Chirality hierarchical transfer in homochiral polymer crystallization under high-pressure CO₂

Abstract

Ordered phase transitions are commonly correlated to symmetry breaking, while disordered phase transitions are characterized by symmetry restoration. Nevertheless, this study demonstrates that these correlation relations are not always applicable in chiral polymers under high-pressure Carbon Dioxide. Without racemization, homochiral Poly (lactide acid) can generate two vortex-shaped dendritic crystals with opposite spiral chirality, and snowflake-shaped dendritic crystals without spiral chirality. The transition from homochiral molecules to achiral crystals signifies the chiral symmetry restoration during the ordering process. The primary elements responsible for the various hierarchical transfers of homochiral Poly (lactide acid) are related to chain tilt, surface stress, and frustrated structures of Poly (lactide acid) crystals. Here, we show the entropy impact of Carbon Dioxide can be utilized to programmatically regulate the morphological chirality of crystal superstructure and crystal form of homochiral Poly (lactide acid).

Introduction

Chirality refers to the inherent characteristic of a molecule or item that cannot be overlapped with its mirror copy. The natural reserve’s imbalance of chirality and regulation of chiral substances play a crucial role in the emergence of life and ensuring medical safety. In the field of condensed state physics, it becomes essential to comprehend the hierarchical transfer of chirality from individual molecules to macroscopic assemblies for manipulating the characteristics and functionalities of materials^1,2,3.

There are several ways in which chirality may be transferred. The helical chirality of the crystalline stem in polymers is influenced by the chirality of the predominant enantiomer in an unequal copolymer consisting of two enantiomers with opposing chirality^4,5. This winner-takes-all mode enables chiral amplification from the microscale to the macroscale. Minor alterations in the chemical composition of a homochiral polymer may cause a reversal in the twist orientation of the resultant crystals, known as chiral reversal^6,7. In addition, it has been shown that achiral inorganic materials or liquid crystal polymers can generate crystals that exhibit chiral characteristics. This phenomenon is referred to as chiral symmetry breaking^8,9,10. Based on the understanding of chirality hierarchical transfer mechanisms, scientists have developed many techniques to manipulate the macroscopic chirality of polymers. The prevailing chemical techniques for manipulating the chirality of polymer crystals or assemblies include the incorporation of chiral segments⁵, chiral nanoparticles³, and chiral dopants¹¹. Some physical techniques, such as light stimulation¹², electric fields¹³, magnetic fields¹⁴, and opto-electro-magnetic multi-field effects¹⁵, have also gained widespread attention in the manipulation of chirality. Currently, there is a significant focus on developing flexible, dependable, and programmable methods for controlling the chirality of chiral materials.

This work has identified a mechanism for hierarchically transferring chirality and has put forward a technique for manipulating the crystal chirality of chiral polymers using high-pressure Carbon Dioxide (CO₂). The model polymers used are Poly(D-lactide acid) (PDLA) and Poly(L-lactide acid) (PLLA) (Supplementary Fig. 1), which are synthesized using ring-opening polymerization and have a high level of purity. The crystallization process in high-pressure CO₂ is recorded using a self-developed visualization system^16,17. Aluminum alloy’s three-dimensional (3D) printing technology is used to manufacture conformal cooling tunnels (Supplementary Figs. 2 and 3) to enhance temperature control precision. It is found that a homochiral Poly(lactide acid) (PLA) can generate two distinct forms of vortex-like dendrites with opposing spiral orientations, namely two different chiral crystal morphologies, under high-pressure CO₂. Furthermore, it is worth noting that the homochiral PLA can also generate a dendritic with a snowflake-like structure without chirality under high-pressure CO₂. The transition from homochiral molecules to achiral crystals implies the restoration of chiral symmetry during the ordering process. This phenomenon seems to contradict the phase transition principle, which states that ordering transformations are associated with symmetry breaking. The primary objective of this paper is to understand this phenomenon and elucidate the chirality hierarchical transfer mechanisms in the crystallization process of homochiral PLA under high-pressure CO₂.

Results

Chirality transfer from molecule to crystal morphology

Figure 1 illustrates the crystal morphology of homochiral PDLA and PLLA in high-pressure CO₂. Each homochiral PLA exhibits three distinct morphologies of dendritic crystals: (1) The dendritic crystals generate a vortex-like pattern with a spiral chirality matching the molecule chirality (Fig. 1a, b); (2) snowflake-like dendrites without spiral chirality (Fig. 1c, d); (3) The dendritic crystals generate a vortex-like pattern with a spiral chirality contrary to the molecular chirality (Fig. 1e, f). Supplementary Discussion gives the definition rules of morphology chirality of these PLA crystals. Supplementary Figs. 4–10 provide in situ images of the growth process of crystals with mentioned morphologies. The morphologies of these three types of dendrites mean the occurrence of chiral amplification, chiral disappearance, and chiral reversal in the chirality transfer from molecules to macroscopic crystals, respectively. Among them, the chirality disappearance means that chirality symmetry restoration is realized in the crystallization process of homochiral polymer.

