFs are for PDA members only.
Note to pda.org users
The PDA and PDA bookstore websites (www.pda.org and www.pda.org/bookstore) are separate websites from the PDA JPST website. When you first join PDA, your initial UserID and Password are sent to HighWirePress to create your PDA JPST account. Subsequent UserrID and Password changes required at the PDA websites will not pass on to PDA JPST and vice versa. If you forget your PDA JPST UserID and/or Password, you can request help to retrieve UserID and reset Password below.