Introduction

Organic electronic devices based on polymeric semiconductors have attracted significant interest from both academic and industrial sectors over the last three decades. The primary attraction of these devices is the promise of conformable and flexible form factors, which are incompatible with more traditional inorganic technologies. Furthermore, these devices are amenable to large-scale fabrication through cost-effective printing methodologies, such as roll-to-roll processes and ink-jet printing. During such processing procedures, the solvent evaporates rapidly (in a few seconds), which often hinders the arrangement of polymer chains into crystals. Consequently, active layers of organic electronic devices frequently contain large amounts of non-crystallized, disordered glassy regions. For instance, the active layers of OLEDs, which are utilized in billions of cell phone and television displays, are composed exclusively of glassy phases1; and many high-performing donor:acceptor blends used in organic photovoltaic cells consist predominantly of glassy phases with residual or no crystalline phases present in the devices2,3,4. Consequently, glassy phases must play a pivotal role - if not often dominate - the function of many organic electronic devices. However, little attention is usually paid to glassy phases of organic semiconductors5,6,7,8,9. For instance, while numerous processing and post-processing treatments for organic electronic devices are focused on optimizing materials’ crystalline phases (e.g., enhancing the degree of crystallinity, increasing the size of crystals, reducing the crystal-lattice disorder, etc.)10,11,12, the optimization of the thermodynamics of the pervasive glassy phases remains rather unexplored. While there have been several attempts to control the glass transition temperature, Tg, in polymeric semiconductors through geometrical confinement13, chemical structure (such as Mw)14 or by multicomponent mixing15, control of the glass thermodynamic state itself for a given material and geometry has thus far been elusive.

In the present manuscript, we show the significant potential of manipulating the thermodynamic state of glassy phases to regulate the optoelectronic properties of polymer semiconductors. It is demonstrated that the thermodynamic state - and consequently the solid-state microstructure - of the glassy phase of a polymer semiconductor, in this case poly(9,9-di-n-octylfluorenyl-2,7-diyl) (PFO), can be readily manipulated via (i) different vitrification kinetics and (ii) the glass formation from different liquid states, e.g., an isotropic liquid and a liquid crystalline mesophase. A comprehensive examination of both strategies is provided in Fig. 1 panel a and Supplementary Fig. S1 of the Supplementary Information. In addition, we demonstrate that the optical properties of polymer semiconductors, specifically the photoluminescence behavior, can be readily controlled via fine-tuning the structure of glassy phases, which opens up a plethora of possibilities for the design of more efficient post-processing treatments for optoelectronic devices.

Fig. 1: Scheme of glass thermodynamics.
Fig. 1: Scheme of glass thermodynamics.The alternative text for this image may have been generated using AI.
Full size image

((Upper panel) Typical temperature dependence of the enthalpy or volume for a glass cooled at high (β1) and low cooling rate (β2), and (lower panel) corresponding first derivatives.

Results

Assessing the glass thermodynamic state

The thermodynamic state of a material defines its condition by a set of macroscopic properties that describe its energy and phase. These properties typically include volume, entropy, enthalpy, internal energy, etc. For equilibrium condensed phases like crystals and liquids, the set of thermodynamic properties at a given temperature and pressure is well-defined. In contrast, in non-equilibrium phases like glasses, the thermodynamic state depends on their previous history, specifically on the kinetic path followed to form them. Fig. 1 schematizes the temperature-dependent thermodynamic properties at constant pressure (upper panel) and their derivatives (lower panel) for an amorphous material undergoing a cooling/heating cycle. When a liquid is cooled at a fast rate (β1), it deviates from the equilibrium liquid state at a certain temperature, addressed as vitrification or glass transition temperature, Tg, forming a low-density, high-energy glass. Conversely, when the cooling rate, β2, is substantially reduced, the system has more time to rearrange and, therefore, it retains the equilibrium state down to lower temperatures. As a result, the system vitrifies in a denser glass with lower enthalpy. During subsequent heating, the thermodynamic properties follow a non-equilibrium line until the material gains sufficient molecular mobility to approach the equilibrium liquid state, undergoing the glass to liquid transition. The latter is accompanied by an abrupt increase in enthalpy, H, and volume, V, more pronounced in low-energy glasses, known as equilibrium recovery. For differential properties (Fig. 1, lower panel), such as the specific heat capacity, cp = ∂H/∂T∣P, the coefficient of thermal expansion, α = 1/V∂V/∂T∣P, the glass to liquid transition is observed as an abrupt step from the glass to the liquid line, on top of which the rapid recovery of equilibrium of low energy glasses is observed as a superimposed peak (see scheme in Fig. 1).

