Low‐Temperature Sintering of Low‐Loss Millimeter‐Wave Dielectric Ceramics Based on Li‐Kosmochlor, LiCrSi2O6

Ceramic dielectrics with particularly low levels of dielectric loss and low permittivity are of growing interest toward millimeter‐wave applications. Based on composition and structural considerations, lithic kosmochlor is expected to meet these requirements while exhibiting low densification temperatures, allowing its integration also in co‐fired circuits. Herein, it is found that the spark plasma sintering of LiCrSi2O6 ceramics facilitates densification at temperatures more than 200 °C lower than conventionally processed materials in only a fraction of the time. A spark plasma sintering duration of 10 min is necessary in order to achieve a controllable process with a repeatable high‐performance product material exhibiting a fine‐grained microstructure. Characterization millimeter‐wave frequencies reveal that LiCrSi2O6 materials produced in this manner exhibit excellent dielectric properties (Qf = 80 700 GHz and εr = 7.5 at 134.24 GHz) following densification at only 950 °C with a relative density of 97.4%. These results represent an unprecedented combination of low dielectric losses and processing temperatures for ceramics processed without the addition of sintering aids. The materials and methods explored here present a promising pathway toward high‐performance millimeter‐wave and co‐firable systems.


Introduction
Earth's crust exhibits a silicic nature and is dominated by silicate minerals, which are composed of assemblies of SiO 4 tetrahedra. As a result of their terrestrial abundance, silicates have formed the basis for many widely used structural ceramics and are even proposed toward use on other planets, e.g., Mars. [1] More recently, diverse silicate ceramics of controlled structure and composition are increasingly being studied in terms of their functional properties. Here, noteworthy advancements have taken place in the use of silicates for biomedical, [2] battery electrode, [3] or membrane applications. [4] Toward rapidly emerging applications utilizing radiation in the millimeter-wave (mmW) region of the electromagnetic spectrum, including 5G communication and automotive radar, silicates have been identified as attractive candidate dielectric materials. This is chiefly due to their low levels of permittivity (i.e., dielectric constant ε r ) and achievable low levels of dielectric loss. [5] The requirements for dielectric materials for components at mmW frequencies, 30-300 GHz, differ somewhat compared to lower radiofrequencies. In the MHz and low GHz range, medium to high ε r values in the range 60-140 are commonly necessary to keep resonating component sizes sufficiently small. [6] However, using higher permittivity materials typically incurs higher dielectric losses and lower signal propagation speeds. At mmW frequencies, shorter wavelengths mean that lower permittivity dielectrics are generally more suitable, with a significant priority being given to minimization of dielectric losses, which become more problematic at higher frequencies. In this context, silicate ceramics have shown considerable promise as they can combine low permittivity and low dielectric loss as well as great substitutional diversity that allows for the tuning of those properties.
Among silicates, the pyroxene family, also known as inosilicates, are structurally distinguished by chains of [SiO 4 ] tetrahedra, connected via their apices, which extend indefinitely along the crystallographic c axis. Pyroxene-type materials possess the general form M2M1Si 2 O 6 with M2 and M1 having the summed oxidation state of þ4. Favorable high-frequency dielectric properties have been reported for pyroxenes including CaMgSi 2 O 6 , Mg 2 Si 2 O 6 , and Mg 1.5 Co 0.5 Si 2 O 6 . [7][8][9] One notable obstacle toward future implementation of these materials as dielectrics in mmW devices is their rather high sintering temperature, commonly of DOI: 10.1002/pssa.202200685 Ceramic dielectrics with particularly low levels of dielectric loss and low permittivity are of growing interest toward millimeter-wave applications. Based on composition and structural considerations, lithic kosmochlor is expected to meet these requirements while exhibiting low densification temperatures, allowing its integration also in co-fired circuits. Herein, it is found that the spark plasma sintering of LiCrSi 2 O 6 ceramics facilitates densification at temperatures more than 200°C lower than conventionally processed materials in