A chemical sensor for CBr4 based on quasi-2D and 3D hybrid organic–inorganic perovskites immobilized on TiO2 films

Pavlos Nikolaou a, Anastasia Vassilakopoulou ab, Dionysios Papadatos a, Emmanuel Topoglidis *a and Ioannis Koutselas *a
aMaterials Science Department, School of Natural Sciences, University of Patras, Patras, 26504, Greece. E-mail: etop@upatras.gr; ikouts@upatras.gr; Tel: +302610969928
bCenter for Technology Research & Innovation (CETRI) Ltd, Thessalonikis 1 Str., Nicolaou Pentadromos Center, Limassol 3025, Cyprus

Received 30th November 2017 , Accepted 15th January 2018

First published on 16th January 2018

The effective immobilization of hybrid organic–inorganic semiconductors (HOIS) or their blends over standard mesoporous, transparent, semiconducting TiO2 films in order to study their electrochemical behavior as sensors through a cyclic voltammetry technique (CV) has been reported for the first time. The CV technique is used in order to acquire a comprehensive understanding of the electron transport dynamics within the TiO2 film, loaded with perovskites, while the structural and optical properties can be easily extracted from X-ray diffraction, optical absorption and photoluminescence measurements. The electrochemical behavior of the adsorbed perovskites was studied as a function of the potential applied to the semiconductive film, which induces transition from an insulating to a conductive state. The localized traps in the bandgap of the composite semiconductive film predominantly govern the electron transfer process. The HOIS employed are either 3D or mixtures of quasi-2D perovskites, for which there is strong evidence that the lead bromide and chloride based HOIS exhibit cathodic peaks at positive voltages. In addition, exploiting the capability of ions to migrate and interact with the HOIS lattice, a perovskite-based electrode was successfully used as an (electro-)chemical sensor for CBr4. All electrochemical processes concurring on the composite hybrid film vary with respect to the CBr4 concentration in the tested solution, leading to a CBr4 detectability range of the order of 20 ppb mol mol−1.

1. Introduction

In recent years, mesoporous (mp) structured perovskite solar cells (PSCs) based on low dimensional (LD) or 3D HOIS1 have attracted much attention, as the power conversion efficiencies of PSCs have increased sharply from 4% to over 20%.2–4 HOIS are easily processed and low cost and their power conversion efficiencies exceed those of polycrystalline silicon solar cells.

In mp structured PSCs, semiconducting hosts such as TiO2 films are employed as electron transport layers or scaffolds of the perovskite layer.5 TiO2 is used due to its excellent physicochemical properties, such as its large band gap, large surface area, chemical stability, unique electrical and optical properties, non-toxicity and low cost. The TiO2 film acts as not only the scaffold of the perovskite layer but also the pathway for electron transport.6 It has been reported that the structural properties of the TiO2 film, such as particle size thickness and porosity, have a significant influence on the performance of PSCs.7 Highly porous TiO2 films promote the easy entrapment of perovskites, which subsequently fills the pores. A higher deposition of perovskites on the TiO2 film results in increased light absorption and a higher current density.8 Moreover, the interface between the mp-TiO2 film and the perovskite film plays a key role in determining the overall conversion efficiency of PSCs. Increasing the specific surface area and porosity of the TiO2 film can promote a deeper infiltration of perovskites into TiO2 films. The efficient contact between TiO2 and perovskites enhances electron transport.9

mp-TiO2 films have also been used for other applications ranging from dye-sensitized solar cells10 to chemical11–13 and biochemical sensors,14 electrochromic windows,15 photo- and/or electrocatalysis16 and energy storage devices such as rechargeable batteries17 and electrochemical supercapacitors.18 Most of these applications rely on electro- or photo-induced electron transfer reactions of redox-active molecules immobilized within the mesoporous network of the TiO2 films, which are deposited on conductive active substrates such as ITO or FTO glass.19,20,21 By applying a cathodic potential the TiO2 film turns into a conductive substrate when a sufficient number of electrons are injected into it and to an insulator under reverse bias. This is due to their n-type semiconductive properties and their wide band gap, ∼3.2 eV, allowing the adsorbed redox molecules to be functional. Electron transport can occur via hopping between adjacent redox molecules, or through the conduction band of a semiconductive TiO2 matrix.19 The latter occurs only when the film is rendered sufficiently conducting by applying the necessary cathodic potential and is a random-walk diffusion process. This electron diffusion has been shown to be ambipolar because it is coupled electrostatically with ions, present in the electrolyte.22 Another widely reported observation is a trap limited electron displacement due to the existence of an exponential distribution of localized trap states extending into the bandgap from the conduction band edge.19,23 Electrons move between band-gap states via the conduction band according to a trapping–detrapping process, a phenomenon that explains the dispersive electron transport in mp TiO2 films.24,25 In addition, the interfacial electron transfer at the underlying uncovered conductive substrate (e.g. ITO glass) and the interfacial charge transfer taking place at the electronically conducting TiO2 surface may also affect or limit the electrochemical conversion rate of the redox molecules adsorbed (entrapped) in the film's mesoporous network.26–28

HOIS, commonly known as perovskites, are multifunctional materials with interesting electronic and photonic properties for applications in photovoltaics,29 optoelectronics such as room temperature LEDs30 and electrochromism,31 while most of their properties are due to the strong excitonic state. Also, ionic migration within the lattice of the perovskite compounds allows for a variety of applications in electrochemical devices. HOIP should not be confused with perovskite oxides, which are also an interesting class of materials that have been used in high temperature sensors,32 and recently, interest in using perovskites to develop other types of electrochemical sensors has been increasing.33–35 In addition, native defects in hybrid lead halide perovskite materials are able to migrate within the perovskite structure because of the soft character of the perovskite compounds. How ions interact with the hybrid perovskite structure during the charging process is still unknown. Recently, it has been shown that perovskite based electrodes exhibit high stability upon electrochemical cycling without severe distortions of the crystal structure (drastic structural alterations).36

