2025 Volume 8 Issues 1–3
|
INEOS OPEN, 2025, 8 (1–3), 166–170 Journal of Nesmeyanov Institute of Organoelement Compounds Download PDF |
|
Poly(3-hydroxybutyrate) Crystallization: Investigation by Physicochemical Methods
and Molecular Dynamics Simulations
M. A. Hossain,*a G. A. Bonartseva,b V. V. Voinova,a A. S. Pryadko,c
M. A. Surmeneva,c R. A. Surmenev,c and A. P. Bonartseva
a Faculty of Biology, Lomonosov Moscow State University, Leninskie Gory 1, str. 12, Moscow, 119234 Russia
b Research Center of Biotechnology, Russian Academy of Sciences, Leninskii pr. 33, str. 2, Moscow, 119071 Russia
c National Research Tomsk Polytechnic University, pr. Lenina 30, Tomsk, 634050 Russia
Corresponding author: M. A. Hossain, e-mail: asifhossain38@yahoo.com
Received 2 February 2025; accepted 30 March 2025
Abstract

Poly(3-hydroxybutyrate) (PHB) is a polymer of a polyhydroxyalkanoate family which when synthesized using the biotechnological methods is biocompatible and biodegradable. However, the complex crystallization behavior of this polymer makes its physical properties difficult to predict accurately. We performed the Raman spectroscopy and surface free energy calculation on five samples of different molecular weights, which did not reveal any significant differences among the samples. In addition, the molecular dynamics simulations using 100 monomer long chains were performed. Our simulation boxes of the polymeric material were found to behave appropriately in response to changes in temperature, and the Young modulus of the simulation boxes were found to be 2.08, 1.84 and 2.9 for the production, annealing and quenching runs, respectively.
Key words: polymer, polyhydroxyalkanoate, molecular dynamics.
Introduction
Poly-3-hydroxybutyrate is a naturally occurring biopolymer, a member of the polyhydroxyalkanoate (PHA) family, which has received significant attention in biomedical applications owing to its biocompatibility, biodegradability, and mechanical properties. Derived from various bacterial sources, PHB is commonly used in medical fields for applications such as tissue engineering scaffolds [1]. Despite its promising features, a detailed understanding of its molecular behavior, in particular, under physiological conditions remains essential for optimizing its use in medical applications.
One key property of PHB that influences its performance is its crystallization behavior [2]. PHB can form semi-crystalline structures, where the crystalline regions provide mechanical strength, and the amorphous regions contribute to flexibility. The crystallinity of PHB strongly depends on its processing conditions and has a significant impact on its biodegradation rate, thermal stability, and mechanical performance. Understanding the crystallization process at the molecular level is crucial to tailor its properties for specific biomedical uses.
Molecular dynamics (MD) simulations offer a powerful computational technique for probing the atomic and molecular behavior of biopolymers like PHB, including their crystallization patterns. By simulating the interactions between atoms over time, MD provides insights into the structural stability, crystallization dynamics, and mechanical properties of PHB under various environmental conditions. Such simulations can help predict the polymer behavior in physiological media, in turn, helping researchers design more effective medical materials.
In this work, we present a combined experimental and molecular dynamics study of PHB, exploring its structural properties, crystallization behavior and mechanical performance. Our goal was to deepen the understanding of the PHB molecular behavior, providing a foundation for its enhanced application in medical fields.
Results and discussion
Unlike when modelling proteins, while simulating materials we start with a structure that is totally artificial. After randomly adding a number of polymer chains inside a box, the system inside the box exists in an unnatural configuration, which does not resemble the polymeric material. To turn the system into a material, the polymer chains have to be allowed to diffuse and settle into a more natural configuration. Forcefields such as GROMOS and GaFF were created with proteins and nucleic acids with molecular biology and biopharmaceutics in mind, rather than for polyesters as polymeric materials. As Xiang et al. showed [3] that for molecular dynamics simulations it is imperative to consider the motivation behind a forcefield with respect to the simulation objective.
The COMPASS (Condensed-phase Optimized Molecular Potentials for Atomistic Simulation Studies) forcefield was created with polymers and soft materials in mind, especially when dealing with crystallization and phase transitions [4]. COMPASS is parameterized for a wide range of polymers and is good for modeling both intra- and intermolecular interactions. This forcefield has been shown to be reliable for both natural polymers and inorganic-organic hybrid polymers [5, 6]. This is why Material studio was the platform of choice for our experiments, as it gives us access to the COMPASS forcefield.
An important question about the validity of the simulated polymeric system is concerned with the chain length. The length of polymer chains in the simulated model is very short at just 100 monomers, while natural PHB has chain lengths that can range from 400 to well over 1000 monomer repeats. We wanted to assess if large internal structural differences arise due to chain length differences. For this reason, in our experimental part we used the samples that have molar masses ranging from 348 kDa to well over 1300 kDa. A PHB monomer has molar mass of 104 Da [7], and the experimental value if polydispersity index (PDI) of PHB to be 1.85. Then for a 384kDa polymer, considering:

and the degree of polymerization DP to be,

where Mw is a molecular weight (348 kDa), Mn is the average molecular weight, and M0 is the molar mass of the PHB monomer (104 Da [7]). Therefore, for the 384 kDa PHB material, the calculated DP is 1996, which means that 384 kDA PHB have chains that are approximately 1996 monomers long. Simulating chains of such length is far beyond what is feasible with current computational hardware.
Raman and FTIR spectroscopy were used to study if significant differences arise due to large variations in the molecular weight. Spectroscopic results did not show any significant difference among the samples. The analysis of the samples by Raman spectroscopy showed similar peaks of different heights (Fig. 1). The FTIR spectra also revealed coinciding peaks (Fig. 2). The surface free energy (SFE) plays a key role in the process of crystallization according to the classical theory of nucleation [8], and in our samples, we found that large differences in chain length (molecular weight) did not significantly affect the value of SFE (Fig. 3). Hence, in theory, the simulations using chains that are 100 monomers long should be viable, as 100 monomer repeats give us a good tradeoff between the simulation calculation time and the complexity of the system.

Figure 1. Raman spectra of the samples showing similar peaks of different heights.
The full spectra are provided in Fig. S1 in the Electronic supplementary information (ESI).

Figure 2. Fragments of the FTIR spectra for five samples showing the overlapping peaks. For the full spectra, see Fig. S2 in the ESI.

Figure 3. Surface energies of the samples and their respective molecular masses.
When it comes to simulating PHB, an important point to consider is that it is semi-crystalline in nature. Crystallization occurs when the temperature of a system of molecules drops below its melting point. This is why we performed two cycles of simulated annealing, where the temperature of our system was slowly (in 5 steps) raised to 500 K and then reduced to 300 K, which is well below the melting point of PHB. We observed that the density of our system decreases with increasing temperature and vice versa (Fig. 4), which is in line with the natural behavior. We compared the Radial Distribution function of the atoms in the system before and after annealing (Fig. 5). A slight increase in the height of the first peak and well as secondary peaks indicate the system is now in a more compact state.

Figure 4. Changes in the temperature during simulated annealing for 2 cycles (top).
Evolution of the density due to changes in the temperature during annealing (bottom).

Figure 5. Radial distribution of the simulation box before and after simulated annealing.
The Quench task in Material studio's Forcite module allows us to optimize and decrease the potential energy of a system while keeping the temperature constant. It is a classical molecular dynamics calculation, just that structure is quenched every specific number of frames. During quenching, the space of a possible configuration of atoms is sampled, and the lowest energy configuration is chosen artificially, and the simulation is resumed. Our simulation box was quenched 20 times across a trajectory of about 200 ns. We can see that, as the structure is being quenched, the total energy of the system goes down, while the density of the system increases (Fig. 6), indicating it is becoming more closely packed. This alludes to the secondary crystallization of PHB, during which as the PHB film ages, its degree of crystallinity increases, implying that the material becomes more brittle and stable (harder to biodegrade) [9, 10]. The only difference is during Quench optimization, this is happening really fast compared to taking place across months in the case of secondary crystallization.

