An effective tellurium surface modification strategy to enhance the capacity and rate capability of Ni-rich LiNi_0.8Co_0.1Mn_0.1O₂ cathode material

Abstract

LiNi_0.8Co_0.1Mn_0.1O₂ (NCM) layered oxide is contemplated as an auspicious cathode candidate for commercialized lithium-ion batteries. Regardless, the successful commercial utilization of these materials is impeded by technical issues like structural degradation and poor cyclability. Elemental doping is among the most viable strategies for enhancing electrochemical performance. Herein, the preparation of surface tellurium-doped NCM is done by utilizing the methodology solid-state route at high temperatures. Surface doping of the Te ions leads to structural stability owing to the inactivation of oxygen at the surface via the binding of slabs of transition metal-oxygen. Remarkably, 1 wt% of Te doping in NCM exhibits enhanced electrochemical characteristics with an excellent discharge capacity, i.e., 225.8 mAh/g (0.1C), improved rate-capability of 156 mAh/g (5C) with 82.2% retention in capacity (0.5C) over 100 cycles within 2.7–4.3V as compared to all other prepared electrodes. Hence, the optimal doping of Te is favorable for enhancing capacity, cyclability along with rate capability of NCM.

Highlights

• •Tellurium-modified LiNi_0.8Co_0.1Mn_0.1O₂ cathode material is prepared via a solid-state route.
• •Surface modification using Te ions leads to structural stability.
• •1 wt% of Te doping in NCM exhibits enhanced 225.8 mAh/g capacity.
• •1 wt% of Te doping in NCM shows improved rate capability of 156 mAh/g (5C).
• •1%-Te NCM delivers improved capacity retention of 82.2% (0.5C) over 100 cycles.

1. Introduction

Lithium-ion batteries (LIBs) are treated as pervasive energy-storing technology for portable devices, automobile industry, and home appliances, attributed to high energy and power density, cost-effectiveness, long cyclic life, negligible memory loss, eco-friendliness, and low self-discharge [[1], [2], [3], [4]]. However, further enhancement in the electrochemical performance is expeditiously required to fulfill the ever-growing demands for the system of energy storage on a commercial scale [5]. The cathode component of LIBs is the most vital part of manufacturing electrochemically enhanced LIBs [6].

Ni-rich layered oxide (LiNi_aCo_bMn_cO₂, a≥0.8, a+b + c = 1), wherein Ni composition is dominant over Mn and Co, are auspicious cathode candidates due to greater specific capacity, i.e., more than 200 mAh/g in 3.0–4.3 V, low cost, and exceptional energy density [[7], [8], [9]]. Unfortunately, the application of these materials still faces several challenges, including cation disorder and oxidation of Ni³⁺ to Ni⁴⁺ during cycling that can cause detrimental side reactions with electrolyte and formed Ni–O impurity phase [[10], [11], [12]]. Moreover, the conversion of R $\bar{3}$ m to Fd $\bar{3}$ m and then to Fm $\bar{3}$ m phase will give rise to anisotropic lattice contraction during the de-lithiation state that causes further structural degradation [[13], [14], [15]]. Hence, the abovementioned challenges should be resolved before the commercial-scale implementation of Ni-rich-based cathodes.

The electrochemical characteristics of these types of materials have been alleviated via several modification works, like doping, coating, pre-oxidation, and concentration gradient [16,17]. Among these modification strategies, element doping is the most feasible way to overcome structural instability and augment electrochemical stability [[18], [19], [20]]. To date, doping of the various elements, including Ti [21], Nb [22], Zr [23], Mo [24], V [25], Al [26], and so on, has been extensively examined to ameliorate electrochemical features.

The doping elements with dominantly stronger metal-oxygen bond energy can offer multiple advantages. Firstly, the dopant acts as a pillar by preventing the John-Teller distortion and ensures structural stability during cycling [27,28]. Secondly, cation disorder is prevented by restraining the evolution of oxygen [23,29]. In addition, the dopant with high bonding energy can also impede the reactivity between the electrode and electrolyte interface [10].