To assess the thermodynamic state attained by a certain glassy phase after a given thermal protocol, we have resourced to the use of the so-called fictive temperature, Tf. For glassy systems, this concept, introduced by Tool16, is typically used as a metric to describe the glass thermodynamic state. In particular, Tf is obtained from the intersection of the glass line drawn from the thermodynamic state of a given glass to the extrapolated line of the melt (see Fig. 1, upper panel). The Tf defines the glass thermodynamic state, that is, its enthalpy, entropy or density, thereby embodying a meaningful parameter to understanding glass behavior. A high Tf corresponds to a less dense glass with higher enthalpy and entropy, whereas a low Tf is the signature of a dense and stable glass exhibiting lower internal energy.

To unveil the thermodynamic state of a glass via the Tf, differential techniques such as calorimetry are commonly used, which allows measuring the cp. The typical protocol to determine the Tf relies on performing a calorimetric heating scan performed at the end of glass processing, that is to say, the kinetic path used to obtain it. By assuming that the glass thermodynamic state remains unchanged until the glass-to-liquid transition, Tf can be calculated using the Moynihan method or area matching method17. From the definition of Tf, this approach entails the equality of the following integrals:

$${\int }_{{T}_{f}}^{{T}_{high}}({c}_{p,l}-{c}_{p,g})dT={\int }_{{T}_{low}}^{{T}_{high}}({c}_{p}-{c}_{p,g})dT,$$
(1)

where Tlow and Thigh are temperatures far below and far above the glass to liquid transition, respectively, and cp,l and cp,g are the liquid and glass specific heats, respectively. Graphically, the Tf, is determined when the area of the trapezoid formed between the liquid and glass lines (from Tf to Thigh) equals the area underneath the experimental specific heat scan and the glass line (from Tlow to Thigh).

In the present study, PFO was selected for investigation due to its propensity for straightforward vitrification from the liquid melt state18. A salient feature of this material is its demonstration of two distinct liquid states: an isotropic (ISO) liquid and a nematic (NEM) liquid, from which the formation of the glass can occur (see sketch in Supplementary Fig. S1 of Supplementary Information).

Photoluminescence v s glass thermodynamic state

Photoluminescence (PL) analysis was conducted as a proxy for the optoelectronic behavior of PFO glassy phases (see details in the Experimental Section). It is acknowledged that PL exhibits sensitivity to the molecular organization/arrangement (i.e., solid-state structure) of materials19,20. Consequently, it is considered a suitable optoelectronic phenomenon for reflecting structural differences, such as density differences, among glasses. For instance, the temperature-dependent PL of PFO can be used to probe thermal expansion in the material, as well as to monitor the glass, cold crystalline and melt phase transitions21. Another example would be the changes in the PL of PFO when varying the molecular weight of the polymer22. Furthermore, the PL spectra of PFO films with different solid-state microstructures have been the subject of extensive research23,24,25,26.

To evaluate the effect of the different glasses on the optical properties, we have prepared four samples formed by cooling from the isotropic (ISO) melt and the nematic (NEM) melt at 4000 K s−1 and at 0.01 K s−1, respectively. These samples exhibited Tf values of 333, 351, 323 and 343 K, respectively, as determined by fast scanning calorimetry (FSC) (see Methods for details regarding this technique) through the Moynihan method17 (Fig. 2 panel a). We have chosen to monitor the photoluminescence (PL) as a key optical property for OLEDs, which, moreover, is highly sensitive to microstructural changes. Figure 2, panel b provides a comparison of the photoluminescence spectra excited at 355 nm for four different 100 %-glassy PFO films. Each of these spectra corresponds to the average of ca 100 measurements on a film. The data sets presented herein demonstrate unequivocally a monotonic red-shift of the 0-0 PL band as the Tf of the glass is reduced (see Fig. 2 panel c). This outcome is consistent with glasses exhibiting a lower Tf having higher densities and thus a larger refractive index, which results in the shift of the 0-0 PL band to larger wavelengths due to stronger gas to solid shift19. This is consistent with the redshifts observed in the PL of polyfluorene upon application of hydrostatic pressure27,28, and with the blue shifts observed in the PL upon thermal expansion21. Besides the overall bathochromic shift, variation between the intensities of the 1-0 and 0-0 bands can also be observed, as well as differences in the intensity of the green band, ca 500 nm. These variations result in changes in the color coordinates, going from a purely blue emission towards a slightly greener gamut. Here, it is worth pointing out that during the PL measurements, we also investigated the homogeneity of the PL emission in the PFO glassy samples and the thermal/photo stability of the samples in relation to their PL response. As shown in Supplementary Fig. S3 of the Supplementary Information and its corresponding discussion, our results demonstrated that PFO samples are notably homogeneous and that glassy samples do not suffer from rejuvenation or any chemical or physical changes when exposed to the laser beam. Therefore, glasses exhibit the required stability to perform this analysis.

Fig. 2: Connection between glass thermodynamics state and PL.
Fig. 2: Connection between glass thermodynamics state and PL.The alternative text for this image may have been generated using AI.
Full size image

a Specific heat temperature scans and (b) photoluminescence (PL) spectra for spun cast 100 % glassy PFO films vitrified at 0.01 and 4000 K/s from the isotropic (ISO) and the nematic (NEM) melts, from which their Tfs are determined. c Position of the 0-0 PL transition plotted vs Tf.

Having provided compelling evidence that the thermodynamic state of the glassy phase has a direct influence on the optoelectronic behavior of polymer semiconductors, we start by focusing on how the thermodynamic state of the glassy phase can be manipulated. In this work, two approaches were used to achieve glasses with different Tf: (i) the variation of the vitrification rate and (ii) the vitrification of liquids/melts with different degrees of order, i.e., an isotropic liquid and a nematic mesophase.

Thermotropic landscape of PFO

To establish an experimental thermal protocol enabling systematic exploration of the strategies outlined above, it is first necessary to define the thermotropic phase landscape of PFO. This analysis allows identification of the temperature ranges corresponding to the isotropic melt, nematic melt, semicrystalline phase, and glassy state, as well as the corresponding phase transition temperatures. To this end, we subjected a PFO film to the thermal protocol presented in Fig. 3a18,29.

Fig. 3: Sampling PFO thermodynamic state.
Fig. 3: Sampling PFO thermodynamic state.The alternative text for this image may have been generated using AI.
Full size image

a Thermal protocol employed for the FSC experiments. b FSC heating traces (at 4000 Ks−1) following the isothermal step of 1 h at temperatures ranging from Ta = 153 to 473 K, in steps of 10 K. Endothermic peaks shadowed in blue, green and brown, indicated by corresponding arrows with the same color, correspond to the enthalpic relaxation of the glassy phase, the melting of crystals and the nematic-isotropic transition, respectively.

The material was initially heated to a temperature well above the nematic-isotropic transition (e.g., 723 K) to erase any prior thermal history. Subsequently, the samples were rapidly cooled at 4000 Ks−1 to a selected isothermal annealing temperature (Ta), varied between 153 and 473 K, and held at this temperature for 30 minutes. During the isothermal stage, the material evolves toward thermodynamic equilibrium by minimizing its free energy. When Ta lies within the temperature range corresponding to a non-equilibrium glassy state, physical aging of the glass is expected. If Ta falls within the supercooled liquid region, the system undergoes cold crystallization. In contrast, when Ta is located within the stability range of the nematic melt, the isotropic liquid transforms into an orientationally ordered nematic state. The structural evolution occurring at each Ta can be probed during the subsequent heating scan. Physical aging, crystallization, and nematic ordering during the isothermal treatment give rise to characteristic endothermic peaks upon reheating, reflecting the enthalpic changes associated with each transformation. The resulting data indicate that the isotropic glassy phase forms below 343 K, corresponding to the vitrification onset of PFO at a cooling rate of 4000 Ks−1. Between 343 and 383 K, supercooled PFO undergoes crystallization. The transition to the nematic disordered phase occurs in the range 383–483 K, while above 483 K the isotropic melt represents the thermodynamically stable phase.

PFO molecular mobility

Once the thermotropic phase behavior of PFO has been elucidated, we now focus on the production of PFO glasses with different Tf via vitrification of ISO and NEM melts at different cooling rates, ranging from 0.1 to 4000 K s−1. Before investigating vitrification kinetics, we first characterized spontaneous thermal fluctuations in both isotropic liquid and liquid crystal, specifically the cooperative α relaxation, which is known to play a prominent role in vitrification30,31,32. Hence, the information conveyed by this characterization is of utmost importance to unveil the molecular mechanisms underlying PFO vitrification both in the ISO and NEM states. To this aim, we have employed FSC implementing step-response protocols as detailed in the Methods section33,34. By applying a small temperature change, which guarantees linearity of the perturbation, this methodology allows determining the linear thermal susceptibility, conveying information on the typical time scale of spontaneous thermal fluctuations. Figure 4, panel b shows the temperature and frequency dependence of the reversing specific heat: \({c}_{p,rev}={({c}_{p}^{{\prime} 2}+{c}_{p}^{{\prime\prime} 2})}^{1/2}\); which approximately coincides with \({c}_{p}^{{\prime} }\), being the latter typically more than one order of magnitude larger than \({c}_{p}^{{\prime\prime} }\)35,36. On increasing the temperature, cp,rev exhibits a step at the glass transition resulting from the activation of rotational and translational degrees of freedom when the glass transforms into the liquid. Inspection of Fig. 4, panel b reveals that the cp,rev step shifts to higher temperature when increasing the frequency. Furthermore, such a step generally takes place at higher temperatures in the ISO liquid with respect to the NEM phase, underlying slower dynamics of the former. Via the temperature at the mid-point of cp,rev determined at different angular frequencies ω, we can obtain the temperature-dependent relaxation time: τ = ω−1. The latter, plotted as a function of temperature in the right panel of Fig. 4, panel d (empty symbols), indicates that both isotropic liquid and liquid crystal dynamics are highly activated and their temperature dependent can be described through the Vogel-Fulcher-Tammann (VFT) equation (lines in Fig. 4, panel d), which underlines super-Arrhenius temperature dependence: \(\tau={\tau }_{0}\exp (DT/(T-{T}_{0}))\); where τ0 is the pre-exponential factor; D a temperature independent parameter; and T0 the so-called Vogel temperature, at which the relaxation time would diverge. This super-Arrhenius temperature dependence indicates that the dynamics detected by linear calorimetric experiments actually delivers information on the cooperative α relaxation.

Fig. 4: PFO vitrification kinetics and dynamics.
Fig. 4: PFO vitrification kinetics and dynamics.The alternative text for this image may have been generated using AI.
Full size image

a Thermal protocols employed to interrogate polymer segmental mobility in step-response experiments (top panel), and vitrification kinetics at the different rates in the ISO and NEM melts (central and bottom panels, respectively). b Reversing specific heat as a function of temperature and frequency for the ISO and the NEM systems (frequencies analyzed = 2–50 Hz). c Specific heat scans on heating at 4000 Ks−1 after cooling the ISO and the NM melts at the indicated rates. d Corresponding activation plot showing the cooling rate dependent Tf rescaled via the de FKR parameter C to the relaxation time obtained from the complex specific heat in linear step response measurements. Continuous lines are VFT fits to the ω vs temperature data, from step response analysis with D = 7.0 and 7.7; T0 = 273.7 and 264.9 K for the isotropic liquid and liquid crystal, respectively; and log τ0 = − 13 for both systems.

History dependent glass thermodynamic state

Lastly, we turn the focus on the central part of our investigation, that is, the kinetic transformation of the supercooled liquid, both isotropic and liquid crystal, into glass. Figure 4, panel a (central and bottom panels) illustrates the thermal protocols employed to investigate vitrification kinetics from the ISO and the NEM melts, respectively. Glasses vitrified from the ISO melt were prepared by quenching from 573 K to 353 K at − 4000 K s−1 followed by a subsequent cooling to 223 K at different rates (see details in “Methods”). In the case of vitrification from the NEM phase, cooling from 353 K to 223 K at different rates was conducted immediately after its formation from the ISO phase and cooling at 4000 K s−1. According to the findings of the present study, the formation of the NEM phase was maximized when samples were first isothermally crystallized at 363 K and then molten at 543 K. In both cases, cooling at 4000 K s−1 down to 353 K was required to avoid crystallization.

The thermodynamic state attained after vitrifying ISO and NEM melts at a given rate was analyzed from heating scans determining the glass Tf via Eq. (1). The results of these heating scans are presented in Fig. 4, panel c. As expected, cooling at progressively smaller rates results in the development of an increasingly intense endothermic overshoot, signifying the access to increasingly low energy glasses with lower Tfs36,37,38. The outcome of this analysis is presented in the activation plot reported in Fig. 4, panel d (solid symbols), together with temperature-dependent α relaxation times obtained from step response analysis. Our data shows that NEM glasses generally exhibit lower Tfs, and consequently lower enthalpy and free energy values, than isotropic glasses when subjected to the same cooling rate. This outcome is consistent with the hypothesis that the precursor NEM melt exhibits lower enthalpy and free energy values than the ISO melt (Supplementary Fig. S1 of the Supplementary Information). This is also consistent with the smaller α relaxation time of NEM with respect to ISO glasses.

It is noteworthy that, in the elevated temperature regime, the experimental Tfs demonstrate a marked super-Arrhenius temperature dependence, indicative of a vitrification process driven by the α relaxation. This means that, in this range, vitrification is exclusively triggered by the α relaxation, an outcome implying that τ and the cooling rate, β, fulfill the proportionality addressed by the Frenkel-Kobeko-Reiner (FKR) relation39: τ = Cβ−1, with C temperature-independent constant equal to 33 for both ISO and NEM glasses. In contrast, in the low-temperature regime, approximately below 333 K and 323 K for the ISO and the NEM systems, respectively, the vitrification kinetics deviates from the VFT law. Here, it is worth noting that the available α relaxation data are restricted to the high-temperature regime (see Fig. 4d). In principle, vitrification kinetics data could therefore be included in the VFT fit to extend the temperature range. However, incorporating these data leads to an artificially weak temperature dependence of the relaxation time. When this dependence is quantified in terms of Angell’s fragility concept, using the steepness index: \(m=-{\left.\frac{\partial \log \omega }{\partial ({T}_{g}/T)}\right|}_{T={T}_{g}}\), values in the range m = 60–70 are obtained for both ISO and NEM40. By contrast, fits based exclusively on α-relaxation data yield significantly higher fragilities, m > 80, for both glasses (see SI for details). The latter values are consistent with those reported for polymers with strong steric constraints, such as PFO, which contains an aromatic ring in the main chain. Conversely, fragility indices in the range m = 60–70 are unphysically low for such systems, indicating that inclusion of vitrification kinetics data in the VFT analysis distorts the intrinsic temperature dependence of the α relaxation. Hence, the deviation from the VFT law using exclusively purely relaxation data implies that the experimentally Tf values are lower than expected based solely on the α relaxation. This finding suggests the involvement of molecular mechanisms, bearing potential for glass equilibration41,42, beyond the α relaxation in the vitrification process, which leads to a delay in the onset of the vitrification to lower temperatures. This finding is independent of the initial liquid state, ISO or NEM, used to form the glass and is in line with recent findings on bulk metallic glasses36 and confined polymeric34,35, metallic43, and molecular glasses44.

The primary implication of this result is that secondary molecular mechanisms—traditionally viewed as minor contributors to glass formation—can in fact be harnessed as a powerful tool to steer vitrification toward deeper thermodynamic basins. While the α relaxation defines the dominant cooperative rearrangement of the system, these secondary pathways likely involve localized or sub-cooperative rearrangements that permit structural optimization within the frozen matrix. Such processes could entail non-cooperative small displacements inducing density fluctuations45,46, which remain kinetically active even way below the conventional glass transition temperature. Their contribution to equilibration may thus provide an additional lever to fine-tune both local packing and intermolecular interactions. While the potential of equilibration of conventional secondary relaxations is of relevance47 but might be limited, great potential is borne by the slow Arrhenius process (SAP), whose universal presence has been proved for polymers41, small molecules42, and molecular dynamic simulations48,49; and modeled via a collective small displacements approach45.

Importantly, accessing lower-enthalpy glassy states through these secondary mechanisms has direct implications for optoelectronic functionality. In polymer semiconductors such as PFO, the degree of structural relaxation governs charge transfer states23. By kinetically navigating the energy landscape to favor deeper, more relaxed glassy configurations, one may induce a shift in photoluminescence spectra. In this light, the ability to exploit channels mediated by secondary mechanisms represents a transformative strategy for materials design: rather than being an unavoidable consequence of cooling, the glass transition becomes a tunable process that can be directed to achieve desired optoelectronic outcomes.

More broadly, these findings suggest a new paradigm in the processing-structure-property relationships of functional polymer glasses. The discovery that distinct supercooled precursors (isotropic vs. nematic) and secondary kinetic mechanisms jointly dictate the thermodynamic and microscopic configuration of the resulting glass establishes the foundation for what may be termed kinetic thermodynamic engineering of polymeric semiconductors. By judiciously designing thermal pathways and exploiting non-equilibrium dynamics, it becomes possible to program the glassy state’s free-energy landscape, thereby tailoring key functional responses such as charge transport, exciton diffusion, and emission efficiency. These insights extend beyond PFO and may apply broadly to other conjugated polymer systems, organic light-emitting diodes, and photovoltaic architectures, where glassy-phase optimization can play a decisive role in advancing device performance and stability. Indeed, while our results have been obtained for a single conjugated polymer, PFO, the analogy with a wide range of glasses, including conventional polymers and metallic glasses, in what concerns the decoupling of vitrification for the α relaxation32 implies that similar findings to those of the present work are expected for other conjugated polymers. However, we point out that the extension to other conjugated polymers might be challenging due to the difficulties in detecting the glass transition by calorimetry in these systems50, which requires the application of complex thermal protocols29.

Discussion

To summarize, in this study, we demonstrate the efficacy of glassy-phase-tuning in modifying the optoelectronic properties of polymer semiconductors. This represents a paradigm shift in the methodology employed for the optimization of optoelectronic properties of molecular semiconductors and devices. Conventional post-processing protocols for organic electronic devices have heretofore focused on the optimization of the material’s crystalline phase. However, the results of this study demonstrate that the manipulation of the thermodynamic state of glassy phases is also an efficient approach and open up a plethora of possibilities for the design of more efficient post-processing treatments for optoelectronic devices. Importantly, here we show that the modification of the glass state does not require any manipulation of the polymer semiconductor nor mixing of different compounds. The PL response of the semiconducting polymer PFO is strongly dependent on the thermodynamic state of the glass, which is connected to its structure. Specifically, the 0-0 PL band undergoes a red-shift as the glass Tf is decreased, a direct consequence of the increase of the refractive index of the material as the glass is densified. To obtain glasses with different thermodynamic energy states (and structures), two approaches were exploited: the application of different vitrification rates and the vitrification from melt states having different degrees of molecular order. These two simple approaches enabled the production of PFO glasses with Tfs that differ by 35 K between each other, which implies a large change in the glass energy. Furthermore, glasses achieved from nematic melts at a certain vitrification rate exhibited lower Tf, and thus lower energies, than their isotropic counterparts. More intriguingly, irrespective of the molecular order of the vitrifying melt, the implementation of low vitrification rates resulted in glasses characterized by anomalously low energy states. While the process of vitrification at elevated rates is predominantly mediated by the α relaxation, a molecular mechanism, mildly activated and thereby extending beyond the α relaxation, is also operative at low rates, causes a delay in vitrification to lower temperatures, thereby enabling the formation of low-energy thermodynamic glassy states.

Methods

Materials

Poly(9,9-di-n-octylfluorenyl-2,7-diyl) (PFO) was obtained from De Mello’s group. The number average molecular weight (Mn) of the PFO employed was 20.7 kg/mol, as determined by size exclusion chromatography in combination with multi-angle light scattering (SEC-MALS). The polydispersity was 2.03, obtained by size exclusion chromatography calibrated with polystyrene (SEC-PS), respectively.

