Phase separation and its relevance to pharma glass packaging

Hello everyone – this post is meant to expand on some statements that I previously made during my “Properties of Pharma Glass” series.?I’ve mentioned the idea of phase separation in glass in my posts on permeability and color/optical transmission.?Phase separation is a universal phenomenon.?However, the specific chemical mechanisms that drive phase separation in an inorganic glass are very different than something like a vinaigrette for your salad (see Footnote 1).?We’re not going to get bogged down in these differences in this article.?Let’s focus instead on a general description of phase separation and how it applies to glasses used as packaging for parenteral drugs.

We begin by considering a solution – i.e., a homogenous mixture comprised of two or more substances.?What does it actually mean for a mixture to be homogenous??Imagine a container filled with a liquid that is a homogeneous mixture.?Now let’s assume you have a way to accurately measure the composition of that liquid at multiple points within the volume.?You should find that the results of a compositional analysis converge around a fairly constant value, irrespective of sampling location within the homogeneous mixture.?Phase separation, also called “immiscibility”, is the loss of homogeneity – the system literally separates into chemically distinct phases.?This can be easy to observe with low viscosity liquids such as vegetable oil and water that readily undergo phase separation into two distinct layers.?However, things can become a little more subtle as the driving force for separation decreases and/or the viscosity of the separating phases is relatively high.

The phase separation that occurs between vegetable oil and water is an example of what is called “stable immiscibility”.?This occurs when phase separation is observed above the “liquidus” – i.e., a boundary that defines the temperature at which a system is 100% liquid under equilibrium conditions.?There is also “metastable immiscibility”, which is when phase separation is observed below the liquidus. ?Figure 1 is a binary phase diagram system Na?B?O?? (also written Na?O·4B?O?) and SiO?.?Let’s break that down – a phase diagram is just a graphical depiction of the various phases (solids, liquids, and/or gases) that co-exist under a given set of conditions.?We’re going to assume that the pressure of the system is fixed at standard atmospheric pressure (see Footnote 2). ??We can vary two parameters in this case – the temperature (the ordinate axis) and the composition (the abscissa axis).?It is a binary phase diagram because there are two components in the system (Na?B?O?? and SiO?).

Understanding how the composition works in a phase diagram can be a little confusing for the uninitiated.?The first important rule is that the sum of your components must always add up to 100%, meaning that a system containing 50% Na?B?O?? must also contain 50% SiO? (ignoring trace impurities).?Next, the compositional axis is a sort of lever.?Assume the fulcrum of the lever is at the extreme left of the diagram – this represents a system that is entirely comprised of Na?B?O??.?Similarly, a fulcrum placed at the extreme right is 100% SiO?.?Based on this information, it’s fairly intuitive to understand how a fulcrum placed at the middle of the axis represents a system containing equal amounts of both components.?Now imagine a composition represented by a point at 80% – this is a system consisting of 20% Na?B?O?? and 80% SiO?.???The fulcrum (located at the 80% mark) divides the compositional axis into two parts.?The larger portion to the left of the fulcrum represents the majority component (SiO?).?This is where the analogy of a lever becomes important – the majority component has greater “mechanical advantage”, so to speak.

You can see the direction of increasing temperature.?As temperature goes up, you first see the appearance of heterogeneous mixtures of a liquid phase and a solid phase (e.g., high quartz + liquid – see Footnote 3).?However, increasing the temperature eventually places the system at the liquidus line indicated in red.?Note that only liquid is present above the liquidus, irrespective of the composition.?The yellow line marks the boundary of metastable immiscibility within this system.?Any composition at a temperature below and inside ?this boundary is expected to exhibit phase separation at equilibrium.?This is another point where things can become a little confusing.?Note that I haven’t recently mentioned glass – that’s because by definition glass is not an equilibrium material.?You will not see “glass” appear anywhere on a phase diagram.?Glasses are instead formed by cooling a liquid sufficiently fast to avoid crystallization.?So why are we discussing this??Because supercooled liquids can still undergo phase separation and form a glass upon cooling.?Under the right conditions, the resulting glass can retain the phase separation that was established at a higher temperature.

The next thing we need to know is that the three-dimensional structure of phase separation within a glass depends on where you are within the region of immiscibility.?A detailed discussion of this topic is beyond the scope of this article.?Just know that compositions near the edges of the immiscibility region tend to exhibit a “droplet-in-matrix” morphology, while compositions near the center tend to exhibit an “interconnected” morphology (see Footnote 4).?Figure 2 shows examples of what I mean by these two morphology types.?As the name suggests, a droplet-in-matrix morphology consists of generally spherical droplets of one glass composition dispersed in a continuous matrix of another glass composition.?An interconnected morphology is comprised of two interpenetrating networks, each having their own distinct glass composition. ?The morphology of phase separation can have a significant impact on the properties of a glass (we’ll come back to this point later).