The hybrid organic–inorganic perovskites used in this work are of the form CH3NH3PbX3, where X = Cl, Br, and these perovskites are the active sensing semiconductors immobilized on the surfaces of the TiO2 films, in some cases modified with stabilizing amine. These perovskites exhibit a 3D framework of corner sharing connected MX6 (M = Pb, X = Cl or Br or mixed halogen content) octahedra, with protonated methylammonium cations located between them.1,37,38 These cations serve as stabilizers to the inorganic network, and also affect the optoelectronic properties of the inorganic network.1,38 Moreover, a mixture of CH3NH3PbBr3 and CH3NH3PbCl3 has been treated with phenethylammonium chloride (PhE·HCl) in order to further stabilize it, also inducing some structural changes to the former, either by replacing the Br with Cl and by inducing a phase transformation of the 3D to a blend of quasi-2D perovskites, each with the structural formula (CH3NH3)n−1(PhE-H)2Pbn(BryCl1−y)3n+1, where n = 1, 2, 3… and 0 ≤ y ≤ 1; when n = 1 and n = ∞ these are considered 2D and 3D HOIS, respectively.1 The band gap values (Eg) increase successively as n decreases.1 For given y, the various n crystalline structures exhibit distinct optical properties.

Organic halides (RX), such as CBr4, are widespread environmental pollutants which typically contain a carbon–halogen bond and are resistant to natural degradation. Halogenated hydrocarbons are used in the textile industry, and are in agricultural or domestic use. Accidental or deliberate release of these materials into soil, groundwater, etc., can exert long-term toxic effects and present serious health risks, and therefore their prevalence in ground water is of considerable environmental concern.39,40 Analysis of RX pollutants in water has relied on the use of liquid chromatography (LC) and gas chromatography-mass spectrometry (GC-MS), but both these methods are time-consuming and costly.41 In this work a new functional electrochemical sensor is proposed based on HOIS entrapment onto TiO2 thin films, and it has been observed that using particular lead halide based HOIS it is possible to observe the sensing of CBr4 in the ppb scale, while also unreported experiments show subtle interaction of this sensor with light irradiation and will be reported elsewhere.

2. Materials and methods

Soda lime glass slides (1.1 mm × 25 mm × 75 mm size) with 15 ohms sq−1 indium tin oxide (ITO) coating were obtained from PsiOTec, UK. Commercial 90-T transparent TiO2 paste with an average nanoparticle size of 20 nm was purchased from Dyesol, Australia. Tetrabutylammonium hexafluorophosphate (TBAH), acetonitrile (AcN), absolute ethanol, dichloromethane (DCM), anhydrous N,N′-dimethylformamide (DMF), lead(II) bromide (PbBr2, 99.999% trace metals basis), methylamine (MA −40% in water), phenethylamine (PhE, 99%), hydrobromic acid (HBr, ACS reagent, 48%) and anhydrous toluene (99.8%) were obtained from Sigma-Aldrich Chemical Co. Carbon tetrabromide (CBr4) was obtained from Alfa-Aesar. Dimethyl sulfoxide (DMSO) was obtained from Fisher reagents. All chemicals were of reagent grade and used without any further purification.

2.1 Preparation of mesoporous TiO2 film electrodes

The Dyesol TiO2 nano-product (90-T) is a commercially obtained paste, produced by a sol–gel/hydrothermal method. It contains highly dispersed and stable anatase nanoparticles of 20 nm size and was used without any dilution in ethanol or terpinol. This paste was used to prepare thin films of TiO2 on ITO glass slides, which were first cleaned in a detergent solution using an ultrasonic bath for 15 min and then rinsed with 18 MΩ water and ethanol. The TiO2 paste suspension was then applied to the surface of the conducting glass by the “doctor blade” technique. Masking each glass slide with 3 M Scotch Magic tape (type 810, thickness 62.5 μm) controlled the thickness and width of the paste spread area. One tape layer was employed for each TiO2 film deposition, which provided a film thickness of 3.1 μm and size of 1 × 1 cm2. The spread suspension was then allowed to dry before being sintered for 20 min at 450 °C. The resulting ITO substrates with the deposited TiO2 films were cut into 25 mm × 10 mm pieces.

2.2 Preparation and immobilization of the perovskite on the TiO2 films

For the typical synthesis of the 3D perovskite (CH3NH3)PbBr3, 5.4 mmol of PbBr2 was dissolved in 3 ml of AcN with the addition of 3 ml of HBr, and stirred until an optically clear solution was obtained. 13.3 mmol of methylamine solution was dissolved in 3 ml of AcN with 0.910 g of HBr by stirring. These two solutions were slowly mixed and dried afterwards. A similar procedure was followed for (CH3NH3)PbCl3; however, DMSO was used instead of DMF. The perovskites were adsorbed on the mp films, as such, or using mixtures of those Br and Cl analogues in a 1[thin space (1/6-em)]:[thin space (1/6-em)]1 weight ratio. In the last case, when converted into moles, the ideal resultant compound would be CH3NH3Pb(Br0.42Cl0.57)3. This final mixture compound (denoted DP209) will be analyzed further below with EDX for its halogen composition. Mixing 20 μl of a solution containing DP209 (200 mg in 200 μl of DMF and 200 μl of toluene) and protonated phenethylammonium hydrochloride (PhE·HCl) at 3 M concentration, in 600 μl of DMF, yielded a solution from which the final films were prepared. The latter solution should theoretically dry to the mixed halide 3D perovskite, DP209, and include blends of quasi-2D HOIS. Should there be a preferential replacement of the Br atoms by the Cl atoms or phase separation due to the limited solubility of the Cl analogue in DMF, then the perovskite's chemical formula would not have been CH3NH3Pb(Br0.42Cl0.57)3 but richer in Br. It is useful to add that CH3NH3PbCl3 is not easily soluble in DMF, yet in the presence of the Br analogue, those two HOIS can be dissolved. Perovskite immobilization was achieved by adding 20 μl micrometric volumes of the resultant mixed perovskite solution on the surface of a TiO2 film. The films were then heated at 60 °C for 30 minutes. The optical transparency of the TiO2 film allows the adsorption process to be monitored by UV-vis absorption spectroscopy. Perovskite adsorption on the TiO2 films was monitored by recording the UV/visible absorption spectra of the films at room temperature.