Fig. 1: Crystal morphology and spiral curvature statistic of PLA enantiomer observed by in situ high-pressure visualization system.

a PLLA dendrites with left-hand spiral (Crystallization temperature T_c = 80 °C, crystallization pressure, i.e., CO₂ pressure P_c = 500 Psi). b PDLA dendrites with right-hand spiral (T_c = 80 °C, p_c = 500 Psi). c PLLA snowflake crystals without spiral (T_c = 60 °C, P_c = 1400 Psi). d PDLA snowflake crystals without spiral (T_c = 60 °C, Pc = 1400 Psi). e PLLA dendrites with right-hand spiral (T_c = 60 °C, P_c = 1250 Psi). b, f PDLA dendrites with left-hand spiral (T_c = 60 °C, P_c = 1250 Psi). The scale bars in (a–f) represent 100 μm, and the spiral directions of dendrites are marked in the bottom left corner of the images. g Spiral chirality and curvature of PLA dendritic crystal after treatment under different crystallization temperatures and different CO₂ pressures. The curvature of bent branch crystals in Supplementary Figs. 11 and 12 is measured by the Kappa plugin of ImageJ software, with 12 randomly measured branches for each image. IQR stands for inter-quartile range. Here we define left-handed spiral (curving clockwise) as positive curvature, while right-handed spiral (curving counterclockwise) as negative curvature.

Generation conditions are the key to understand the hierarchical transfer of chirality. From Fig. 1g and Supplementary Fig. 12, PLLA crystals exhibit a vortex shape with a left-handed spiral under a temperature of 70–80°C or a CO₂ pressure of 250–750 Psi. Conversely, under a temperature of 50–60°C and a CO₂ pressure of 1000 Psi, they exhibit a vortex shape with a right-handed spiral. Supplementary Fig. 13 shows that PLA mainly generates spherulites without spiral features under atmospheric pressure. So, high-pressure CO₂ is the necessary condition to reveal the morphology chirality of PLA dendritic crystal. From Supplementary Fig. 14, with the increase of CO₂ pressure at 60 °C, spiral chirality of PLA crystal undergoes chiral amplification, chiral disappearance, chiral reversal, and chiral disappearance again. The non-spiral snowflake crystals can be found between two opposite chiral morphologies or under high-pressure conditions (e.g., P_c = 1600 Psi). Thus, the effect of CO₂ on the chiral characteristics of PLA dendrites should be primarily investigated.

The phase field simulation of Gránásy et al.¹⁸ indicated that impurity particles can change the growth direction of achiral polymer dendrites, but it leads to dizzy dendrites with random directions. Prud’homme et al.¹⁹ found that unequal mixing of PLLA and PDLA also can generate vortex-like dendrites. Although there is a correlation between molecular chirality and crystal chirality, molecular chirality is not a decisive factor of crystal chirality. Figure 2 gives the chirality information of PLA monomer and conformational chirality of PLLA molecular chains in crystals with different spiral chirality formed under high-pressure CO₂.

Fig. 2: Chirality information at different levels.

a Right-handed spiral crystals of PLLA (T_c = 50 °C, P_c = 1000 Psi). b Left-handed spiral crystals of PLLA (T_c = 70 °C, P_c = 750 Psi). c Achiral dendritic crystal of PLLA (T_c = 60 °C, P_c = 1000 Psi). The scale bar lengths of (a–c) are 250 μm. d, e Electronic circular dichroism (ECD) and ultraviolet-visible (UV–vis) absorption spectra of PDLA and PLLA in Acetonitrile (AcCN) solution, PDLAcd means the CD signal of PDLA solution, PDLAabs means the UV-vis absorption signal of PDLA solution. The unit mdeg means millidegree which equals to 0.001 degree. f–i Vibrational circular dichroism (VCD) and corresponding Fourier transform infrared (FT–IR) absorption spectra of crystalline PLLA solid sample, the sample number represents the crystallization conditions, for example, 50–1000 means a crystallization temperature of 50 °C and a CO₂ pressure of 1000 Psi. The Y-axis of (f) and (h) means the difference in absorption of left-handed and right-handed polarized light by samples.

From Fig. 2d, PLLA and PDLA solutions show a positive and a negative Cotton effect at 210 nm, respectively. According to Ho’s research^20,21, the ECD signals result from the interaction of chiral entities in chiral polymers. In this way, the absolute configurations of chiral monomers in PLAs can be identified. Here, PLLA was selected to be treated under high-pressure CO₂. From Fig. 2f, g, the three PLLA crystalline samples show a similar split-type Cotton effect with an inflection point at 1760 cm⁻¹ corresponds to the characteristic absorption of the C=O stretching motion of the ester group in PLLA, whose electric transition moment perpendicular to the helical axis of the chiral PLLA chain. Based on the coupled oscillator model and the signatures of the split-type Cotton effect in the VCD spectra²⁰, the Cotton effect in the three crystalline PLLAs is identified as negative chirality. Namely, the helical conformations of the three crystalline PLLAs are left-handed. The induced VCD signals of the absorption bands of the C–O–C vibration ranged from 1000 to 1250 cm⁻¹ (Fig. 2h, i) also show the same trend and point to a left-handed helical conformation of PLLA chains²¹. These results mean that the three PLLA samples shown in Fig. 2a–c with different morphology chirality have the same helical conformation chirality of molecular chain in crystal. So, the reversal and disappearance of macroscopic spiral chirality of PLLA crystals are not decided by the helical chirality of crystalline chains.

