In situ probing of the crystallization kinetics of rrP 3 HT on single layer graphene as a function of temperature †

We studied the molecular packing and crystallization of a highly regio-regular semiconducting polymer poly(3-hexylthiophene) (P3HT) on both single layer graphene and silicon as a function of temperature, during cooling from the melt. The onset of crystallization, crystallites’ size, orientation, and kinetics of formation were measured in situ by synchrotron grazing incidence X-ray diffraction (GIXD) during cooling and revealed a very different crystallization process on each surface. A favored crystalline orientation with out of plane p–p stacking formed at a temperature of 200 1C on graphene, whereas the first crystallites formed with an edge-on orientation at 185 1C on silicon. The crystallization of face-on lamellae revealed two surprising effects during cooling: (a) a constant low value of the p–p spacing below 60 1C; and (b) a reduction by half in the coherence length of face-on lamellae from 100 to 30 1C, which corresponded with the weakening of the 2nd or 3rd order of the in-plane (k00) diffraction peak. The final ratio of face-on to edge-on orientations was 40% on graphene, and 2% on silicon, revealing the very different crystallization mechanisms. These results provide a better understanding of how surfaces with different chemistries and intermolecular interactions with the polythiophene polymer chains lead to different crystallization processes and crystallites orientations for specific electronic applications.


Introduction
3][4] Another advantageous property of graphene is that it can be combined with a thin layer of an organic semiconductor to produce organic hybrid optoelectronic devices. 5,6Efficient transfer of charges at the interface between graphene and a thin layer of conjugated organic semiconductor, has been shown to occur through p-p interactions. 4,7,8It was moreover recently demonstrated that combining graphene and a semiconducting polymer such as poly-3-hexylthiophene (P3HT) produced enhanced charge transport in the vertical direction, and could help the development of faster and more efficient opto-electronic devices such as OPVs and OLEDs. 9wever, the crystallinity of the polymer film strongly influences charge transport in the active layer of an electronic device, which will finally impact its overall efficiency.Therefore, in order to predict and control the electronic properties of a single sheet graphene/semiconducting polymer film layer, it is important to understand how the polymer crystallizes on a mono-layer of graphene, and how different crystallization on graphene is from a typical substrate, e.g.silicon or silicon oxide.Indeed, the crystallization of a semiconducting polymer, such as P3HT, depends on several factors: interaction with the substrate, annealing temperature, rate of cooling, etc.Although P3HT has been widely studied on more common (and often weakly interacting) surfaces, e.g.1][12][13] It has previously been shown that the crystalline order of P3HT can be affected by changing the chemistry of the substrate, as was previously shown by spin coating the polymer on substrates treated with either hexamethyldisilizane or octadecyltrichlorosilane. 14ere, we show a detailed in situ study of the kinetics and process of crystallization of a 85 nm thick P3HT film spun from orthodichlorobenzene (o-DCB) on both graphene and silicon (with native oxide) as a function of temperature during slow cooling from the melt.We discuss our results in the light of a previous study of thin P3HT films (20-25 nm) spun from chloroform onto silicon, where the (200) layer spacing and peak width were measured in situ during heating and cooling. 16Our findings reveal several differences with this previous study, but also clearly show that the way crystallites form on graphene is fundamentally different than on the less interacting silicon surface, which results in different crystallite orientations, kinetics of crystallization, and a very different final ratio of face-on to edge-on lamellae, which imply largely different opto-electronic properties.The influence of the surface chemistry on the crystalline structure of polymer films was previously observed. 14

Materials
Poly-3-hexylthiophene (P3HT) (98% RR, M w = 32 kD) was purchased from American Dye Source Inc. Single layer graphene was produced by CVD on a copper foil and subsequently transferred to a solid substrate (silicon or glass).Details of the synthesis and transfer procedure of graphene, and other materials used can be found in the ESI.†

Sample preparation and characterization
Thin films of regio-regular P3HT (98%) (85 nm AE 3 nm as measured by AFM and shown in Fig. S1 in ESI †) were deposited on both graphene and silicon (with native oxide) substrates by spin-coating of a dilute solution of P3HT in ortho-dichlorobenzene (o-DCB).They were then placed in an inert atmosphere (helium) chamber and heated up to 240 1C and maintained there for 5 min, before slow cooling (2-3 1C min À1 ), where they were analyzed in situ by synchrotron X-ray diffraction at the 11-3 beamline at the Stanford Synchrotron Radiation Lightsource (SSRL), as schematized in Fig. 1A and B. This method gives accurate information about the crystalline structure, the crystallites orientation and their coherence length inside the films.

Results and discussion
Sheets of graphene synthesized by chemical vapor deposition were transferred onto the silicon substrates by a floating method as detailed elsewhere. 9,17These were recently characterized by Fig. 1 Preparation of the graphene/P3HT samples.(A) Thin films of P3HT were spun from solution onto either silicon or a single layer graphene surface, heated up to 240 1C for 5 min in inert atmosphere, and slowly cooled down to room temperature.(B) During cooling down, the films were characterized by grazing incidence X-ray diffraction (GIXD), and their crystallization monitored at various temperatures using a planar detector.(C) 2D grazing incidence diffraction patterns of P3HT films on Si and on graphene, showing q z , q xy and the angle w.
This journal is © the Owner Societies 2017 Raman spectroscopy which confirmed that these sheets were monolayers of graphene. 9he thin films of P3HT which was spun on silicon and graphene substrates were heated to 240 1C, above their melting point (T m E 230 1C), in order to ensure complete vanishing of crystallites formed during spin-coating.Fig. S2 (ESI †) shows grazing incidence X-ray diffraction (GIXD) snapshots of both films on silicon (Si) and graphene (G) at different temperatures during cooling.At 240 1C, no diffraction was detected in either film, indicating a disordered state.It is possible that some partial ordering is still present in the form of small aggregates in the film.
Upon cooling, crystallites started to form and became visible in the diffraction patterns.Close to room temperature, at 30 1C, the crystalline structure in both films showed clear differences (Fig. 1C) with different crystalline orientations and in various amounts.Both samples contained edge-on lamellae (characterized by the strong (100) diffraction peak along q z ), as schematically showed in Fig. 2B, where the insulating hexyl side chains shown Fig. 1A are oriented perpendicularly to the substrate.3][24][25] The solvent and time of film growth were found to be important, and usually longer crystallization times and high boiling point solvent have been reported to favor edge-on growth on silicon.This observation is consistent with our experimental conditions which were meant to simulate a thermodynamic equilibrium in the melt with very slow cooling.][28] Face-on lamellae, with p-p stacking pointing along the [010] direction (see Fig. 2C) were also present mostly on the graphene sample (strong diffraction spots in (100) xy in xy plane, and in (010) z near the z axis).0][31][32] Recently, the use of single layer graphene was shown to promote face-on orientation and to enhance out-of-plane conductivity and mobility in P3HT compared to silicon, reaching a mobility m E 2.8 Â 10 À4 cm 2 V À1 s À1 . 9ig. 3A represents the evolution of the amount of edge-on lamellae during cooling at a 2-3 1C min À1 rate on Si and G, and showed that these grew much faster, and reached a much larger amount, on the silicon substrate.By comparison, the final amount of edge-on lamellae on graphene were E19% of that on silicon at 30 1C.We defined the face-on ratio (F) as the intensity of the (100) xy peak divided by the sum of the intensities of both the (100) xy and (100) z peaks.The evolution of F as a function of temperature is plotted in Fig. 3B.Peak intensities were extracted from fitting peak areas of line cuts from the 2D diffraction patterns.This panel in particular shows that the film on graphene started crystallizing with face-on lamellae only (characterized by 100% face-on ratio), before the first edge-on crystallites started to appear around 185 1C.This indicates that the P3HT chains orientate themselves cofacially with the substrate due to the p-p interactions with the graphene layer, instead of the preferred edge-on orientation encountered on silicon.This result is consistent with simulations of crystallization of P3HT on graphene, which showed that the strong p-p interaction between the thiophene backbone and the graphene layer results in higher binding energy for the face-on orientation compared to edge-on. 10 In the case of the silicon substrate, hydrophobic interactions between the substrate and the hexyl side chains results in an edge-on configuration which minimizes the free energy for crystallization compared to face-on crystallites. 33,34The final face-on ratio was E40% on graphene, whereas it was only E2% on silicon, highlighting the very different crystallization processes taking place on both surfaces.
The pole figure showed in Fig. 3C, also clearly shows the difference in the final crystalline orientations at 30 1C for both films on silicon and graphene, and represents the (100) normalized diffraction intensity (area) (see Fig. 1C).It is clearly visible that the film on silicon is mostly edge-on, while the film on graphene has a much larger proportion of face-on crystallites.
It is also clearly seen that at higher w (60 to 881), the amount of misoriented face-on lamellae is more pronounced on graphene compared to silicon (Fig. 3C).
A more detailed analysis of the nucleation and growth of crystallites on silicon (Si) and graphene (G) is presented in Fig. 4.During heating up, all diffraction peaks disappeared at 240 1C on both substrates, indicating loss of ordered crystallites in the films.During slow cooling, crystallites started to form again between E200 and 185 1C, but with a different crystalline orientation on silicon and on graphene.The temperatures given take into account the accuracy of reading from the thermocouple (0.1 1C) and the resolution during cooling (one measure every 5 1C for graphene and every 10 1C for silicon in the range 240 to 180 1C).As the films were slowly cooled down, the formation small crystallites became apparent in the form of weak diffraction peaks in (100) at q = 0.35 Å À1 for the graphene substrate, q = 0.31 Å À1 for the silicon substrate.Since it was difficult to precisely determine the onset of crystallization due to the weak diffraction intensity at temperatures close to the melting point, and for consistency, we arbitrarily set a normalized intensity value of 2% of the maximum intensity as the threshold for the onset of crystallization in the samples.
On graphene, the first crystallites with an edge-on orientation appeared between 185 and 180 1C, as shown by the small diffraction peak at q E 0.324 Å À1 along the z axis (Fig. 4A).By contrast, the first face-on crystallites appeared at a higher temperature between 205 and 200 1C, as seen by the diffracted intensity at q E 0.354 Å À1 along the xy axis in Fig. 4B, and indicate a smaller interlayer d spacing in the xy plane compared to the z direction perpendicular to the plane.The nucleation and subsequent growth of these lamellae during cooling is clearly visible in the plots of Fig. 4.
On silicon, the first crystallites appeared with the (100) edgeon lamellar orientation (diffraction spot at q E 0.327 Å À1 along the z axis) at a temperature below 195 1C, and at E180 1C the (100) z peak is already very well defined (Fig. 4C), and much more intense and sharper than on graphene (Fig. 4A).During further cooling, the position of this peak shifted to q E 0.360 Å À1 on both surfaces, corresponding to a contraction of (100) interlayer spacing (see Fig. S3 of ESI †).Moreover, it is interesting to note that contrary to the evolution of the (100) xy peak's width on graphene, the (100) z peak's width actually decreased considerably during cooling, indicating lamellar growth and a much higher coherence length as can be seen in Fig. 6.A detailed study of growth dynamics, interplanar spacing and coherence length on both surfaces is presented thereafter.
The shift in q shown Fig. S3 in ESI, † indicates a reduction in interplanar spacing d z on both silicon and graphene of the edge-on lamellae along the z axis.As illustrated in Fig. 5A, this reduction occurred for d z E 2.02 to 1.75 nm, and is consistent with a previous study of P3HT performed on silicon where a reduction in d spacing from E1.83 down to 1.66 nm was observed. 16A shift in 100 xy peak position to higher q from 0.355 to 0.380 Å À1 on graphene and from 0.359 to 0.380 Å À1 on silicon was also observed during cooling, which indicates a reduction of interlayer d xy spacing of the face-on lamellae along the (100) direction in the xy plane, from 1.78 to 1.66 nm on graphene, and follow a similar trend on silicon with nearly identical values.The values measured for d xy are consistently smaller by 0.1 to 0.2 nm compared to that measured for d z likely This journal is © the Owner Societies 2017 due to refraction effect. 35Another interesting observation concerns the reduction in d z of the (010) face-on lamellae on graphene.As can be seen in Fig. 5C, the (010) interlayer spacing decreases in the z direction during cooling due to thermal contraction.However, surprisingly d z stopped decreasing and remained at a constant value of d z E 0.374-0.375nm below E80 1C.One would expect a continuous decrease in interplanar spacing until cooling to room temperature.7][38] The low d value measured here on graphene potentially indicates faster charge transport along the p-p stacking in the vertical direction, compared to larger p-p interplanar spacing.
We showed that as the polymer was cooled, the interlayer spacing decreased.It is also possible to estimate the coherence length of the crystallites from the width of the diffraction peaks as a function of temperature.The coherence length L of the crystallites was then calculated from the peak FWHM b using the Scherrer's formula: where K is the shape factor, estimated to 0.9, l the X-ray wavelength (0.974 Å) and y the Bragg angle.Fig. 6 shows the evolution of the coherence length for (100) z lamellae oriented in the z direction, for (100) xy along the xy plane, and for (010) z lamellae oriented at E81 from the z axis on graphene.
On silicon, edge-on crystallites grew relatively quickly, and reached L z 100 E 30 nm at a temperature of 180 1C, just E15 1C below the onset for crystallization of the first edge-on lamellae.The coherence length L z (100) of the crystallites remained between E31-37 nm during further cooling down to E60 1C, and then slightly decreased below 60 1C reaching their lowest value of E29 nm at room temperature.Note that no value of L 100 could be analyzed at temperatures above 180 1C due to a weak and undefined diffraction peak.On graphene, L z 100 increased from zero at 180 1C to E28 nm at 115 1C where it remained approximately constant until 60 1C where it slowly decreased to reach E25 nm at room temperature.The evolution in L z 100 is very similar on both silicon and graphene from 115 to 30 1C, only with slightly lower values by 2-3 nm on graphene.Moreover, on graphene from E200 1C, face-on lamellae with a coherence length L z 010 E 9.5-12.5 nm (measured at an angle of E81 from the z axis) were present until E100 1C.Below this temperature, the face-on coherence length quickly decreased to its final lowest value of E5.0 nm at E30 1C, representing a 50% reduction in L z 010 .The kinetics of crystallite formation are therefore very different on both surfaces, with P3HT crystallizing at higher temperatures on graphene compared to silicon.This suggests that the graphene layer acted as a seed for crystallization with a preferential face-on orientation, while a P3HT film on silicon favored an edge-on configuration.Face-on crystallites form preferably on graphene, compared to silicon, because of the p intermolecular forces between the graphene sheet and the P3HT molecules.This molecular arrangement enables a more efficient charge transport across the thickness of the film in the direction of the p-p stacking. 39On silicon, the first (100) diffraction occurred at E195 1C (AE5) along the z axis (Table 1), consistent with edge-on lamellae formation.However, on graphene the first diffraction peak appeared from (100) along the xy axis (Table 1), indicating the formation of face-on lamellae.Moreover, the face-on crystallites formed at a higher temperature (E200 1C (AE5)).With further cooling to 100 1C, the diffraction intensity I 100 (xy) became much more intense on graphene (E45 times more at 100 1C), whereas it stayed very weak on silicon, indicating strong vertical p stacking on graphene.During cooling, edge-on lamellae grew and became more numerous on both samples, but I 100 (z) was much more intense on silicon at all temperatures, reaching E10 times more intensity at 100 1C compared to its value at 195 1C.Moreover, the overall degree of crystallinity of the P3HT film, as the integrated intensity of the 100 peak on the whole w range (from the xy plane to the z axis), was much lower on graphene (E40 a.u. at 30 1C) compared to the one on silicon (100 a.u. at 30 1C).It should also be noted that film thickness is also an important parameter which will likely change the crystallization dynamics, and therefore the results obtained here may not be representative of films which are much thicker or much thinner than the ones used here.