2.3 Characterization of the perovskite–TiO2 samples

X-ray powder diffraction (XRD) measurements of the perovskite–TiO2 films on ITO glass were performed with a Bruker D8 advance X-ray Diffractometer using Cu (λ = 1.54178 Å) radiation with an anode current of 40 mA and an accelerating voltage of 40 kV at a scanning speed of 0.015 deg s−1. The diffraction patterns were indexed by comparison with the Joint Committee on Powder Diffraction Standards (JCPDS) file numbers 21-1276 and 21-1272 for rutile and anatase TiO2, respectively. The morphologies of the samples and the thicknesses of the TiO2 thin film electrodes were analyzed using a ZEISS EVO MA 10 Scanning Electron Microscope equipped with an EDS analyzer (Oxford) with 127 eV resolution. Au sputtering on the specimen's surface was employed in some cases to increase the samples’ surface conductivities.

UV-visible optical absorption (OA) spectra were recorded on a Shimadzu 1800 spectrophotometer in the 300–800 nm range, in sampling steps of 0.5 nm at 1.5 nm slits, using a combination of halogen and deuterium lamps as sources. The photoluminescence (PL) and photoluminescence excitation (PLE) measurements were carried out on a Hitachi F-2500 FL spectrophotometer, employing a xenon 150 W lamp and an R928 photomultiplier. The excitation and detection slits were set at 2.5 nm and the accelerating voltage was set to 700 V. All spectra were recorded at room temperature (RT).

2.4 Electrochemical measurements

Electrochemical measurements were performed using an Autolab PGStat 101 potentiostat. The electrochemical cell was a 6 ml, three-electrode cell with quartz windows, employing a platinum mesh flag as the counter electrode, a Ag/AgCl in a 3 M KCl reference electrode and a perovskite–TiO2 film on ITO glass as the working electrode.

The electrolyte solution used was DCM with (Bsol) or without 0.1 M TBAH. All potentials are reported against Ag/AgCl and all experiments were carried out at RT. After the adsorption of perovskites on the nanostructured TiO2 films, we undertook an electrochemical study of their properties. Cyclic voltammograms (CVs) of perovskites on TiO2, in some cases as stabilized with phenethylammonium chloride (PhE·HCl), were acquired among other spectroscopic and structural data, while the same experiments were performed using CH3NH3PbCl3. In all experiments, the detectability of CBr4 was achieved by detecting variable electrochemical signals. PhE·HCl was synthesized as described by Vassilakopoulou et al.42 The HOIS used here are not soluble in DCM. A CBr4 stock solution (CSS) was prepared by dissolving 0.33 mg in 10 ml of DCM, which was used for introducing CBr4 in Bsol in multiples of 10 μl of CSS; 10 μl added are 10 moles of CBr4 per 109 moles of DCM (10 ppb mol mol−1).

2.5 Sample acronyms

Table 1 lists the acronyms used for sample naming along with some information.
Table 1 Acronyms for the reported samples
Acronym Specification
DP209 Solid obtained from (100 mg of CH3NH3PbBr3 + 100 mg of CH3NH3PbCl3 in 200 μl of DMF) + 200 μl of toluene
Bsol 10 ml of DCM with 0.1 M TBAH
Pn1 ITO with a thin layer of sintered TiO2 (3.15 μm thickness)
Pn11 20 μl of a solution synthesized from 20 μl of DP209 with 3 M PhE·HCl in 600 μl of DMF; deposited on Pn1
Pn15 30 μl of 50 mg of CH3NH3PbCl3 in 121 μl of DMSO; deposited on Pn1

3. Results and discussion

3.1 Structural characterization

Fig. 1 presents the XRD patterns of the ITO/glass substrate (1a), a similar substrate coated with a sintered thin TiO2 layer (1b, Pn1), CH3NH3PbCl3 (1c), CH3NH3PbBr3 (1d), and TiO2 films with immobilized perovskites DP209/PhE·HCl (1e, Pn11) and with CH3NH3PbCl3 (1f, Pn15).
image file: c7qm00550d-f1.tif
Fig. 1 Left: XRD patterns of (a) ITO/glass, (b) Pn1, (c) Pn15, (d) CH3NH3PbBr3, and (e) Pn11. Right: (a) ITO/glass, (b) Pn1, (c) CH3NH3PbCl3 and (f) Pn15. The inset shows Pn1, Pn11 and Pn15 patterns for comparison purposes.

The patterns have been arranged in two parts to help the reader relate the sample XRD measurements. The substrate's XRD pattern (Fig. 1a) revealed four characteristic peaks which correspond to the (222), (400), (440) and (622) planes of the ITO lattice structure and are consistent with the reported values of the JCPDS ITO card (6-416); the angle values are 21.7°, 30.65°, 35.68°, 51° and 60°. Also, a similar pattern (1b) revealed extra characteristic 2θ peaks at 25.28°, 37.8° and 48.05°, which correspond to the (101), (004) and (200) planes of anatase TiO2, but as expected no characteristic peaks of the rutile TiO2 phase were observed since the commercial Dyesol 90-T TiO2 paste is only anatase. The XRD pattern of the Pn11 sample (1e) coincides mostly with that of CH3NH3PbBr3; however, some of the latter's pattern peaks are shifted or missing such as the 33.38° and 38.6°, implying that the addition of CH3NH3PbCl3 (1c) has been successfully integrated with CH3NH3PbBr3 (1d), giving rise to a nominal compound described as DP209/PhE·HCl. Other quasi-2D moieties that may have formed do not always appear in the XRD patterns, since they are considered to be nano-sized particles. However, they certainly exist since the PhE molecule can form 2D HOIS, such as the (PhE-H)2PbX4 or other quasi-2D forms.39 The XRD patterns of the films have also been measured after the CV measurement, observing only a small decrease of the XRD peak intensities. The peaks of the pristine CH3NH3PbX3 (X = Br, Cl) are in accordance with other works.43 Similarly, the pattern of Pn15 (1f) coincides with that of CH3NH3PbCl3 (1c), confirming the successful entrapment of the latter on the TiO2 film, and its stability, as the pattern has remained the same before and after the CV measurements. The Scherrer analysis of the XRD pattern peak of TiO2 shows that the average diameter of the TiO2 particles is 14.8 nm, suggesting the porosity of the TiO2 film.

