Atomic-scale combination of germanium-zinc nanofibers for structural and electrochemical evolution

Alloys are recently receiving considerable attention in the community of rechargeable batteries as possible alternatives to carbonaceous negative electrodes; however, challenges remain for the practical utilization of these materials. Herein, we report the synthesis of germanium-zinc alloy nanofibers through electrospinning and a subsequent calcination step. Evidenced by in situ transmission electron microscopy and electrochemical impedance spectroscopy characterizations, this one-dimensional design possesses unique structures. Both germanium and zinc atoms are homogenously distributed allowing for outstanding electronic conductivity and high available capacity for lithium storage. The as-prepared materials present high rate capability (capacity of ~ 50% at 20 C compared to that at 0.2 C-rate) and cycle retention (73% at 3.0 C-rate) with a retaining capacity of 546 mAh g−1 even after 1000 cycles. When assembled in a full cell, high energy density can be maintained during 400 cycles, which indicates that the current material has the potential to be used in a large-scale energy storage system.


Comment 2: Why Zn is selected? If other metallic elements have similar effect?
Response 2: Thanks to the reviewer for asking this great question. Based on phase diagram of binary metal system, elements that can form alloy with Ge are largely classified into two groups. First, for example, Cu or Fe metal can form alloys with various phases like Cu 3  Ag-Ge, Au-Ge, Sb-Ge, and Sn-Ge. However, Ag and Au metals are unfortunately expensive and the conditions are restrained owing to sensitive metal precursors. Besides, Sb has the difficulty in synthesizing via electrospinning method because the solubility of Sb precursor in water and mixing condition with Ge precursors make them not suitable for this system. Sb-Ge can be a good candidate for structural novelty but it is not suitable for our synthetic system. As another example, Sn element features almost similar thermodynamic behavior when mixed with Ge and implies the feasibility of homogeneous structure with atom-level distribution, which is similar to Zn-Ge system. In experimental result in Fig. RA-3e, Sn-Ge nanofibers show overall amorphous phase after calcination with broad carbon position. Then, after reduction, those had been reduced to Ge, Sn, and GeSn with partial germanium oxide or tin oxide. As compared with Ge-Zn system, tin oxide has lower thermodynamics of reduction under hydrogen gas atmosphere than zinc oxide. So, tin should be emerged after reduction and some alloying with Ge. Furthermore, tin can melt down at given reduction temperature so Sn element reacts with Li ion as many as Li x Sn (0<x<4.4) so that it can trigger larger volume expansion and then structural failure during long-term cycle test, whereas Li y Zn (0<y<1) provides limited volume expansion. Based on these reasons, we lastly chose Zn elements to make atomically distributed Zn-incorporated Ge nanofiber considering various parameters.
Based on these points below, we have added sentences in the revised manuscript, with a figure in the supplementary information.
(Revised manuscript, page 3, line 22-30) Among various kinds of metals, Zn was carefully selected as some metals (such as Cu and Fe) that form alloys with Ge, verified from X-ray diffraction (XRD) analysis before and after reduction ( Supplemnentary Fig. 1a). Moreover, other kinds of metals (such as Ag, Au, Sb, and Sn) can be combined with Ge similar to Zn, but other limitations are present. For Ag and Au, they are very expensive and synthetic conditions are difficult due to their sensitive precursors; for Sb, it is difficult to be mixed together with Ge in electrospinning solution; for Sn, it can be easily mixed with Ge and form various kinds of phases, from GeSn to partial Ge oxide and Sn oxide ( Supplemnentary Fig. 1b). As a result, Zn was selected as the most optimal element to carry out further experiments to synthesize O-dGZNFs.

Comment 3:
The sintering temperature is higher than 450 °C, why there is still much carbon remained in the materials?

Response 3:
We appreciate the reviewer's comment, which will strengthen our manuscript.
As noted by the reviewer, it is necessary to explain how much carbon is remained in the materials. Based on the previous literature (Mater. Res. 2015, 18, 509-518), it can be observed that not all carbon is decomposed at 450 °C, and even at the temperature of 500 °C, some carbon remained. Especially, in the presence of metal precursor together with PVP, the decomposition of PVP is expected to be slower and occurs at higher temperature, as the crystallization of metal oxides formed from metal precursors also takes place.

Comment 4:
The statement and analysis of the bonding between Ge, Zn and O are ambiguous, please improve it. Some statement seems inconsistent with the data, for example the peak at 750 cm-1 is almost sightless in O-dGZNFs, please modify or give some explanation.

Response 4:
We appreciate the reviewer's comment, which will strengthen our manuscript.
The dashed line in Fig. 2a was not exactly matched with real position. The shoulder exists in 750 cm -1 in Raman spectra. We re-aligned the peak position. We elaborated further in the manuscript to clarify some statements of the bonding between Ge, Zn, and O. In addition, to examine the effect of oxygen content in the electrochemical performance, charge capacity retention tests ( Supplementary Fig. 15) were carried out for O-dGZNFs with different Ge/Zn ratio, where higher portion of Zn leads to higher oxygen content. When the oxygen content is higher, it resulted in stable cycle retention owing to high ionic/electronic conductivity, but showed poor reversible capacity. On the other hand, when oxygen content was lower, it resulted in higher initial capacity but capacity fading was more prominent. (black) and 650°C (red) in air.

Comment 7:
The annotation in Fig. 4c seems to have some mistake.

Response 7:
We appreciate the reviewer's comments, which correct our mistakes. We modified the annotation in Fig. 4c of the revised manuscript, as shown below: (Revised manuscript, Fig. 4c Furthermore, the conductance of O-dGZNF in state iii was still 31 times higher than that of O-dGNF even after complete lithiation of both NFs -this is attributed to the higher intrinsic conductivity of Zn (1.7 x 10 7 S m -1 ) compared with that of Ge (2 x 10 3 S m -1 ) in bulk and lithiated Ge also exhibits enhanced electrical properties. (Revised manuscript, page 7, line 9-11) When the Li ion diffusion coefficients were further compared after the 1 st , 10 th , and 50 th cycle, they gradually increased, which are proportional to the increased amount of lithium composites ( Supplementary Fig. 17 and Supplementary Table S1).

Comment 1: Are there carbon and nitrogen element in Ge-Zn nanofiber? Carbon and
nitrogen element can be seen in Fig. 1e. This is very important.

Response 1:
That is an excellent question. Because we used PVP polymer to fabricate onedimensional nanofiber, carbon and nitrogen elements have to be included in our samples. So, we confirmed the distribution of carbon and nitrogen atoms in Fig. 1e. Further, we specified the weight percent of each carbon and nitrogen using EDS/EA analysis as shown in Supplementary Fig. 2. These data can be helpful to understand the existence of carbon and nitrogen in nanofibers.

Comment 2:
The carbon derived from PVP also improves the electronic conductivity.

Comment 3: Why the capacitive contribution of o-dGNFs is higher than o-dGZNFs? Why
the capacitive contribution was increased after 50th cycles? Please check the particle size before and after 50th cycles.

Response 3:
We thank the reviewer to raise this great point. At first, both electrodes have higher capacitive currents rather than faradaic currents due to quite high surface area of onedimensional nanofibers. In contrast, after 50 cycles, metallic bonds are developed in shape of Ge-Ge or Ge-Zn in the electrode with Li 2 O and Li 2 CO 3 formation. It means that these bonds can increase the kinetics of charge transfer under same condition of electrode before cycling.   The Zn portion in electrode can perform high electronic conductivity and this factor directly affects to high current density stability. That is why O-dGZNFs electrode at 2.0 C-rate shows quite stable electrochemical results (Fig. 3d) while O-dGNFs eventually lose their reversible capacity upon cycling due to poor electronic conductivity even though they have stable SEI layer on the surface of electrode.

Synthesis of O-dGNFs and O-dGZNFs. As-synthesized O-iGNFs and O-iGZNFs underwent
reduction process in quartz furnace (OTF-1200X-II, MTI corporation) filled by argon (Ar). In this process, furnace under Ar was heated to 600°C at a ramping rate of 5 °C min -1 and once the temperature reached an expected value, the atmosphere was changed to Ar and hydrogen gas mixture (Ar/H 2 (96/4, v/v)) and maintained for 1 h. Afterward, the furnace was filled again by Ar instead of Ar/H 2 and cooled down spontaneously to room temperature.

Comment 2:
In the Fig. 3d, the O-dGNFs shows a significant capacity gain from 50 cycles to  Figure RA and B. Then, we modified and added the images in Fig. 1. Please see the revised manuscript.   Here, to clearly optimize and confirm the structural evolution, we added charge/discharge curve at various cycle states. As shown in Figure RC, initial 10 cycles have slight plateau, attributed to conversion reaction for the oxide reformation, related to equation (R1). But, the plateau is eventually reduced and finally the shape is changed to linear slope at followed cycles while alloying reaction parts still remain reversibly, assigned to equation (R2). Thus,

Response 3:
We appreciate the reviewer's great point, which will strengthen our manuscript.
As shown in Fig. 3