3.1. Structure and Morphology of the [email protected] Composites
Structure and morphology of the initial and the thermally treated
[email protected] composites were investigated by SEM and TEM.
Figure 1a,d depicts representative top view images of initial
[email protected] composites. The spherical ZnO NPs with an average particle size of (4.7 ± 0.7) nm are homogeneously distributed on both sides of the EG sheets without observable aggregation. The corresponding histogram, which was obtained from at least 100 ZnO NPs on EG sheets, shows a very narrow size distribution. It should be noted that the ZnO particles are strongly attached to EG sheets—even by ultrasonic treatment they could not be removed [
33], indicating a strong interaction between ZnO particles and EG.
Figure 1b,c,e,f shows SEM and TEM images of
[email protected] °C and
[email protected] °C composites with corresponding particle size distributions. The observed particle sizes of the adsorbed ZnO NPs increase in both cases from ca. 5 nm (untreated) to approximately 42 nm at 600 °C and to over 77 nm at 700 °C, indicating further growth of the NP during the thermal post-treatment on the EG sheets.
The XRD patterns of initial
[email protected] composites (abbreviated with RT) and thermally treated
[email protected] composites at different temperatures are shown in
Figure 2a,b. The diffraction reflexes of EG in the XRD pattern are indexed with black triangles. EG exhibits four intensive diffraction reflexes at 2θ = 26.4°, 42.3°, 44.5°, and 50.6° corresponding to the (002), (100), (101), and (004) reflections of graphite (ICDD, PDF No. 00-056-0159). The characteristic graphitic reflex clearly appears at 2θ = 26.4° [
37].
The
[email protected] composites do not display the diffraction reflexes of graphite only, but also further diffraction reflexes located at 2θ = 31.7°, 34.3°, 36.1°, 47.4°, 56.5°, and 62.7°, attributed to (100), (002), (101), (102), (110), and (103) reflections of Wurtzite hexagonal structured ZnO (ICDD, PDF No. 04-008-8198), respectively. They are represented by the open circles in
Figure 2a. The broadening of the ZnO diffraction reflexes in the pattern of the initial
[email protected] composites indicates the low crystallinity of the nano-sized ZnO particles due to the absence of any long-range order [
38]. After thermal treatment, the three main ZnO reflexes (100), (101), and (002) sharpen with increasing temperature, confirming that the average ZnO crystallite size and the crystallinity are significantly increased (
Figure 2b). Nevertheless, the XRD pattern of
[email protected] °C composites shows the absence of the pronounced ZnO reflexes. Furthermore, the XRD patterns reveal no additional phases, confirming the high purity of
[email protected] composites, and no changes of the basic structure, even after the thermal post-treatment processes.
In order to estimate the amount of ZnO in the composite after thermal treatments at different temperatures, inductively coupled plasma optical emission spectrometry (ICP-OES) was carried out.
Figure 2c reveals a constant ZnO amount on the EG sheets up to 700 °C, followed by an abrupt decrease of the ZnO content at ≥720 °C, accompanied by an increase of the ZnO particle sizes. In order to confirm the ZnO content of the composites during the thermal post-treatment, TGA, EDX and XRD were applied (
Figure 2d). All methods show a similar ZnO content. For all samples, EDX measurements show slightly higher values due to the local limitations of this method. However, ZnO particles cannot be detected on the EG sheets after a thermal treatment at 800 °C. The corresponding SEM images (
Figure 3) support these findings and provide a deeper insight into the
[email protected] composites after thermal post-treatment at different temperatures, i.e., 600 °C (
Figure 3a), 700 °C (
Figure 3b), and 800 °C (
Figure 3c). The thermal treated
[email protected] composites show, on the one hand, increasing ZnO particles size with increasing temperature and, on the other hand, an exfoliated graphite without ZnO NP coverage after thermal treatment at 800 °C (
Figure 3c).
We attribute this to a carbothermal reduction of ZnO to Zn according to Equations (4) and (5) [
39,
40,
41].