Figure 6. Quench optimization of PHB material, showing that a steady decrease in the total energy (top)
is accompanied by an increase in the density (botoom).
A fresh PHB film is produced when a solution of PHB in chloroform is stirred and then poured onto a Petri dish. Then the solvent is allowed to evaporate. As in our previous work [9], we showed that the Young modulus of freshly poured PHB films is found to be around 1.8–2.2 GPa (Table 1). In a fresh PHB film, the PHB chains go from a highly mobile state (in solution) to a restricted configuration (solid). This is why when simulating polymers, we first equilibrate the system at high temperature to let the chains move around. As a freshly produced PHB film ages, its Young modulus increases up to about 4.7 GPa in a week [9]. In our simulations, we did equilibration run at 500 K, then the production run at 300 K for 100 ns. After the production run, the Young modulus was calculated to be 2.08 GPa, which very nicely corresponds to that of fresh PHB film.
Table 1. Comparison of the Young moduli of various simulation boxes
|
Source
|
Young modulus, GPa
|
|
Ref. [9]
|
1.8 ± 0.6 [9]
|
|
After production run
|
2.08
|
|
After annealing
|
1.84
|
|
After quenching
|
2.9
|
Crystallization is the process where a group of atoms attain a highly ordered configuration. When we look at the simulations of semi-crystalline polymeric chains after 200 ns, the configuration still looks chaotic and unorderly to the naked eye. In reality, an easy estimate of the degree of crystallinity is to check how brittle the PHB film is (higher Young modulus). The less crystalline the material, the more it flexible or ductile—this is the reasoning behind creating copolymers with 3-hydroxyvalerate (PHB-HV). The HV units disrupt the crystallization process of homogenous PHB; this gives rise to a more ductile material (lower Young modulus) [9]. We calculated the Young modulus of each of our simulation boxes (production, anneal, quench) (Table 2). The simulation boxes can be thought of as a polymeric material cubic cross-section, the dimensions of which is approximately the length of 100 monomers. The purpose of calculating mechanical properties such as the Young modulus is to give us an idea of what is happening inside what looks to the naked eye like a chaotic configuration of polymeric chains. Like how after a freshly poured PHB film goes from a mobile configuration of chains (in solution) to a restricted configuration (solid), it has a relatively low Young modulus. Then as the film ages, its Young modulus steadily goes up over time due to secondary crystallization. Similarly, PHB chains that were equilibrated at 500 K, after the production run at 100 K for 100 ns, the simulation box has a relatively low Young modulus. If we push the simulation box to a lower energy configuration (Quench optimization), the Young modulus increases to about 2.9 GPa. It seems like decreasing internal (potential) energy of a mixture of PHB chains results in an increase in the Young modulus. Why the Young modulus goes down after simulated annealing is something we need to further investigate, as according to our prediction, it should be increasing. The material is supposed to become more crystalline after reheating and cooling.
Table 2. Samples and their molar masses
|
Sample name
|
Molar mass, kDa
|
|
PHB-348
|
348
|
|
PHB-1095
|
1095
|
|
PHBV-5.9%
|
1272
|
|
PHBV-9.0%
|
1336
|
|
PHBV-17.6%
|
600
|
Experiment and calculations
Experimental analysis
Five biologically synthesized PHB samples were used in the experimental part of our research (Table 1). The samples had different molecular weights, and three of the films were PHB-HV copolymers.
The investigations were performed to assess the relationship between the molecular weight and crystallinity. The surface free energy of materials plays a key role in the crystallization. According to the classical theory of nucleation (CNT), the free energy required for the formation of a critical nucleus (ΔG*) is given by the following equation

where γ is the surface free energy per unit area, and ΔGv is the volume free energy change associated with the phase transformation [11]. The surface free energy of the samples γ was measured using the contact angle method, with water and diiodomethane as probe liquids. The Owens–Wendt–Rabel–Kaelble (OWRK) method was applied to calculate the surface energy by measuring the contact angles of these liquids on the sample surface [12].
The OWRK method is based on the separation of the surface energy into its polar (γp) and dispersive (γd) components. Using the contact angles of water and diiodomethane θ, the total surface energy γs is calculated through the following equation:

where the polar and dispersive components are determined by solving:

where, γld and γlp represent the dispersive and polar components of the liquid surface tension, respectively. By using both water (polar liquid) and diiodomethane (non-polar liquid), the surface free energies of the samples were determined and compared against their molecular weights (Fig. 3).
The Raman spectra were acquired using a confocal Raman microscope integrated with a scanning probe optical unit (NTEGRA Spectra, NT-MDT, Russia) and equipped with a 20× objective. The excitation source was a semiconductor laser operating at a wavelength of 532 nm, with a maximum output power of 150 mW.
The FTIR spectra were collected with a PerkinElmer Spectrum Two FT-IR Spectrometer (USA) equipped with a room-temperature LiTaO3 (lithium tantalate) MIR detector with a SNR of 9300:1 and an optical system with KBr windows for data collection over a spectral range of 550–2000 cm–1 at the best resolution of 0.5 cm–1.
To study the mechanical properties of polymers, the samples of polymer films were cut in the form of blades with a working part length of 10 mm, a width of 3 mm, and a thickness of 0.03–0.04 mm. The mechanical properties of the polymer films (strength, elongation at break, elastic modulus) were determined on an Instron Zwick/Roell BZ 25/TN1S universal dynamometer (Zwick Roell, Germany) with the following parameters: preliminary load 0.05 N, preliminary load speed 1 mm/min, start and end of determination of the Young modulus 0.1 and 0.15 N, respectively, the speed of movement of the crosshead for determining the Young modulus 2–5 mm/min, grip type Vulkollan. The tests were carried out at 20 ℃.
The Young modulus of the PHB films were determined using an Instron 1121 tensile testing machine (Zwick Roell, Germany).
Molecular dynamics analysis
The experiments were carried out in Material Studio's Forcite module, which is just classical molecular dynamics, but additionally it has the propriety COMPASS forcefield. The polymer chains containing 100 3-hydroxybutyrate (3HB) monomer repeats was used for our simulations. The polymer molecule was replicated 50 times in the simulation box to create an amorphous cell. The size of simulation box was chosen such that the density of the simulation box (mass of 50 chains/volume of box) corresponds to the natural density of PHB, which is around 1.18 g/cm3. The energy minimization was carried out to remove steric clashes and atomic overlaps. The preliminary simulations were carried out using a very small timestep to equilibrate the structure until the energy of the system reached stable and natural values. Afterwards, the production simulations runs were carried out using a timestep of 1 fs under NPT ensemble at 300 K temperature to let the polymer chains diffuse and create a configuration that resembles PHB material.
Simulated annealing is a computational method that mimics the gradual heating and cooling of a material. Our PHB material from the production runs were gradually heated to 500 K and then cooled back down to around 300 K. The heating and cooling were done in 5 steps, such that in each step the temperature of the material was raised or reduced by one-fifth of the target temperature. The radial distribution function (RDF) of the material before and after simulated annealing were compared.
The Quench task in the Forcite module allows us to search conformational space for low energy structures by sampling at fixed points along a classical trajectory. Usually, the system of molecules reaches low energy conformations by random thermodynamic fluctuations; however, this can take a very long time, considering the timescale of molecular dynamics simulations. The quenching simulation basically forces the system to follow an artificial optimized path towards low energy conformations. A simulation of 200 ns was carried out, where our structure was quenched every 10 ns. The Young modulus of the material simulation boxes were calculated using the Mechanical Properties task in Material Studio.
Young modulus calculation
The Young modulus of a simulation box is calculated using the Constant Strain Method. A series of small strains ε are applied to the system in different directions. For the strained systems, stress tensor components σ are calculated according to Virial Theorem equation:

where mi, vi, ri, Fi denote the mass, velocity, position, and force of particle i, respectively. The set of σ and ε are fitted according to equation:
σ = Cε (6)
where C is the elasticity matrix. The Young modulus E can be extracted from the elasticity matrix by:

where S11 is an element of the compliance matrix, which is the inverse of the stiffness matrix.
Conclusions
Raman spectroscopy, FTIR spectroscopy, and the surface free energy calculations were performed on five samples of naturally synthesized PHB and PHB-HV films of molecular weights ranging from 348 kDa to 1336 kDa where no significant relationship based on the molecular weight was discovered among the samples. The simulations done using the COMPASS forcefield on 50 molecules of 100 monomers long PHB chains initially yielded a material with the Young modulus of 2.08 GPa. Further reheating the material to 500 K and then cooling back down to 300 K for two cycles reduced the Young modulus to 1.84 GPa. Quenching the structure by decreasing the total energy caused the material to become denser and brittle, as was evidenced by an increase in the Young modulus up to 2.9 GPa.
Acknowledgements
This work was performed within the state assignment of Lomonosov Moscow State University. The research was carried out using the equipment of the shared research facilities of HPC computing resources at Lomonosov Moscow State University. The equipment of the Multi-Access Equipment Centers of Lomonosov Moscow State University was used in the work. Raman spectroscopy experiments were performed on the infrastructure and facilities of Tomsk State University and Institute of Physics of Strength and Materials Science (Tomsk, Russia).
Electronic supplementary information
Electronic supplementary information (ESI) available online: the Raman and FTIR spectra of samples, detailed simulation parameters, and output of the calculation of the mechanical properties. For ESI, see DOI: 10.32931/io2566a.
References
- A. V. Volkov, A. A. Muraev, I. I. Zharkova, V. V. Voinova, E. A. Akoulina, V, A. Zhuikov, D. D. Khaydapova, D. V. Chesnokova, K. A. Menshikh, A. A. Dudun, T. K. Makhina, G. A. Bonartseva, T. F. Asfarov, I. A. Stamboliev, Yu. V. Gazhva, V. M. Ryabova, L. H. Zlatev, S. Y. Ivanov, K. V. Shaitan, A. P. Bonartsev, Mater. Sci. Eng. C, 2020, 114, 110991. DOI: 10.1016/j.msec.2020.110991
- A. J. Owen, J. Heinzel, Ž. Škrbić, V. Divjaković, Polymer, 1992, 33, 1563–1567. DOI: 10.1016/0032-3861(92)90139-N
- Z. Xiang, C. Gao, T. Long, L. Ding, T. Zhou, Z. Wu, J. Phys. Chem. B, 2025, 129, 1864–1873. DOI: 10.1021/acs.jpcb.4c08471
- H. Sun, J. Phys. Chem. B, 1998, 102, 7338–7364. DOI: 10.1021/jp980939v
- T. S. Asche, P. Behrens, A. M. Schneider, J. Sol-Gel Sci. Technol., 2017, 81, 195–204. DOI: 10.1007/s10971-016-4185-y
- S. Saha, A. K. Bhowmick, J. Appl. Polym. Sci., 2019, 136, 47457. DOI: 10.1002/app.47457
- https://pubchem.ncbi.nlm.nih.gov/compound/3-Hydroxybutyrate
- F. Zhang, E. Mohammadi, X. Luo, J. Strzalka, J. Mei, Y. Diao, Langmuir, 2018, 34, 1109–1122. DOI: 10.1021/acs.langmuir.7b02807
- V. Zhuikov, A. Bonartsev, D. Bagrov, A. Rusakov, A. Useinov, V. Myshkina, T. Mahina, K. Shaitan, G. Bonartseva, Solid State Phenom., 2017, 258, 354–357. DOI: 10.4028/www.scientific.net/SSP.258.354
- V. A. Zhuikov, A. P. Bonartsev, D. V. Bagrov, S. G. Yakovlev, V. L. Myshkina, T. G. Makhina, I. V. Bessonov, M. N. Kopitsyna, A. S. Morozov, A. A. Rusakov, A. S. Useinov, K. V. Shaitan, G. A. Bonartseva, Mol. Cryst. Liq. Cryst., 2017, 648, 236–243. DOI: 10.1080/15421406.2017.1302580
- D. Turnbull, J. C. Fisher, J. Chem. Phys., 1949, 17, 71–73. DOI: 10.1063/1.1747055
- D. K. Owens, R. C. Wendt, J. Appl. Polym. Sci., 1969, 13, 1741–1747. DOI: 10.1002/app.1969.070130815