3.2 Electrochemical characterization

The Pn15 sample has been formed by immobilizing CH3NH3PbCl3 on the TiO2 film of the ITO-glass substrate. This active material is based on methylammonium lead chloride, which is assumed to be sensitive to bromine that could be offered by the CBr4 molecule, which serves as the target detection molecule. The final composite has been studied by CV, using DCM with 0.1 M TBAH as an electrolyte solution, while the same measurements have also been performed on a blank TiO2 film. Fig. 2 presents the CV scans for the bare Pn1 and Pn15 samples and that of Pn15 after the addition of CBr4 to the electrolyte solution.
image file: c7qm00550d-f2.tif
Fig. 2 CV scans for (a) a blank thin TiO2 film on an ITO glass substrate, (b) the same as in (a) with immobilized Pn15 and (c) the latter in the presence of 60 μl of CSS in 6 ml of DCM (all acquired at 0.1 V s−1).

The CV, presented in Fig. 2a, recorded for a bare TiO2 film is characteristic of the charging/discharging current associated with the injection of electrons into the sub-bandgap states of the TiO2 film. The charging of the TiO2 film usually begins at −0.3 V (in PBS buffer),8 while in our case the presence of the salt in the organic electrolyte serves as an inert electrolyte over a wide potential range. The current exponentially grows during the forward cathodic scan and then for the backward scan exponentially decreases during the reverse sweep. This behavior is assigned to the transition from an insulating to a conductive TiO2 state. No redox peaks are observed.

Fig. 2b shows the CV of the perovskite CH3NH3PbCl3 (Pn15) immobilized on a TiO2 film. The different electrochemical behavior of the Pn15 sample, compared to the bare film (Pn1), can be clearly observed. The CV of the Pn15 sample shows, in addition to the charging/discharging currents for the bare TiO2, two irreversible cathodic peaks at 0.48 V and −0.07 V. The first peak observed at 0.48 V is due to the perovskite's presence and especially the second peak observed at −0.07 V occurs because of the perovskite's chlorine.

The first peak is a single, not very sharp, big reduction (cathodic) peak located at 0.48 V, while in the reverse scan there is no clear re-oxidation (anodic peak), maybe a very broad and ill-defined wave of very small amplitude, indicating a very sluggish re-oxidation. The absence of a diffusion controlled reversible wave strongly indicates that the perovskite cannot move by diffusion through the TiO2 nanostructure for exchanging maybe electrons to the underlying ITO substrate. Also, it suggests the absence of electron hopping between neighboring perovskite molecules. Due to their molecular size and or their affinity to TiO2, we managed to add 20 μl of 3 M perovskite onto the TiO2 surface, resulting in high coloration of the film and suggesting a high surface coverage, which could contribute to the drastic decrease of the mobility of the perovskite molecules in the mp-TiO2 film. BET measurements of the pristine TiO2 film have been performed, yielding a specific surface area of 73 m2 g−1, a total pore volume of 0.793 ml g−1 and an average pore size of 21 nm; however, BET measurements with the adsorbed perovskite showed smaller values for the specific surface area, which is hard to evaluate due to the nanosized particles of the perovskite itself.

Furthermore, Fig. 3 presents the effect of the potential scan rate on the electrochemical behavior (CVs) of the perovskite/TiO2 for the Pn11 and Pn15 samples. The cathodic peak potentials of the films shifted negatively with increase of the scan rate. The observed relationship between the cathodic peak current and the scan rate in the 10–100 mV s−1 range is linear, indicative of a surface-confined electrochemical process.

image file: c7qm00550d-f3.tif
Fig. 3 CV scans for (left) Pn15 and (right) Pn11 at different scan rates; the effect of the potential scan rate on the electrochemical behavior of the perovskite/TiO2 is depicted.

An explanation for the observed peak shift with respect to the scan rate is that when the CV is carried out in the cathodic direction the electron Fermi level is shifted upward and traps are filled with electrons, while the conduction band is filled at more cathodic biases. The CV therefore displays a first cathodic peak (capacitive) induced by the filling of the traps by the electrons followed either by a faradaic current from the conduction band or by the capacitive current. In the anodic direction, because charge transfer from the traps is slow with respect to the trap charging velocity, traps are emptied relatively according to the detrapping rate. If detrapping is very slow with respect to the trapping charging velocity, charges are accumulated in the traps in the cathodic direction. The voltage of the cathodic peak depends on the scan rate and shifts toward the cathodic direction as the scan rate increases.

3.3 Sensor development

In Fig. 4, the CV scan of the Pn15 sample with increasing concentration of CBr4 in the DCM electrolyte solution is depicted. When 60 μl of CSS is added to the 6 ml DCM solution in the electrochemical cell, the first peak of the perovskite/TiO2 film begins to differentiate into two overlapping peaks, the first occurring at 0.47 V and the second appearing at 0.31 V, for which the current increases upon the addition of increasing concentrations of CBr4. The second peak occurring at negative potentials increases in value and shifts towards negative voltage values upon increasing the amount of the detecting molecule. On the reverse scan, the current exhibits greater values than in the previous case without using CBr4. The CBr4 changes the structure of the immobilized perovskite and allows the electrons passing in layers to be easily detectable by our sensor. The Pn11 sample is the sample where the perovskite DP209/PhE·HCl is deposited on a thin mp-TiO2 film. This composite has been used to obtain CV scans when immersed in an electrolyte solution containing CH2Cl2 and 0.1 M TBAH.
image file: c7qm00550d-f4.tif
Fig. 4 CV scans for Pn15 with increasing concentration of CBr4 in the Bsol solution. The scan rate was 0.1 V s−1.

Fig. 5 presents the CV scans of blank TiO2 and Pn11 and of the latter with CBr4 added to the solution. Measurements have been acquired with the pristine TiO2 film, observed in Fig. 5a (black), where there are no detectable peaks. The blank CV recorded at a perovskite free TiO2 film is characteristic of the charging/discharging current associated with the injection of electrons into the TiO2 film.

image file: c7qm00550d-f5.tif
Fig. 5 CV scans of (a) blank TiO2, (b) Pn11, and (c) Pn11 with 60 μl CSS, all acquired with 0.1 V s−1 scan rate.