Moreover, dopants with high valence can be utilized to alleviate the electronic conductivity and mitigate polarization. Such metal ions present in transition metal slabs supply the additional charge and increase the force of repulsion among the layers. Consequently, the diffusivity of the Li⁺ is increased by the increment of lattice and interlayer spacing. Hence, the cyclability, as well as rate capability, can be considerably ameliorated [25,30]. Furthermore, the dopants having large ionic radii compared with Ni, Mn, and Co can boost the rate capability by expanding the diffusion channels of Li⁺ [26].

Various doping elements with high valence and bond dissociation energy have been explored. Li et al. [3] Nb-modified LiNi_0.8Co_0.1Mn_0.1O₂, the higher Nb and oxygen bond energy, less cation mixing, and expanded Li⁺ diffusion paths lead to structural as well as cyclic stability. Park et al. [28] stated that B-doped LiNi_0.90Co_0.05Mn_0.05O₂ has 91% retention of capacity at 55 °C (100 cycles) by forming an optimized microstructure that can cause the slight release of internal strain during cycling. Jamil et al. [31] reported that Ta-doped LiNi_0.88Co_0.09Al_0.03O₂ exhibited enhanced rate capability and cyclability attributed to the robust Ta and oxygen bond dissociation energy. Shang et al. [32] illustrated that W-doped NCM811 at 4.5 V (100 cycles) exhibited 7.9% capacity loss, which was much lower than unmodified material owing to structural stability and lesser impedance value. Therefore, the utilization of Te⁶⁺ with high valence and 548 kJ/mol bond energy of Te–O [33] as a dopant in cathode materials can raise their electrochemical characteristics. Recently, Huang et al. [34] analyzed the impact of Te doping in LiNi_0.88Co_0.09Al_0.03O₂. The existence of a stronger bond between Te and oxygen inhibited the oxygen evolution and enhanced the phase reversibility (H2 → H3), which led to the stable structure along with the advancement in electrochemical characteristics.

In this study, Te-doped NCM cathodes were prepared using a two-step process involving hydroxide co-precipitation along with a solid-state technique with varying amounts of Te. This investigation focuses on a mechanistic perception of impact of tellurium doping on electrochemical characteristics of NCM.

2. Experimental details

2.1. Methodology

Hydroxide co-precipitation methodology was utilized to produce Ni_0.8Co_0.1Mn_0.1(OH)₂. Aqueous solutions (1 M) of NiSO₄·H₂O, MnSO₄·H₂O, and CoSO₄·6H₂O were pumped simultaneously with concentrated NaOH and NH₃. H₂O solution. Later on, 3 M NaOH was utilized to adjust pH. This mixture was vigorously stirred and kept at 50 °C, and hydroxide precursor was accumulated by filtration, washing, and then dried at 120 °C.

The LiNi_0.8Co_0.1Mn_0.1O₂ cathode material was obtained via hand grinding Ni_0.8Co_0.1Mn_0.1(OH)₂ with LiOH.H₂O (TM: Li = 1:1.03 mol ratio) using anhydrous ethanol (solvent) and finally pre-sintered (500 °C for 6 h) and after that sintered (750 °C for 20 h) in flowing oxygen environment marked as NCM.

The modified sample marked 1% Te-NCM was made via mixing the desired quantity of H₆TeO₆ in deionized water, then adding as-synthesized Ni_0.8Co_0.1Mn_0.1(OH)₂ in it so doping amount is 1 wt%. The excess water was removed by overnight heating at 80 °C. Then yield was blended with LiOH.H₂O and sintered under the same sintering conditions.

2.2. Material characterization

Structural investigation was conducted via X-ray diffraction (XRD) with Bruker D8 Advance instrument (λ = 0.158 nm). The microstructure was examined with high-resolution transmission electron microscopy (TEM) using G2 520 FEI, Tecnai instrument. Moreover, the element valence present on surface of as-prepared materials was examined via X-ray photoelectron spectroscopy (XPS) with ULVAC-PHI 5000 VersaProbe instrument via Al-K_α radiation hν = 1.4866 keV.