Calorimetric measurements

Fast scanning calorimetry measurements were performed by means of a Mettler Toledo Flash DSC 1, equipped with a Huber TC100 intracooler, allowing to operate at a temperature ranging between 173 and 723 K. MultiSTaR UFS1 (24 × 24 × 0.6 mm3) chip sensors were employed, following the conditioning and temperature correction protocols recommended by the manufacturer. Temperature calibration was performed with indium at a heating rate of 1000 K s−1. The sample was prepared by spin-coating a thin film of PFO directly onto the backside of the chip sensor. The procedure included three steps: (i) applying a masking layer by placing a drop of glucose solution on the reference area on the backside of the chip; (ii) spin-coating a 10 mg/mL PFO solution in THF at 2000 rpm for 60 seconds; and (iii) removing the masking agent by carefully rinsing the reference area with water. Prior to any calorimetric scan, the physical state of PFO was defined by a thermal pre-treatment. To obtain PFO in an isotropic (ISO) state, the sample was kept at 573 K for 0.1 s and quenched by cooling at 4000 K s−1 to 188 K. On the other hand, to form the nematic (NEM) state, the sample was kept at 573 K for 0.1 s, cooled at 353 K at 4000 K s−1 and isothermally crystallized for 30 min. After the crystallization step, the sample was melted by heating to 443 K at 4000 K s−1 and, finally, cooled at 188 K at 4000 K s−1.

The temperature dependence of spontaneous fluctuations for the ISO and NEM states of PFO was assessed through step response analyses33,34. After choosing the initial phase of PFO (ISO or NEM) two different oscillating protocols were applied from − 50 °C up to 200 °C (sketched in Fig. 4, panel a, upper panel). The first, consisting in a loop of up-jumps of 2 K at a heating rate of 2000 Ks−1 followed by a 0.1014 s isotherm, resulted in a perturbation period tp = 0.1024 s and a base frequency (f0) of about 10 Hz. The second makes use of a heating rate (β) of 200 Ks−1 for the 2 K up-jump and an isotherm of 1.014 s, to access lower f0 of about 1 Hz (tp = 1.024 s). The frequency dependent complex specific heat, \({c}_{p}^{*}\), was calculated by Discrete Fast Fourier Transformation (DFFT) of the recorded heat flow, HF(t), and the instantaneous heating rate, β(t), for f0 and higher harmonics f = kf0(k = 2, 3, …):

$${c}_{p}^{*}(\omega )=\frac{{\int }_{0}^{{t}_{p}}HF(t){e}^{-i\omega t}dt}{{\int }_{0}^{{t}_{p}}{\beta }_{h}(t){e}^{-i\omega t}dt}$$
(2)

where ω is the angular frequency (f = 2πω). DFFT was repeated for each period of oscillation tp. Accessing higher harmonics allows assessing \({c}_{p}^{*}\) in the frequency range between 2 and 50 Hz. Finally, the frequency-dependent glass transition temperature Tg,dyn(ω), which is assessed in linear response, is determined at the inflection point of the glass transition step in the reversing specific heat cp,rev(ω), i.e., the modulus of \({c}_{p}^{*}(\omega )\).

The vitrification kinetics was determined for PFO in both ISO and NEM states at cooling rates (βc) ranging between 0.1 and 4000 Ks−1. The temperature programs for both samples are shown in the middle and lower panels of Fig. 4 panel a. After the pre-treatment for the desired PFO phase, a fast cooling was performed up to 353 K at 4000 Ks−1, that is slightly above Tg. Starting from 353 K, variable cooling rates (0.1 ≤ βc≤ 4000 Ks−1) were applied down to 188 K. For each βc, a subsequent heating scan was recorded at a fixed rate of 1000 Ks−1. The cooling rate dependent fictive temperature, Tf(βc), was calculated on heating scans via the Moynihan method (see eq. (1)).

Photoluminesce measurements

Photoluminescence spectra were acquired using a Witec alpha300RA confocal system, excited using a 355 nm solid state laser through a 40x UV-prepared objective. The PFO samples on the FSC chips were imaged over an area of 90 μm × 90 μm, taking 18 × 18 spectra, with an integration time of 100 ms per point measured on the fly and a laser power of 100 μW. Subsequently, a cluster analysis was employed to group the measured spectra and differentiate the region of the film that was heated than that of the surroundings (e.g., on top of the electrodes). The spectra shown correspond to the average over 100 experimental spectra taken at the heated area.