In contrast, in Fig. 5b, which is the CV of Pn11 in the same electrolyte solution, it appears that there is a cathodic current peak at 0.58 V and another smaller one at 0.75 V in the reduction area. In the reverse scan there is no peak detected. In Fig. 5c, the same measurement is performed for Pn11 after 60 μl of CSS is introduced into the solution. In the presence of CBr4 the previous cathodic peak shifts towards smaller voltage values and the maximum current also increases. The peak now appears to occur at 0.3 V, and the maximum current is −68.6 μA, while in Fig. 5b the current is only −27.3 μA. In both cases the perovskite appears to be reduced at these voltages by electron transfer from the conduction band of the TiO2 film. In the case of Fig. 5c CBr4 most probably invaded the perovskite's structure and changed it. It is possible that methylamine by being reduced could escape to the environment, thus, forcing the remaining perovskite to form other perovskite-like chemical moieties based on low dimensional arrangement of PbBr6 octahedra, rather than PbBr2.

In more detail, we have varied the CBr4 content in the previous test solution in order to check the capability of the Pn11 sample to detect CBr4 from the CV graphs. We observed, from the curves in Fig. 6, that the shift of the peak towards negative potentials depends on the quantity of CBr4 introduced into the solution. The initial peak is located at −27 μA and 0.58 V, while it shifts towards the point of 0.30 V and −69 μA. The inset of Fig. 6 shows the cathodic peak dependence on the added CSS volume, which appears to be fitted by a limiting exponential, with the exception of the second point which may indicate some abrupt transition taking place in the sample upon the CBr4 addition. Control CVs of perovskite-free TiO2 films exhibited a negligible (or none) dependence upon the addition of CBr4. Finally, all the reported results can be systematically reproduced; however, use of the same film in repetitive scans exhibited slightly reduced current values. Rinsing the film with electrolyte solution, free of CBr4, returned the sensor to its initial state. This reversibility allowed the voltammetric detection of CBr4 to be repeated at least three times using the same film, while after that point the sensor's replacement cost is minimal. Calculating the limit of CBr4 detection is not straightforward due to the form of the plotted data in the Fig. 6 inset; however, the definite detection limit with the presented data is reported to be 20 ppb mol mol−1. It must be stated that further addition of CBr4, not shown here, did not induce further shift of the electrochemical peak.

image file: c7qm00550d-f6.tif
Fig. 6 CV scans for Pn11 in the presence of increasing amounts of CSS in the Bsol solution, at a scan rate of 0.1 V s−1. The inset shows the dependence of the cathodic peak on the CSS volume added.

3.4 Optical and morphological properties

Fig. 7 shows the OA spectra of the Pn1 (7a) and Pn11 samples before (7b) and after (7c) the CV measurements, in order to evaluate the degradation of the adsorbed perovskite after its exposure to the electrochemical environment. The Pn1 sample exhibits no OA peaks in the 375–700 nm region, while below 350 nm the glass/ITO absorbs intensely and OA data from this region have not been included here; therefore, all OA peaks detected in the Pn11 and Pn15 samples are due to the adsorbed perovskites. The 3D perovskite semiconductor CH3NH3PbBr3 shows an OA excitonic absorption peak at 2.33 eV;44,45 therefore, since the adsorbed HOIS in Pn11 does not appear to be arising from nanosized particles as deduced from the sharp XRD patterns which would have had a higher band gap than the 3D compound,1 one may conclude that the 513 nm (2.41 eV) OA peak is due to the mixed Br–Cl content of the material deposited since the peak at 513 nm is at higher energies than the 3D CH3NH3PbBr3 perovskite's OA peak. Also, the OA peak at 513 nm is due to the mixed halogen content since Br replacement by Cl increases the band gap energy. In fact, given that the excitonic OA peak of CH3NH3PbCl3 falls at 3.12 eV, while that of the Br analogue is at 2.33 eV,45 and that there is a linear relation between the energy band gap (Eg) values of mixed halide CH3NH3PbBr3−xClx perovskites,44 and for their excitonic OA peak positions, i.e. E(x) = (1 − x) ECl + xEBr, with respect to the content x, 0 ≤ x ≤ 1, it is easy to calculate that the excitonic OA peak at 513 nm corresponds to a Cl content of 10%; at least this appears to be the dominant phase in Pn11. After the CV measurements, including the exposure to CBr4, the excitonic peak of the film has shifted slightly to 505 nm, implying that some Br has been removed from the deposited perovskite or that some degradation of the 3D perovskite has occurred towards lower dimensionalities or that the PhE·HCl additive has contributed, apart from stabilizing the 3D HOIS nanoparticles, to slightly alter the 3D phases by transforming those towards quasi-2D phases; the latter exhibit higher energy band gaps than the 3D phases.
image file: c7qm00550d-f7.tif
Fig. 7 OA spectra of (a) Pn1 and Pn11 before (b) and after (c) the CV scans.

Fig. 8 shows the respective OA of the Pn15 sample before (8b) and after (8c) the CV measurements, including the CBr4 measurements, along with the OA spectrum of Pn1. It appears that no changes take place in the OA peak position at 390 nm nor does the OA spectrum alter after the CV scans, which have exposed Pn15 to the electrochemical environment. The appearance of the broad OA peak of CH3NH3PbCl3 at 3.12 eV, or 397 nm, is in accordance with previous results44,45 for both Fig. 8b and c. The exciton peak for the adsorbed perovskite is not that pronounced as in the case of the Br based perovskites.

image file: c7qm00550d-f8.tif
Fig. 8 OA spectra of (a) Pn1 and Pn15 before (b) and after (c) the CV scans.