One last important piece of information is that the size of the morphological features within a phase separated glass are not necessarily fixed over time.?Holding a glass at an appropriate temperature and time can cause the average size of features to increase.?This change in size is driven by the favorable energetics of reducing the interfacial surface area between the two chemically distinct phases (see Footnote 6).?Figure 3 shows an example of this change in scale caused by holding an alkali borosilicate glass (one that happens to exhibit a droplet-in-matrix morphology) at an elevated temperature for increasing amounts of time.

Before we move on to our final topic, I also need to mention that I am intentionally ignoring one other aspect of this discussion.?I’m not going to cover how to determine the relative amounts and compositions of the two phases that form within the region of immiscibility.?The process is akin to the idea of using a “lever” on the compositional axis and is reasonably straightforward in a binary system.?The process becomes more complicated in a three component (i.e., a ternary) system.?Knowing how to do these calculations isn’t needed to appreciate what comes next.?Instead, I’m going to make a generalization for the context that we care about (glasses suitable for parenteral packaging) – when phase separation occurs, it will tend to produce one phase that is alkali borate-rich and another that is silica-rich.

Figure 3 is a pseudo-ternary diagram for the Na?O-B?O?-SiO? system.?What makes this a “pseudo”-ternary??We’re not actually using this diagram anymore to show the various liquid and solid phases that co-exist as a function of composition and temperature.?The smaller inset triangle shows the entire system, but we’re only concerned with the portion highlighted in yellow in this case – this is the same portion shown as an expanded trapezoid.?Moving from the representation of a binary to a ternary system involves a change in perspective.?We are now looking down at the projected image of a three-dimensional composition/temperature landscape.?The temperature axis is still there, but it’s perpendicular to your screen and the direction of increasing temperature pointing straight out at you.?Just like a topographical map, we can draw contours to make sense of the projected image.?These contours represent lines of equal temperature instead of equal elevation.?We could in principle add further features to Figure 3 to represent the liquidus surface, regions of solid/liquid mixtures, etc. but the diagram would become overly complex and obscure the portion that we care about – the contours describing the metastable immiscibility “dome”.

Similar to the binary diagram in Figure 1, a composition plotted at the extremes of the ternary (now a triangle corner instead of the edge of a rectangle) represents 100% of a given component. ?A compositional point in the exact middle of the triangle contains 1/3 Na?O + 1/3 B?O? + 1/3 SiO?, and so on.?I mention this because the region of immiscibility is decidedly skewed towards compositions that are relatively low in Na?O and relatively high in SiO? – i.e., a composition region that is very similar to alkali borosilicate glasses that have historically been used for parenteral packaging.?I’ve plotted two compositional points on the ternary that roughly correspond to average compositions for Type IA and IB glasses.?These are rough approximations because no pharma grade borosilicate glasses contain just three components.?They also contain components such as K?O, MgO, CaO, Al?O?, etc. I’ve therefore made some simplifying assumptions by aggregating components that play similar roles within the structure of the glass.?Any alkali or alkaline earth oxides are summed together as “Na?O”, Al?O? and B?O? are summed together as “B?O?”, and SiO? stands alone.?Putting aside some issues with this approach (see Footnote 7), we see that both borosilicate glass types are located within the metastable immiscibility region of the pseudo-ternary diagram.

We’ve established that Type IA and IB borosilicate glasses used in parenteral packaging reside within a region of metastable immiscibility.?I also briefly mentioned before that the presence of phase separation can impact glass properties.?Which properties? – a whole lot of them, including anything that depends on mass transport such as viscosity, conductivity, permeability, and chemical durability.?In the case of Type IA glasses, we know that a droplet-in-matrix type of phase separation is present.?My previous generalization still applies – the droplets are alkali borate-rich while the continuous matrix is mostly silica.?In effect, the least chemically durable components of the glass are trapped within another glass that has exceptional durability.?This is what provides Type IA borosilicates with the chemical resistance required for storing parenteral drugs.?And here’s the magic of this compositional family -- the alkali and boron are “unshackled” at temperatures above the immiscibility dome to fully participate in affecting the properties of the material.?They help to reduce melting temperatures and viscosity when you need them, and then go back into their droplet phase upon cooling (see Footnote 10).?The specific details of the Type IB borosilicates are less well defined.?I’ve surprisingly found only one piece of evidence so far in the open literature indicating that a commercially available Type IB borosilicate glass shows the barest hints of being phase separated (presumably also droplet-in-matrix – see Footnote 11).