Hierarchical chirality transfer mechanism

At present, four reasons are mainly considered as the causes of polymer growth deviating from the crystallographic direction²²: rhythmic growth²³, self-induced compositional field²⁴, successive screw dislocations²⁵, and unbalanced stresses on fold surface^26,27. In Fig. 3a, b, bamboo-like dendrites appear in both the left-handed and right-handed spiral crystals of PDLA. The bamboo-like structure is formed by rhythmic growth²⁸, so rhythmic growth is not a factor determining the spiral chirality of PDLA. In addition, in Fig. 3c–e, no matter which kind of spiral chirality, similar melt fields are self-induced by the crystal growth at the front of the dendrites. These self-induced melt fields are all shown as melt enrichment states, providing a high concentration of melt supply for crystal growth. This means that the self-induced field is also not the determining factor for the spiral growth of dendritic crystals.

Fig. 3: Microscopic morphology and structural characteristics of PLA crystal.

a, b Bamboo-like characteristics formed by the rhythmic growth of left-handed and right-handed spiral PDLA crystals. c–e Melt enrichment zones at the edges of three crystal morphologies imaged by the white light interferometer (WLI). f–i Atomic force microscopy (AFM) height images of Bi-layered lamellae originated from self-induced nucleation and spiral dislocations, as well as their height curves at the red dashed line. j, k Height images and peak force error images of multi-layer lamellae. The rotation angle between the top lamella and the bottom lamella is marked in (k).

The PLA dendrites are composed of multilayer stacked lamellae (Fig. 3f–i), Some of them (Fig. 3f) originated from self-induced nucleation²⁹, and others (Fig. 3g) from screw dislocation^30,31. In Fig. 3g, there are two screw dislocations with opposite helical directions in the same dendrite. Therefore, there is no one-to-one correspondence between the helical direction of screw dislocations and the dendritic spiral direction. When the layer number of stacked lamellae increases, the multi-layered structure will be rotated³² and the azimuth angle between the top and bottom layer will be deflected by 1°–3° (Fig. 3j, k). This cumulative deviation will only be reflected in the upper layer of the multi-layer structure; however, the growth direction of dendrites is determined by the basal layer lamellae. From Fig. 4a, the bending of multi-layer layers originated from screw dislocations or self-induced nucleation lags the basal layer lamellae. The growth direction of the upper layer crystals and the entire dendrite is dominated by the basal layer crystal. The research of Shen et al.³³ also supports this viewpoint.

Fig. 4: Crystal structure of PDLA spiral dendrites.

a AFM height image of left-handed spiral crystals (T_c = 50 °C, P_c = 1250 Psi). b AFM height image of right-handed spiral crystal (T_c = 70 °C, P_c = 1000 Psi). c, d AFM height image of non-spiral crystal (T_c = 60 °C, P_c = 1500 Psi). e–h Height curves at the lines in (a–d), separately. Two-dimensional (2D) Grazing incidence wide-angle X-ray scattering (GIWAXS) patterns of PDLA crystal generated at (i) T_c = 60 °C, P_c = 0 Psi, (j) T_c = 50 °C, P_c = 1250 Psi, (k) T_c = 60 °C, P_c = 1500 Psi, and (l) T_c = 70 °C, P_c = 1500 Psi, respectively. m One-dimensional (1D) GIWAXS data. n Transmission electron microscopy (TEM) images of PDLA crystal generated at T_c = 60 °C, P_c = 1250 Psi. o One-heating differential scanning calorimetry (DSC) data of PLLA samples (Number average molecular weight M_n = 6452 g mol⁻¹, Polydispersity Index PDI = 1.58) crystallized in high-pressure CO₂, the sample number represents the crystallization conditions, for example, 50–500 means a crystallization temperature of 50 °C and a CO₂ pressure of 500 Psi.

After excluding rhythmic growth, self-induced field, and screw dislocation mentioned above, the unbalanced stress on the fold surface becomes the possible factor inducing the spiral growth of the basal layer of PDLA spiral dendrites. The unbalanced stress is caused by the differential congestion on the fold surface. Based on Fritzsching’s research³⁴, amorphous chain segments can cause density anomalies on the lamellae surface and further induce chain tilt. From Fig. 4b, f, it is obvious that the two lateral sides of the lamellar crystal of the right-handed spiral crystal exhibit different tilt angles, and Supplementary Fig. 16b, d shows another similar example. This indicates that the molecular chains arranged along the thickness direction of lamellar crystal are tilted. On the surface of polymer lamellar crystal, cilia (chain ends), tie molecules, and folding chains constitute the amorphous layer. These disordered chain segments need more space than the ordered crystalline chain which will cause density anomalies of the crystal surface, resulting in surface stress. Surface stress induces chain tilt^33,35, which in turn leads to an asymmetric growth on the lateral side of the lamellar crystal, finally resulting in the formation of bent dendrites with spiral morphology.

