Porous Ge@C materials via twin polymerization of germanium( II ) salicyl alcoholates for Li-ion batteries † 4 , 2705 – 2719

The germylenes, germanium( II ) 2-(oxidomethyl)phenolate ( 1 ), germanium( II ) 4-methyl-2-(oxidomethyl) phenolate ( 2 ) and germanium( II ) 4-bromo-2-(oxidomethyl)phenolate ( 3 ) were synthesized and their thermally induced twin polymerization to give organic – inorganic hybrid materials was studied. The compounds 1 – 3 form oligomers including dimers, trimers and tetramers as a result of intermolecular coordination of the benzylic oxygen atom to germanium. The structural motifs were studied by single crystal X-ray di ﬀ raction analysis and DFT-D calculations. Thermally induced twin polymerization of these germylenes gave hybrid materials based on germanium-containing phenolic resins. Carbonization of these resins under reductive conditions resulted in porous materials that are composed of germanium and carbon (Ge@C materials), while oxidation with air provided non-porous germanium dioxide. The porous Ge@C materials were tested as potential anode materials for rechargeable Li-ion batteries. Reversible capacities of 540 mA h g (cid:1) 1 were obtained at a current density of 346 mA g (cid:1) 1 without apparent fading for 100 cycles, which demonstrates that germanium is well accessible in the hybrid material.


Synthesis and characterization of germylenes
The germylenes 1-3 were prepared according to the synthesis protocol of 4 (ref. 68) starting from Ge[N(SiMe 3 ) 2 ] 2 and the respective salicyl alcohol with yields in the range from 64% to 73%. Poor solubility in non-polar solvents was observed for germylene 3, whereas the compounds 1 and 2 are soluble in all common organic solvents. 1 H NMR and 1 H 13 C{ 1 H} HSQC NMR spectroscopic analysis of freshly prepared solutions of the germylenes 1-3 in d 8 -THF at ambient temperature gave broad resonance signals (ve signals for 1 and 2 in each case, and four signals for 3) indicating a dynamic coordination behavior including oligomeric species, presumably dominated by their monomers. Single crystals suitable for X-ray diffraction analysis were obtained from saturated diethyl ether solutions of the germylenes 1 and 2, from a saturated 1,4-dioxane solution of 3 and from the cloudy reaction mixture (7 : 4 mixture of npentane and diethyl ether) containing compound 1 aer ltration and slow evaporation of the volatile solvents, respectively. In case of germylene 1, a trimer (crystallization from diethyl ether) and a tetramer (crystallization from the reaction mixture) were obtained in the solid state, whereas the compounds 2 and 3 form dimers. Noteworthy, a trimer was previously determined for germylene 4 upon crystallization from a saturated diethyl ether solution, whereas in CDCl 3 solution an equilibrium between the trimer and the dimer/tetramer was observed. 68 Thus, the presence of different oligomers of the compounds 1-3 in solution is presumed. DFT-D calculations support this assumption, which is in accordance with the relative formation energies of their oligomers (Scheme 2). The formation of the tetramers was calculated to be energetically slightly favored in comparison with the corresponding trimer formation, whereas the dimers are energetically less favored. However, the small energy differences between the oligomers, especially between their trimers and tetramers, support the assumption that they are in a chemical equilibrium in solution.
Compound 1 crystallizes in the monoclinic space group P2 1 / n and forms trimers (1) 3 in the solid state. The molecular structure is given in Fig. 1 and selected bond lengths and bond angles are presented in the caption of this gure. Details of the structure determination are summarized in Table 3.
The tetramer [2$(1) 4 $n-pentane] crystallizes in the tetragonal space group I 4 exhibiting two crystallographically independent tetramers, a-(1) 4 and b-(1) 4 , which show similar structural motifs (Fig. S1 †) and thus only a-(1) 4 is discussed. The molecular structure of a-(1) 4 is given in Fig. 2 and selected bond lengths and bond angles are presented in the caption of this gure. Details of the structure determination are summarized in Table 3.
Tetramer a-(1) 4 consists of four equivalent monomeric building blocks of 1 that can be superimposed as a result of a S 4 axis (Fig. 2b). The core units of the tetramer shows an eightmembered -[Ge-O-] 4 ring in which the germanium atoms span a distorted tetrahedron (Fig. 2b). The germanium atoms are three-coordinate by one terminal phenolic oxygen atom and two bridging benzylic oxygen atoms. The Ge1-O1 [1.987(3)Å] bond length is shorter than the Ge1-O1 0 [2.025(3)Å] bond indicating secondary bonding of O / Ge that stabilizes the low valent germanium atoms. The small deviations in the bond lengths and bond angles between a-(1) 4 and b-(1) 4 may result from crystal packing due to the presence of one molecule n-pentane in the unit cell. Please note, when 2$(1) 4 $n-pentane is slowly heated, a reconstruction of the framework to give (1) 3 and release of the n-pentane were observed by temperature-dependent PXRD and thermogravimetric analysis (TGA), respectively ( Fig. S2 and S3 †). In addition, 1 H NMR spectroscopic analysis in CDCl 3 at À60 C gave two sets of signals that are assigned to the trimer (1) 3 and the tetramer (1) 4 . The observation of only one set of signals for the tetramer of 1 indicates that a-(1) 4 and b-(1) 4 relax in their geometries to result in a single tetrameric structure in solution (Fig. S4 †).
The dimer (2) 2 crystallizes in the triclinic space group P 1. The molecular structure of (2) 2 is depicted in Fig. 3 and selected bond lengths and bond angles are presented in the caption of this gure. Details of the structure determination are summarized in Table 3. A similar structure to (2) 2 was determined for (3) 2 ( Fig. S5 †), and therefore, only the molecular structure of (2) 2 is briey discussed.
The dimer (2) 2 exhibits C i symmetry with a center of inversion located in the plane spanned by the Ge1, O2, Ge1 0 and O2 0 atoms. The equivalent benzylic oxygen atoms O2 and O2 0 bridge the germanium atoms, which exhibit a threefold pyramidal coordination. The germanium oxygen bond lengths and bond angles are in agreement with the values determined for (1) 3 ,(1) 4 and (4) 3 and thus are not further discussed. 68

Twin polymerization
Thermally induced twin polymerization of the germylenes 1-3 resulted in the formation of germanium-containing phenolic resin hybrid materials. According to the concept of TP the idealized structural motif of the hybrid material is illustrated by eqn (1).  The molecular structures of the compounds 2 and 3, exhibiting a substituent in para position to the phenolic oxygen atom, favor the ortho/ortho 0 connectivity of the bridging methylene groups interconnecting the benzene rings in the phenolic resin as obtained aer polymerization. By contrast, precursor 1 allows additional ortho/para 0 connectivity. In general, the phenolic resins that were prepared by twin polymerization show a high degree of cross linkages of the bridging methylene groups and are insoluble. It is noteworthy that our attempts to obtain polymeric hybrid materials by proton induced twin polymerization as reported for other metal salicyl alcoholates (M ¼ Si, Ge and Sn) 64,65,67 failed. However, thermally induced twin polymerization is feasible as it is outlined in the following section.
2.2.1 Studies on the reactivity. Studies on the reactivity of the compounds 1-3 regarding the process of thermally induced TP were performed by differential scanning calorimetry (DSC) measurements up to 300 C. The thermograms starting from amorphous materials are depicted in Fig. 4.
The compounds 1-3 do not melt and the DSC curves show exothermic processes, which are assigned to thermally induced TP. The onset temperatures are 102 C(1), 163 C(2) and 151 C (3). Two distinguishable minima at 110 C and 145 C for 1, 183 C and 194 C for 2, and 161 C and 200 C for 3 indicate that different formation processes take place at specic  (14) and O2-Ge1-O1 0 94.14 (12). Symmetry transformations used to generate equivalent atoms: 0 ¼ x À 0.5, Ày + 1.5, Àz + 0.5; 00 ¼À x +2 ,Ày +1 ,z; 000 ¼Àx + 1.5, y + 0.5, Àz + 0.5. temperatures during the polymerization processes e.g., formation of the phenolic resin and/or condensation to give the nal germanium-containing resin. The observation of two minima in all DSC experiments indicate that the polymerization follows similar reaction processes for 1-3, but the compounds show ad i fferent reactivity. However, the order of the onset temperatures does not follow the electronic and steric features of the substituents as was expected based on previous reports for spirocyclic silicon salicyl alcoholates. 65 Thus, we conclude that the nature of the substituents in 1-3 plays a minor role for the initiation process here. This assumption is further supported by comparison of the onset temperatures of the polymerization process for amorphous 1 (102 C) and the crystalline materials (1) 3 (142 C) and 2$(1) 4 $n-pentane (119 C) (Fig. 5).
The different onset temperatures indicate that the initiation of the polymerization process depends on structural parameters of the germylenes e.g., structure of the oligomer and/or crystal packing effects. The crystalline forms of germylene 1 gave higher onset temperatures and narrower temperature ranges (D15 C) for the exothermic processes than observed for the amorphous material. The exothermic processes gave a dQ dm value of À345 J g À1 for amorphous 1, À377 J g À1 for crystalline (1) 3 and À340 J g À1 for crystalline 2$(1) 4 $n-pentane. Thus, the ascalculated molar reaction enthalpies (DH) that are referred to one monomeric unit of 1 (M ¼ 194.73 g mol À1 ) are À67 kJ mol À1 for amorphous 1, À73 kJ mol À1 for crystalline (1) 3 and À66 kJ mol À1 for crystalline 2$(1) 4 $n-pentane. As the dQ dm values and as a consequence the molar reaction enthalpies (DH) are comparable for the amorphous and the crystalline materials of 1, the reaction energies do not rely on the structures of the precursors. It is noteworthy that the DSC curve of 2$(1) 4 $ n-pentane exhibits a second minimum at 146 C as it is also observed for amorphous 1 (145 C). These minima are close to the single minimum detected for the twin polymerization of (1) 3 (143 C) indicating that at least some formation processes in order to form the hybrid material e.g., formation of the germanium oxide, take place at higher temperatures. We conclude that amorphous materials possess smaller reaction barriers for the initiation process of thermally induced twin polymerization. The initiation of TP for the crystalline materials is aggravated presumably due to crystal packing. However, as the polymerization process is initiated, the thermally induced TP of the germylenes is facilitated by a pre-organized arrangement of the precursors considering the narrow temperature ranges of the polymerization process for the crystalline forms of 1. 2.2.2 Characterization of the hybrid materials obtained by thermally induced twin polymerization. Thermally induced twin polymerization of 1-3 was carried out by bulk phase experiments under inert atmosphere. Polymerization temperatures of 200-230 C were chosen according to the results of the DSC measurements of the germylenes 1-3. The as-obtained germanium-containing phenolic resin hybrid materials were characterized by 13 C{ 1 H} cross polarization magic angle spinning (CP-MAS) NMR and IR spectroscopy, PXRD, CHN analysis, XPS (exemplarily carried out for HM-1) and EDX spectroscopy. A summary of the polymerization temperatures, the product identities and the yields of the as-obtained hybrid materials is given in Table 1. Fig. 6 depicts the 13 C{ 1 H} CP-MAS NMR spectra of these materials.
All expected signals for the phenolic resin of the hybrid materials HM-1-HM-3 were observed with typical chemical shis as reported for hybrid materials obtained by thermally induced TP. 55,65 The broad resonance signals with chemical shisofd ¼ 20-30 ppm and the resonance signals between d ¼ Fig. 4 Differential scanning calorimetry of the amorphous materials of the germylenes 1 (black curve), 2 (red curve) and 3 (blue curve); heating rate 10 K min À1 ,N 2 atmosphere, N 2 volume flow 50 mL min À1 . that are referred to one monomeric unit of 1 of the exothermic processes are depicted in black for amorphous 1, red for crystals of (1) 3 and blue for crystals of 2$(1) 4 $n-pentane.
110-120 ppm (unsubstituted ortho carbons: d ¼ 113-117 ppm (ref. 72)) indicate the presence of nonequivalent bridging methylene groups in the phenolic resin for all hybrid materials. [72][73][74] Nevertheless, the resonance signals assigned to the bridging methylene groups (d ¼ 20-30 ppm) exhibit the most intense resonance signals centered at d ¼ 23 ppm indicating that an ortho/ortho 0 connectivity of the bridging methylene groups is prevalent in the phenolic resin even for HM-1. These values match with reference data reported for ortho/ortho 0 bridging methylene carbons of phenol-formaldehyde resins (d ¼ 20-27 ppm). 73 In addition, a trend for the content of unsymmetrical connectivity patterns of the bridging methylene groups with respect to electronic and steric features of the substituents is not observed, which is in contrast to our studies of the phenolic resin-silica hybrid materials obtained by thermally induced TP of spirocyclic silicon salicyl alcoholates. 65 The latter and the prevalent ortho/ortho 0 connectivity of the bridging methylene groups indicate that the connectivity motif of the phenolic resin in the hybrid materials is only marginally inuenced by the nature of the substituents at the aromatic moieties of the germylenes. It is noteworthy that the resonance signals, which are assigned to the bridging methylene groups, exhibit shoulders with chemical shis centered at d ¼ 16 ppm in all 13 C { 1 H} CP-MAS NMR spectra. This observation is assigned to the presence of small amounts of Ge-CH 2 connectivity within the hybrid materials (Fig. S6 †). The latter is a result of an oxidative insertion reaction of the germylenes as reported previously for compound 4, which slowly converts into 2,4,6, S6 †). 68 Compound 5 serves as a model compound for XPS analysis that was exemplarily carried out for HM-1 due to its Ge-CH 2 groups. The XPS data indicate that the germanium atoms in HM-1 exhibit a similar chemical environment as it was determined for 5 ( Fig. S7 †). However, the XPS data are also consistent with data reported for germanium oxides. [75][76][77][78][79][80][81][82][83][84][85] Thus the XPS data do not allow to differentiate between the presence of Ge-CH 2 connectivity and germanium oxides within HM-1. However, in accordance with the 13 C{ 1 H} CP-MAS NMR data, we suggest that thermally induced TP of the germylenes provides hybrid materials consisting of germanium oxide incorporated in a phenolic resins, which contains small portions of Ge-CH 2 connectivity.  Typical IR absorption bands for n(O-H) (3600-3200 cm À1 ), n(C]C) (1600-1400 cm À1 ), n(C-O) (1250-1200 cm À1 ), aromatic backbone vibrational modes (850-700 cm À1 ) and Ge-O vibrational modes (430-650 cm À1 ) were observed in the FT-IR spectra of all hybrid materials (Fig. S8 †). An additional absorption band maximum at 1096 cm À1 was observed in the IR spectrum of HM-3, which is assigned to the C-Br stretching vibrational mode. CHN analyses revealed carbon contents of the hybrid materials that are slightly lower than the values (DC < 1.1%) calculated for complete conversions of the germylenes. The latter is in agreement with the determined minor weight losses (Fig. S3 †) during the polymerization processes. However, the rather small deviations in the determined carbon and hydrogen contents in comparison with their calculated values indicate a high amount of conversion of the germylenes into the hybrid materials. This is further supported by EDX analysis and the high yields as obtained for the hybrid materials aer the work up procedures.

Synthesis and characterization of porous materials
Conversion of phenolic resin-silica hybrid materials obtained by TP of spirocyclic silicon salicyl alcoholates into microporous carbon and into mesoporous silica was reported by us recently. 65 Variations of the molecular structure of the precursors e.g.,d i fferent substituents at the aromatic moieties were shown to affect the properties of the porous materials. 65 Expecting qualitatively similar results, we carried out the conversion reactions of the germanium-containing phenolic resin hybrid materials (HM-1-HM-3) under reductive and oxidative conditions in order to obtain porous materials consisting of germanium incorporated in a carbon matrix (Ge@C) and porous GeO 2 , respectively (Scheme 1). The product identities of the as-obtained Ge@C and GeO 2 materials, their BET surface areas and in case of the Ge@C composites the pore volumes determined by nitrogen adsorption measurements in addition with their germanium content are summarized in Table 2.
The as-prepared Ge@C materials exhibit large BET surface areas with C-3 (418 m 2 g À1 )o ffering the largest value. The adsorption isotherms of the Ge@C composites are assigned to type I isotherms showing hysteresis loops (Fig. S9 †), which may result from bottleneck pores. Analysis of the adsorption isotherms using QSDFT gave 14.9% (C-2), 25.1% (C-1a), 34.5% (C-1b) and 46.7% (C-3) micropore contents (Fig. S10 †). CHN analyses gave carbon contents of 32.66% (C-1a), 34.41% (C-1b), 31.58% (C-2) and 60.89% (C-3). In addition, EDX spectroscopy revealed germanium contents of (46.5 AE 2.8)% (C-1a), (39.8 AE 2.4)% (C-1b), (36.6 AE 2.0)% (C-2) and (26.5 AE 1.6)% (C-3). Bromine was not detected by EDX analysis of C-3. The materials C-1a, C-1b and C-2 exhibit low carbon but high germanium contents, whereas a low germanium and a high carbon content were determined for C-3. The mass losses during the synthesis of C-1a, C-1b and C-2 were about 50%, whereas 68% was observed for C-3. The latter, in addition with the high carbon value, low germanium and lacking bromine content as determined for C-3, indicate that large amounts of germanium and almost all bromine were removed during the synthesis of C-3 resulting in a material exhibiting a much higher carbon content (C: 60.89%) as initially present in the starting material HM-3 (C: 30.10%) or germylene 3 (C: 30.40%). As carbon is mainly the porous component of the Ge@C materials, the higher specic surface-to-mass ratio of C-3 results from larger proportions of carbonized material in addition with the higher micropore content that may result from decomposition processes ( Fig. S3 and S11 †) causing the removal of all bromine and some germanium. Raman spectroscopy using a confocal micro Raman system revealed for Ge@C areas that exhibit either high germanium but low carbon or high carbon but low germanium contents (Fig. S12 †). The latter is indicative that phase separation between germanium and carbon on the mm-scale occurred during the synthesis of the Ge@C composites under reductive conditions, which was conrmed by scanning electron microscopy in combination with EDX mapping (SEM/EDX) (Fig. S13 †). However, PXRD analyses revealed that the Ge@C composites exhibit crystallites with average primary particle size of (51 AE 6) nm (C-1a), (13 AE 1) nm (C-1b), (39 AE 3) nm (C-2) and (38 AE 3) nm (C-3) (Fig. S14 †). TEM images exemplarily recorded for C-1a and C-1b of selected carbon-rich areas revealed the presence of nanoscaled areas (<10 nm) consisting of crystalline germanium (Fig. S15 †). Notably, smaller crystallites were obtained at 600 C [(13 AE 1) nm for C-1b] as compared to 800 C [(51 AE 6) nm for C-1a], while the BET surface area does not differ signicantly. The as-prepared GeO 2 samples exhibit low BET surface areas, which is indicative for a non-porous material. This is further supported by the adsorption isotherms that were assigned to type II isotherms, which is typical for non-porous and/or macroporous materials (Fig. S9 †). The isotherms show pronounced hysteresis down to p/p 0 < 0.05, which may result from hampering adsorption and/or hindered desorption of nitrogen due to interactions between the adsorbate and the adsorbent. PXRD analysis and Raman spectroscopy revealed the formation of hexagonal GeO 2 for all oxidized materials. However, minor portions of tetragonal GeO 2 were also detected for Ox-1 and Ox-2 ( Fig. S12 and S14 †).
In conclusion, porous Ge@C composites and non-porous GeO 2 were obtained following the synthesis concept of TP (Scheme 1). The properties of the as-obtained Ge@C and GeO 2 materials is only marginally inuenced by the molecular structure of the precursors with the exception that bromine within the hybrid material (HM-3) results in larger loss of germanium by removal of all bromine during the synthesis of the Ge@C material (C-3). This results in higher carbon contents and larger BET surface area for Ge@C composites (C-3 vs. C-1/C-2). The nally obtained non-porous GeO 2 materials indicate that the proposed germanium oxide networks within the hybrid materials [eqn (1)] is not stable with respect to further conversion processes. However, the formation of the as-obtained microporous Ge@C composites containing germanium nanocrystals indicates that the hybrid materials consisting of phenolic-resin and germanium oxide as dominant germanium species exhibit indeed phase nanodomains.

Electrochemical measurements
In order to provide a proof of principle, the Ge@C materials C-1a and C-1b were tested as potential anode material for rechargeable Li-ion batteries. Electrodes were prepared by mixing the respective Ge@C sample with carbon black (CB) and carboxymethyl cellulose (CMC) with the mass ratio of 8 : 1 : 1 using a planetary ball-mill. Electrochemical tests were carried out in Li-ion half cells using 1 M LiPF 6 in a mixture (1 : 1 by weight) of ethylene carbonate (EC) and dimethyl carbonate (DMC) as electrolyte. Fluoroethylene carbonate (FEC) was used as electrolyte additive to improve cycling stability. 43 The results of the electrochemical measurements for C1-a and C1-b are shown in Fig. 7. Cells were cycled in the potential range of 0.005-1.0 V using a current of 346 mA g À1 or 1384 mA g À1 respectively.
For the rst discharge cycle high irreversible capacities are obtained due to the formation of the solid electrolyte interface and the reduction of surface oxides according to GeO 2 + 4Li + + Table 3 Crystallographic and experimental data of the single crystal X-ray diffraction analyses of (1) 3 ,2$(1) 4 $n-pentane, (2) 2 and (3) Largest diff. peak and hole 0.632 and À0.531 eÅ À3 0.301 and À0.265 eÅ À3 1.495 and À1.544 eÅ À3 1.108 and À0.649 eÅ À3 4e À / Ge + 2Li 2 O. However, during subsequent cycling in all cases the coulombic efficiency, the ratio between charge and discharge capacity, increased to $99%. For cycling at 346 mA g À1 composite C1-b delivers stable capacities of 540 mA h g À1 with no fading for at least 100 cycles. In contrast, composite C1-a only provides capacities of approximately 400 mA h g À1 . Even at a current of 1384 mA g À1 both samples of porous Ge@C material afforded stable capacities for 500 cycles with C1-b showing clearly better rate capability delivering a capacity of $390 mA h g À1 compared to $220 mA h g À1 for C1a. It is important to note that in all cases currents and capacities were related to the total mass of the Ge@C composite.
Considering that EDX-analysis yields a Ge-content of 46.5 AE 2.8% for C-1a and 39.8 AE 2.4% for C-1b the corresponding Ge-content in the electrodes can be estimated at 37% and 32% respectively. Relating currents and capacities only to this fraction of Ge in Ge@C, composite C1-b therefore reaches stable capacities of 1360 mA h g À1 for 100 cycles at a rate of 870 mA g À1 and 980 mA h g À1 for 500 cycles at a rate of 3480 mA g À1 . Despite the fact that direct comparison of electrochemical results has to be done carefully due to typically differences in electrode composition (e.g. Ge-content) or electrochemical cycling (e.g. current, potential range), we note that the performance of the herein presented Ge@C composite C1-b is indeed comparable with recent reports on Ge/C materials (Table S1 † for detailed  comparison). 37,48,50,86,87 These good electrochemical properties can be attributed to several factors. Due to nanoscale dimension of crystalline germanium in the Ge@C composites, and appropriate porosity, the massive volume changes occurring during lithiation/delithiation can be effectively accommodated, leading to excellent cycling stability. Further, due to the small size of germanium domains, intimate contact with carbon and porous structure of Ge@C, the ionic and electronic conductivity are enhanced allowing fast charging/discharging. In this regard, the signicantly superior rate capability obtained for composite C1-b compared to C1-a can be attributed to the smaller size of crystalline primary germanium particles [(13 AE 1) nm vs. (51 AE 6) nm, see Table 2] and therefore shorter diffusion distances. The fact that the capacity does not reach the theoretical maximum at ambient temperature of 1384 mA h g À1 (assuming formation of Li 15 Ge 4 ), might be attributed to the relatively high currents used and possibly fractions of germanium oxides as indicated by EDX measurements (Table S2 †), which leads to lower capacities. For comparison, whereas germanium has a theoretical capacity of 1384 mA h g À1 the reversible theoretical capacity of GeO 2 is 961 mA h g À1 . Fig. 7c and d show the This journal is © The Royal Society of Chemistry 2016 galvanostatic charge/discharge curves of composite C1-a and C1-b. In both cases the average delithiation potential is relatively low ($0.5 V) indicating high suitability of such Ge@C materials for future full cell applications as a high performance alternative to commercial graphite anodes.

Conclusions
The germanium(II) salicyl alcoholates, germanium(II) 2-(oxidomethyl)phenolate (1), germanium(II) 4-methyl-2-(oxidomethyl) phenolate (2) and germanium(II) 4-bromo-2-(oxidomethyl) phenolate (3), were synthesized by the reaction of Ge [N(SiMe 3 ) 2 ] 2 and the respective salicyl alcohols. Dimeric, trimeric and tetrameric structures of the germylenes were determined in the solid state and equilibriums between oligomers of these compounds were observed in solution. Twin polymerization of the compounds 1-3 that was thermally induced by heating up to 230 C gave germanium-containing phenolic resin hybrid materials. In contrast to spirocyclic silicon salicyl alcoholates, no inuence following the steric and electronic features of their substituents [R ¼ H(1), Me (2) and Br (3)] was observed on the process of thermally induced twin polymerization. Moreover, analyses of the as-obtained hybrid materials and bulk syntheses of Ge@C composites and GeO 2 support this observation. However, a signicant dependency on the solid form of the germylenes e.g., amorphous vs. crystalline, was observed for the process of thermally induced twin polymerization. Its initiation is hampered by a densely packed structure in the rst place but as the polymerization process is initiated, a pre-organized arrangement of the precursors facilitates the polymerization. Porous Ge@C composites and non-porous GeO 2 were obtained by either conversion of the as-obtained hybrid materials under reductive or oxidative conditions. Testing the obtained Ge@C composites as anode materials for rechargeable lithiumion batteries in order to provide a proof of principle on our concept revealed that excellent cyclability, rate capability and high coulombic efficiency were obtained, especially for the Ge@C composite possessing smaller crystalline germanium primary particle sizes [(13 AE 1) nm]. Despite the fact that the present cost for germanium compounds is possibly too high for implementing germanium-based anodes in commercial lithium-ion batteries, the herein presented concept of twin polymerization provides a general strategy to prepare highperformance electrode materials by embedding the active material efficiently into a conductive carbon matrix and holds considerable potential to be expanded to further systems in the future.

Experimental section
All reactions were performed under argon using Schlenk techniques or in a glovebox. Solvents were puried and dried by applying standard techniques. The reactions were carried out with freshly distilled, dried solvents. 1 H, 13 C{ 1 H} and 1 H 13 C{ 1 H} HSQC NMR spectra were recorded with a Bruker Avance III 500 spectrometer. Solid-state NMR spectra were measured at 9.4 T with a Bruker Avance 400 spectrometer equipped with doubletuned probes capable of magic angle spinning (MAS). 13 C{ 1 H} CP-MAS NMR spectra were measured at 100.6 MHz in 3.2 mm standard zirconium oxide rotors (BRUKER) spinning at 15 kHz. Cross polarization (CP) with a contact time of 3 ms was used to enhance sensitivity. The recycle delay was 5 s. The spectrum was referenced externally to tetramethylsilane (TMS) as well as to adamantane as secondary standard (38.48 ppm for 13 C). All spectra were collected with 1 H decoupling using a two-pulse phase modulation sequence. ATR-FT-IR spectra were recorded with a BioRad FTS-165 spectrometer. FT-IR were recorded with a Nicolet IR 200 spectrometer from Thermo Scientic (Thermo Fisher Scientic Inc., Waltham, MA, USA) in a KBr Matrix. Raman spectra were collected on a LabRam HR800 confocal mirco Raman system equipped with a helium-neon-laser (l ¼ 632.8 nm, P ¼ 3.08 mW) without usage of any lter (D 0 )a tftyfold magnication. Energy-dispersive X-ray spectroscopy (EDX) and scanning electron microscopy (SEM) were performed using a NovaNano SEM from FEI with the following parameters: pressure ($10 À5 mbar), work distance (5 to 7 mm) and acceleration voltage (18 kV) using a Si Dri Detector XFlash 3001 from Bruker AXS. Transmission electron micrographs were obtained by a 200 kV high resolution transmission electron microscope [HRTEM, CM 20 FEG, Co. Philips (FEI Europe, Europe NanoPort, Eindhoven, The Netherlands)] with imaging energy lter from Gatan (GIF). Melting points were determined with a melting point B-540 apparatus from Büchi. CHN analyses were determined using a FlashEA 1112 NC Analyzer from Thermo Fisher Scientic. DSC experiments were determined with a Mettler Toledo DSC 30 using 40 mL aluminum crucibles. The measurements were performed up to 400 C with a heating rate of 10 K min À1 in N 2 atmosphere and a volume ow of 50 mL min À1 . TGA/DSC experiments were determined with a Mettler Toledo TGA/DSC1 1600 system with a MX1 balance. The measurement was performed from 40 to 800 C with a heating rate of 10 K min À1 in Ar atmosphere and a volume ow of 60 mL min À1 . Nitrogen physisorption isotherms were obtained at À196 C using an Autosorb IQ2 apparatus from Quantachrome. All samples were activated in vacuum at 150 C for 3 h prior to the measurements. Specic surface areas were calculated applying the BET equation (p/p 0 ¼ 0.150 AE 0.002). The micropore content was estimated according to a QSDFT model (QSDFT: Quenched Solid Density Functional Theory, model for slit and cylindrical pores using the adsorption branch) for carbon samples using the Autosorb 1.56 soware from Quantachrome. [88][89][90][91][92][93][94][95] The specic micropore and total pore volume were also calculated by the above mentioned DFT models. For X-ray photoelectron spectroscopy (XPS) measurements, the powder samples were pressed on indium foil (99.99%, from MaTeck GmbH) that was cleaned with standard solvents in an ultrasonic bath. Measurements were performed in ultra-high vacuum (UHV) employing a hemispherical electron analyzer (PHOIBOS 150 from SPECS GmbH) and an Al-K a X-ray source (XR50M from SPECS GmbH) equipped with a monochromator (FOCUS from SPECS GmbH). The pass energy was set to 10 eV for acquisition of the core-level and Auger electron spectra. The estimated error in binding energies is AE0.2 eV. No charge compensation was applied. Powder X-ray diffraction (PXRD) patterns were collected using a STOE-STADI Pdiffractometer from STOE with Cu-K a radiation (40 kV, 40 mA) and a Ge(111) monochromator. The crystallite size was estimated using the Scherrer equation: s ¼ Kl/b cos q, where s is the volume weighted crystallite size, K is the Scherrer constant here taken as 1.0, l is the X-ray wavelength, q is the Bragg angle in and b is the full width of the diffraction line at half of the maximum intensity (FWHM, background subtracted). The FWHM is corrected for instrumental broadening using a LaB 6 standard (SRM 660) purchased from NIST (National Institute of Standards and Technology). The value of b was corrected from (b measured 2 and b instrument 2 are the FWHMs of measured and standard proles): Germanium(IV) chloride and 1,1,1,3,3,3-hexamethyldisilazane were purchased from ABCR GmbH & Co KG. 2.5 M n-butyllithium and 5-bromo-2-hydroxybenzaldehyde were purchased from Merck Schuchardt OHG (Hohenbrunn). Single crystals suitable for X-ray diffraction analysis were obtained (i) directly from the ltrate of the reaction mixture aer ltration of the precipitate to give 2$(1) 4 $n-pentane, (ii) by evaporation of the solvent at ambient temperature of a saturated solution of 1 in diethyl ether and (iii) by cooling a saturated solution of (1) 4 in diethyl ether to À18 C to give (1) 3 , respectively.

Electrode fabrication, cell assembly and electrochemical measurements
Ge@C samples, carbon black (CB, Super C65, TIMCAL), and carboxymethyl cellulose (CMC, Grade: 2200, Daicel Fine Chem Ltd.) were mixed in the ratio 8 : 1 : 1 with deionized water using a Fritsch Pulverisette 7 classic planetary mill (1 hour at 500 rpm). The resulting aqueous slurries were coated onto Cu foil (9 mm, MTI Corporation) and subsequently dried at 80 C for 12 hours under vacuum. Electrochemical measurements were conducted in air tight coin-type cells assembled in an Ar-lled glove box (O 2 < 0.1 ppm, H 2 O < 0.1 ppm) using elemental lithium as both counter and reference electrode. 1 M LiPF 6 in a 1 : 1 mixture by wt of ethylene carbonate (EC) and dimethyl carbonate (DMC) (Merck, battery grade) with 3% uoroethylene carbonate (FEC, Hisunny Chemical Co., battery grade) served as electrolyte. As separator glass ber (GF/D, Whatman) was used. Galvanostatic cycling tests were carried out at ambient temperature on MPG2 multi-channel workstation (BioLogic).