2.3. Electrochemical testing

To fabricate cathode, active material (80 %), conductive agent (10%), and binder (10%) were LiNi_0.8Co_0.1Mn_0.1O₂, conductive carbon, and polyvinylidene fluoride, respectively. N-methyl pyrrolidone (NMP) was taken as a solvent and slurry was pasted on an aluminum-foil sheet and subsequently vacuum-dried at 80 °C for overnight. Around 3 mg/cm² was the material loading on the cathode pellets. 1 M lithium hexafluorophosphate (LiPF₆) added in 1:1:1 vol % of EC/DMC/EMC mixture was utilized as an electrolyte. Li metal and Celgard 2325 (porous polypropylene S4 separator membrane) were taken as anode and separator, respectively. A glove box having pure argon was utilized to assemble CR2025 cells. To evaluate electrochemical performance at ambient temperature, coin cells were activated for four consecutive cycles (0.1C) and then continued at 0.5C and 2.7–4.3V for 100 cycles by a BST8 MTI battery tester.

3. Results and discussion

Structural investigation of NCM and Te-doped NCM materials via X-ray diffractometer (XRD) is demonstrated in Fig. 1 and Figure S1. The as-prepared samples display similar diffracted peaks that are associated with a hexagonal structure (α-NaFeO₂, R $\bar{3}$ m space group) (JCPDS #00-009-0063) and have no impurity peaks (Fig. 1(a) and Fig. S1 a) [35,36]. These results suggest that the Te addition did not vary the structure of NCM. Moreover, the magnified XRD graphs in the selected portion of 2θ (63–66°) (Fig. 1(b) and Fig. S1b) reveal that the peak splitting becomes less evident with the addition of Te. This may be associated with the conversion of phase near the surface of the materials [32,37]. Further analysis of the structural characteristics of pristine and Te-doped NCM is done by the Rietveld refinement method, and Fig. 1(c,d), Figure S1 (c,d), and Table 1 and table S1 present the corresponding outcomes. It can be perceived from profile refinements that the calculated curves are highly matched with the observed curves, and the R_wp value for all the samples is <4%, which indicates the credibility of the refined data [3,38]. The c/a value is 4.9 for all the samples, which suggests that Te doping does not influence their well-ordered layered structure [16,31].

Fig. 1.

(a) XRD profiles and (b) enlarged regions for (018)/(110) peaks and Rietveld-refinement of (c) NCM, and (d) 1% Te-NCM.

Table 1.

Structural properties of NCM and Te-doped NCM from Rietveld refinement.

Sample	a (Å)	c (Å)	V(Å3)	c/a	I(003)/(104)	Rwp (%)	Goodness of fit (GOF)
NCM	2.8743	14.212	101.687	4.94	1.38	3.32	1.46
1%Te-NCM	2.8745	14.215	101.714	4.94	1.19	3.07	1.38

Furthermore, peak intensity ratio R(I_(003)/(104)) value is a crucial parameter in determining the structural order of cathode materials because (104) reflection does not depend on the disordering of structure. In comparison (003), reflection can be influenced by cation disorder in the layered structure [39,40]. A greater R(I_(003)/(104)) value suggests lower cation mixing [41,42]. Table 1 shows that doping of Te aggravates cation mixing, which can be explained based on charge conservation. The doping of Te⁶⁺ ions with a high valence state into layer of transition metal conserves charge by producing Ni²⁺. Hence, the existence of more Ni²⁺ having similar ionic radii to Li⁺, i.e., r_Ni²⁺ = 0.69 Å, r_Li⁺ = 0.72 Å, facilitates the movement of Ni²⁺ into Li slab and promotes rock-salt phase at the surface [32,37,43]. However, prior research has shown that a suitable quantity of rock-salt phase hinders the side reactions and enhances electrochemical performance [44,45].

The morphology and microstructural differences between Te-doped and pristine NCM are investigated through HRTEM. HRTEM images with associated fast Fourier-transformation (FFT) patterns of pristine NCM, as well as 1% Te-NCM, are presented in Fig. 2 (a-d). Furthermore, Fig. 2(e and f) presents homogenous distribution of the elements in NCM and 1% Te-NCM. The HRTEM micrographs of both samples (Fig. 2 b,d) reveal a high crystalline nature represented by distinct fringes. The inter-atomic spacings in regions I & II of pristine NCM attribute to the (003) plane along with the adjacent (101) and (104) planes, manifest by FFT images, which confirms the existence of layered structure (R $\bar{3}$ m space group) [3,14,16]. However, 1% Te-doped NCM exhibits the layered structure in region I (bulk), while a rock salt phase (Fm $\bar{3}$ m) exists in region II (surface). The occurrence of Fm $\bar{3}$ m phase at surface, in conjunction with an increased Ni²⁺/Ni³⁺ ratio (Fig. 1), indicates that the Te doping causes the creation of Fm $\bar{3}$ m phase on surface [24]. The rock-salt phase on the surface becomes active electrochemically to alleviate capacity, rate capability and retain the structural integrity of layered structure [35,41,46,47].