The corresponding STA-MS analyses, which were conducted from 30 °C to 800 °C in argon, support these findings and provide a deeper insight into the combustion process of the initial
[email protected] composites (
Figure 4a). N
2 as carrier gas would not have been appropriate due to the overlap of the signals of N
2 and CO in the MS. The ion signal exhibiting a pronounced intensity at about 420 °C is assigned to CO
2 (
m/
z 44) and indicates the ZnO reduction according to Equation (5) (
Figure 4b). The reduction of ZnO to Zn continues up to 700 °C (
Figure 4c). The CO signal (
m/
z 28) appears first at 800 °C. It indicates the combustion of carbon (
Figure 4d). Thus, the remaining almost 90 wt% at 800 °C are attributed to graphite, while the loss of ca. 10 wt% demonstrates a good correlation with the initial amount of adsorbed ZnO NPs. However, up to a temperature of 800 °C, a Zn signal at 65
m/
z (as it is expected when Zn evaporates) was not detected, presumably due to the condensation of gaseous Zn at cold parts in front of the detector. The
m/
z 39 signal may be an indication for the hydrocarbon residues resulting from either solvent or defects in the graphite structure. Widespread peaks indicating Ar appear at 40
m/
z (Ar
+), 20
m/
z (Ar
2+) and 36
m/
z (
36Ar isotope). There is also a peak at 38
m/
z assigned to the third stable isotope of argon (
38Ar isotope) [
42,
43].
The carbothermal reduction during the thermal process should provide a lack of oxygen in the ZnO crystals. In order to prove this assumption, cross sections of thermally treated
[email protected] composites were investigated by EDX to determine the zinc and oxygen amount in the ZnO crystals. It is obvious from the SEM cross section of
[email protected] °C composites that a layered structure is formed, consisting of alternating arrangement of ZnO particles and EG sheets (
Figure 5a). It is evident that the ZnO particles (white particles) are sandwiched between the EG sheets (black sheets). The “wavy” signal background may stem from the embedding material consisting of EpoFix Resin and EpoFix Hardener (25:3).
The larger particles (treatment temperature ≥ 550 °C) facilitate a quantitative characterization of the ZnO stoichiometry due to the absence of the C signal originating from EG sheets (
Figure 5b). The ZnO particles of
[email protected] °C composites show a higher amount of oxygen compared to the amount of zinc. In contrast, the
[email protected] °C composites show a lack of oxygen supporting the assumption of carbothermal reduction. The amount of oxygen decreases with increasing temperature caused by ongoing carbothermal reduction.
3.2. Electrochemical Performance
The increase of the Zn-content in the ZnO particles (which can also be written as ZnO
1−x expressing the formation of oxygen vacancies in ZnO) is accompanied by the formation a donor level between the valence and conduction band and therefore by an increased density of conduction electrons [
44,
45,
46]. This should lead to an increased electrical conductivity of the electrodes, with positive effects on the electrochemical behavior. However, the increase of the particle size leads to longer diffusion paths for Li
+ ions insertion [
21,
47].
These composites are electrochemically investigated by galvanostatic measurements.
Figure 6a compares the cycle performance of EG and
[email protected] composites before and after the thermal treatments. The capacity of the initial
[email protected] composites is very low with only 21.7 mAh/g after the 50th cycle, which is below the threshold of initial EG with a value of 274.6 mAh/g. As obvious from
Figure 6a, the capacity of
[email protected] °C composites (218 mAh/g) is much higher compared to the initial
[email protected] composites, whereas the capacity value of pure EG has still not been achieved. Only if the initial
[email protected] composites are thermal treated above 600 °C, a higher capacity occurs. The
[email protected] °C composites show the best cycle performance among the four tested composites. A capacity of 343.5 mAh/g is maintained after the 50th cycle, which is a significant improvement of +25.1% compared to the initial EG substrate.
The enhanced electrochemical performance of thermally treated
[email protected] composites may be attributed to their increased conductivity. The electrochemical conductivity of the
[email protected] composite anodes is measured via EIS to confirm the relationship between the increased Zn stoichiometry in the ZnO particles and the enhanced electrochemical performance.