It is important to note that the use of PhE·HCl has a twofold action. First, it stabilizes any 3D HOIS nanoparticles by capping those providing even some H2O resistance, and secondly, it induces a change towards a mixture of quasi-2D species (CH3NH3)n−1(PhE-H)2Pbn(BryCl1−y)3n+1, which, for given y, have distinct OA and PL peaks. However, mixtures of varying y values may form a continuous set of peaks from the highest to the lowest possible excitonic energy (i.e. from y = 0/n = 1 to y = 1/n = ∞). The broad OA spectra in Fig. 7 may well enclose a broad absorption profile from such quasi-2D species.

HOIS also show strong photoluminescence (PL) signals due to the stable excitonic states with increased oscillator strength, which can be used to identify excitonic states and the inorganic network dimensionality.1 For example, in Fig. 9, the PL signal of the substrate Pn1 (λexc = 300 nm), at 380 nm, is plotted for comparison to the signal of the Pn11 sample, i.e. after the HOIS has been deposited, and that of the pristine perovskite materials. In particular, Pn11 (before and after the CV scans as stated in the OA spectral analysis) shows the same broad PL band with a peak at 463 nm, and a strong shoulder at 500 nm, which is almost identical to the pristine DP209/PhE·HCl material's spectrum, all of which have been obtained with λexc = 400 nm.

image file: c7qm00550d-f9.tif
Fig. 9 PL spectra of (a) Pn1 and Pn11 (b) before and (c) after the CV scans, and (d) DP209 stabilized/modified with PhE·HCl.

Normally, it would be expected that Pn11 will show a slightly red-shifted PL peak relative to the OA peak at 513 or 505 nm; however, a broad spectrum is observed, which means that the PL spectrum is composed not of a mixed halide of the form CH3NH3PbCl3 or CH3NH3PbBr3 or CH3NH3Pb(Br0.42Cl0.57)3, since this would have entailed three different distinct PL peaks. It seems that many types of halides have formed with PL varying from 420 to 530 nm, which could be, for example, 3D HOIS of the form CH3NH3Pb(Br3−yCly)3 with varying values of y. The sample DP209, as it has been treated with PhE·HCl, has provided also, in Pn11, a blend of perovskite compounds, which is a set of quasi-2D perovskites, with the following formula (CH3NH3)n−1(C6H5CH3CH3NH3)2PbnX3n+1; X = Br, Cl, for n = 1, 2, 3,…,∞. Such blends have odd properties and the most intriguing is the energy funneling properties of excitonic energy towards the moieties with lower band gap energy, i.e. the cases of high n values.42,43,46 Also, their OA and PL spectra are usually composed of broad peaks, yet in many cases the individual OA and PL peaks can be distinguished. Thus, the observed PL signal at 463 nm is due to two factors: (1) the quasi-2D mixture that has formed with the phenethylamine cation perturbing the 3D structure of DP209 to quasi-2D phases, yielding a PL signal below the 532 nm of the CH3NH3PbBr3 peak, and (2) the replacement of Br by Cl which induces a blue shift to the 532 nm peak of CH3NH3PbBr31 and to the quasi-2D phases. Finally, in this PL analysis context, it is reasonable that the Pn11 OA spectrum shows no features below 513 nm, as it is composed of multiple phases. Similarly, Fig. 10 shows that the Pn15 exhibits, before and after the CV scans, a PL signal almost identical to that of CH3NH3PbCl3, which is composed of a broad PL peak covering the region from 425 to 475 nm. The high energy peak is attributed to the pristine material's excitonic PL, where the low energy side of the PL spectrum is attributed to defects. These results are in accordance with the broad PL reported in ref. 45; however, the peak at 410 nm is shifted here to 420 nm. This type of small fluctuation in the 3D perovskite optical peak OA/PL positions has been observed before.

image file: c7qm00550d-f10.tif
Fig. 10 PL spectra of (a) Pn1, (b) CH3NH3PbCl3, Pn15 (c) before and (d) after the CV scans.

The surface morphologies and thicknesses of the TiO2 films were analyzed by SEM not only for determining their morphological characteristics but also to ascertain if the samples change under the CV scan and their exposure to the particular solution and to determine their halogen content, which in turn controls their optical properties. All films consist of a rigid, porous network of TiO2 nanoparticles that are well distributed and all bonded together through a sintering process. In Fig. 11 (a and b, magnifications of 5K and 62K, respectively), the particles’ sizes composing the film electrodes were determined to be 20–30 nm for the Dyesol–TiO2 electrode. SEM was also used to determine the thickness of the aforementioned films to be 4 μm for 1 layer of Dyesol–TiO2. These results confirm a sponge like structure for all films and their pore sizes are sufficiently large enough for the spherical particles composed of 3D perovskite structures to diffuse throughout the porous structure, while other larger perovskites formed will adsorb on the TiO2 surface. In fact, a 5 × 5 × 5 unit cell of CH3NH3PbX3, for X = Br, Cl, is sufficient to behave as a 3D system, which would have a size of 3 × 3 × 3 nm. The surface areas of all three films are greatly enhanced over flat electrode surfaces.

image file: c7qm00550d-f11.tif
Fig. 11 SEM images at 15 kV of (a) the TiO2 film profile, (b) the surface morphology, (c) Pn11, and (d) Pn15. The film thickness is measured to be 3.1 μm in (a).

SEM images for the perovskite loaded TiO2 films have been acquired after the CV measurements. The perovskite particles that have formed on the surface of the film can be observed in Fig. 11c (magnification 6K) and Fig. 11d (magnification 10K) for Pn11 and Pn15, respectively. Most particles for Pn11 are regular uniform cuboids with an average size in the micrometer scale, while for Pn15 the particles are smaller than the micrometer.

EDX measurements after the CV scans have shown for the Pn11 sample a large variation in the Cl[thin space (1/6-em)]:[thin space (1/6-em)]Br mole ratio, from 1[thin space (1/6-em)]:[thin space (1/6-em)]19 to 1[thin space (1/6-em)]:[thin space (1/6-em)]2.3, possibly due to the varying phases that coexist on the sample surface due to the mixed halide environment, while for the Pb[thin space (1/6-em)]:[thin space (1/6-em)]X (X = halide) molar ratio, values of 2.7 to 3.3 have been observed. The Cl[thin space (1/6-em)]:[thin space (1/6-em)]Br ratios are in better qualitative agreement with that same ratio determined from the OA peak position. Also, the variation in the Cl to Br ratios implies that indeed the Pn11 sample is composed of a multitude of phases, where some are in majority; thus, an OA peak at 513 nm is observed. Finally, in the case of EDX results for Pn15 a Pb[thin space (1/6-em)]:[thin space (1/6-em)]Cl molar ratio of 1[thin space (1/6-em)]:[thin space (1/6-em)]3 was found, which is in agreement with the chemical formula and procedure used for Pn15.