only a fraction of the time. A spark plasma sintering duration of 10 min is necessary in order to achieve a controllable process with a repeatable high-performance product material exhibiting a fine-grained microstructure. Characterization millimeterwave frequencies reveal that LiCrSi 2 O 6 materials produced in this manner exhibit excellent dielectric properties (Q f = 80 700 GHz and ε r = 7.5 at 134.24 GHz) following densification at only 950°C with a relative density of 97.4%. These results represent an unprecedented combination of low dielectric losses and processing temperatures for ceramics processed without the addition of sintering aids. The materials and methods explored here present a promising pathway toward high-performance millimeter-wave and co-firable systems.
1300°C or higher, making their production expensive and limiting their utilization in the rapidly growing field of low-temperature co-firable ceramics (LTCC). Although sintering additives are routinely and successfully employed to decrease the processing temperatures of ceramics, their use tends to impair the relevant dielectric properties immensely. [10,11] Based on the literature summarized in ref. [5], the identification of pyroxene ceramics combining low permittivity with low dielectric loss that can be densified at low temperatures without the addition of additives poses an application-driven approach toward the development of new ceramic dielectrics for diverse mmW elements.
From a perspective of low processing temperature, due to the strong fluxing effect of alkali M2 site occupancy, lithic silicates are a rational choice, and there are reports of single phase materials of this type with good high-frequency dielectric properties, sintered at comparably low temperatures. [12,13] Surprisingly, to our knowledge there have yet been no in-depth studies evaluating the sintering and microwave dielectric behavior of lithic or other alkali pyroxene ceramics.
The permittivity of a ceramic can be interpreted in terms of cation polarizability. Because of the low dielectric polarizabilites of both Li þ and Cr 3þ ions, as indicated in Figure 1, [14] the compound LiCrSi 2 O 6 (Li-kosmochlor) is expected to present a very low dielectric permittivity, suitable for implementation in the rapidly growing field of mmW dielectrics. Although the compound's interesting high pressure behavior and multiferroic character have been discussed, [15] its room temperature high-frequency dielectric properties remain unexplored. The magnetoelectric effect observed in LiCrSi 2 O 6 and other alkali pyroxenes, whereby the permittivity can be adjusted by the application of a magnetic field, further motivate the study of this system toward emerging mmW applications. In this study, we make use of recent advancements acquired for the synthesis of silicate ceramics [16] to produce phase pure LiCrSi 2 O 6 ceramic powders. Although we expect this material to densify in diverse geometries by conventional sintering at relatively low temperatures (compared to most other silicate ceramics), we employ spark plasma sintering (SPS) here as a promising approach toward the implementation of silicates in mmW dielectric and low-temperature technologies such as LTCCs. The choice of this field-assisted sintering technique stems from several aspects: 1) the method has proven to produce very dense ceramic specimens at lower sintering temperatures and durations, saving time and consuming only 20%-33% of the energy compared to conventional sintering; [17] 2) fast heating rates (up to several hundred°C min À1 ) with SPS suppress particle coarsening by bypassing the nondensifying mechanism of surface diffusion and thus enhance densification while avoiding potentially problematic abnormal grain growth; [18] and 3) it has been reported that SPS can lead to cleaner grain boundaries (e.g., direct grain boundaries and less amorphous or impurity phase layers) and thus improved functional properties in ceramics. [19,20] Although the feasibility of SPS for producing dense ceramics has been demonstrated manifold in the literature, its effect on crystalline silicate ceramics is so far rather unexplored. Based on an earlier study on steatite ceramics, in which the sintering temperature could be reduced by 200°C by employing SPS, [21] this study aims at achieving a similar improvement in terms of processing temperature reduction, with relevance toward LTCC dielectrics. The samples prepared here via SPS are compared to specimens obtained from conventional furnace sintering.