Figure 6b shows Nyquist plots (plots in the complex plane) of initial
[email protected] composites and of composites that were thermally treated at 600 °C and 700 °C. The plot consists of semicircles in the high frequency range that is presumably related to the charge transfer resistance and a straight line in the low frequency range that may reflect the Li
+ transport in the electrolyte. The intercept on the Re{Z} axis at high frequencies can be considered as the internal resistance (
Ri). Obviously, the diameter of the semicircle for initial
[email protected] composites is larger than that of the thermally treated
[email protected] composites, confirming the facilitated electron transport and of thermally treated composites [
24]. The internal resistance
Ri of initial
[email protected] composites is (6.3 ± 0.5)
, while the thermally treated composites show a decrease of
Ri down to (1.7 ± 0.2)
with increasing temperature (
Figure 6c). The increased electrical conductivity of the electrodes can be attributed to the reduction of the ZnO particles, indicating an important effect on the improved capacity utilization of the composites.
Figure 6d shows the relationship between the Zn stoichiometry in the ZnO particles for different particle sizes. The achieved capacity behaves similarly for different particle sizes (
Figure 6e). For better comparison, the capacity is indicated in % due to the slightly different ZnO amounts within the composites (
Figure 2d). Therefore, 100% indicate the theoretical capacity of each
[email protected] composite. It is shown that the increase of the Zn stoichiometry in the ZnO (or the increase of oxygen vacancies in ZnO that lead to increased electronic conduction) particles runs analogously to the course of the capacity, whereas the ZnO particle size does not seem to influence the capacity.
One may ask why the electrochemical performance of samples treated at 600 °C and 700 °C are so similar (
Figure 6a). A possible explanation for this behavior could be the similar physical properties of both samples. They exhibit comparable internal resistances (
Figure 6c) and an increased loss of oxygen in the ZnO particles was found (
Figure 5b). Nevertheless, the larger particles size of 700 °C sample negatively affects the stability of electrochemical performance. The thermal treated composite at 700 °C shows a capacity fading during electrochemical cycling.
It should be noticed that the capacities of the composites are not yet satisfactory due to the initial low capacity and rate performance (
Figure S1) of the used substrate. By transferring the presented concept to other types of graphite offering a higher initial capacity, the electrochemical performance can be further improved.
Galvanostatic discharge/charge curves of anode materials provide a deeper insight into the lithiation/de-lithiation process and help to better understand the reason for the poor electrochemical performance of the initial
[email protected] composites.
Figure 7 presents galvanostatic discharge (Li
+ ions intercalation) and charge (Li
+ ions de-intercalation) curves of EG, initial
[email protected] and
[email protected] °C composites measured between 0.05 V and 2 V at 0.05 C-rate for the 1st, the 2nd, the 3rd, and the 30th cycle.
Figure 7a shows a typical graphite profile containing long plateaus below 0.25 V indicating Li
+ ions intercalation and de-intercalation into the substrate [
48]. During the lithiation, Li
+ ions intercalate stepwise into the interlayer space of graphite, forming graphite intercalation compounds [
49]. The plateaus located at 0.20 V, 0.11 V and 0.08 V indicate the Li
+ ion intercalation steps into the graphite [
50]. In addition, the short plateau at approximately 0.7 V reveals electrolyte decomposition caused by the irreversible formation of an SEI layer in the first charge process [
51]. After the first cycle, the slopes and the curves are similar in shape indicating a good reversibility of the redox processes. The staging behavior becomes clearer in the differential specific capacity plots (d
c/d
U,
Figure 7b), which has been directly derived from the discharge/charge curves by differentiation. It is worth noting that the plateaus in the voltage curves coincide with the peaks in the d
c/d
U-profiles. The reduction peaks show the Li
+ ion intercalation steps into the graphite, whereas the oxidation peaks reveal the de-lithiation of graphite.
Figure 7c shows the galvanostatic discharge/charge curves of the initial
[email protected] composites. Compared with the EG electrode, it is obvious that the initial
[email protected] composites exhibit shorter plateaus as well as lower discharge and charge capacities. Additional information regarding the mechanism of the lithiation/de-lithiation process are revealed by the d
c/d
U-curves (
Figure 7d). In the first cathodic scan, there is a broad peak in the range between 0.6 V and 0.8 V. We attribute it to the first electrochemical process of the composite, which contains the SEI layer formation and the reduction of ZnO to Zn (Equation (1)). The potentials of these reactions are very close, thus only one broad peak is detected. The same occurs for Li
+ ion intercalation into the graphite and the formation of the LiZn alloy. However, the cathodic peak at 0.05 V is attributed to Li
+ ion intercalation into the substrate, whereas the cathodic peak at 0.1 V indicates the formation of the final LiZn alloy stage according to Equation (3). In the first anodic scan, only two peaks at 0.21 V and 0.27 V are detected. They indicate the partial de-lithiation of EG and the de-alloying of ZnLi to the next Li
2Zn
3 intermediate stage. The absence of peaks at 0.66 V and approximately 1.4 V indicates the incomplete de-alloying process of ZnLi to Zn and the irreversible ZnO reduction to Zn and Li
2O matrix. The irreversible reaction of some metal oxides to metal and the inactive Li
2O matrix is already well known [
52].
In summary, this may lead to the following picture: During electrochemical cycling, Li
+ ions react irreversibly with ZnO reducing ZnO to Zn and Li
2O matrix, then Li
+ ions alloys further with Zn. The electrochemical active species such as Zn and EG are partially involved in the reaction with Li
+ ions. The low electrical conductivity of the composite (
Figure 6c) involves kinetic inhibition making it impossible for Zn to pass through all alloying stages with Li
+. Hence, the reversible cycling occurs only between ZnLi and Li
2Zn
3 stages.
Peaks indicating the lithiation/de-lithiation process with graphite are either missing or are not as pronounced as in the case of EG. This implies that ZnO NPs adsorption takes place on the EG surface as well as on the EG edges preventing Li
+ ion insertion into the graphite layers and resulting in low capacity utilization. This seems to be the reason for the poor electrochemical performance of the initial
[email protected] composites.
Figure 8a shows a schematic illustration for the Li
+ ion insertion in the initial
[email protected] composites according to the above-mentioned findings.
Compared with the initial
[email protected] composite electrode, it is obvious that the
[email protected] °C composites exhibit longer plateaus, higher discharge and charge capacities (
Figure 7e), as well as pronounced cathodic/anodic peaks (
Figure 7f). Furthermore, two additional anodic peaks appear at 0.53 V (Li
2Zn
5) and 0.67 V (Zn) indicating the reversible de-alloying process of ZnLi to Zn. After the 1st cycle, the curves become almost identical in terms of shape and peak position indicating a reversible intercalation of the Li
+ ions into the composite.
[email protected] °C composites exhibit better electrochemical performance than the initial
[email protected] composites due to the better electrical conductivity (
Figure 6c) caused by the carbothermal reduction during the thermal post-treatment. The reduction changes the stoichiometry leading to an increased number of oxygen vacancies (
Figure 5b) and to a higher electrical conductivity (
Figure 6b) making it possible for Zn to pass through all alloying stages with Li
+ ions (Li
xZn
y), whereas the ZnO reduction is still irreversible. It is known from literature that the present Li
2O matrix acts as a “glue” to keep the alloying particles together and to reduce the volume change within the particles [
52,
53]. Hence, in the present work, both Li
2O matrix and EG contribute to improved cycle stability.
The thermal treatment introduces growing ZnO NPs on the EG sheets. Consequently, the ZnO particles do not block the paths for Li
+ ion intercalation inside the graphite layers. The accessibility of Li
+ ions into the EG is confirmed by the sharp and pronounced peaks below 0.25 V. Thus, two electrochemical active species, namely Zn and EG, are completely involved in the reaction with the Li
+ ions (
Figure 8b) providing an enhanced capacity of over 25% as compared to the pure EG anode.