Core–shell carbon-polymer quantum dot passivation for near infrared perovskite light emitting diodes

A 27.7 mg formamidinium iodide (FAI) purchased from GreatCell Solar, 33.2 mg lead iodide (PbI₂) obtained from Sigma-Aldrich, and 7.7 mg 5-aminovaleric acid (AVA) purchased from Sigma-Aldrich dissolved in 1 ml of anhydrous N,N-dimethylformamide (Sigma-Aldrich). The precursor solution was stirred for 2 h at 80 °C in an argon-filled glove box before use.

Patterned indium-tin oxide (ITO) coated glass substrates with sheet resistance 8 Ω square⁻¹ (Ossila) were used for PeLEDs fabrication. The substrates were cleaned in deionized (DI) water, acetone, and 2-propanol, for 10 min and dried with N₂ before they further treated by UV-ozone for 20 min. A solution of colloidal dispersion of tin(IV) oxide in DI water (15%, Alfa Aesar) was spin-coated on ITO substrates at 3000 rpm for 30 s. The SnO₂ films annealed at 150 °C for 30 min. For the BDP modified SnO₂ films, a solution of 0.5 mg ml⁻¹ in chlorobenzene (CB) was spin-coated at 2000 rpm for 40 s atop SnO₂ film. The perovskite precursor solution was spin-coated at 3000 rpm for 1 min, followed by annealing at 100 °C for 16 min. As the hole transport layer (HTL), poly-TPD (poly(N,N'-bis(4-butylphenyl)-N,N'-bisphenylbenzidine), Sigma-Aldrich) was spin-coated from CB solution (13 mg ml⁻¹). Molybdenum oxide (MoO₃, 10 nm) and aluminum (Al) were thermally evaporated to serve as the hole injection layer and anode electrode, respectively. For the device characterization, the current density–voltage–luminance (J–V–L) curves, radiance, and EQE were all carried out with an Oriel spectrometer coupled with a fiber integration sphere (FOIS-1) and a Keithley 2400 source meter.

Fourier-transform infrared (FT-IR) spectroscopy was performed using a Bruker Tensor 27 spectrophotometer with a DTGS detector. ¹H NMR spectra were recorded on a Bruker Avance 500 MHz spectrometer as solutions in deuterated DMSO-d₆ using the residual solvent signal as the internal standard and were processed with Topspin 4.0.8. X-ray photoelectron spectrum (XPS) was recorded by a Leybold EA-11 electron analyzer operating in constant energy mode at a pass energy of 100 eV. The x-ray source was a nonmonochromatized Al K line at 1486.6 eV (12 keV with 20 mA anode current). The work function (W_F) of pristine and BDP-modified SnO₂ films were estimated after recording the ultraviolent photoemission spectra (UPS) of the films deposited on ITO substrates used for PeLEDs fabrication. For the UPS measurements, the He I (21.22 eV) excitation line was used and a negative bias of 12.28 V was applied to the sample in order to separate sample and analyzer high binding energy (BE) cut-offs and estimate the absolute work function value. Atomic force microscopy (AFM) images were captured using an XE7 microscope (Park Systems) by operating in tapping mode. The crystallinity of perovskite films was investigated using an x-ray Siemens D-500 606 diffractometer. UV–Vis absorption spectra of the polymer solutions (concentration 10⁻⁵ M) and the prepared films were recorded with a Shimadzu UV-1900 spectrometer. Steady-state PL measurements were performed with a commercial platform (ARKEO—Cicci Research): the substrate was illuminated with a diode-pumped solid-state Nd:YVO4 + KTP laser (peak wavelength 532 nm ± 1 nm, optical power 1 mW on a circular spot of 2 mm of diameter: 31 mW cm⁻²) at an inclination of 45°. The fluorescence on the opposite side of the substrate is focused on a bundle of fibers (10 mm in diameter) with an aspheric lens close to the substrate to maximize the PL. The bundle sends the signal to a CCD-based spectrometer. Integration time and the number of averaging was maintained the same to better compare the results. Time-resolved PL (TRPL) spectra were measured with an FS5 spectrofluorometer from Edinburgh Instruments. A 478.4 nm laser was used as an excitation source. Scanning electron microscopy (SEM) images of the perovskite films were recorded with a JEOL 7401f FESEM.

The quantum dots (QDs) used in this study were synthesized through a bottom-up hydrothermal method, according to Ladomenou et al [29]. In brief, citric acid (CA) and ethylene diamine (EDA) were used as starting materials and were reacted in an autoclave, followed by heating at 180 °C for 8 h (figure 1(a)). A purification step took place for the final isolation of the desired product. This method has been previously reported to yield pure carbon quantum dots when heating above 200 °C [30, 31]; however, we intentionally applied here lower temperature during the synthesis in order to receive intermediate products with different degrees of polymerization [32]. We aimed to isolate shell-core polymer-carbon QDs having the molecule 5-oxo-1,2,3,5-tetrahydroimidazo[1,2-α]pyridine-7-carboxylic acid (IPCA) as the polymeric shell. The mixture or assembly IPCA-carbon dots (PCDs) was purified in a 1000 Da dialysis membrane and the fluorophore IPCA has not been removed, as shown in ¹HNMR spectra (figures 1(b) and S1). Figure 1(b) (red line) presents a spectrum of QDs that lacks starting materials. The aliphatic protons of carbon cores at 4.2–2.5 ppm are overlapped with two characteristic aliphatic triplet peaks. Then, two peaks at 6.0–5.6 ppm characteristic of double bond confirm the presence of IPCA in the carbon dots solution. Many differences between PCDs and simple carbon dots produced from CA were observed (figure S2).

Zoom In Zoom Out Reset image size

The existence of carbon core in the PCDs is evident by the transmission electron microscopy (TEM) measurements, presented in figure 1(c). We can conclude that the product contains a carbon core with a diameter of around 2.5 nm. The existence of an extensive amine network in the polymeric shell of these dots is expected to offer bulk and surface passivation in the perovskite through effective coordination with undercoordinated lead cations (Pb²⁺), which will benefit the device performance and stability.

Besides the emissive layer, the charge injection/transport and interlayer materials also play a crucial role in the device performance, as they dictate efficient charge injection and transport within the device layers and heterointerfaces; they can also modify the nanomorphology of the perovskite emitter deposited atop. Therefore, we first investigated the optoelectronic and structural properties of SnO₂ as-deposited and coated with a very thin BDP interlayer (from a 0.5 mg ml⁻¹ solution in CB).

The formation of a thin BDP interlayer on top of SnO₂ is evidenced by the XPS measurements, which showed the existence of nitrogen (N) in the BDP covered oxide (termed as the SnO₂/BDP sample). Figure 2(a) shows the N1s core level peaks of the XPS spectra taken in pristine SnO₂ and BDP/SnO₂ [33]. We observe that this BDP has covered the surface of SnO₂ having as a result a surface work function (W_F) reduction. From the shift towards higher BE in the electron cut-off of the ultra-violet photoelectron spectroscopy (UPS) measurements of BDP/SnO₂ (figure 2(b)), a W_F reduction of 150 meV (from 4.15 eV to 4.0 eV) is concluded. This shift is due to the formation of a negative interfacial dipole arising from the amino groups, which push electrons toward the SnO₂ and effectively reduce its surface W_F [34, 35]. This is expected to improve electron injection towards the conduction band of the perovskite emitter [36].

Zoom In Zoom Out Reset image size

On the other hand, a characteristic peak appears near the onset of the VB, in particular, at around 2.8 eV, in the UPS spectrum of BDP/SnO₂. This peak can be attributed to BDP as evidenced by the UPS spectrum of a pristine BDP layer coated (from a 5 mg ml⁻¹ solution in CB) on ITO/glass (figure S3). Figures 2(c)–(e) show the UV–Vis absorption, the derived Tauc plots and transmittance spectra of SnO₂ as-deposited and coated with BDP. Whereas the presence of BDP is evident by the characteristics absorption peak at around 500 nm (figure S4), it does not alter significantly the transmittance spectrum of the transparent bottom electrode. In addition, the energy bandgap value (E_g) was calculated from the Tauc plots to be 3.6 eV in both cases.

The surface topography of SnO₂ before and after the BDP coverage was further investigated with AFM measurements. Figures 2(f) and (g) show the surface topographies of as-deposited and BDP-modified SnO₂, respectively, where it is seen that the BDP modifier has covered the surface of SnO₂. The root-mean-square (RMS) surface roughness slightly increased from 0.35 to 1.14 nm, which is not expected to significantly alter the perovskite overlayer. Interestingly, the BDP-modified SnO₂ exhibited significant differences in water contact angle (figure 3(a)) and adhesion in DMSO solvent (figure 3(b)), with respect to the pristine SnO₂ substrate. In particular, the pristine oxide surface exhibits high hydrophilicity due to the high amount of hydroxyl groups present on the oxide's surface. This, however, has been proven detrimental for the device operation as these negative hydroxyl groups may attract positively charged ions within the perovskite layer [37]. A layer of such positive ions may thus built-up at the respective interface hence upward shifting the vacuum level and increasing the electron injection barrier.

Zoom In Zoom Out Reset image size

Fourier-transform infrared measurements were performed for the SnO₂ thin film, BDP thin film and BDP coated on top of SnO₂ (figure 3(c)). In the FT-IR spectrum of BDP, the bending modes of N–H can be detected at 1624 cm⁻¹ whereas the stretching modes appear at around 3500 cm⁻¹. The characteristic stretching band of SnO₂ made by the vibrations of the Sn–O–Sn bond is centered at 640 cm⁻¹. However, tin oxide exhibits a strong broad peak in the 3630–3220 cm⁻¹ area; this is the characteristic stretching band of hydroxyl group (O–H stretching). This band also appears centered around 3430 cm⁻¹ in the SnO₂/BDP spectrum but with lower intensity. One possible explanation of this is that in the pure SnO₂ film, the surface is rich with hydroxyl groups and with the addition of BDP these groups form extensive hydrogen bond networks with the –NH₂ groups of BDP thus being inactivated. As a result, reduced hydrophilicity is obtained for the BDP modified oxide surface, which increases the surface tension and also the measured contact angle for a DMSO droplet. We expect that this will directly influence the nanomorphology of the perovskite overlayer.

To gain an insight into the effect of the SnO₂ surface modification on the structural properties and crystallinity of the perovskite film coated atop, x-ray diffraction (XRD) measurements were performed. Figures 4(a) and (b) presents the XRD patterns of FAPbI₃ films (without AVA) spin-coated on SnO₂ without and with BDP surface modification, respectively. It becomes evident that the characteristic perovskite peaks are more pronounced in the SnO₂/BDP/FAPbI₃ sample. More interestingly, the full-width-half-maximum of the crystallization peak at ∼13.9° corresponding to the (001) plane is reduced indicating improved film quality with low roughness and smaller overall crystal size of 48.78 nm compared with the 56.57 nm for the FAPbI₃ spin-coated on the pristine SnO₂. Furthermore, the increased intensity of the same (001) crystallization peak also suggests that the surface modifier of SnO₂ enhances the preferred orientation of the perovskite film deposited on top and suppresses the formation of PbI₂.

Zoom In Zoom Out Reset image size

The nanomorphology of FAPbI₃ perovskite was also altered when coated onto the BDP modified SnO₂. AFM 10 × 10 μm² topographies of a 50 nm thick FAPbI₃ deposited on pristine and BDP-modified SnO₂ on glass/ITO substrates are shown in figures 4(c) and (d), respectively. In the reference sample, interconnecting grains are observed, which form a 'network' of long perovskite islands. However, in the case of perovskite deposited directly onto BDP, these grains are smaller and separated one from the other. The same conclusion can be drawn from SEM topographic images of the same samples (figure S5). This can be explained by taking into account the reduced hydrophilicity of the BDP-coated SnO₂ sample. In hydrophilic surfaces water molecules adsorbed onto them act as nucleation sites for the perovskite formation. As a result, extended perovskite grains are formed in hydrophilic surfaces which are highly desired for perovskite solar cells [37]. However, the situation is different for PeLEDs where smaller grains are necessary in order to confine both charges and excitons and increase the radiative recombination rate [38]. We therefore argue that the decrease in grain size in the perovskite coated onto the BDP modified substrate could be beneficial for LED operation [39]. Additionally, a reduction in surface roughness of the perovskite film deposited onto the BDP-modified ETL (from 22.48 nm for the reference sample SnO₂/FAPbI₃ to the 19.51 nm for the FAPbI₃ deposited on SnO₂/BDP) is indicative of a better contact quality between the perovskite emitter and the HTL deposited on top of it.

Moreover, the interfacial electron transport between the perovskite and SnO₂ in the absence and presence of the BDP surface modifier was studied by using steady-state PL and transient PL (TRPL) measurements. As shown in figure 4(e), the FAPbI₃ coated on the ITO/SnO₂/BDP substrate exhibits a 44% reduction in the PL intensity, indicating a faster electron extraction from the perovskite to the metal oxide, which can be due to a reduction in the electron injection/extraction barrier at the interface. This indicates that enhanced electron injection can also be obtained upon the application of a forward bias during the device operation. The same conclusion can be drawn from the decrease in carrier lifetime shown in transient PL (TRPL) spectra of figure 4(f). A red shift of about 10 nm the PL peak is also observed for the BDP/FAPbI₃ sample, which can be attributed to changes in the lattice parameters; in particular, a small shift from 13.89° to 13.83° for the (001) XRD peak of BDP/FAPbI₃ is also observed. These changes may originate from coordination of the amino groups of the BDP underlayer with undercoordinated I⁻ or Pb²⁺ within the perovskite lattice resulting in trap filling and better interface quality. Notably, the absorption spectra of the two samples were nearly similar (figure S6).

To passivate defects present in the perovskite bulk, we next applied PCD as the passivation agent embedded into the FAPbI₃ perovskite emitter. We first tested their effect in the perovskite microstructure (without AVA) and the nanomorphology of the resultant perovskite layer (without AVA). Figures 5(a) and (b) shows the SEM images of the perovskite derived from the spin coating (onto glass/ITO substrates) of the precursor solution without and with PCDs (in a concentration of 1 mg ml⁻¹ in the precursor). It becomes evident that PCDs largely affect the layer morphology resulting in the formation of well-defined nanograins of smaller size compared to the reference sample which consists of longer perovskite 'islands'. These small grains can strongly confine excitons and charges therein thus increasing the radiative recombination, which is crucial for perovskites that generally exhibit low exciton BEs of few meV [40]. These grains are slightly decreased in size when coated on top of the BDP layer. Figures 6(a)–(d) depicts the AFM topographies of FAPbI₃ films embedding PCD additive in a concentration of 0.4 mg ml⁻¹ and 1 mg ml⁻¹ (named hereafter as FAPbI₃ + PCD-0.4 and FAPbI₃ + PCD-1, respectively) on top of pristine SnO₂ (figures 6(a) and (b), respectively) and on BDP-modified SnO₂ substrate (figures 6(c) and (d), respectively). It becomes evident that both layers, FAPbI₃ + PCD-0.4 and FAPbI₃ + PCD-1, consist of smaller grains when deposited on the BDP modified SnO₂ exactly as in the case of pristine perovskite coated on as-deposited and BDP-modified oxides. Additionally, the FAPbI₃ + PCD layers deposited on the SnO₂/BDP exhibited lower RMS surface roughness compared with the perovskite film coated on the pristine SnO₂. Especially, in the case of the FAPbI₃ + PCD with concentration of 1 mg ml⁻¹, the RMS was 14.5 nm, while the SnO₂/FAPbI₃ + PCD-1 exhibited higher RMS of 23.3 nm. This significant improvement in the microstructure of the perovskite embedding PCD and of the nanomorphology of the perovskite film deposited on BDP coated SnO₂ is expected to be beneficial for the device performance.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Furthermore, an increase in the intensity of the steady-state PL spectrum of the perovskite film (coated on glass) with a 0.4% concentration in PCDs compared to the reference film indicates the passivation effect of the polymerized carbon dots (figure 6(e)). However, when inserted in a higher concentration these dots induce large decrease in the PL emission of the hybrid perovskite:carbon dot film. Notably, the UV–Vis absorption spectra of all samples were very similar (figure S7). The same conclusions can be drawn from TRPL decay curves (figure 6(f)) that also suggest a passivation effect of PCD with the lower concentration (i.e. 0.4 mg ml⁻¹). PL and TRPL measurements taken in perovskite samples without and with the carbon dots coated on as-deposited and BDP-modified SnO₂ also indicated enhanced charge injection, when using the BDP interlayer, as well as, the passivation effect of PCDs (figure S8).