However, there is no obvious difference between the two lateral sides of lamellar crystal in the left-hand spiral dendrites formed by PDLA, and the molecular chains are not significantly tilted (Fig. 4a, e and Supplementary Fig. 16a, c). So, the chiral reversal may no longer be attributed to surface stress or chain tilt. Figure 4i–l shows the 2D GIWAXS images of PDLA crystals. It shows that crystals of all the samples have a preferred orientation in the in-plane direction, and the lamellar crystals are mainly flat on the substrate. This is consistent with the AFM results shown in Figs. 3 and 4 and Supplementary Fig. 15. Figure 4m shows 1D GIWAXS data of all samples. We found that β crystals appeared in the PDLA crystal samples in addition to the common α and α’ (δ) crystals. For samples crystallized at 60 °C under atmospheric normal pressure, the characteristic peaks mainly appear at 16.4° and 18.9°, corresponding to the (200)/(110) crystal face and (203) crystal face of α’ (δ) crystal type, respectively³⁶. For the right-handed spiral dendrite formed at 70 °C under 1500 Psi pressure, the characteristic peaks mainly appear at 16.8° and 19.1°, corresponding to the (200)/(110) crystal face and (203) crystal face of α crystal respectively³⁶. For the left-handed spiral dendrite samples crystallized at 50 °C at 1250 Psi, the characteristic peaks mainly appear at 17.1° and 19.6°, corresponding to the (200) crystal face and (203) crystal face of β crystal, respectively^37,38.

Figure 4n shows the TEM diffraction pattern of the left-handed spiral dendrite of PDLA. Two features can be observed from the pattern: (1) The diffraction spots are hexagonally symmetric, indicating that the unit cell is trigonal; (2) Two landmark reflections are indexed 120 and 210 and are located between the 300 and 030 reflections (Supplementary Fig. 17). These two features indicate that the observed PLA crystals are β crystals^39,40,41.

In addition, Fig. 4o presents the DSC curves of PLLA crystalline samples during the first heating process. From the curves, the main melting peaks of the 4 curves occur around 155 °C. However, in the 50–1500 and 60–1500 curves, the second melting peak appears around 147.8 °C. Based on the crystal transition mechanism discussed by Ru et al.³⁸, the relatively low temperature melting peak can be attributed to β crystals. As the temperature increases, the β crystals melt and recrystallize to form α crystals, followed by the melting peak of α crystals at a higher temperature. The phenomenon of dual melting peaks again proves the presence of β crystal.

The above findings suggest that the shift in spiral chirality within vortex-like dendrites coincides with a transformation in crystal structure. Specifically, the right-hand spiral PDLA dendrites correspond to the α crystal form, while the left-hand spiral PDLA dendrites correspond to the β crystal form. PDLA snowflake dendrites lacking spirals contain a mixture of both α and β crystals (or α’, α, and β). Consequently, the direct growth of achiral snowflake crystals arises from the neutralizing of opposing spiral directions corresponding to the two distinct crystal forms.

To further validate the correlation between the spiral chirality of PLA dendrites and the β crystal form, Fig. 5 illustrates the crystallization outcomes of PLLA with varying molecular weights under conditions of 50 °C and 1000 Psi. As the molecular weight increases, the proportion of β crystals also increases. In the PLLA1 sample (M_n = 577 g mol⁻¹), predominantly α crystals are formed with only a small fraction of β crystals exhibiting weak bending and almost no discernible spiral morphology. Conversely, in PLLA2 and PLLA3 with higher molecular weights, the presence of β crystal peaks becomes pronounced. Correspondingly, the crystal morphology displays distinct counterclockwise bent dendrites, representing the right-handed spiral morphology. This evidence further substantiates the correlation between chiral reversal and the β crystal structure.

Fig. 5: The effect of molecular weight on the crystal structure and spiral morphology of PLLA crystalized in 50 °C and 1000 Psi CO₂.

a GPC data. b 1D GIWAXS data. c–e 2D GIWAXS of PLLA1, PLLA2, and PLLA3. Crystal microscopy images of PLLA1, PLLA2, and PLLA3, and the lengths of scale bars of three microscopy images (f–h) are 250 μm.

Furthermore, upon comparing Fig. 5d, e, it is observed that the PLLA2 sample predominantly displays in-plane diffraction spots, with dendrites primarily composed of flat-on lamellar crystals. In contrast, the PLLA3 sample exhibits diffraction rings, indicative of a relatively disordered arrangement of lamellar crystals within the sample. This arrangement includes not only flat-on lamellae but also edge-on or inclined lamellar crystals. Interestingly, despite the disordered arrangement of crystallites, there is no discernible impact on the curvature of PLLA dendrites. This implies that while the disorderly arranged lamellar crystals may cover and conceal the underlying lamellar crystals, they do not influence the bending direction of the underlying lamellar crystals.

Entropy effect of high-pressure CO₂

High-pressure crystallization experiments have demonstrated that PLA can generate β crystal under mechanical pressure exceeding 100 MPa and the temperature near the melting point (160–180 °C)^38,42. A question arises: how can CO₂, with a pressure of about 10 MPa, induce PLA β crystal at low temperatures (50–60 °C)? The Flory-Huggins binary interaction parameter χ of PLA and CO₂ indicates that CO₂ is a poor solvent for PLA⁴³. In addition, considering that the molecular weights of PLA in this study are lower than or close to the critical entanglement molecular weight⁴⁴ of PLA, the molten PLA molecule chain can be assumed as a rod-shaped molecule with high orientation entropy. According to Onsager theory⁴⁵ and entropy-induced phase transition theory⁴⁶, the thermal motion of poor solvents will provide entropy increase to compensate for the ordered arrangement (orientation entropy decrease) of rod-shaped molecules. In the crystallization process of the PLA/CO₂ system, the energy change induced by CO₂ mainly includes the adsorption energy between CO₂ and PLA, and the translational energy of CO₂ molecules desorbed from PLA crystal.

For PLLA/CO₂ systems, there are two specific interactions between CO₂ and electron-donating groups (carbonyl and ether) or some polymers with Lewis base properties⁴⁷. The interaction energy between CO₂ and carbonyl and ether (${E}_{{{{\rm{CO}}}}_{2}-{{\rm{carbonyl}}}}$, ${E}_{{{{\rm{CO}}}}_{2}-{{\rm{ether}}}}$) is approximately 2.2 kcal mol⁻¹ obtained by ab initio calculation⁴⁷. Here, we assume that the binding probabilities of CO₂ molecule with carbonyl and ether of PLA molecular chain are equal, and then the CO₂ adsorption energy change $\triangle {G}_{{\mbox{a}}}$ that needs to be overcome for 1 mol PLA monomer to form crystals is approximately as:

$$\triangle {G}_{{\mbox{a}}}=\frac{1}{2}{n}_{1}\left({E}_{{{\mbox{CO}}}_{2}-{\mbox{carbonyl}}}+{E}_{{{\mbox{CO}}}_{2}-{\mbox{carbonyl}}}\right)$$

(1)

where ${n}_{1}$ is the molar number of CO₂, which can be calculated by ${n}_{1}=\left(72{w}_{1}\right)/\left[44\left(1-{w}_{1}\right)\right]$. ${w}_{1}$ is the mass fraction, which can be calculated by Sanchez–Lacombe (S-L) equation of state^48,49, and the calculation process is given in detail in the Supplementary Discussion. The translational energy ${\triangle G}_{{{{\rm{t}}}}}$ of CO₂ desorbed from PLA crystal can be calculated by ${\triangle G}_{{{{\rm{t}}}}}={n}_{1}({G}_{{{{\rm{m}}}},{{{\rm{t}}}}}-{G}_{{{{\rm{m}}}},{{{\rm{t}}}}}^{{{{\rm{\theta }}}}})$, where ${G}_{{{{\rm{m}}}},{{{\rm{t}}}}}$ means the molar translational energy of CO₂, ${G}_{{{{\rm{m}}}},{{{\rm{t}}}}}^{{{{\rm{\theta }}}}}$ means the molar translational energy of CO₂ at ideal state ($P$ = 101,325 Pa and $T$ = 273.15 K). Based on Sackur–Tetrode formula⁵⁰,${G}_{{{{\rm{m}}}},{{{\rm{t}}}}}$ of CO₂ can be calculated as:

$${G}_{{{{\rm{m}}}},{{{\rm{t}}}}}= {H}_{{{{\rm{m}}}},{{{\rm{t}}}}}-T{S}_{{{{\rm{m}}}},{{{\rm{t}}}}}=\frac{5}{2}{RT}-{RT}\left({{\mathrm{ln}}}\frac{{q}_{{{{\rm{t}}}}}}{{N}_{{{{\rm{a}}}}}}+\frac{5}{2}\right) \\= -{RT}\left\{{{\mathrm{ln}}}\left[\frac{{\left(2\pi {mkT}\right)}^{\frac{3}{2}}}{{{N}_{{{{\rm{a}}}}}h}^{3}}{V}_{{{{\rm{m}}}}}\right]\right\}$$

(2)

where ${H}_{{{{\rm{m}}}},{{{\rm{t}}}}}$ is the molar translational enthalpy, ${S}_{{{{\rm{m}}}},{{{\rm{t}}}}}$ is the molar translational entropy, $R$ is the gas constant, ${N}_{{{{\rm{a}}}}}$ is the Avogadro constant, ${q}_{{{{\rm{t}}}}}$ is the translational partition function, $h$ is the Planck constant, $k$ is the Boltzmann constant, and the molar volume of CO₂ can be estimated as ${V}_{{{{\rm{m}}}}}=\frac{{RT}}{P}$. The mass $m$ of CO₂ molecules can be determined by $m=M/{N}_{{{{\rm{a}}}}}$, where $M$ is the molar mass of CO₂. The values of parameters in the above model and references are listed in Supplementary Table 1.

According to density functional calculations, the energy of monomer in the unit cell of PLA β crystal exceeds that of PLA α crystal by 0.7 kcal mol^−1 51. Figure 6a, b shows the calculation results of the above equations. Under low temperature and high-pressure conditions, the difference between the translational energy and the adsorption energy (${\triangle G}_{{{{\rm{t}}}}}-{\triangle G}_{{{{\rm{a}}}}}$) caused by CO₂ is larger than the energy gap required to generate PLA β crystal, theoretically (Fig. 6a). This result proves possibility that high-pressure CO₂ induces PLA β crystal at low temperatures. Furthermore, the energy contribution of CO₂ translational entropy ${T\triangle S}_{{{{\rm{t}}}}}$ is more significant than the adsorption energy change ${\triangle G}_{{{{\rm{a}}}}}$ and the translational enthalpy ${\triangle H}_{{{{\rm{t}}}}}$ of CO₂ (Fig. 6b). This result means that translational entropy increase of CO₂ plays a key role in the formation of PLA β crystal.

Fig. 6: The energy calculation results and control of chirality transfer of PLA in high-pressure CO₂.

a Calculation results of energy contribution of CO₂ to PLA crystallization. b Adsorption energy change ${\triangle G}_{{{{\rm{a}}}}}$, translational enthalpy change ${\triangle H}_{{{{\rm{t}}}}}$ and energy contribution of translational entropy ${T\triangle S}_{{{{\rm{t}}}}}$ caused by CO₂ at a crystallization temperature of 50 °C, where ${\triangle H}_{{{{\rm{t}}}}}$ and ${\triangle S}_{{{{\rm{t}}}}}$ were calculated by ${\triangle H}_{{{{\rm{t}}}}}={n}_{1}({H}_{{{{\rm{m}}}},{{{\rm{t}}}}}-{H}_{{{{\rm{m}}}},{{{\rm{t}}}}}^{{{{\rm{\theta }}}}})$ and ${\triangle S}_{{{{\rm{t}}}}}={n}_{1}({S}_{{{{\rm{m}}}},{{{\rm{t}}}}}-{S}_{{{{\rm{m}}}},{{{\rm{t}}}}}^{{{{\rm{\theta }}}}})$, respectively. The superscript ${{{\rm{\theta }}}}$ means the ideal state whose $P$ = 0.1 MPa and $T$ = 273.15 K. c Changing the CO₂ pressure from 1250 to 750 Psi at 60 °C induces a reversal in the spiral chirality of PDLA crystals. d Simultaneously changing the temperature from 50 to 70 °C and adjusting the CO₂ pressure from 1300 to 1000 Psi induces a change in the spiral chirality of PDLA crystals.

Given that high-pressure CO₂ serves as the primary factor inducing chiral reversal in PLA dendrites, we can exert control over the chirality and morphology of PLA dendrites through manipulation of CO₂ pressure. As depicted in Fig. 6c, d, the chirality of PDLA dendrites undergoes immediate reversal upon alterations in CO₂ pressure or crystallization temperature, resulting in the S-shaped or Z-shaped dendrites with opposite chirality. Additionally, Supplementary Figs. 18–20 provide further insights into the real-time changes in dendrite chirality and crystal morphology. Thus, the crystal morphology can be flexibly programmed via high-pressure CO₂, giving a straightforward and cost-effective approach to regulating the pattern of crystal superstructure (spiral structure with the growth in three dimensions) and crystal chirality of PLA films.

Discussion

In the crystallization process of PLA, molecular chirality indeed plays a crucial role in the morphology chirality of crystals. As depicted in Fig. 1 and Supplementary Figs. 11 and 12, PLLA and PDLA crystals exhibit mirror-symmetric spiral chirality. However, while molecular chirality is significant, it is not the sole determinant. Through adjustments in temperature and CO₂ pressure settings, homochiral PLA was manipulated to generate two vortex-like dendrites with opposing chirality, as well as a snowflake-like dendrite without spiral chirality. These three crystal morphologies correspond to chiral amplification, chiral reversal, and chiral symmetry restoration, respectively. These findings underscore that there is no strict one-to-one correspondence between molecular chirality and crystal morphology chirality. This assertion is further validated by the VCD results illustrated in Fig. 2.

Additionally, this research, as depicted in Figs. 3–5, reveals that the difference in macroscopic morphology chirality is intricately linked to variations in crystal structure. The occurrence of chiral amplification (left-handed spiral dendrites of PLLA or right-handed spiral dendrites of PDLA) is commonly associated with PLA α crystals, while the manifestation of chiral reversal (right-handed spiral dendrites of PLLA or left-handed spiral dendrites of PDLA) directly correlates with the formation of β crystals.

The curved growth of dendritic α crystals, attributed to chain tilting induced by surface stress of lamellar crystals, has been widely acknowledged^{26,27,33,34,35}. The presence of tilted lamellae, as depicted in Fig. 4f and Supplementary Fig. 16d, supports this perspective. However, in the case of crystals undergoing chiral reversal, we did not observe any obvious tilted lamellae, as illustrated in Fig. 4e and Supplementary Fig. 16c. This suggests that the curved growth mechanism of crystals with a β crystal structure may differ from that of α crystals.

The PLA β crystal structure is characterized as a frustrated structure featuring three trigonal symmetric helical chains arranged in a North-South-South arrangement within the unit cell^39,40. In this arrangement, one side of the unit cell’s (1 1 0) plane appears relatively flat, while the other side of the layer exhibits a distinct notch between the two helices oriented in the South direction. Deposition of an initial stem on the notched side of the layer requires lower energy consumption compared to the flat side. Lotz and Okihara^41,52 proposed that this asymmetric topology results in an imbalance in the growth rate of the two symmetric crystal planes. This imbalance could potentially explain the curved growth observed in PLA dendrites composed of β crystals.

To generate a non-spiral (achiral) snowflake crystal of homochiral PLA, two key conditions should be met: the coexistence of α and β crystals, and the presence of a highly ordered structure (Supplementary Fig. 13).