Fig. 2.

HRTEM images of (a) NCM (b) enlarge view as well as FFT images; (c) 1% Te-NCM (d) enlarge as well as FFT images; EDS mapping of Ni, Co, Mn of (e) NCM and EDS mapping of Ni, Co, Mn and Te of (f) 1% Te-NCM.

The investigation of the variation of the valence state of elements in NCM and 1% Te-NCM samples was done via XPS, as presented in Fig. 3. Binding energy peaks related to Li (1s), Co (2p), O (1s), Ni (2p), and Mn (2p) are observed for both samples, while a peak corresponding to Te (3d) is also detected in 1% Te-NCM (Fig. 3(a)). In Fig. 3(b), Ni 2p spectra of samples contain couple of prominent peaks that are associated with Ni 2p_3/2 and Ni 2p_1/2, together with satellite peaks. The deconvolution and fitting of peak Ni 2p_3/2 reveal Ni²⁺ as well as Ni³⁺ existence on the surface. Ni 2p_3/2 peak is found at 854.7 eV for NCM and 856.3 eV for 1% Te-NCM. So, after the Te doping, the Ni 2p_3/2 peak moves from a lower to a greater value of binding energy, implying the likely rise of Ni²⁺ near the surface to conserve the charge according to the previous reports of doping of high oxidation state metal ions [32,48].

Fig. 3.

XPS profiles (a) NCM and 1% Te NCM samples. Deconvoluted spectra (b) Ni 2p, (c) Co 2p, (d) Mn 2p, (e) O 1s, and (f) Te 3d for NCM and 1%NCM.

Moreover, it can be observed from spectra that Ni²⁺/Ni³⁺ is greater for 1% Te-NCM as compared to NCM. Hence, also in accordance with XRD results, it could be deduced that the enrichment of Te ion close to the surface results in the conversion of Ni³⁺ into Ni²⁺ and originates cation mixing [32,48,49]. However, more Ni²⁺ produces a rock-salt phase on surface that may enhance structural stability [50]. The peaks in Co 2p spectra of both samples belong to Co 2p_3/2, as well as Co 2p_1/2, as illustrated in Fig. 3(c). The diversion of Co 2p_3/2 peak to the elevated binding energy after Te doping may be attributed to Co³⁺ to Co²⁺ ion adjustment to retain the charge equilibrium [34]. In Mn 2p spectra (Fig. 3(d)), the Mn 2p_3/2 is positioned at 641.8 eV and 642 eV for NCM and 1% Te-NCM corresponding to Mn⁴⁺ [51]. Moreover, in O 1s spectra, as displayed in Fig. 3(e), two peaks belong to lattice oxygen (Mn/Co–O bonds) and absorbed oxygen (LiOH and Li₂CO₃). Peak intensity of the lattice oxygen increases after Te doping, implying that the strong Te–O bonds produce more lattice oxygen and enhance structural stability [31]. As illustrated in Fig. 3(f), peaks positioned at 576.3 and 587.3 eV are linked with Te 3d_5/2 and Te 3d_3/2, accordingly, revealing that valence state of Te is +6 [34].

The electrochemical characteristics of NCM and 1% Te-NCM cathodes are demonstrated in Fig. 4(a-d). Fig. 4(a) presents 1st cycle voltage curves within 2.7 and 4.3 V (0.1C). NCM and 1% Te-NCM discharge capacity are 188.4 and 225.8 mAh/g, correspodingly. Increased discharge capacity is ascribed to appropriate doping of Te on the surface because, during discharging, Li ions can go back to their respective positions without cation mixing. Furthermore, cycling curves of NCM and 1% Te-NCM at 0.5C between a potential of 2.7–4.3 V are depicted in Fig. 4(b-c). Observably, an preliminary potential drops during discharge and a voltage plateau throughout cycling are evident which are accredited to the impedance of the cells. The NCM has a potential drop of 0.15 V which is very less than 0.33 V of 1% Te-NCM after 100 cycles. Moreover, electrochemical features of NCM with 1% Te-NCM is further compared by calculating the discharge plateau retention (DPR) using the formula in Eq. (1) [52]:

$$\mathit{D}\mathit{P}\mathit{R}=\frac{\mathit{P}\mathit{C}}{\mathit{D}\mathit{C}}$$ (1)

Where the PC and DC are discharge capacities from preliminary discharge potential to potential of 3.5 and 2.7 V, respectively. The DPR of 1% Te-NCM is 80.1% which is far higher than 66.7% of NCM at the 100th cycle.