Synthesis
Ceramic powders with a nominal composition of LiCrSi 2 O 6 were produced using a modified chelation-polymerization-based solgel approach developed by Pechini, [22] a method in which steric entrapment of cations allows for the homogenous mixing of elements in product materials, thus avoiding phase segregation. [23] This method has proven to be very reliable in the production of diverse phase pure ceramic silicate powders with well-controlled stoichiometry, [16] which is often difficult by standard solid-state routes. Due to the high amounts of LiCrSi 2 O 6 powders required for this study, the synthesis method described in ref. [16] has been scaled up here for several batches.
A total metal ions:citric acid:ethylene glycol molar ratio of 1:6:6 was used for the syntheses in this work. Initially, SiC 8 H 20 O 4 (tetraethylorthosilicate, Carl Roth, Germany, 99%) was added dropwise into a stirred aqueous citric acid solution (pH = 3). When the solution was completely clear, the respective metal precursors of LiNO 3 (Carl Roth, Germany, 99.5%), Cr(NO 3 ) 3 ·9H 2 O (Carl Roth, Germany, 98%), NaNO 3 (Carl Roth, Germany, 99%), and Al(NO 3 ) 3 ·9H 2 O (Merck, Germany, 98.5%) were added. A pH value of 3 was adjusted for optimal binding of citrate to metal ions and to avoid precipitation of metal hydroxides. [24] The mixture was then converted into a covalent polymer network by the addition of ethylene glycol and the incipient transesterification reaction between citric acid and dropwise added ethylene glycol (Merck, Germany, 99%) upon heating to %80°C. Continued heating reduced the solution into a thick resin, which was heated with 3°C min À1 to 400°C and held at this temperature for 6 h in a muffle furnace. The resulting black powder was ground in an agate mortar, heated with a rate of 3°C min À1 in ambient atmosphere to 1000°C, and then calcined for 3 h at this temperature. The calcination Figure 1. Dielectric polarizabilities of uni-and tricharged cations, relevant for alkali pyroxenes, based on the data in ref. [14].
www.advancedsciencenews.com www.pss-a.com temperature of 1000°C was chosen based on initial tests explained in Section 3. The calcined powders were then ball-milled in polyoxymethylene jars with 12 mm ZrO 2 balls and deinonized water to facilitate mechanical activation of surfaces. Subsequently, slurries were dried in air at 80°C to obtain the ceramic powder product.

Conventional Sintering
For the conventional sintering of materials, 4.9 g of calcined powder was mixed thoroughly with 15 drops of a 10% polyvinyl alcohol solution and hydraulically pressed in a 30 mm steel pressing die. The powder was precompacted with a pressure of 5 MPa for 3 min. After carefully releasing this pressure, a second run was performed at 10 MPa for 3 min. The product pellets were sintered at different peak temperatures within the furnace sintering range of 1175-1275°C for LiCrSi 2 O 6 , as determined via hot-stage microscopy (HSM). All samples were heated in a muffle furnace in air at 3°C min À1 to 600°C and held at this temperature for 6 h to allow for binder removal. Subsequently, the pellets where heated at 3°C min À1 to the peak temperature and sintered for 6 h. Sintering temperatures of 1175, 1200, 1225, 1250, and 1275°C, respectively, were chosen. Cooling of the pellets was conducted with a rate of 10°C min À1 .

SPS
Densification of powders by SPS (Dr. Sinter Lab Jr. 211Lx, Fuji Electronic Co., Japan) was carried out under vacuum, as is most commonly done in SPS. Here, 2.4 g quantities of calcined materials were sintered in a 20 mm cylindrical graphite die having a 20 mm inner diameter. Before heating, a uniaxial pressure of 40 MPa was applied slowly for initial compaction; this pressure was then maintained during the sintering and densification process. After heating, the die was passively cooled. The punches were kept at constant pressure throughout the cooling and were only slowly released below 100°C to avoid pressure relief-induced cracking. At the same time, the chamber was vented with ambient air. Residual carbon on sample surfaces was removed mechanically by grinding. The following parameters and combinations thereof have been employed for our experiments: 1) peak temperatures: 900, 950, 1000, 1050, and 1100°C; 2) heating rates: 10 and 100°C min À1 ; 3) Holding times: 1 and 10 min. The peak temperatures have been chosen based on initial HSM results (Figure 2b) that are discussed in Section 3.1, and considering that optimal sintering temperatures for a material are significantly lower in SPS compared to conventional furnace sintering. Sample names are composed as follows: peak temperature [°C]heating rate (°C min À1 )holding time (min), e.g., the sample 950-10-10 was heated to 950°C with 10°C min À1 and sintered at this temperature for 10 min.