• (1) Coexistence of α and β crystals: based on statistical data from the Cambridge Structure Database regarding crystalline substances and crystal structures, it’s established that a homochiral enantiomer cannot produce an achiral crystal. Furthermore, there is no corresponding group space⁵³. Considering the GIWAXS results in Fig. 4m, we propose that the achiral snowflake crystal does not consist of a single crystal form but rather arises from chiral neutralization in the assembly of α and β crystals.

• (2) Highly ordered structure: the role of CO₂ as a solvent influences the growth dynamics of crystals. High-pressure CO₂ assists polymer chains in acquiring more free volume, reducing interaction with adjacent chain segments, or minimizing entanglements with other chains. Consequently, polymer molecular chains gain enhanced mobility, facilitating the formation of highly ordered crystal structures. These highly ordered structures maintain the crystallographic characteristics of PLA crystals by promoting regularity and symmetry. It’s worth noting that crystals formed under high pressure may also consist of α (or α') and β crystals, as evidenced by the 1D GIWAXS data of the 60–1500 sample (Fig. 4m).

Finally, we present a hierarchical transfer mechanism illustrating the transfer of PLA chirality from the monomer level to the helix chain, then to the crystal structure, and ultimately to macroscopic assembly, as depicted in Fig. 7.

Fig. 7: Hierarchical transfer mechanism of PLA chirality from monomer to helix chain, to crystal structure, and to macroscopic assembly.

This schematic uses PDLA as an example to illustrate the hierarchical transfer process of chirality in high-pressure CO₂, PLLA has the opposite chiral transfer process but the same transfer mechanism. The drawing style of PLA β crystal is referenced from the literature ⁴¹. It should be pointed out that the vertical helical chains in PLA β crystals, which are formed under high-pressure CO₂, indicate that they are not obviously tilted compared to the helical chains in α crystals.

The transition from homochiral molecules to achiral crystals signifies the restoration of chirality symmetry during the ordering process. Identifying this phenomenon in the crystallization of chiral inorganic compounds could pose challenges. However, in chiral polymers, the crystallization process may coincide with surface stress, chain tilt, and the formation of frustrated structures due to the flexible nature of the polymer’s long-chain molecules. The existence of metastable ordering behaviors and polymorphism of crystal structures, provides opportunities for chiral polymers to undergo diverse modes of hierarchical transfer of chirality.

The study offers two potential applications: firstly, the growth direction of PLA dendrites can be flexibly controlled using high-pressure CO₂, enabling programmable crystal morphology. This method, easily implemented in both laboratory and factory settings, realizes the simulation predictions by Gránásy et al.¹⁸ without the need for substrate-embedded oriented particles, external electromagnetic field rotation, or laser pulse angular momentum control. It provides a straightforward and cost-effective approach to controlling the condensed state of crystal superstructure in polymer films.

Secondly, as PLA with β crystal exhibits superior tensile strength and heat resistance compared to α crystal, this study presents a possible method to produce PLA β crystal using low-cost and environmentally friendly CO₂. Figure 5 and Supplementary Figs. 21–23 further confirm that PLA β crystal formation is achievable not only in low molecular weight or thin film samples but also in high molecular weight PLA and block samples under high-pressure CO₂. This suggests the potential scalability of the high-pressure CO₂ crystallization method. Compared to methods involving mechanical pressures exceeding 100 MPa^38,42, utilizing the entropy effect of CO₂ enables the generation of PLA β crystal at temperatures of 50–60 °C and pressures of 5–10 MPa, significantly reducing energy consumption and production equipment costs associated with PLA β crystal preparation.

Methods

Materials

CO₂ with a purity of 99.99% was supplied by Jinan Deyang Special Gas Co., Ltd., China. PLA samples (Ester-capped PDLA and Ester-capped PLLA) were provided by Jinan Daigang Biomaterial Co., Ltd., China. The original samples were firstly purified two times by reprecipitation method using chloroform as solvent and methanol as precipitant and then purified two times by solution suction filtration using excess ethyl acetate as solvent. The polymer was dried for 10 days at 50 °C in a vacuum drying oven to remove any moisture before formal experiments. By using Gel Permeation Chromatography, the number-average molecular weight, and polydispersity of the purified samples were determined as: for PDLA, M_n ≈ 1700 g mol⁻¹, PDI = 1.18; for PLLA, M_n ≈ 1750 g mol⁻¹, PDI = 1.21. The melting points of purified samples were measured in the atmosphere by the secondary heating process, T_m (temperature of the melting peak) of PDLA = 122.8 °C, and T_m of PLLA = 124.7 °C. The original data of the abovementioned measurements are given in Supplementary Fig. 1. Furthermore, the specific rotation degree of purified samples was measured by a Specific Optical Rotation instrument, and the values were determined as [α] = 152.2° for PDLA and [α] = 151.3° for PLLA. This means that the two types of PLA samples have high purity. Other material information involved in this study will be annotated in the text.

Film preparation

This research used the Silicon (Si) wafer as the substrate to prepare PLA films. The Si wafer was cleaned twice by a plasma cleaner (Model: CPC-A) to remove any organic contamination. The gas pressure, power, and working time of the plasma cleaner were set as 50 Pa, 200 W, and 5 min, respectively. This treatment is to improve the polymer film stability and avoid dewetting phenomenon in high-pressure CO₂. The PLA sample was dissolved in Dichloromethane, then the PLA solution was dropped onto a hydrophilic Si wafer at room temperature. After solvent evaporation, PLA film can be obtained, and the film thickness was controlled by the concentration and volume of PLA solution.

