Sayan
Dey
a,
Sumita
Santra
b,
Anupam
Midya
c,
Prasanta Kumar
Guha
a and
Samit Kumar
Ray
*bc
aDepartment of Electronics & Electrical Communication Engineering, Indian Institute of Technology, Kharagpur, India
bDepartment of Physics, Indian Institute of Technology, Kharagpur 721302, India. E-mail: physkr@phy.iitkgp.ernet.in
cSchool of Nanoscience & Technology, Indian Institute of Technology, Kharagpur, India
First published on 10th November 2016
Nickel oxide, and non-stoichiometric and Cu-doped variants (NiO, Ni2O3 and CuxNi(1−x)O) possessing porous coral-like nanostructures, were prepared by a facile, low temperature, hydrothermal approach. Structural and morphological characterization was performed by X-ray diffractometry, X-ray photoelectron spectroscopy and FESEM imaging. The Cu-doped oxide was found to possess lattice strain which resulted in the induction of surface defects. The prepared materials were used as sensing materials for toxic Cr(VI) ions in aqueous medium. A novel sensing technique was proposed based on adsorption, which was found to be more effective in terms of reproducibility and reusability of the sensor than the conventional electrochemical sensing technique. The sensing mechanism was explained based on the phenomenon of adsorption. This makes the efficiency of the sensor surface-dependent rather than chemical reactivity-dependent, thereby making it a non-destructive sensing technique. The Cu-doped NiO nanostructure (10% Cu doping) was found to show maximum sensitivity (252.62 at 1 ppm) and high selectivity, together with a low response time (∼2 s), towards Cr(VI) ions relative to Cd(II), As(V) and Pb(II) ions. This was due to its enhanced surface properties compared with its un-doped variants. The effect of Cu doping on the nanostructure morphology and consequently on sensor efficiency was also studied. The limit of detection was found to be 1 ppm (1 mg L−1) of Cr(VI) ions in aqueous solution. This material, along with the technique proposed, can thus be advantageous for the cost-efficient monitoring of water quality and drinking water standards.
Environmental significanceA CuxNi(1−x)O coral-like nanostructure was synthesized using a low-temperature, environmentally friendly hydrothermal technique and then used for toxic heavy metal (Cr(VI)) ion detection at ultra-low levels in water-quality monitoring. Introduction of a dopant (Cu) into the NiO lattice generated strain which enhanced the effective number of surface defects (adsorption sites). The novel adsorption–desorption-mediated sensing technique facilitates fabrication of ultra-fast (∼2 s) re-usable heavy metal sensors which, until now, have not been reported. The sensor is exceedingly selective towards Cr(VI) ions, due to the precise material engineering achieved through metal doping. This results in effective tuning of the electrical double layer formed in solution with Cr(VI), relative to other analytes. |
Conventional electrochemical sensing includes a redox reaction (either reduction or oxidation or both) at the electrodes (sensing material) in order to sense the presence of heavy metal contaminants in water. However, this leads to the destruction of the sensing material, thereby making it unusable for repeated use.15 Hence, reusability of such sensors is an issue. Non-destructive techniques have not yet been developed for heavy metal ion sensors.
To the best of our knowledge, metal-doped variants of NiO or its higher oxide (Ni2O3) coral-like hierarchical nanostructures have not yet been synthesized and used for heavy metal detection in aqueous solution. In this study, we synthesized NiO, Ni2O3 and Cu-doped NiO coral-like nanostructures by a low-temperature wet chemical synthetic route and tested their efficacy for Cr(VI) ion sensing in aqueous solution by non-destructive adsorption/desorption electrochemical techniques. We also compared the performance of the materials so prepared, and studied the effect of doping of metal ions in the NiO structure on its sensing properties. Moreover, extensive reproducibility and reusability tests of our sensors are presented, illustrating that the sensors can be used to develop a real-time flowing water monitoring system, which will in turn facilitate the development of a smart water remediation system.
