EPR spectroscopic technique is very well suited for understanding the role of native point defects since it provides a direct way to observe various paramagnetic defect states. Thus, it complements other experimental methods that are giving information on the electronic structure such as Raman spectroscopy. Combined studies of Raman and EPR spectroscopy always give extensive information on the defect structures in particular for SiCN39, MgB240, 41, ZnO42, and C-dots43, 44. On the other hand, in B4C sample systems, due to lack of analytical characterization such as data from advanced EPR techniques, which preclude the unambiguous determination of defect states, the role of native defects on B4C properties remains unclear and is still a matter of debate. This is mainly due to the fact that at present most of the researchers who investigate B4C properties are trying to improve phase stability while overlooking the effect of defect structures at a microscopic level. For instance, one of the carbon inclusions in polycrystalline B4C studies by EPR unambiguously showed that point defects played a significant role in achieving a much higher surface-to-volume ratio by the inclusion of free C at the surface layer of the material45. In another study, the dependence on temperature and thermal treatment of the B4C powders were studied in detail by EPR spectroscopy by just probing the intrinsic defect structures. It has been demonstrated that native defects of B4C and conduction electrons are responsible for EPR absorption13.
As a result, in the enlightenment of previous studies, we employed here both Raman and EPR spectroscopy and deduced the valuable information as follows: Fig. 3 presents the Raman spectra of the graphite and active C raw materials. The first peak is the D-band (pseudo-Voigt) around 1340 cm-1, a defect-induced Raman mode of graphite that is not observed in perfect graphitic structures46. The G-band (pseudo-Voigt) around 1580 cm-1 belongs to defect-free graphite47. There are three peaks in the Raman spectra of graphite presented in the wavenumber range of 100 to 3000 cm-1, which correspond to D- (~1344 cm-1), G- (~1571 cm-1), and 2D- (sometimes called G') (~2700 cm-1) bands, respectively. The G-band appeared at a slightly higher wavenumber (~1592 cm-1) in the spectra of active carbon than in graphite. This shift may result from the crystallite size differences between different carbon domains.
The intensities of these bands are related to the amount of graphitization degree, where the intensity of D-band is proportional to the amount of disordered sp3 carbon, and the intensity of the G-band is proportional to the amount of ordered graphitic (sp2) carbon contained in the sample. The 2D-band is also related to the amount of disordered sp3 carbon, but its intensity usually affects the property of the used laser. The 2D-band around 2700 cm-1 was not observed in the spectra of active carbon (Fig. 3). Since this is related to the performance or intensity of the used laser, 2D-band peak is not used for analyzing the hybridization behavior or conductivity of the samples. Therefore, we compared here the intensities of the two characteristic main peaks of D- and G-bands, to determine the graphitization degree of two different carbon precursors.
The D-band, which arises from the defects and disorders in the carbon lattice, significantly increased in the Raman spectra of active carbon48. Based on the origins of the D- and G-bands, the intensity ratio of the D- to G-band (I(D)/I(G) ratio) can be used to estimate the defect density of carbon materials 48. According to the Raman spectra in Fig. 3, the I(D)/I(G) ratio increased from ~0.25 to ~0.99, as the carbon material changed from graphite to active carbon. The high I(D)/I(G) ratio in the spectra of active carbon proves its highly defected and disordered structure. On the other hand, a low I(D)/I(G) ratio in the spectra of graphite would mean the graphitization is high, which might be lead to a better electrical conductivity44.
EPR spectroscopy is one of the superior magnetic resonance techniques which give valuable information on the local electronic configuration in the crystal lattice. In particular, EPR is extremely sensitive to paramagnetic metal ions such as Fe3+, Mn2+, and Cr3+ or defect centers of diamagnetic materials where the electrons are trapped and become ionized hence paramagnetically active. In this context, materials like semiconductor ZnO49, superconductive MgB240 and, high-dielectric Ta2O550 show EPR active defect centers although they are diamagnetic. The formation energy of defect centers mainly determines the concentration of the defects that mainly exist in the material. The main defect structure discussion in B4C material is whether the EPR signal arises from the free-carbon or localized-carbon defects. The discussion in literature mainly focuses on these two main defect centers based on the interexchange of carbon and boron atoms and the possible existence of B12, B11C, and B10C2 icosahedra, as well as the permissibility of the C–B–B, C–B–C, and C–C–C chains. In the present work, high-grade B2O3 was used together with either Active C or graphite for obtaining B4C. The defect structures will be explained by monitoring the defects in starting materials, basically the carbonaceous ones. Finally, their defect contribution will be analyzed in B4C. In Fig. 4(a), X-Band (9.64 GHz) room temperature EPR spectra are given for Active C, graphite, B2O3, and the final products of B4C were presented. As expected, diamagnetic B2O3 in EPR inactive therefore no significant EPR signal was detected. This also shows that the B2O3 starting material is defect-free and does not possess any impurities. In general, high ceramic materials such as PbTiO351, BaTiO352, or PbZrTiO353 are EPR inactive because their crystal field is too high (in the order of GHz) compared to Zeeman energy. For instance, the crystal field energy of ZnO is in the order of MHz and it always gives EPR signal due to intrinsic defect centers such as oxygen or zinc vacancies/interstitials54. On the other hand, carbonaceous materials of Active C and graphite both revealed EPR signal with completely different features. The g-factor and the EPR linewidth of both carbonaceous materials were different indicating