Finally, as explained before, scan rate detectability depending on the CV scan's peak position offset was not found to vary with faster scan rates. It is possible, however, that the diffusivity of the ions changes, depending on the TiO2 pore size,47,48 but the investigation on the dependence of the detection limit on the structural properties of the TiO2 film is still in progress. In fact it is possible that the porosity of the nanomaterial itself could create a different set of micropores. The complex interaction behavior of the large and smaller pores with molecules and in particular the hindering of the diffusion of large molecules, has been reported in the past and also very recently47,48 and might be the case also in this work, not only for CBr4 but also for methylamine and small structural units of PbX64− that may shift during the redox processes.

Future work regarding the detectability of CBr4 in realistic cases must be properly addressed with further research in (i) possible cross detection of other halide contaminants by this type of reported sensor, (ii) deterioration of the sensor due to other contaminants and finally (iii) the effect of water on the perovskite itself. The first two issues will most probably serve for numerous further publications; however, the third can be addressed here. One method for hardening perovskites, especially the 3D ones of the form CH3NH3PbX3, where X = I, Br, Cl, is capping with long amines or oleic acid, or using 2D or quasi-2D perovskites of the form (CH3NH3)n−1(long-amine)2PbnI3n+1. Moreover, in other publications it has been shown that the entrapment of perovskites in mesoporous organic or inorganic matrices is possible with total protection from H2O, even when rinsed with large quantities of H2O.49,50 Moreover, it is possible to use crosslinkable and doped fullerene silane-functionalized for stabilization of perovskites,51 or invoke chitosan or Nafion polymers which are stabilization membranes to support perovskites avoiding water degradation.

4. Conclusions

In summary, we have developed a new (electro)chemical sensor by designing a composite material of a mesoporous TiO2 film on an ITO substrate with entrapped 3D or mixtures of 3D/quasi-2D dimensional hybrid organic–inorganic perovskites, both of which can be used for CBr4 detection at 20 ppb mol mol−1 levels. The perovskites used in this work have been based on lead chloride or bromide or with a mixed halogen content. Using the CV technique, it was observed that upon addition of CBr4 the CV signal exhibits shifted cathodic peaks towards lower voltages, which is explained by subtle reversible changes in the perovskite's structure, without removing or degrading the perovskite from the composite film, while the CBr4 detection allows for increased cathodic currents. It is expected that modification of this technique might lead to future advancements in electrochemical sensors, involving self-assembled semiconductor systems (perovskites), while their strong interaction with light might also yield more fruitful sensors.

Conflicts of interest

The authors declare that they have no conflicts of interest.


This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors.