Fig. 4.

(a) Initial cycling graphs of NCM and 1% Te-NCM (0.1C, 2.7–4.3 V). Charge-discharge graphs (b) NCM (c) 1% Te-NCM (0.5C, 2.7–4.3 V) at 1st, 25th, 50th, and 100th cycle. (d) Cyclic stability of pristine NCM and 1% Te-NCM between 2.7 and 4.3 V (0.5C, 100 cycles).

The NCM cathode has variations in charge/discharge curves (Fig. 4(b and c)) at extended cycles owing to the polarization of the electrode, which is due to structural degradation. However, 1% Te-NCM maintains similar profiles, a small potential drop, and has a slower capacity decay due to the strong Te–O bonding that can effectively reduce degradation of structure [34]. Fig. 4(d) illustrates cyclic behavior of NCM and 1% Te-NCM at 0.5C. 1% Te-NCM has 82.2% retention in the capacity which is higher than 74.6% of NCM over 100 cycles. Enhanced cyclability of 1% Te-NCM can be accredited to the surface doping of tellurium ions and the inactivation of oxygen at the surface by binding the slabs of TM-O through the presence of strong bonding between the Te ions and oxygen that facilitates the structural stability upon cycling [34].

Fig. 5(a-b) depicts cyclic voltammetry (CV) profiles at 0.1 mV/s and 2.7–4.3 V. NCM as well as 1% Te-NCM electrodes have three cathodic and anodic peaks, related to transition of phase from (i) hexagonal (H1) to monoclinic (M) phase (3.7–3.9 V), (ii) monoclinic (M) to hexagonal structure (H2) (3.9–4.1 V) and (iii) hexagonal (H2) to hexagonal (H3) phase during Li-ions insertion/extraction. In the process of phase transformation from H2 → H3, the more extraction of the lithium ions will contract the c parameter, and expedite a large variation volume as well as micro stress [53]. As a result, layered structure will collapse and degrade cyclability of material [54]. Fig. 5 indicates the smaller H3 peak of 1% Te-NCM compared to NCM, which infers the restraining effect of Te on the occurrence of the H3 phase and, as a result, the enhancement of structural stability. In addition, the overlapping of CV curves is better for 1% Te-NCM than NCM, which indicates better reversibility [55].

Fig. 5.

The CV profile (a) NCM (b) 1% Te-NCM (0.1 mV/s, 2.7–4.3 V).

At several C-rates from 0.1 to 5C after every four-cycle and then restored to 0.1C, rate capability of NCM, and 1% Te-NCM cathode materials are demonstrated in Fig. 6(a). Both samples have a decreasing trend of capacity with the increment in the C-rate, which is caused by polarization. Furthermore, the specific capacities of the 1% Te-NCM sample have superior values in comparison with NCM. It can also be observed from the results that the 1% Te-NCM still offers ∼156 mAh/g (5C), which is considerably greater than pristine NCM (∼123.1 mAh/g). Also, when C-rate is restored to 0.1C, 1% Te-NCM still provides a capacity of 212.3 mAh/g, over NCM (193.2 mAh/g). Hence, 1% Te-NCM cathode attains greater capacity and enhanced rate capability owing to its structural retention after the migration of Li-ions.

Fig. 6.

(a) Rate capability (b) Nyquist plots of NCM and 1% Te-NCM cathodes (c) plots of Z′ versus ω^−1/2.

The interfacial characteristic of both prepared samples was investigated after the 3rd and 100th cycles via electrochemical impedance spectroscopy (EIS) along with the Nyquist curves depicted in Fig. 6(b). The region of high-frequency provides insight into surface film resistance (R_s) as well as charge transfer resistance (R_ct) present at interfacial region. On the other hand, region of low frequency is ascribed to Warburg impedance which gives valuable information about Li⁺ diffusion kinetics [56,57].