We’ve discussed how phase separation can improve chemical durability – can it also make it worse??Most definitely, sometimes in very useful ways.?The classic example is what I will generally refer to as Controlled Porous Glass (CPG). ?Shifting the initial glass composition towards higher B?O? content can move the system into a region of immiscibility that now exhibits an interconnected morphology (see Footnote 12).?The phases are still alkali borate-rich and silica-rich, but the low durability phase is now accessible to the external environment and can be readily dissolved by acidic solutions.?A highly porous silica-based material, otherwise known as CPG, is left behind after the alkali borate phase is leached away.?Furthermore, the pore size distribution of this sponge-like material can be engineered by controlling the relative proportions of the two phases and appropriate heat treatments within the immiscibility region.?CPG materials find use in a variety of applications, including drug delivery, catalyst supports, size exclusion chromatography, and enzyme immobilization.

Writing this post was a bit of nostalgia for me.?I used to teach a course to undergraduate students that focused on phase diagrams during my days in academia.?It turns out that phase diagrams can provide a surprising amount of insight into the manufacturing and properties of all sorts of materials.

Have any questions??Please comment below or feel free to contact me directly.

?

Footnotes

1.?The culinary relevance of phase separation doesn’t end at vinaigrettes.?If you’re not already familiar with it, try looking up the “ouzo effect”.

2.?We often ignore the influence of pressure for many oxide material systems – i.e., the phase diagrams may not change much as a function of pressure.?There are of course exceptions.?For example, oxides containing variable redox state ions such as iron can be strongly impacted by the partial pressure of oxidizing and reducing gases.

3. ?High quartz, tridymite, and cristobalite – these are all various crystalline solid forms or “polymorphs” of SiO?.

4.?You will see multiple terms used in the literature to describe the mechanisms leading to the two different morphologies of phase separation.?For example, droplet-in-matrix and interconnected are called “nucleation and growth” and “spinodal decomposition”, respectively.?While these terms are more formally correct, they’re not terribly descriptive if you’re not familiar with the subject.

5.?These illustrations are mean to be idealized versions of these two morphologies.?In reality, hybrid morphologies that straddle the line between these two extremes can be observed depending on the composition, temperature, and time conditions that produce a particular microstructure.?Sequential phase separation events can also occur in some cases – i.e., two phases that form at some elevated temperature can each undergo their own separation event upon cooling to produce complex, multi-phase systems.

6.?Going back to the oil and water example – we intuitively know that shaking the two together initially creates a dispersion of the two components.?Over time, the two phases separate out into two distinct layers.?It’s the same thermodynamic driving force at work.?The amount of interfacial area between the oil and water is at a minimum when the two phases revert from small droplets into two individual layers.

7.?This process of aggregating components is problematic because the assumption being made is not fully accurate.?For example, the addition of Al?O? to alkali borosilicate glasses has been shown to suppress phase separation to some extent.?Just remember that the approach being used here is purely for the convenience of visualizing a multi-component system on a two-dimensional image.

8. Reference: Haller W, Blackburn DH, Wagstaff FE, Charles RJ (1970).?Metastable immiscibility surface in the system Na?O-B?O?-SiO?. Journal of the American Ceramic Society, 53: 34-39.

9.?Reference: Bartl MH, Gatterer K, Fritzer HP, Arafa S (2001).?Investigation of phase separation in Nd3+ doped ternary sodium borosilicate glasses by optical spectroscopy. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 57: 1991-1999.

10.?This raises an important point – the phenomenon of phase separation in glass was not truly understood until the mid-20th century.?As is often the case in science, our ability to understand a phenomenon is in part dictated by the tools that are available to probe and characterize a system.?It took the development of electron microscopy to positively identify the presence of phase separation.?But this lack of understanding didn’t stop people from leveraging the technological benefits to first develop heat- and chemical-resistance borosilicate glasses in the late 19th and early 20th centuries through trial-and-error experimentation.

11.?Reference: Kreski PK, Varshneya AK (2012).?Microstructural phase separation and delamination in glass for pharma applications. Processing, Properties, and Applications of Glass and Optical Materials: Ceramic Transactions, Volume 231, Varshenya AK, Schaeffer HA, Richardson KA, Wightman M, Pye LD (Editors), Wiley-American Ceramic Society, 85-90.

12.?Reinforcing Footnote 10 – this effect was first discovered in 1920 but not fully understood.?Reference: Turner WES, Winks F (1926).?The influence of boric oxide on the properties of chemical and heat-resisting glass. II. The resistance to chemical reagents. Journal of the Society of Glass Technology, 10: 102.

?

The original version of this article was published on June 23, 2023.