High-pressure visualization

We have independently developed a high-pressure visualization system. Supplementary Fig. 2 shows the schematic diagram of the control system of the high-pressure system. To observe the opaque sample, e.g., the PLA film coated on the Si wafer of this research, we have replaced the transmission light source with a coaxial light source, which is a cold point light source that can minimize the impact of the light beam on the sample. The camera used in this system is Basler acA2440-75uc. Supplementary Fig. 3 shows the photo and schematic diagram of the sample cell of the system. Compared to the old sample cell structure, the material of the sample cell has been replaced with 7000 series aluminum alloy. We designed a set of conformal cooling channels, and a large number of turbulence columns with a diameter of 2 mm have been arranged inside the channel, further improving the heat exchange of the sample cell. The volume of high-pressure fluid in the sample cell has been reduced from 9 ml to 1 ml to avoid critical opalescence, further improving the pressure control accuracy of the sample cell. The sample cell was manufactured by using 3D printing technology (the selected laser melting equipment is RenAM 500E). The high-pressure visualization system can accurately control the gas pressure within the range of 0.1–40 MPa, and the control accuracy is ±0.02 MPa.

Crystallization in CO₂

The PLA film coated on a Si wafer was installed in a sample cell of the visualization system to implement melt isothermal crystallization. The sample cell was heated to 150 °C from room temperature. Under the action of a high-pressure pump connected with the sample cell, CO₂ pressure was held at a setting pressure (250–1500 Psi) with a fluctuation of 1 Psi in the whole crystallization process. After a holding time of 3 min with constant temperature and constant pressure, the cell was cooled down to the crystallization temperature T_c by using anhydrous alcohol with a temperature of −60 °C. Afterward, the system temperature was maintained at T_c to induce crystallization. After the crystallization was finished, the vent valve was opened to expel CO₂ from the cell. Meanwhile, the cell was quickly cooled to 20 °C by using the absolute ethyl alcohol with a temperature of −60 °C. Finally, the cell was opened, and the sample was taken out for subsequent characterization.

Differential scanning calorimeter (DSC)

The glass transition temperature and melting point of PLA were measured using a NETZSCH STA 4495F5 apparatus. The samples with an average weight of 10 mg are quenched from 190 °C to −20 °C at a rate of 60 °C min⁻¹ and reheated at a rate of 10 °C min⁻¹ under a nitrogen atmosphere.

Gel permeation chromatography (GPC)

GPC characterizations were conducted using the Waters GPC system with a 1525 binary HPLC pump, and a Waters 2414 refractive index detector was used. PLA solutions were prepared in tetrahydrofuran (THF) and prefiltered on a filter disk (hydrophobic polytetrafluoroethylene, 0.45 μm pore size) before injection. All measurements were performed using THF as the carrier solvent with a flow rate of 1.0 ml min⁻¹ at 30 °C. Standard monodispersed PS (Shodex standard, Kawasaki, Japan) were used for calibration.

Specific optical rotation (SOR)

The specific rotation value of PLA was measured at 25 °C in chloroform at a concentration of 0.8 g dL⁻¹ with a Rudolph spectropolarimeter (Autopol IV-T) using a quartz cell of 100 mm path length at a wavelength of 589 nm.

White light interferometer (WLI)

The altitude and morphology of polymer film were measured by WLI (Veeco NT9300) at atmospheric condition, which can take the place of AFM when the size of the observed object is bigger than 50 μm.

Atomic force microscopy (AFM)

AFM was used to measure the thickness and morphology of crystals at atmospheric conditions on Digital Veeco Instruments, operating with silicon cantilevers in the PeakForce Tapping mode. The peak force error image was obtained to reflect the phase morphology of the polymer sample.

Transmission electron microscopy (TEM)

TEM (JEOL JEM-2100F) was used to obtain selected area electron diffraction (SAED) patterns of crystallized samples. The operation voltage was 200 kV, beam current was 105 μA. To prepare the sample for characterization, a drop of 10% HF solution was dropped onto the sample surface, and the films were separated from Si wafer surface. The PLA film was transferred to a copper grid for TEM observations.

Grazing incidence wide angle X-ray scattering (GIWAXS)

GIWAXS measurements were performed at room temperature on Xeuss 2.0, Xenocs, France. The incident angle was set as 0.15°. Xeuss 2.0 is equipped with a copper target, the light tube power is 30 W, and the wavelength is 1.54189 Å. The detector is Pilatus 3R 300 K and the single pixel size is 172 μm. The sample detector distance was 150 mm, and the exposure time was 300 s. Data were analyzed with the software Fit 2D.

Electronic circular dichroism (ECD)

UV−vis absorption and CD spectra were acquired using a Chirascan V100 spectrometer. Solution samples were placed in a cylindrical quartz cell. The concentration of the solution was 0.1 mg ml⁻¹ for PLA in AcCN.

Vibrational circular dichroism (VCD)

FT-IR absorption and VCD spectra were acquired using a JASCO FVS-6000 spectrometer. 0.2 mg PLLA solid samples were crystallized between two Calcium fluoride (CaF₂) wafers (Diameter = 10 mm) to avoid dewetting of PLA film at melting temperature.