The sensing measurements were carried out using a Potentiostat Interface 1000 Gamry setup. Ag/AgCl and Pt were used as reference and counter electrodes, respectively, and the material was deposited on a 0.5 mm-thick aluminium sheet (3 cm × 3 cm) which was used as the working electrode for the half cell. 0.1 M NaCl solution was used as an electrolyte solution for the cyclic voltammetry experiment operated between −1 V to +1 V. Potassium dichromate solution at calculated concentrations was taken as the Cr(VI) ion source, which was dissolved in 0.1 M NaCl electrolyte solution (taken as the control for the experiment) for the subsequent measurements in 0.1 M NaCl electrolyte solution.
Fig. 1 (a) X-Ray diffractograms for crystalline NiO, Cu(1−x)NixO and Ni2O3 hierarchical nanostructures. (b) Peak shift and intensity reduction for CuxNi(1−x)O in comparison with pure NiO. |
Table 1 presents the strain energy created within the NiO lattice when Cu atoms are doped. The strain energy in case of CuxNi(1−x)O is more than that in the pure NiO lattice. This may imply the possible replacement of Ni by Cu atoms within the crystal lattice, resulting in doping of NiO with Cu.
Sample | Williamson–Hall plot | Size-strain plot | ||||
---|---|---|---|---|---|---|
Strain | Particle size (in nm) | r 2 | Strain | Particle size (in nm) | r 2 | |
NiO | 9.42 × 10−2 | 31.80 | 0.927 | 3.16 × 10−1 | 36.09 | 0.998 |
CuxNi(1−x)O | 2.2 × 10−1 | 15.72 | 0.856 | 4.24 × 10−1 | 21.45 | 0.993 |
The chemical composition of the doped oxide was extracted from XPS and estimated to have 42.3% nickel, 7.4% copper and 50.3% oxygen. The full spectrum and individual peaks of the respective elements are shown in the ESI† (Fig. S1). The analysis of the peaks confirms the copper doping in the NiO lattice from the obtained Cu peaks in the XPS spectra.
Analysis of the FTIR and Raman spectra reveals the bond nature and structure of the prepared materials. From the FTIR spectrum in Fig. 2(a), for all the three optimised samples, peaks are obtained at ∼450 cm−1 due to the presence of M–O (where M = Ni, Cu) bonds. Peaks between 1000 cm−1 and 1500 cm−1 are due to the –OH group bending vibrations, whereas peaks above 2000 cm−1 are due to water molecule (−OH) stretching vibrations. A peak is obtained at ∼1000 cm−1 in the case of Ni2O3 and is due to the formation of Ni–O–Ni bonds, which is different from the remaining two samples. The Raman spectra in Fig. 2(b) reveal two broad peaks at around 500 cm−1 and 1100 cm−1 corresponding to the coupling interaction of Ni–O oscillations. It can be seen that there are slight shifts in both peaks in the case of the CuxNi(1−x)O (10% doped) sample. This may be attributed to the possible doping of Cu ions in the NiO lattice resulting in the formation of strained hereto structures. A considerable amount of shift may have occurred due to the formation of the Ni–O–Ni bond.
Fig. 2 Spectroscopic investigation for NiO, Ni2O3 and CuxNi(1−x)O (10% Cu) (a) FTIR spectrum (b) Raman spectrum. |
FESEM images in Fig. 3(a)–(c) indicate the formation of a coral-like hierarchical morphology in the NiO, Ni2O3 and CuxNi(1−x)O samples, respectively. The coral-like hierarchical structures possess a higher density of defects, which may enhance the surface properties of the so-prepared oxides. It is also expected that the Cu-doped NiO sample should possess a higher surface activity, due to the strain generated by Cu atoms within the NiO crystal lattice. The formation of these coral-like structures is dependent on the basic structure forming the hierarchical structure. It can be observed from Fig. 3 that the pure oxides are formed from only a single type of structure, such as filaments or lengthened particles (in the case of NiO in Fig. 3(a)) and flakes (in the case of Ni2O3 in Fig. 3(b)). However, from Fig. 3(c) it can be seen that the hierarchical structure is a mixture of both filaments (and spherical particles) and flakes. This may be due to the possibility that the doping has controlled the nucleation rate and the orientation of growth, owing to the presence of two cations in the solution, thereby effectively increasing the surface area available for adsorption of chemical species (active adsorption sites). The EDAX spectrum and its analysis also support the XPS analysis in terms of chemical composition of the Cu-doped NiO sample (Fig. S2 in ESI†).
It is observed in Fig. 3 (nanostructures of each oxide type are magnified and presented) that, as the nickel oxide is varied from NiO to Ni2O3 and then to CuxNi(1−x)O, the surface morphology changes. From a visual perspective, it may be concluded that the doped variant has a much rougher surface as compared with the pure oxides of nickel. This can be attributed to the enhanced adsorption of Cr(VI) ions in aqueous solution, thereby leading to effective sensing of Cr(VI).
The microstructures so formed were studied by transmission electron microscopy (TEM) imaging. The lattice image and fringe pattern were also investigated. The doped oxide was expected to be larger in size compared with the pure variants. This may be attributed to the probable inflation of structure due to the incorporation of Cu (atomic number 29) having a fully filled d orbital (3d10 configuration) in place of Ni (atomic number 28) having a 3d9 configuration. Moreover, the surface roughness is greater in the case of CuxNi(1−x)O in comparison with the pure oxide variants, which provides more active adsorption sites. When comparing these three images, it can be established from Fig. 4(c) that the Cu-doped sample hierarchical structure is a mixture of flakes and filaments as primary growth units whereas, in the case of NiO (Fig. 4(a)), it is homogenous filaments, and in Ni2O3 (Fig. 4(b)), it is homogenous flakes. Thus, the presence of two basic units contributes a differential surface area for adsorption, which is expected to be greater than for those structures formed from a single basic unit (as in Fig. 3(a) and (b)). The lattice images and selected area electron diffraction (SAED) patterns of the three structures are provided in the ESI† (Fig. S4). The lattice spacing for all three samples was found to be ∼0.28 nm, confirming the formation of nickel oxides.
In order to study the effect of doping concentration on the structure, the doping amount was varied from 5% to 20% Cu. From the FESEM images shown in Fig. 5, it is seen that, on doping with Cu, the hierarchical structure is generated by spherical nanoparticles instead of flake-like nanostructures. As a result, the effective surface area increases and hence the number of effective adsorption sites also increases. However, on increasing the doping amount from 5% to 20%, the porosity of the surface of the nanostructure is seen to decrease (from Fig. 4(c)) and the spherical-like particle forms a coral-like nanostructure. As a result, surface porosity decreases, and hence the sensitivity and limit of detection are expected to decrease.
Fig. 5 Change in morphology of CuxNi(1−x)O based on percentage of Cu doping (% by weight) (a) 5% (b) 10% (c) 20%. |
The coral-like structure is selected because it is expected to provide more adsorption sites, owing to the induced surface roughness and intentionally generated surface defects. The surface morphology depends largely on the precursors used in the synthesis. In this regard, it should be mentioned that urea is used to produce an alkaline environment, which favours the heterogeneous nucleation responsible for the growth of coral-like mesoporous nanostructures.18 Hydrolysis of urea results in the formation of NH3 (NH4+ in solution) and CO32− at a temperature above 90 °C, which makes the solution alkaline and thus favourable for heterogeneous nucleation. Also, the bubble templating effect due to urea hydrolysis can be considered to be a synergistic cause of formation of hierarchical nanostructures. This takes place by the association of hydrogen bubbles in situ, thereby resulting in effective heteronucleation19 on PEG chains as templates to form directional Ni(OH)2 growth.
To investigate the sensing of Cr(VI) ions in aqueous solution by the prepared sensing materials, cyclic voltammetry was performed within the range −1 V to +1 V. It is observed from Fig. 6(a)–(c) that the current decreases with a decrease in the concentration of Cr(VI) in water. After a certain concentration of Cr(VI), it is observed that the current starts to increase and eventually the curve (i.e. current) becomes almost equal to the curve without an analyte (i.e. Cr(VI) ions in solution), which is shown in Fig. 6(a). Thus, the limit of detection can be considered to be the concentration at which a detectable reduction in current or an increase in resistance is maximum. The percentage increase in resistance is considered to be the sensitivity of the sensor. Moreover, the limit of detection is seen to depend on the electrode material, and this improved from NiO to Ni2O3 and then to CuxNi(1−x)O. The resistance is calculated from the measured current at 1 V. Table 2 represents the current (in mA), percentage change in current (from reference), resistance (at 1 V) and the sensitivity (percentage increase in resistance) for the three sensing oxide materials (NiO, Ni2O3 and CuxNi(1−x)O).
Fig. 6 C–V plots of (a) NiO (b) Ni2O3 and (c) Cu(1−x)NixO hierarchical nanostructures (−1 V to +1 V). |
Concentration of Cr(VI) ions | NiO (coral-like nanostructure) | Ni2O3 (coral-like nanostructure) | CuxNi(1−x)O (coral-like nanostructure) | ||||||
---|---|---|---|---|---|---|---|---|---|
(ppm) | Sensed current (base current = 2.5 mA) | Increase in resistance (kΩ) | Response | Sensed current (base current = 1.3 mA) | Increase in resistance (kΩ) | Response | Sensed current (base current = 7 mA) | Increase in resistance (kΩ) | Response |
200 | 2.45 | 23.557 | 57.78 | 0.425 | 1.559 | 1.30 | 1.1 | 0.766 | 5.36 |
150 | 2.47 | 35.326 | 87.14 | 0.31 | 2.432 | 2.31 | 0.9 | 0.968 | 7.21 |
100 | 2.48 | 52.370 | 129.65 | 0.148 | 5.963 | 6.63 | 0.734 | 1.220 | 8.54 |
50 | — | — | — | 0.092 | 10.076 | 11.77 | 0.118 | 8.332 | 58.32 |
10 | — | — | — | 0.037 | 26.380 | 32.27 | 0.093 | 10.610 | 74.27 |
1 | — | — | — | — | — | — | 0.0276 | 36.089 | 252.62 |
The complete sensing mechanism by adsorption can be subdivided into the stages shown below.
So, ΔI = I0 − Isense = K·r (where ‘Isense’ is the current after adding the analyte and ‘K’ is the constant of proportionality)
From Ohm's law,
The response of the sensor (S) can be expressed as a percentage by the following relation:
(1) |
i.e. (in times if more than 100%)
The adsorption of Cr(VI) ions by NiO, Ni2O3 and Cu-doped NiO coral-like nanostructures has been tested by the batch adsorption technique. The details of the batch adsorption technique are explained in ESI.† The variation of adsorption rate with concentration is shown in Fig. 8.
From Fig. 8, it is observed that, as the concentration of Cr(VI) ions is increased, the adsorption rate is increased until a certain concentration depending on the chemistry of the adsorbent, which determines the number of adsorption sites and the surface potential (which are the determining factors for the adsorption of heavy metal ions). After this critical concentration, the rate of adsorption again decreases. It may be considered that, at very low concentrations, the number of heavy metal ions is less than the total number of available adsorption sites. The rate of adsorption is less and hence the current generated in the cathode by the electrochemical cell is very close to the base current (without analyte). After a critical concentration at which the adsorption is maximal, all the available adsorption sites are saturated. Hence, any concentration above the critical concentration would result in a decreased adsorption rate relative to the concentration. As the number of Cr(VI) ions in the electrolyte solution increases with increase in concentration, the current would saturate very close to the lowest current. Thus, for any concentration below the critical concentration of Cr(VI) ions Ccritical, if any current above the minimum current is observed then it can be due to the synergistic effect of excess Cr(VI) and Na+ ions in solution. However, this current is negligible and within the error bars for the experiments carried out. The possible reduction of current from the base current takes place because some of the sites having unsatisfied electrons (dangling bonds) act as adsorption sites and adsorb few Cr(VI) ions, and hence electrolyte discharge is not possible on those sites. When the concentration of Cr(VI) ions is greater than Ccritical, the total current contributed by the electrochemical cell has a synergistic effect on the excess un-adsorbed Cr(VI) and Na+ ions in solution. However, due to saturation of all the dangling bonds in the electrode, the current is still much lower than the base current and very close to the current corresponding to the concentration of maximum adsorption (though slightly higher than the minimum current). Hence, the concentration with the maximum adsorption (a Gaussian-type curve is expected for variation of adsorption rate) at which the minimum current (as well as the saturation current) is observed can be considered to be the limit of detection achieved by the sensor (Fig. 8).
Adsorption by a nanostructure is a surface phenomenon. Thus, it is expected to depend largely on the surface morphology of the prepared nanostructure. It is observed from Fig. 2 and 3 that, due to possible Cu doping, the coral-like structure so formed is composed of two types of basic units, i.e. filaments (and elongated particles) and flakes. When compared with the pure oxides, it is found that the coral-like structures are made of only one type of structure (either filament or flakes). Thus, the effective surface area of the surface available for adsorption is expected to increase due to the possible Cu doping when compared with pure oxides, as a result of which there is an increased adsorption of Cr(VI) ions. Doping with a more electron-rich species (i.e. Cu in comparison with Ni) is expected to increase the electron richness of the surface, thereby providing more negative centres for the effective adsorption of positive Cr(VI) ions (e.g. cationic centres of hydrated dichromate radicals at the mild ambient pH of waste water). This is evident from the zeta potential (surface charge) measurement of doped and pure NiO samples. The positivity of the surface is found to decrease ∼6.6 times in the doped sample in comparison with the pure NiO sample (5.51 mV for CuxNi(1−x)O and 36.3 mV for NiO coral-like nanostructures) (graphs provided in ESI,† Fig. S5 and S6).
To study the porosity of the optimized sample, a N2 adsorption–desorption experiment (BET/BJH) was performed for the CuxNi(1−x)O (10%-doped) sample (Fig. 9). The outgassing time was 180 min and the temperature was 150 °C. The outgassing time and temperature were recorded prior to exposure of the material to the adsorbent. Significant changes (if any) in the weight of the sample were recorded prior to the N2 adsorption–desorption experiment.27 It was observed that the surface area for adsorption for the sample was 44.543 m2 g−1 with a pore volume of 0.29 cc g−1 and an average pore radius of 15.718 Å (1.572 nm). Thus, from the BET study, it can be concluded that the prepared materials are very porous, with a higher surface area available for adsorption as compared with the pure NiO nanostructures (surface area ∼33 m2 g−1),27 and hence they show a comparably higher adsorption rate from their pure oxide counterparts. From the nature of the isotherm, it can be considered to be a typical mesoporous material. However, the pore size is slightly less as compared with those of mesoporous materials (∼2 nm to 50 nm). Thus, the material can be considered to be a pseudo-mesoporous material, which is microporous but tends to be mesoporous.
Concentration of Cr(VI) ions | 5% Cu-doped NiO | 10% Cu-doped NiO | 20% Cu-doped NiO | ||||||
---|---|---|---|---|---|---|---|---|---|
(in ppm) | Sensed current (base current = 34 mA) | Increase in resistance (Ω) | Response | Sensed current (base current = 7 mA) | Increase in resistance (kΩ) | Response | Sensed current (base current = 44 mA) | Increase in resistance (Ω) | Response |
200 | 30.8 | 3.06 | 0.104 | 1.1 | 0.766 | 5.36 | 41.6 | 1.287 | 0.05 |
100 | 28.7 | 5.43 | 0.18.46 | 0.9 | 0.968 | 8.54 | 31.5 | 8.996 | 0.396 |
50 | 24.3 | 11.74 | 0.3992 | 0.734 | 1.220 | 58.32 | 25.2 | 16.81 | 0.74 |
25 | 23.1 | 13.88 | 0.472 | 0.118 | 8.332 | 61.25 | — | — | — |
10 | 17.9 | 26.45 | 0.899 | 0.093 | 10.610 | 74.27 | — | — | — |
1 | — | — | — | 0.0276 | 36.089 | 252.62 | — | — | — |
From the results, it was observed that the sensitivity increased dramatically for 10% Cu-doped NiO in comparison with the other two variants. Also, the limit of detection was greater compared with the other two variants. This may be attributed to the effective adsorption of Cr(VI) on the Cu-doped NiO and the dependence of adsorption rate on the doping amount of Cu on the NiO. It also explained the morphological dependency of adsorption rate, which influenced the sensitivity of the sensing material. The response time (time required to detect a measurable change in Cr(VI) ion concentration in aqueous medium) for Cr(VI) ions also decreased with respect to the surface morphology, and was seen to be least for 10% Cu-doped NiO with respect to the other two variants (shown in Fig. S3 in ESI†).
Desorption cycle | Initial current (before adsorption) (in amperes) | Final current (after desorption) (in amperes) | Percentage recovery |
---|---|---|---|
Cycle 1 | 0.0266 | 0.0259 | 97.3 |
Cycle 2 | 0.0266 | 0.0264 | 99.24 |
Cycle 3 | 0.0266 | 0.0264 | 99.24 |
From Table 4, it can be concluded that after two desorption cycles, complete recovery of the sensing material can be obtained with a percentage error of 0.75% and hence the material can be re-used for sensing.
To investigate reproducibility, Cr(VI) ion-sensing experiments were carried out periodically for up to 6 months and the percentage variation was analyzed. The reproducibility results are presented in Fig. 10. It is observed that the sensitivity error for the first month, for all concentrations, is within ∼0.5%. The error increases with time, showing a maximum of ∼2.5% at the end of 3 months (90 days) and ∼7.5% at the end of 6 months (180 days).
Fig. 11 Selectivity of a Cu doped NiO (10% Cu by weight) sensor for Cr(VI) over Cd(II), As(V) and Pb(II) ions. |
The effect of the K+ ion (sourced from K2Cr2O7) was also considered and the material was tested for any possible sensitivity to K+ ions that might affect the overall sensitivity. In a typical experiment, KCl (as source of K+) was used as an analyte and sensing measurements were performed. However, no significant change in base current was observed (less than 0.1% change). Thus, it may be considered that the sensitivity observed with Cr(VI) ions (sourced from K2Cr2O7) as an analyte may be solely due to Cr(VI) ions.
Negative adsorption sites are formed due to an excess electronic cloud (due to Cu doping) and cationic vacancies, whereas adsorption sites in pure oxides are due only to cationic vacancies. Thus, it should be capable of sharing 6 electrons to satisfy the Cr(VI) ion to its filled configuration. Moreover, the cations and the analyte (i.e. Cr(VI) ions) have comparable atomic radii. Hence, they are more accessible to the cationic defect sites acting as active adsorption sites. Therefore, due to the cumulative effect, the effective adsorption for Cr(VI) is much more than for other heavy metal cations. This results in the selective detection of Cr(VI) over other analytes.
Fig. 12 (a) Response time of a Cu-doped NiO sensor for Cr(VI) ions. (b) Amperometric plot for 10% Cu-doped NiO. |
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6en00285d |
This journal is © The Royal Society of Chemistry 2017 |