  1. G. C. Papavassiliou, Prog. Solid State Chem., 1997, 25, 125–270 CrossRef CAS.
  2. J. Liu, Y. Wu, C. Qin, X. Yang, T. Yasuda, A. Islam, K. Zhang, W. Peng, W. Chen and L. Han, Energy Environ. Sci., 2014, 7, 2963–2967 CAS.
  3. M. M. Lee, J. Teuscher, T. Miyasaka, T. N. Murakami and H. J. Smith, Science, 2012, 338, 643–647 CrossRef CAS PubMed.
  4. M. A. Green, K. Emery, Y. Hishikawa, W. Warta and E. D. Dunlop, Prog. Photovoltaics, 2016, 24, 905–913 Search PubMed.
  5. M. He, D. Zheng, M. Wang, C. Lin and Z. Lin, J. Mater. Chem. A, 2014, 2, 5994–6003 CAS.
  6. H. Lu, K. Deng, N. Yan, Y. Ma, B. Gu, Y. Wang and L. Li, Sci. Bull., 2016, 61, 778–786 CrossRef CAS.
  7. X. Sun, J. Xu, L. Xiao, J. Chen, B. Zhang, J. Yao and S. Dai, Int. J. Photoenergy, 2017, 2017, 4935265 Search PubMed.
  8. A. Staniszewski, A. J. Morris, T. Ito and G. J. Mayer, J. Phys. Chem. B, 2007, 111, 6822–6828 CrossRef CAS PubMed.
  9. C. Renault, L. Nicole, C. Sanchez, C. Costentin, V. Balland and B. Limoges, Phys. Chem. Chem. Phys., 2015, 17, 10592–10607 RSC.
  10. B. O’Regan and M. Grätzel, Nature, 1991, 353, 3737–3740 Search PubMed.
  11. E. Palomares, R. Vilar and J. R. Durrant, Chem. Commun., 2004, 362–363 RSC.
  12. J. Bai and B. Zhou, Chem. Rev., 2014, 114, 10131–10176 CrossRef CAS PubMed.
  13. L. L. Deng, C. X. Zhao, Y. Ma, S. S. Chen and G. Xu, Anal. Methods, 2013, 5, 3709–3713 RSC.
  14. E. Topoglidis, C. J. Campbell, A. E. G. Cass and J. R. Durrant, Langmuir, 2001, 17, 7899–7906 CrossRef CAS.
  15. F. Campusa, P. Bonhôtea, M. Grätzel, S. Heinenb and L. Walderb, Sol. Energy Mater. Sol. Cells, 1999, 56, 281–297 CrossRef.
  16. C. Chen, J. J. Conception, J. W. Jurss and T. J. Meyer, J. Am. Chem. Soc., 2009, 131, 15580–15581 CrossRef PubMed.
  17. A. Singhal, G. Skandan, G. Amatucci, F. Badway, N. Ye, A. Manthiram, H. Ye and J. J. Xu, J. Power Sources, 2004, 129, 38–44 CrossRef CAS.
  18. Z. Weng, H. Guo, X. Liu, S. Wu, K. Yeung and P. K. Chu, RSC Adv., 2013, 3, 24758–24775 RSC.
  19. J. Bisquert and A. Zaban, Appl. Phys. A: Mater. Sci. Process., 2003, 77, 507–514 CrossRef CAS.
  20. P. E. de Jongh and D. Vanmaekelbergh, Phys. Rev. Lett., 1996, 77, 3427–3430 CrossRef CAS PubMed.
  21. H. Kim, M. C. M. Gilmore, A. Piqué, J. S. Horwitz, H. Mattoussi, H. Murata, Z. H. Kafafi and D. B. Chrisey, J. Appl. Phys., 1999, 86, 6451–6461 CrossRef CAS.
  22. N. Kopidakis, E. A. Schiff, N. G. Park, J. van de Lagemaat and A. J. Frank, J. Phys. Chem. B, 2000, 104, 3930–3936 CrossRef CAS.
  23. P. E. de Jongh and D. Vanmaekelbergh, Phys. Rev. Lett., 1996, 77, 3427–3430 CrossRef CAS PubMed.
  24. J. Nelson, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 15374–15380 CrossRef CAS.
  25. J. van de Lagemaat and A. J. Frank, J. Phys. Chem. B, 2000, 104, 4292–4294 CrossRef CAS.
  26. J. Bisquert, F. Fabregat-Santiago, I. Mora-Sero, G. Garcia-Belmonte and S. Gimenez, J. Phys. Chem. C, 2009, 113, 17278–17290 CAS.
  27. S. Ardo and G. J. Meyer, Chem. Soc. Rev., 2009, 38, 115–164 RSC.
  28. E. Topoglidis, T. Lutz, R. L. Willis, C. J. Barnett, A. E. G. Cass and J. R. Durrant, Faraday Discuss., 2000, 116, 35–46 RSC.
  29. A. Kojima, K. Teshima, Y. Shirai and T. Miyasaka, J. Am. Chem. Soc., 2009, 131, 6050–6051 CrossRef CAS PubMed.
  30. I. Koutselas, P. Bampoulis, E. Maratou, T. Evagelinou, G. Pagona and G. C. Papavassiliou, J. Phys. Chem. C, 2011, 115, 8475–8483 CAS.
  31. T. B. Song, Q. Chen, H. Zhou, C. Jiang, H. H. Wang, Y. Yang, Y. Liu, J. You and Y. Yang, J. Mater. Chem. A, 2015, 3, 9032–9050 CAS.
  32. J. W. Fergus, Sens. Actuators, B, 2007, 123, 1169–1179 CrossRef CAS.
  33. G. L. Luque, N. F. Ferreyra, A. G. Leyva and G. A. Rivas, Sens. Actuators, B, 2009, 142, 331–336 CrossRef CAS.
  34. M. Sivakumar, K. Pandi, S.-M. Chen, Y.-H. Cheng and M. Sakthivel, New J. Chem., 2017, 41, 11201–11207 RSC.
  35. Y. Shu, Y. Yan, J. Chen, Q. Xu, H. Pang and X. Hu, ACS Appl. Mater. Interfaces, 2017, 9, 22342–22349 CAS.
  36. N. Vicente and G. Garcia-Belmonte, J. Phys. Chem. Lett., 2017, 8, 1371–1374 CrossRef CAS PubMed.
  37. K. T. Butler, J. M. Frost and A. Walsh, Mater. Horiz., 2015, 2, 228–231 RSC.
  38. C. Li, Y. Zhou, L. Wang, Y. Chang, L. Etgar, G. Cui, N. P. Padture and S. Pang, Angew. Chem., 2017, 129, 1–6 CrossRef.
  39. T. Ito and G. J. Meyer, Environ. Eng. Sci., 2007, 24, 31–44 CrossRef CAS.
  40. N. Nakashima, T. Tokunaga, H. Owaki, H. Murakami and T. Sagara, Colloids Surf., A, 2000, 169, 163–170 CrossRef CAS.
  41. E. Pitach, T. Portoles, J. M. Marin, M. Ibanez, F. Albarran and F. Hernandez, Anal. Bioanal. Chem., 2010, 397, 2763–2776 CrossRef PubMed.
  42. A. Vassilakopoulou, D. Papadatos, I. Zakouras and I. Koutselas, J. Alloys Compd., 2016, 692, 589–598 CrossRef.
  43. G. C. Papavassiliou, A. P. Patsis, D. J. Lagouvardos and I. B. Koutselas, Synth. Met., 1993, 57, 3889–3894 CrossRef CAS.
  44. I. B. Koutselas, L. Ducasse and G. C. Papavassiliou, J. Phys.: Condens. Matter, 1996, 8, 1217–1227 CrossRef CAS.
  45. G. C. Papavassiliou and I. B. Koutselas, Synth. Met., 1995, 71, 1713–1714 CrossRef CAS.
  46. M. Yuan, L. N. Quan, R. Comin, G. Walters, R. Sabatini, O. Voznyy, S. Hoogland, Y. Zhao, E. M. Beauregard, P. Kanjanaboos, Z. Lu, D. H. Kim and E. H. Sargent, Nat. Nanotechnol., 2016, 11, 872–877 CrossRef CAS PubMed.
  47. Y. Ma, C. X. Zhao, L. L. Deng, H. Yan, S. S. Chen and G. Xu, Electrochim. Acta, 2013, 102, 127–132 CrossRef CAS.
  48. S. Ha and J. H. Moon, J. Phys. Chem. C, 2017, 121, 12046–12052 CAS.
  49. A. Vassilakopoulou, D. Papadatos and I. Koutselas, Appl. Surf. Sci., 2017, 400, 434–439 CrossRef CAS.
  50. A. Vassilakopoulou, D. Papadatos and I. Koutselas, Microporous Mesoporous Mater., 2017, 249, 165–175 CrossRef CAS.
  51. Y. Bai, Q. Dong, Y. Shao, Y. Deng, Q. Wang, L. Shen, D. Wang, W. Wei and J. Huang, Nat. Commun., 2016, 7, 12806 CrossRef CAS PubMed.

This journal is © the Partner Organisations 2018