The resistance values in Table 2 depict a harsh increase in surface film along with R_ctof NCM from 32.29 to 100.5 Ω for the 3rd to 100th cycle, respectively. Whereas 1% Te-NCM represents a slight increment in R_sf + R_ct from 17.24 to 40.87 Ω. The lower impedance values indicate decline of side reactions that occur among cathode and electrolyte along with the facilitation of Li⁺ diffusivity. The Li⁺ diffusion coefficient is listed in Table 2, evaluated via equation (1):

$$D_{{Li}^{+}}=\frac{R^2{T}^2}{2A^2{n}^4{F}^4{C}^2{\sigma }^2}$$ (1b)

Here R is gas constant, n denotes the number of electrons for lithium ions, A stands for geometric area i.e., 0.785 cm² of the electrode, T stands for the kelvin temperature, F denotes Faraday's constant and C is lithium ions' concentration i.e. 0.001 mol/cm³ and also $\sigma$ (Warburg impedance) is determined by equation (2):

$$Z^{\prime }=R_{sf}+R_{ct}+\sigma {\omega }^{-1/2}$$ (2)

Table 2.

EIS fitted values after 3rd and 100th cycles with corresponding lithium-ion diffusion coefficient.

	3rd Cycle		100th Cycle
Samples	Rsf + Rct(Ω)	D Li+(cm2/S)	Rsf + Rct(Ω)	D Li+(cm2/S)
NCM	32.29	$2.55\times {10}^{-12}$	100.5	$2.43\times {10}^{-13}$
1% Te-NCM	17.24	$4.57\times {10}^{-12}$	40.87	$1.32\times {10}^{-12}$

For NCM, calculated Li⁺ diffusion coefficients are $2.55\times {10}^{-12}$ and $2.43\times {10}^{-13}$ cm²/S after tested for 3rd and 100th cycles respectively. In comparison for 1% Te-NCM, $D_{{Li}^{+}}are$ $4.57\times {10}^{-12}$ , and $1.37\times {10}^{-12}$ cm²/S after tested for 3rd and 100th cycles respectively. Hence, 1% Te-NCM exhibits enhanced Li⁺ transportation and stabilized cathode-electrolyte interface. A comparison of the electrochemical characteristics of NCM after doping with various elements is shown in Table 3.

Table 3.

Comparitive electrochemical performance of NCM after doping with various elements.

Doping element	Potential	Discharge capacity (mAh/g)	Retention/Cycle Number	Reference
Ti	2.8–4.3	214.9 at 0.1C	77%/150/1C	[58]
Nb	3.0–4.3	202.3 at 0.1C	90.6%/100/0.1C	[59]
Zr	3.0–4.5	180 at 1C	74%/200/1C	[60]
Te	2.7–4.3	225.8 at 0.1 C	82.2%/100/0.5C	This work

4. Conclusion

In summary, pristine NCM and Te surface-doped NCM cathode materials are prepared via a solid-state route at high temperatures. The optimal Te doping (1 wt %) boosted electrochemical features of cathode material. The presence of strong bonding between Te and oxygen becomes the cause of binding the TM-O slabs and prevents the surface oxygen emission and consequently reduces the structural degradation. Specifically, 1% Te-NCM provides a higher capacity of 225.8 mAh/g (0.1C) and 156 mAh/g (5C) , as compared to pristine NCM (188.4 mAh/g (0.1C), and 123.1 mAh/g (5C) ). Furthermore, retention of capacity for 1% Te-NCM and NCM are 82.2% and 74.6% respectively at 0.5C over 100 cycles. Hence, 1% Te-NCM with remarkable capacity, improved rate capability, along with excellent cyclability, can be a promising LIB cathode.

Data availability statement

No data was used for the research described in the article.

CRediT authorship contribution statement

Annam Butt: Writing – original draft, Investigation, Conceptualization. Sidra Jamil: Writing – original draft, Formal analysis, Data curation. Muhammad Fasehullah: Investigation, Formal analysis. Haseeb Ahmad: Project administration, Investigation. Muhammad Khurram Tufail: Investigation, Conceptualization. Rehana Sharif: Validation, Supervision. Ghulam Ali: Writing – review & editing, Validation, Supervision, Funding acquisition.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix ASupplementary data

The following is the supplementary data to this article: