Large-scale lateral alignment of nanorods (NRs) is of interest for manifestation of their anistropic properties including polarized emission and directional electrical transport. This study investigates the utility of mechanical rubbing for macroscopic scale alignment of colloidal semiconductor NRs. CdSe/CdS seeded-rods, exhibiting linearly polarized emission, are aligned by mechanical rubbing of a spin-coated glass substrate. The dragging force exerted by the rubbing fibers results in deflection and reorientation of the NRs along the rubbing direction. The rubbed samples were characterized by various methods including absorption, polarized emission, optical fluorescence microscopy, atomic force microscopy, and ultra-high resolution scanning electron microscopy. The emission polarization contrast ratio (CR), defined as the ratio between emission intensities parallel and perpendicular to the rubbing direction, was used to characterize the rods alignment. The effects of substrate treatments on the CR were studied, showing that partially hydrophobic surface provides optimal conditions for alignment. Excess organic ligands added to the deposited NR solution strongly affect the extent of alignment. This was studied for a series of NR samples of different dimensions and an optimal additive ratio of ∼3 ligand molecules per 1 nm2 NR surface area was found to yield the highest CR. Average CR values of 3.5 were detected over the entire 6 cm2 substrate area, with local values exceeding 4.5. While samples of rubbed spherical quantum dots and spin-coated films of NRs show no emission polarization, the emission intensity from rubbed NR samples is polarized obeying Malus' law (wherein, the intensity is proportional to cos2(θ)). Mechanical rubbing, well known for its use in LC devices, may be considered as a method for large-scale alignment of NRs on substrates.
Semiconductor nanocrystals (NCs) manifest unique, size-dependent, electronic, and optical properties 1, 2. Advancements in the synthesis of these colloidal NCs yielded high quantum efficiencies 3 and control of the attained morphology 4, 5. Of particular interest are elongated semiconductor NCs, known as nanorods (NRs), exhibiting anisotropic properties such as polarized emission and lasing 6–12. Alignment of NRs on large scales by facile means is therefore of fundamental interest for the study of their anisotropic collective behavior and is also relevant for different optoelectronic applications of NRs. In this work, we examined the use of mechanical rubbing to induce NR alignment on large macroscopic scale areas. Effects of rubbing parameters, the substrate surface functionalization and excess organic ligands on the rubbing process, and the resulting alignment were investigated.
Reports on alignment of NRs can be distinguished between vertical 13–17 and lateral alignments, relative to the substrate's surface. Mechanical rubbing is relevant for lateral alignment and the polarization is characterized by the polarization contrast ratio (CR)
where is the emission intensity measured parallel (perpendicular) to the alignment direction of the NRs. Different methods were reported to induce lateral alignment. Controlled slow solvent evaporation provided areas of smectic alignment but is time consuming 18–20. The “coffee stain effect” was also used for smectic alignment of NRs perpendicular to the radial direction of the formed rings 21. Alignment by external electric fields 22, 23 was used and CR of levels of 2.7 were reported for seeded rods on sub-mm2 scale areas 24. Such technique requires design and production of an electrode structure, which is time consuming and costly, and limits the effective alignment area to relatively smaller regions.
Another approach is the alignment by stretching polymer films, with NRs embedded inside 4, 25, 26 and in this case the rods are within a matrix and not aligned on substrate. The method studied in this work is the alignment by mechanical rubbing, a technique widely used in the liquid crystal (LC) industry for the preparation of polymeric alignment layers 27, 28, which governs the alignment of the LC molecules 29, 30. Only few reports utilized rubbing for the alignment of NRs. Hikmet et al. reported the use of mechanical rubbing, during the fabrication of light emitting diodes, as a method for the alignment of the NRs layers and CR values of 2.2 were measured. Similarly, Janssen's Group 31, 32 reported alignment of NR layers through mechanical rubbing, for the investigation of QD-NR energy transfer interactions, yielding a CR of 1.75. The previous alignment attempts carried out through mechanical rubbing reported polarization CRs of up to 2.2 on areas of mm2. The full potential of the rubbing method was not pursued, and there is still a lack in the understanding of the critical parameters affecting this process. In this work, we have studied the ability to laterally align seeded CdSe/CdS NRs by means of mechanical rubbing. We examined the key aspects influencing the obtained CR allowing the production of oriented samples with CR values of 3.5 over areas of 2 × 3 cm2.
CdSe NCs (cores) were synthesized by injection of a selenium TOP complex into a mixture of cadmium oxide dissolved in ODPA and TOPO. After precipitation of the CdSe cores a CdS shell was grown from cadmium oxide and sulfur in ODPA, TOPO, and HPA following a similar method reported by Manna and co-workers 24. The obtained core/shell NRs and core/shell QDs, synthesized under similar conditions, show an exitonic peak at 625 nm and red fluorescent emission at 632 nm (Fig. 1c, inset).
2.3 Stock solution preparation
The deposited solutions were prepared by dissolving the growth solution in toluene in an ultrasonic bath heated to 60 °C, until a clear homogenous solution was obtained. MeOH was added to the hot solution followed by vigorous stirring and precipitation in a centrifuge. The clear solution was discarded and the precipitant was dried under Ar(g). The previous step was repeated, and the obtained dry precipitant was kept under inert conditions until used.
2.4 Substrate treatments
Super frost® microscope glass slide (Menzel-Glaser) were scribed, using a diamond tip pen, into 2 × 3 cm2 substrates. The substrates were initially cleaned in a 2% Micro-90 chemical cleaning solution (International Products Corporation) in DW by ultrasonic agitation for 30 min at 80 °C. The substrates were thoroughly washed under running DW and transferred into a MeOH (Tech grade) solution and submerged in an ultrasonic bath for 30 min without heating. The substrates were then passed into a toluene (Tech grade) solution and once again put in an ultrasonic bath for 30 min without heating. The substrates were kept in a clean toluene (Tech grade) solution until used, when they were first blown dry using N2(g).
2.5 Addition of organic molecules
A TOPO solution was prepared by dissolving about 200 mg of TOPO (99%) in approximately 5 g of toluene (anhydrous). Calculated volumes of the additive solution were added to the clean precipitant, according to the desired molar ratio as determined by absorption measurements of the NR solution.
2.6 Spin coating
The spin coating of the fluorescent thin films was done using a P6000 spin coater (Specialty coatings Inc.) set at a chuck speed of 2000 rpm, ramp time of approximately 3 s, and a deceleration time of 3 s. Thirty microliters of the stock solution were drop casted onto the dry substrate which was immediately spun.
2.7 Optical measurements
The optical characterization of the sample was done by un-polarized light excitation using a commercial blue LED bulb. The source light was passed through a short-pass (550 nm) onto a diffusive glass slide, generating a broad uniform illumination. The scattered light then passed through a polarization mixer before reaching the fluorescent sample. The emitted photons were collected, through a high-pass (595 nm) filter and a linear polarizer, using a Nikon (D40) CCD camera.
2.8 Microscopic measurements
The microscopic measurements were conducted using a Nikon Diaphot inverted microscope mounted with a monochrome DS2MBWC CCD and a linear polarizer in the optical path. The samples were excited by a blue LED (405 nm) focused through the objective. The emission and excitation were distinguished using a dichroic mirror, and high pass filter (510 nm) on the emission path.
2.9 Structural characterization
Imaging the orientation and surface coverage of aligned or spin-coated NRs was done by MagellanTM 400L high resolution scanning electron microscopy (HR-SEM). In-chamber oxygen plasma was used to reduce the contamination and charging by the organic layer. In addition, surface topographic measurements of the rubbed and spin-coated samples were done by atomic force microscopy (AFM) using a Nanoscope Dimension 3100 scanning probe microscope. Tapping mode was applied to minimize contact with the surface and maintain the particle alignment throughout.
2.10 Mechanical rubbing process
The lateral alignment was achieved by a two step process consisting of spin coating a thin film of fluorescent NRs onto a glass substrate followed by mechanically rubbing the sample using a rayon velvet fabric (Yoshikawa, YA-20 R) with a density of 24 000 fibers × cm−2. The rubbing cloth was mounted onto a cylindrical shaft (20 mm diameter) coupled to a DC motor, rotating at a controlled speed in the range between 200 and 2000 rpm.
3 Results and discussion
CdSe/CdS NRs of various dimensions exhibiting strong fluorescent emission were synthesized following the procedure of Manna and co-workers 24 (see Section 2 for details). The lateral alignment was achieved by rubbing various CdSe/CdS thin films spin coated onto different substrates (see Section 2). A schematic cross-section of the rubbing process is illustrated in Fig. 1a, where the rubbing cloth is mounted on the rotating shaft, the spin-coated sample is mounted on a linear motorized stage, and a micrometric vertical stage controls the pile impression. The relative motion between the sample and the rubbing cloth results in a plough-like action applied on the thin film by the fibers of the rubbing cloth, as they pass over the NRs on the substrate. HR-SEM images of a spin-coated thin film show the isotropic nature of the un-rubbed film, characterized by the random NR orientation (Fig. 1b). Upon rubbing, the NRs are deflected by the fibers of the rubbing cloth producing lateral alignment along the rubbing direction (Fig. 1c). Examination of the rubbed samples revealed that the deflected NRs pile up on the edge of the fiber's trail, while reorienting parallel to the rubbing direction as depicted in Fig. 1b. AFM measurements of a rubbed sample exhibited a uniform linear pattern, which is indicative to most rubbed samples, with a peak-to-peak distance of 2.5 µm, depth of 250 nm, and a full width half maximum (FWHM) of 1 µm for each peak (Fig. 2a, b, and d). Considering the initially planar thin film (Fig. 2c) and its thickness of approximately 70 nm, we conclude that the line pattern is a result of both material depletion (from the trenches) and accumulation (on the lines). This is accompanied by deflection of the NRs induced by the dragging forces exerted by the fibers leading to reorientation of the NRs parallel to the rubbing direction.
The different parameters affecting the rubbing process were divided into mechanical parameters, unrelated to the sample preparation process, and parameters involved in the samples preparation. In order to quantify the influence of the mechanical factors, the rubbing strength (RS) parameter 33–35 was used
where r, n, v, M are the roller radius, revolution speed of the rubbing cloth, linear speed of the sample, and the pile impression, respectively. The pile impression, controlled by the micrometric vertical stage, was determined as the depth of the impression relative to the point of contact (see Section 2). Different pile impressions ranging between 10 and 80 µm were examined for the rubbing of 70 nm thick spin-coated films. A pile impression of 30 µm was found to induce minimal loss of material while generating the linear imprint pattern. Using higher values resulted in extensive material removal from the substrate while lower values were found to be insufficient for inducing the alignment of the NRs.
Varying the rotational speed of the rubbing cloth between 200 and 2000 rpm and the linear velocity of the substrate from 0.8 to 12 mm s−1, while maintaining a constant pile impression of 30 µm, did not significantly affect the measured CR value (see Fig. S1 of Supporting Information, online at: www.pss-a.com). This implies that the response to the rubbing process is mainly governed by intrinsic properties of the thin film, similar to previously reported results 34, 35 indicating minor affects to the alignment properties of LC molecules when applying different RS values.
Comprehensive understanding of the rubbing process and the ability to induce lateral alignment was obtained by focusing on two key aspects involved in the sample preparation process: surface–particle interactions and particle–particle interactions. Good compatibility between the substrate and the deposited medium is a pre-requisite for achieving high uniformity and good surface coverage of spin-coated thin films. Contact angle (CA) measurements were performed in order to assess this compatibility, characterizing the hydrophilic/hydrophobic nature of the substrate 36. The substrates were prepared by sequential washings of the glass slide, starting with washing in an aqueous based detergent solution (micro-90) followed by various solvents including toluene, methanol, isopropyl alcohol, and acetone. The resulting CA, as formed by a static water droplet, gave a qualitative indication to the obtained surface properties as a function of the cleaning method (see Fig. S2 of Supporting Information). The compatibility between the substrate and the NR solution in terms of attraction and repulsion forces were found to directly affect the ability to perform physical manipulations on the pre-deposited NRs using mechanical forces 37. We found that changing hydrophilic/hydrophobic characteristics of the surface resulted in a range of measured CR values from unity (no preferred orientation) up to 2.25, for films deposited using the same stock solution and rubbed at constant conditions (see Fig. S3 of Supporting Information).
This shows that the rubbing alignment is highly sensitive to changes in the surface–particle interactions. This is understood considering the proposed mechanism in which the NRs are dragged by the fibers across the surface inflicting their reorientation thus minimizing their friction with the surface and the fiber. Interestingly, a partially hydrophilic substrate (CA = 35°) exhibited the highest CR (2.25) whilst both the hydrophobic (CA = 90°) and the hydrophilic substrates (CA = 10°) produced a low, close to unity, CR. The physical attributes of a partially hydrophilic surface trigger a fine balance between the repulsive and attractive substrate–particle interactions, allowing the NRs reorientation upon contact with the fiber, while minimizing their removal from the substrate.
The surface conditions found to yield the highest CR values were set as the standard substrate cleaning method for future experiments addressing the role of particle–particle interactions in the rubbing alignment process. The stiffness of the thin film, governed by strong particle–particle interactions, contradicts the particle alignment, for which high mobility of the NRs is crucial. The extent by which these interactions affect the alignment process were studied by changing the organic composition of the pre-deposited NR solution. Figure 3 shows thermogravimetric analysis (TGA) of a 67 nm × 5 nm CdSe/CdS sample. The crude material (Fig. 3, black line) exhibits a 98% weight loss when heated up to 600 °C. This weight loss indicates the high organic content of the crude sample comprised of only 2% NRs (w/w). The inset shows the derivative of the weight loss to temperature plot, providing a more accurate correlation between the rate at which the sample's weight is reduced and the equivalent temperature. Comparing the derivative plot of a TOP/TOPO reference sample (inset dashed line) and that of the crude material (inset solid line) correlates between the substantial weight loss occurring at ∼350 °C and TOP/TOPO extraction.
A procedure for preparing NR solutions with known percentages of excess ligands was developed to allow for reproducible preparation of the spin-coated films. The crude material was cleaned by precipitation using a 4:1 solvent/anti-solvent ratio, yielding a precipitate containing 2% of residual capping ligands (Fig. 3, red line). The clean precipitant was redispersed in different volumes of 0.2% TOPO (w/w at toluene) solutions thus preparing various stock solutions with accurately pre-determined additive ratios, defined as the number of TOPO molecules per NR. This was verified by TGA analysis of a clean precipitant incorporated with 45% w/w TOPO (Fig. 3, green trace), which yielded 47% TOPO content, which is considered as free ligands compared to the bound phosphonic acids 38–40. Therefore, by extracting and reinserting TOPO molecules we do not significantly change the characteristics of the surface of the NRs.
Figure 4a shows CR values for rubbed thin films of 67 nm × 5 nm CdSe/CdS seeded rods, incorporated with different additive ratios (TOPO/NR) from null up to 8000 and for spherical QDs. As expected, for spherical QDs a CR of unity was maintained throughout. For the NRs, a maximal value of CR = 3.0 was obtained for an optimum additive ratio of ∼3000 TOPO molecules per NR. At lower TOPO/NR ratios the film was less responsive to the rubbing exhibiting strong adhesion and shear resistance. Increasing the TOPO/NR ratio beyond the maximal point, led to reduction in CR and visual decrease of the film optical quality as an opaque layer of the excess organic material was observed. A similar dependence of the additive to NR ratio was measured for NRs of different lengths, ranging from 26 to 90 nm (see Figs. S5–S7 of Supporting information). This is summarized in Fig. 4b where a linear dependence was observed between the particle surface area and the additive ratio with a slope of 2.7 TOPO molecules per nm2 surface area.
The excess additive ratio also affects material removal during the rubbing process. The optical density (OD) of spin-coated films of 67 nm × 5 nm NRs was compared before and after rubbing for different additive ratios (see Fig. S4 of Supporting Information). Most of these films had thickness of 70 nm, and we observed that at the optimal additive ratio for high CR values, the material loss was also small – amounting to ∼9% upon rubbing.
The optimal additive ratio common for all NRs lengths may be understood considering the common unit cell of the CdS shell, and the similarities in the synthesis procedure, yielding an identical number of capping ligands per unit surface area of the different NRs. Assuming passivation of all Cd atoms for a 67 nm × 5 nm CdS shell yields approximately four bound capping ligands per nm2. The optimal additive ratio of 2.7 TOPO molecules-nm−2, therefore, corresponds to an excess of nearly 1:1 of the surface capping ligands in the matrix. The ligands play an important role in the particle–particle interactions, and the additives partially mask these interactions providing an environment more suitable for the mechanical manipulation by the rubbing fibers. The observation of a constant value of additives per NR surface unit area provides a predictive recipe for optimal alignment of NRs with different dimensions, while minimizing material removal.
The CR of a pre-treated glass substrate coated with 67 nm × 5 nm CdSe/CdS NRs and rubbed was determined by imaging the fluorescent emission through an analyzer at two perpendicular polarization states, parallel (Fig. 5a) and perpendicular (Fig. 5b) to the rubbing direction, thus providing a visual indication to the macroscopic order, manifested as a change in the observed fluorescent intensity. The corresponding contrast map (Fig. 5c) and the CR distribution (Fig. 5d) both indicate an average CR of 3.5 across the entire sample, with localized areas exhibiting CR values of up to 4.5. The absorption spectra (Fig. 5e) of NRs in solution, after deposition, and after rubbing show no significant change in the optical features indicating that the NRs sustain the physical forces applied during the rubbing. Similar CR values were obtained by rubbing a Si/OTS (octyltrichlorosilane) substrate coated with 67 nm × 5 nm CdSe/CdS NRs (see Fig. S8 of Supporting Information).
Microscopic measurements were also conducted by a fluorescent microscope mounted with a linear polarizer.
Imaging the fluorescence intensity through a ×20 objective revealed the underlying linear rubbing pattern (Fig. 6a and b). Observing the resulting contrast map (Fig. 6c), further emphasizes this pattern depicting high CR values of 4.5 confined in longitudinal lines. The resemblance between the average CR determined on the macroscopic scale (3.5) and that determined on the microscopic scale (3.9) validates the accuracy of both measurement techniques.
Fluorescence microscopy measurements were also performed on the rubbed sample whose AFM characterization was shown in Fig. 2. The pattern shown in Fig. 7a and b is in good correlation with the topographic measurements. This indicates the directional assembly of the NRs in parallel lines, also indicated by the CR map (Fig. 7c), and separated by depleted regions from which no emission is visible. These measurements resulted in high CR values of up to 3.5.
Modeling the emission generated upon excitation of the macroscopic sample, was accomplished by representing the collective behavior of the sample as an ensemble of radiating dipoles for which the elongated axis is oriented at degrees with respect to the Z-axis (the rubbing direction). The dipoles are normally distributed around the rubbing direction, , with an standard deviation (SD) of σ. The resultant vectors, as viewed by the detector, were calculated by super-positioning all the separate dipoles to give three predominant emission components (, (in-plane), (out-of-plane)), which were then projected through the analyzer onto the detector's plane. The emission profile of a rubbed sample was measured and compared to the theoretical behavior of a polarized source, as given by the generalization of Malus' equation 41, 42
where I is the measured emission intensity, the intensity parallel (perpendicular) to the polarized axis of the source, and α represents the angle between the analyzer and the polarization axis of the source.
The disorder degree was determined from Eq. (3) by modeling the intensity as a function of the orientation distribution.
and are the emission dipole components of the NR deduced through single particle measurements indicating up to 75% polarized emission for a seeded CdSe/CdS NR 10. Seventy-five percent of polarized emission, which translates into a CR of 6. Figure 8a shows the intensity versus analyzer angle for various values of the SD σ. The experimental data for the rubbed film is also shown, and matches well the simulated curve with σ = 25°, corresponding to an angular distribution width of ±12.5°. This indicates good orientation uniformity. Figure 8b compares the clear angular dependence for the rubbed NRs with reference samples comprising of a spin-coated film of NRs and a rubbed film of QDs. The latter exhibits isotropic emission.
We presented a study of alignment of semiconductor NRs using mechanical rubbing. The method may be also considered for other types of colloidal NRs such as metallic NRs. The rubbing process was optimized through tuning surface–particle and particle–particle interactions yielding high polarization CRs of 3.5 relative to the previously reported values of 1.75 and 2.2 attained by rubbing NRs layers. Moreover, we report a way to minimize the amount of material loss down to 9%, for 70 nm thick films, by incorporating approximately three excess TOPO molecules per unit surface area, followed by depositing the thin film on a partially hydrophilic substrate. Characterizations of the rubbed samples using optical and physical methods provide a view on the alignment mechanism in which the NRs are deflected from the path of the rubbing fibers forcing the reorientation parallel to the rubbing direction. We presented optical characterizations of the aligned samples revealing good order uniformity with a SD of ±12.5° as derived by model calculations fitting the experimental intensity values. Mechanical rubbing, widely used in the LC industry, may be considered also for the alignment of colloidal NRs with the ability to apply it for large areas.
The project was partially funded by the Farkas Center for Light-Induced Processes, supported by the Minerva Gesellschaft fü r die Forshung GmbH, München. U.B. wishes to thank the Alfred and Erica Larisch Memorial Chair in Solar Energy.