1. Introduction

Colloidal semiconductor nanocrystals (NCs) have been widely reported in recent years for applications including selective biological labeling [1] and optoelectronics [2,3]. The most-studied NCs have been those with a CdSe core and a ZnS shell, which can be surface-functionalised, and tailored by varying their mean size to give narrow photoluminescence (PL), sharp excitonic absorption and high quantum yields across the visible spectrum. Usually they are dispersed in a solvent such as toluene, chloroform or water [4]; however, for many applications such as optoelectronics the NCs are required in solid film phase. Polymers are attractive candidates to be blended with NCs for such applications as they are easily processable materials and are readily transparent in the near UV and visible region provided they do not quench the NC luminescence and keep them in a uniform distribution. Surface modification can be applied to make the NCs compatible with most relevant organic materials [5], but poly(methyl methacrylate) (PMMA) was first reported as a matrix for as-received NCs [6]. Spectral shifting in the photoluminescence was a major drawback in the latter case which has subsequently been addressed using a pre-polymerized polymer [7,8]. In recent developments, epoxy resins have also been used as matrices with single-wavelength or mixed NCs to realise encapsulating color converters for light-emitting diodes (LEDs) [9]. Spectral conversion of light-emitting diodes emission has also been achieved using fluorescent microspheres [10].

We report here a further step in the integration of NCs with LEDs by developing a ‘self-aligned’ technique [11] capable of controllably micro-forming NC/epoxy resin nanocomposites on micro-pixellated AlInGaN LEDs (‘micro-LEDs’) [12]. Micro-LED arrays are very attractive pattern-programmable optoelectronic micro-projection sources being explored for a wide range of display and instrumentation applications [13,14]. They come in many formats, but typically consist of a few hundred to a few thousand individually-addressable micro-pixel LED elements in a device active area of a few square millimetres, emitting at one given wavelength defined by the particular nitride epitaxial structure used. By utilizing devices operating at 370nm, we show that one can selectively fabricate nanocomposite microstructures aligned to the underlying active device structures, thus opening up the benefits of quantum dot color converters and potentially integrated quantum dot laser structures with these sources.

The UV micro-stripe AlInGaN LEDs used as demonstrators in this work are chosen for convenience and for their relevance to producing multi-color versions [15] of the structured illumination microscopy we have explored previously [16]. They involve a similar epitaxial structure and fabrication process to those reported earlier [17]. Rectangular parallel mesa structures, 3600×20µm², are dry-etched down to the n-GaN underlayer using inductively coupled plasma (ICP) etching, creating 120 side-by-side micro-stripe pixels in the LED epilayer with a 34µm centre-to-centre spacing. As a result, all the mesas share a common n-electrode. A p-metallization layer runs on each pixel with an independent connection pad for individual current injection. The emitting area of each pixel is defined by an aperture in the p-metal layer 300µm long and 12µm wide. The NCs used in these experiments are commercially available CdSe/ZnS samples (Evident Technologies Inc.) dispersed in toluene with emission from blue (481nm) to red (610nm). The epoxy resin used is a cross-linkable polymer photosensitive at 370nm.

2. Nanocomposite preparation

The nanocomposites are blends of three components comprising the colloidal NCs in toluene, an epoxy matrix in a pre-polymerised form with photo-initiator which acts as a host matrix and a vinyl ether monomer which is believed to help the dispersion of the NCs in the epoxy matrix. The nanocomposite preparation aims to achieve a solution of 0.2% weight percentage (w/w) of NCs in the epoxy matrix. First, the NCs are mixed with a fixed weight of vinyl monomer for all the NCs in toluene. The solution is then agitated in an ultrasonic bath at room temperature for 30 minutes. The UV-sensitive epoxy pre-polymer is then weighed on a high precision scale and added to the NCs/vinyl solution to achieve the designated concentration. Further agitation in an ultrasonic bath is then carried out for 1 hour. As a consequence, the solution has a uniform dispersion of NCs with a low viscosity due to the dilution of the epoxy pre-polymer by toluene. In addition, at this stage this inorganic/organic blend is UV-curable due to the epoxy pre-polymer but it has no particular sensitivity to temperature. The last step of the nanocomposite preparation is then the evaporation of the toluene and other solvents contained in the epoxy pre-polymer to produce a more viscous solution while preventing the latter from curing. To achieve this, magnetic stirring on a hot plate at 180°C (the toluene boiling point is around 110°C) in air for several hours is performed, with rapid cool down under tap water every hour to stop evaporation and determine the viscosity of the solutions. The solvent evaporation time varies with the type of NCs used in the blend. The evaporation time increases for NCs emitting at shorter wavelengths. Table 1 lists the NCs used to make the respective solutions, along with their concentrations in toluene. As the NC concentration decreases when their emission goes to the blue region of the spectrum, a larger volume of NCs in toluene has to be mixed with the epoxy pre-polymer, for example in the case of the blue emitting NCs compared to the red emitting ones, in order to achieve the same concentration. As a consequence, the red NC solution has less added solvent to evaporate and takes about 5h30 while this time increases to 7h30 in the case of blue emitting NCs. These evaporation times are empirical values and do not result from viscosity measurements; as a consequence, the viscosities of the solutions are all different and are the result of almost complete evaporation of the solvents. The solutions are usually kept in a refrigerator and they are left at room temperature for 30 minutes prior to any processing. No apparent degradation of the nanocomposites has been noticed while samples are kept in these conditions for several months.

3. Integration experiments

Four solutions of 0.2% w/w were prepared with the NCs summarised in table 1. The self-alignment technique uses individual micro-LED elements emitting around (in this case) 370nm to locally photocure the nanocomposite deposited on top which then provides an appropriate aligned micro/nanostructure to color convert the output of the same element. To achieve this, the nanocomposite is initially spin-coated onto the device. The packaged micropixelated LED is loaded in a spin coater and a film of NCs/vinyl/epoxy solution is deposited using a spinning speed of 4000 rpm for 40 seconds achieving 20 to 25µm thick films. The micropixels to be covered are then switched on at a specific driving current for a certain amount of time to give the correct exposure dose to the polymer film that triggers curing and cross-linking. The uncured nanocomposite is then washed off using a toluene bath for 20 seconds followed by rinsing in deionised water. These steps can be repeated several times on the same pixel to build up a thicker encapsulation. The devices have been electrically characterised and they have a typical turn-on voltage of 3.6V. All the self-alignment curing has been done at a driving current of 0.5mA giving an optical power density of 4.4mW/cm² in the spin-coated film. Depending on the nanocomposite considered, different exposure doses are required; the red NCs nanocomposite needs a dose as high as 93mJ/cm² while the blue NCs blend requires only 9mJ/cm².

Table 1. Commercial CdSe/ZnS core/shell NC absorption, fluorescence characteristics, FWHM and concentrations in toluene (from datasheet). Fluorescence peaks are given with an error margin of +/-10nm.

View Table

This difference could be attributed to a photon harvesting competition between the photosensitive initiator of the epoxy pre-polymer and the NCs but also to a modification in the curing kinetics within the matrix itself due to possible polymer chains propagation alteration by the NCs. The difference of exposure doses is directly related to the typical absorption curves of the corresponding NCs. With an excitation wavelength near 370nm, as in our case, red NCs absorb much more strongly than the blue NCs as shown in fig. 1. To separately characterize absorption and emission properties, each nanocomposite was spin-coated on clean glass substrates and spectrally analysed with a UV-VIS Cary spectrophotometer for the absorption and a Perkin Elmer LS 50B for the fluorescence spectra. The film thicknesses were then measured with a stylus profilometer. A reference baseline of a clean glass substrate with a corresponding spin coated and cured vinyl ether/epoxy polymer was also measured in both instruments. The resulting absorbance spectra confirm the assumption of differing absorbance for different types of NC used in the nanocomposite fixed at the same concentration (0.2% w/w). For instance, the red nanocomposite has an absorbance at 370nm of about 65cm^-1 while the yellow, green and blue nanocomposites have lower absorbance of about 42cm^-1, 39cm^-1 and 32cm^-1, respectively. The fluorescence spectra of the different nanocomposites also included in fig. 1 show peaks at respectively 604nm, 555nm, 518nm and 477nm. All the peaks are slightly blue-shifted in comparison with the centred values from the datasheet given in table 1 but within the error margin for sample variability. The measured full widths at half maxima (FWHM) are very similar to the manufacturer’s values: 30.5nm for the red and yellow nanocomposites and 32.5nm for the green and 35.5nm for the blue nanocomposites.

 

Fig. 1. Absorbance and emission spectra of the NC nanocomposite films after photocuring

Download Full Size | PDF

Figure 2(a) is an optical micrograph under white light illumination of an integrated nanocomposite (left) produced by the self-aligned direct writing technique. The patterned nanocomposite covers the entire emitting area of the micro-pixel and only that micro-pixel. A schematic of the integration (centre) and a typical micro-pixel in operation before self-aligned writing are also represented (right).

 

Fig. 2. Optical micrographs of respectively, (a) an integrated nanocomposite microstructure under white light illumination (left), a schematic of the integration (centre) and a representative single operating UV pixel (right) and (b) encapsulated micro-LEDs turned-on showing efficient blue, green, yellow and red colour conversion from the UV-emitting micro-pixels by the NCs nanocomposites.

Download Full Size | PDF

Figure 2(b) shows the full range of nanocomposites microstructures integrated onto one micro-LED array. As a consequence, a full visible range of colours is enabled on a same micro-LED array chip with up to five different colour emitters (including UV). However, a broadening of the microstructures is noticeable while considering the nanocomposites emitting to longer wavelengths. This effect is attributed to the increase of the exposure doses for some blends while the microstripes LEDs are driven at constant injection current (constant optical output power). With a constant optical output power, a longer exposure time is needed for achieving a high exposure dose, which results in active cationic polymerization centres to diffuse out of the illumination area. A simple way to address this issue would be to keep the exposure time constant while changing the micro-LED output to meet the requirement for the exposure dose. A pulsed mode operation for curing could also be employed to achieve uniform polymerisation [17] and is under investigation with our devices.

4. Results

The spectral analysis of these hybrid devices is employed to characterize the colour conversion. Blue, green, yellow and red nanocomposites have been integrated by the self-alignment writing method. The PL spectra of the integrated NCs nanocomposites are measured with a UV-visible fibre-coupled calibrated optical spectrometer (Ocean Optics USB4000) with 800ms integration time and aligned with the considered micro-emitter through a 200µm wide slit placed aligned perpendicularly to the micro-pixel and approximately in the middle so emission from about 50µm on both sides of the micro-pixel is not collected by the set-up. Fig. 3 gives the spectral analysis of the emission of particular pixels driven at an injection current of 3.6mA and normalised in respect to the unconverted UV peak intensity. For clarity, and to display all the results together, only the UV emission corresponding to the optical pumping of the red-emitting nanocomposite is shown on this graph, but it was common to all. In addition, the emission spectra for the yellow, green and blue nanocomposites are vertically offset for clarity. While the micro-LED emission peaks at 370nm is clearly noticeable, peaks at 609nm, 560nm, 524nm and 486nm demonstrate the successful optical pumping by the micro-pixel directly underneath and the colour conversion by the NCs nanocomposites from UV to different colours in the visible, from blue to red.

 

Fig. 3. Emission spectroscopy of a device with blue, green, yellow and red NCs nanocomposites encapsulating single micro-pixel driven at the same injection current of 3.6mA

Download Full Size | PDF

The emission spectrum of the yellow nanocomposite in fig. 3 shows an extra peak at 524nm along with the expected peak at 560nm, which is attributed to an adjacent integrated green-emitting nanocomposite which is optically pumped by scattered UV light from the mesas of the micro-emitters. The absorption properties of these NCs nanocomposites allow a radiative energy transfer of the yellow emission to red-emitting NCs nanocomposites and, generally, from high energy emitting to lower energy emitting nanocomposites (fig. 1). The red-emitting nanocomposite is susceptible to absorbing the emission of all the others but as it was spaced several micro-LEDs elements away on the array from the blue, green and yellow-emitting NCs, only a very weak red-contribution is noticeable in the fluorescence spectra concerned.

The optical output power was then characterised to determine the colour conversion efficiency of such devices. First a bare micro-LED was characterised using a UV-enhanced Si photodetetector placed in close proximity (~2mm) to the emitting pixel. The pixel is turned on and the light output power is collected for different driving current. Fig. 4 plots these results with an emission at 370nm. A bare microstripe with dimensions of 300µm×12µm gives an optical output power as high as 34µW at a driving current of 40mA (~944mW/cm² optical power density at a driving current density of 55A/cm², current injected over the full length of the mesa structure, 3600×20µm²). Subsequent measurements are carried out after the nanocomposite integration on similar microstripes. The same measurement method is used through coloured glass long-pass filters to remove the unconverted UV. For the red-emitting nanocomposite, a cut-off frequency at 520nm and 95% transmission at 610nm is chosen. The results are also plotted in fig. 4; the output power collected at 609nm reaches a maximum of about 6µW at a driving current of 40mA (~166mW/cm² optical output power at 55A/cm² current density). The approximate colour conversion efficiency at 40mA driving current can be estimated for the red-emitting NCs nanocomposite to be about 17.7% (with compensated loss from the filter) (18.2% at 30mA). Yellow (560nm), green (524nm) and blue-emitting (486nm) nanocomposites optical output powers are also plotted in fig. 4 and achieve respectively about 3µW, 0.15µW and 0.05µW at 40mA driving current (~83mW/cm², ~4mW/cm² and ~1.3mW/cm² at 55A/cm²). The respective estimated colour conversion efficiencies are 9%, 0.5% and 0.2% (with compensation from filter loss).

 

Fig. 4. Typical optical powers versus injection current for UV microstripe LED and integrated NC nanocomposites by self-alignment writing.

Download Full Size | PDF

The characterisation of the eventual saturation of the emission of these NCs nanocomposites is an interesting feature to inspect. Figure 5 shows the experimental spectral data relevant to this experiment. Spectra are taken with increase of the driving current in the device. Although only the pixel with the NCs nanocomposite emitting at 609nm is turned on, shoulders emitting at 526nm and 561nm are noticeable, although no radiative transfer is possible in this situation, these peaks are attributed to UV light scattering out from the microstripe mesa structure optically pumping these adjacent nanocomposites. The spectral evolution shows no wavelength shift in the case of the nanocomposite; however, the peaks – both UV and red – broaden. The evolution of the peaks at 370nm and 610 nm were then recorded and the area underneath each was integrated to take into account this broadening effect. The UV peak is integrated on the 350nm to 409nm range and the peak at 610nm from 590nm to 630nm. The ratio of the integrated values of the 610nm peak over the UV emission peak at 370nm (S_NC/S_UV) is then plotted versus the driving current in inset of fig. 5. A linear law could fit this evolution meaning that no saturation of the nanocomposite was observed; however, the conversion coefficient is not constant over the driving current which could be attributed to different spectral broadening of the micro-LED and the NC emissions. The conversion efficiency is approximated to be 18% at a driving current of 30mA. This value reasonably matches the estimation obtained with the optical output power. Further saturation study shows that a focussed laser beam down to a 500µm diameter spot with optical power ranging from 0mW to 15mW does not saturate a 20µm thick nanocomposite film at a concentration of 0.2%w/w with either blue, green, yellow or red NCs. Saturation and lifetime characterisation of these nanocomposites are in progress and will be reported later.

 

Fig. 5. Red-emitting NCs nanocomposite emission spectrum optically pumped by the micro-LED underneath over the driving current. Inset, peak area ratio of the emissions at 370nm and 609nm.

Download Full Size | PDF

4. Conclusion

We have reported the successful use of a self-alignment technique to integrate NC nanocomposites onto micro-scale LEDs, as a means of producing multi-color programmable microstuctured sources for a range of applications in displays, microscopy and optoelectronics. The NC nanocomposites are a mix of the NCs in toluene, a vinyl ether monomer and an epoxy polymer. After solvent evaporation, the solutions present themselves as a viscous, spin-coatable and UV-sensitive solution with NC densities ranging from 64×10¹⁴ cm^-3 to 525×10¹⁴ cm^-3 respectively for the red to blue nanocomposites. The measurements of the absorbance and fluorescence spectra of cured nanocomposites spin-coated onto glass show the successful creation of NC nanocomposites with typical characteristics similar to the colloidal NCs in toluene. The integration of the nanocomposites covering emitting area of single microstripe LEDs shows good color conversion characteristics. Color conversion up to 18.2% is thus achieved in the case of the red-emitting NCs for an injection current of 30mA. Converted optical power density up to 166mW/cm² at 609nm is achieved at a driving current density of 55A/cm² in a micron scale format. This technique and the advantage of using UV devices allow the fast creation of multi-colour micro-LED chips with emission ranging from the UV to the red making this sort of device particularly attractive for lab-on-chip and other instrumentation applications.