that the defect kinds and their environment are different. Also, the EPR intensity of both materials is different indicating different concentrations. In the present case, active C and graphite have two distinct difference i) the isotropic g-factor of active C is 2.0031 and graphite is 2.0098 (refer Table 2) and, ii) the linewidth is much higher for graphitic carbon. Such distinctive features in EPR signal of defects reflect their intrinsic characteristics into the produced end product, B4C. The interesting EPR features can be understood in Fig. 4(b) in accordance with Table 2. By the aid of sophisticated spin counting procedures it is possible to determine the defect concentration from EPR spectrum. The details of spin counting procedure are given in Supp. Mater. Shortly, the area of doubly integrated EPR spectrum is directly proportional to the concentration of paramagnetic species. Hence, the highest defect concentration here has the S1@G3 sample which is based on graphitic carbon. Compared to other graphitic samples S1@G6 which is a longer milled S1@G3 sample has almost factor 2 higher defects. One of the best ways to understand the electronic properties of trapped electrons at the defect sites via EPR is to monitor their EPR intensity change by increasing the microwave (MW) power gradually. The square root of MW power versus intensity profiles which are given here in Fig. 4 (c-h) is the key results to see whether the defects are contributing to the electrical conductivity or not. Such an approach has been already applied very effectively to ZnO and other kinds of materials41, 42. According to the results obtained via microwave power saturation, except for the graphite sample, all other materials revealed unsaturated line shape. Graphite in Fig. 4(d) has a strong deviation from linear dependency and at around 25 mW starts to saturate. On the other hand, graphite and active C contains different species of paramagnetic defects hence they have completely different saturation behavior. On the other hand, all B4C material revealed an unsaturated curve which also indicates the same kind of defect center which is paramagnetic. In short, the physical meaning of saturated and unsaturated curves in EPR is as follows49, 55. Easy saturated systems reveal Gaussian type lineshape indicating inhomogeneous broadening. Such species mostly consist of localized electrons and contribute to conduction whereas non-saturated systems have EPR line shape of Lorentzian that are mostly localized and give strong deviation from the free-electron g-factor which is 2.0023. As it is seen in the present case graphite shows both easy saturation and it has the most deviated g-factor which is 2.0098 as given in Table 2. Thus the B4C samples synthesized based on graphite will give higher conductivity. Hence we concentrated more to the sample of S1@G3 which is made of graphite and having the highest defect concentration according to spin counting. Therefore we have tested its performance in a supercapacitor device and present its electrochemical performance.
Table 2: g-factors, integrated area of EPR signal, and the accurate defect concentration of carbon-based starting materials and B4C samples synthesized via various conditions.
|
g-factor (isotropic)
|
Integrated area
|
Defect concentration (spins/g)
|
Active C
|
2.0031
|
1.23 x102
|
2.15x1017
|
Graphite
|
2.0098
|
2.82 x102
|
4.93x1017
|
S1@G3
|
2.0023
|
4.81 x103
|
8.41x1018
|
S2@G6
|
2.0031
|
2.52 x103
|
4.41x1018
|
S3@A3
|
2.0025
|
5.36 x101
|
9.38x1016
|
S4@A6
|
2.0027
|
4x103
|
7x1018
|
Finally, in Fig. 5 (a-d) and 6 (a-b) electrochemical performance test of B4C materials is presented when they are used as an electrode for the supercapacitor device. The device design and the components of the supercapacitor can be seen in Fig. 2. According to PEIS results given in Fig. 5 (a-d) the pronounced effect of defect centers can be seen as follows: the equivalent series resistance (ESR) which mostly responsible for the charge transfer is too low compared to the other three designs. Based on the Nyquist plot, the resistance is around 80 Ohm which shows a highly conductive system compared to the ones given in Fig. 5 (b-d). This supercapacitor also did not reveal any other circuit elements such as Warburg. However, the resistive Warburg element has been obtained for the other three designs indicating at a certain low frequency the mass transport or ion diffusion starts. This also prevents the system to form a complete semi-circle. But in the case of S1@G3 a complete semi-circle has been observed thus the designed supercapacitor has only the ESR and capacitance elements involved in faradaic reactions which gives a typical Randless cell56. Whereas, the other three cells have modified Randless cell with Warburg element. Furthermore, this optimized behavior of S1@G3 electrode motivated us to perform GCPL measurements on it and test its specific capacity, energy and power density and, Coloumbic efficiency.
With the aid of GCPL at the current scan rate of 2.5 A/g very high energy density values were obtained. A typical supercapacitor device possesses 1 Wh/kg energy and 1000 W/kg power density according to the given Ragone charts57. In this work, the energy density value is beyond the typical value which is around a factor 60 higher. Nonetheless, the power density is somewhat half of the typical supercapacitor which should be enhanced by better designs such as changing the electrolyte or using (reduced) graphene oxide instead of active C for the second electrode. The results show that the B4C materials are promising candidates for supercapacitor devices with their high energy density while the major problems of the supercapacitors are their low energy density. In terms of specific capacity, it has been reached around 2000 mAh/g at the first cycles which is a good value for a supercapacitor (refer Fig. S1). The Coulombic efficiency was obtained by the ratio of charging and discharging curves with respect to cycle number and as it is seen in Fig. 6(b) the efficiency is almost outstanding for the device.
The Raman and EPR results indicate a close relationship between the intrinsic defects, starting materials (synthesis), and electrochemical performance. This has been confirmed by potentiostatic electrical tests.