PeLEDs based on FAPbI₃ perovskite emissive layers with and without PCD additive (0.4 mg ml⁻¹) using pristine SnO₂ or BDP-modified SnO₂ were next fabricated. They consisted of the following layers: glass/ITO (100 nm)/SnO₂ or BDP-modified SnO₂ (40 nm)/FAPbI₃ (+AVA) + PCD (50 nm)/poly-TPD (40 nm)/MoO₃ (30 nm)/Al (100 nm). The corresponding energy levels of each layer are shown in figure 7(a). By inserting the BDP modification interlayer on top of SnO₂, a perfect alignment between the highest occupied molecular orbital of the interlayer and the VB of the perovskite occurs. The insertion of a thin BDP interlayer on top of SnO₂ is expected to enhance electron injection and transport within the device.

Zoom In Zoom Out Reset image size

Figure 7(b) shows the electroluminescence (EL) spectra of the pristine and modified with BDP SnO₂, combined with FAPbI₃ and FAPbI₃ + PCD emissive layers. The devices showed a peak EL at 770 nm, which is blue shifted compared to the PL maxima (figure S8). There are several reasons that can explain the differences between the EL and PL spectra. PL is mainly determined by the optical properties of the material, while the EL is determined by a number of factors such as the optical properties and physical structures of the optically active layers, the electrical properties of two conductive regions, which are used for cathode and anode contacts, and the properties of the electrical contacts through which the electrical current injected. EL is usually performed on the finished devices (such as LEDs) since it needs a device structure to inject current. As LEDs are multilayer structures with thicknesses on the order of the emission wavelength, EL spectra will depend on both the thicknesses and the refractive indices of the individual layers as a result of interference effects in a 'weak' microcavity [41]. The above parameters will influence the position of the recombination zone hence affecting the profile and intensity of the EL spectrum. Furthermore, other phenomena such as suppression of Auger recombination under external bias has been reported to alter the profile of PL and EL spectra [42].

The current density and radiance vs. voltage characteristics show that the devices using the BDP modified SnO₂ open at lower voltage compared to the reference ones (based on pristine SnO₂), which suggests efficient carrier injection from the modified bottom electrode (figure 7(c)). More importantly, a large improvement in the radiance of the PCD-based PeLEDs was achieved. Especially, in the case of the device based on the FAPbI₃ + PCD-0.4 coated on the BDP modifier the measured peak radiance was increased to 92 W sr⁻¹ m⁻², representing a large improvement compared to the reference device without PCDs and BDP (2.5 W sr⁻¹ m⁻²). The device with the FAPbI₃ + PCD-0.4 emitter deposited on the pristine SnO₂ ETL showed also enhanced radiance of 54 W sr⁻¹ m⁻², indicating the beneficial role of the PCD on the device performance. Furthermore, a clear enhancement in the obtained EQE was found, when using the modified ETLs and the PCD PeLED emissive layers. EQE was measured to be 6% and 9% for the FAPbI₃ coated on the pristine and BDP modified SnO₂ ETL, respectively, which was increased to 10% and 14% for the PCD PeLEDs based on the SnO₂ and SnO₂/BDP substrates, respectively (figure 7(d)). Table 1 provides a detailed comparison on the current advancement in the FAPbI₃ perovskites PeLEDs concluding the performance of our device and recent state-of-the-art PeLEDs [43–54]. The optimized devices also showed reduced efficiency roll-off compared to the reference with no modification and additive (figure 7(d)). They were also quite reproducible as shown by the histograms of the device EQE data extracted from a batch of for 49 devices (figure 7(e)). The synergistic effect of our additive and interface engineering approach also resulted in enhanced device lifetime [55–57]. Figure 8 presents the variation of normalized EQE versus time of continuous operation of a reference PeLED (using FAPbI₃ and as-deposited SnO₂) and the optimized device based on SnO₂/BDP and PCD additive engineering approach. Both devices were biased at a current density of 50 mA cm⁻² (initial EQE of 6 and 13% for the reference and champion PeLED, respectively). The estimated T₈₀ lifetimes (the time where the EQE declines to 80% of the initial value) were 6 and 33 h for the reference and engineered device, respectively. This high performance and enhanced lifetime of the PCD PeLEDs is attributed to the passivation effect of PCD and the improved microstructure of the perovskite caused by the BDP and PCDs along with the favorable energy level alignment at the electron injection interface upon the BDP modification of SnO₂.

Zoom In Zoom Out Reset image size