Conclusions
In conclusion, we have performed an in situ study of the crystalline structure of P3HT films on both silicon and graphene substrates as a function of the cooling temperature from the melt.Our results show that p-p interactions between P3HT and graphene promote coplanar molecular arrangement of the P3HT chains on the graphene surface, which favor the nucleation of face-on lamellae at a relatively high temperature between 185-200 1C.On silicon however, crystallization of edge-on lamellae was strongly favored due to the weaker interactions with the substrate.The overall degree of crystallinity of the film on graphene was only 40% of that of the film on silicon, and the final ratio of face-on to edge-on lamellae was only 2% on silicon, and much higher on graphene (E40%).Large differences in the crystallite coherence length were also measured on both surfaces, with a surprising reduction in face-on coherence length by 50% on graphene during cooling.These results provide a detailed study of how semi-conducting polymer films crystallize on graphene, and they may help better control their optoelectronic properties by a better understanding of the crystallization process and kinetics on both a weakly and a more strongly interacting surface.a The values given for d and L at the onset temperature are fitted from the data at 163 1C.

Fig. 2
Fig.2P3HT molecular structure and its two main orientations.(A) Shows the P3HT molecular structure with its thiophene ring and the hexyl side chain.(B and C) Schematics showing the edge-on and face-on configurations, and the different crystallographic orientations.It should be noted that the tilting of the chains within the crystals is unknown and not represented here.15

Fig. 3 (
Fig. 3 (A) Intensity of the 100 peak in the out-of-plane direction as a function of temperature during cooling.(B) Log scale of the face-on ratio (F) showing the diffraction intensity ratio of the signal from face-on crystallites compared to both face-on and edge-on.It is clear that the graphene sample starts crystallizing in a face-on configuration, while the film on silicon shows mostly edge-on crystallites.(C) Log scale of w plots for both the films on graphene and silicon at 30 1C, with w = 01 corresponding to edge-on and w = 901 being face-on, as shown in the inset.The intensities reported in (A) are obtained from peak areas fitted from line cuts of the 2D diffraction patterns.The face-on ratio in (B) is defined as I 100 xy /(I 100 xy + I 100 z ).

Fig. 4
Fig. 4 In-plane (xy) and out-of-plane (z) 100 diffraction peaks of P3HT films on graphene and on Si during cooling.The films were maintained in an inert atmosphere and probed at an incident angle of 0.131.(A) Shows the evolution of the 100 z peak during cooling for P3HT on graphene, and (B) show the intensity for the 100 xy peak during cooling.The intensity of the 100 z peak, and of the 100 xy peak, during cooling on Si is shown in (C) and (D) respectively.

Fig. 5
Fig.5Evolution of the interplanar spacing during cooling for P3HT on graphene (red squares) and Si (black circles) for (A) the out-of-plane 100 peak, (B) the in-plane 100 peak and (C) the out-of-plane 010 peak.All distances decreased with cooling, which can be attributed to thermal contraction.It should be noted that the interplanar spacing for the 010 peak reached a minimum at E80 1C.

Fig. 6
Fig.6Evolution of the coherence length during cooling for (A) the 100 xy , (B) the 100 z and (C) 010 z on both silicon (black circles) and graphene (red squares).The coherence length decreased with temperature for the 100 xy peak on both graphene and Si substrate and for the 010 z peak on graphene, while L 100 z increased until it stabilizes at E120 1C before decreasing slightly when the temperature reached E60 1C.

Table 1
Interplanar distance d and coherence length L for the P3HT films on graphene and silicon at the onset temperature and at room temperature.The crystallization occurs on the graphene substrate at higher temperatures than for the silicon substrate