Characterization
Phase analysis was carried out with a D8 Advance X-ray diffractometer (Bruker, USA), in Bragg-Brentano geometry, equipped with a LynxEye 1D detector and Cu radiation (λ = 1.5406 Å).
Microstructural investigations were conducted using a Gemini Leo 1530 scanning electron microscope (Carl Zeiss Microscopy GmbH, Germany) with acceleration voltage ranging from 3 to 10 keV. In order to reveal grain boundaries, pellets were polished and thermally etched at 100°C below the respective sintering temperature for 30 min, with a heating rate of 5°C min À1 . Average grain sizes have been calculated using the ImageJ software. [25] Inductively coupled plasma optical emission spectroscopy (ICP-OES) was used to determine the elemental concentration of Li, Cr, and Si in the calcined samples with an ICP Ultima2 (Horiba, Japan). To achieve this, the sample powder was digested in an autoclave at 200°C for 5 h with 2 mL HF and 5 mL HNO 3 .
To inform the design of sintering experiments, side-view hotstage microscopy (Hesse Instruments, Germany) was performed on calcined powder samples. The ground ceramic material was spring pressure hand pressed with 1.5 N mm À2 into cylinder with 3 mm width and 3 mm height. The measurements were performed in a tube kiln in air, with a heating rate of 10°C min À1 up to 1400°C with 30 min holding time. This measurement was carried out twice, giving similar results. Details of the method can be found elsewhere. [26] Density and open porosity of materials were determined by Archimedes' method in which samples are immersed in water and the immersed weight, sample volume, and mass are used to determine the density, according to ASTM C-373-18. [27] The open porosity values were used to correct the measured relative permittivity values. This was done using the method by Bruggeman. [28] Thermogravimetric (TG) and differential thermal analysis (DTA) data were obtained from heating in a mixture of synthetic air atmosphere (80% Ar and N 2 , 20% O 2 ) with a STA 449F3 (Netzsch, Germany). The measurements have been performed on the synthesized powder after its pyrolysis step at 400°C. The black powder was heated from 40 to 1200°C with a heating rate of 10°C min À1 .
The dielectric properties of the samples were measured in a mmW corrugated waveguide transmission line setup. [29] A vector network analyzer (VNA) of type Rohde&Schwarz (Germany) ZNA67 was used in conjunction with two ZC110 frequency extension modules to address the frequency range from 75 to 110 GHz (W-Band), and two ZC170 modules for the range 110-170 GHz, respectively. The waveguide-based testbed itself is provided by SwissTo12 (Switzerland). Using the VNA, the reflection and transmission properties of the material samples are measured and error-corrected using a Through-Reflect-Match (TRM) calibration technique. [30] After calibration, two well-known and documented low-loss materials are measured and their dielectric properties are determined: polytetrafluorethylen (PTFE, ε r = 2.01) and HR-Si (ε r = 11.9). From the frequency domain reflection and transmission behavior, the dielectric properties are calculated using the Baker-Jarvis-Method. [31] 3. Results

Thermal Analysis and Calcination
Thermogravimetry reveals that the pyrolyzed synthesis product exhibits a mass loss of 34.8% upon heating up to 700°C (Figure 2a), which can be attributed to the loss of organics and decomposition of nitrates and the formation of NO x and N 2 , typical for Pechini-derived ceramics. [16] Bulk crystallization www.advancedsciencenews.com www.pss-a.com is finished by 900°C, as no further exothermic events are observed after this temperature.
Although it was demonstrated that reaction sintering, often involving combined calcination and sintering steps, can yield well-performing silicate dielectrics, [32] we deliberately chose to avoid this additional parameter here in order to gain more control over the sintering process in this study. Hence, it was our goal to obtain a single-phase starting material after calcination. Therefore, calcination temperatures in the range 500-1200°C were tested (Figure 3). The first XRD reflections of the target phase LiCrSi 2 O 6 can be observed after calcination at 600°C. Crystalline secondary phases after heat treatment at 900°C are, in order of their appearance, Cr 2 O 3 , LiCrO 2 , SiO 2 (cristobalite), Li 2 Si 2 O 5 , and SiO 2 (Quartz). According to Izquierdo and West, all these phases can coexist with LiCrSi 2 O 6 under different conditions. [33] No more reflections of those phases can be observed after calcination at 1000°C, indicating the reaction of all secondary materials to form LiCrSi 2 O 6 , and thus the optimal calcination temperature. Reflections of cristobalite and Cr 2 O 3 reappear at calcination temperatures above 1000°C, possibly as the result of volatilization of Li and destabilization of LiCrSi 2 O 6 . The elemental ratio (obtained by means of ICP-OES) of Li:Cr:Si of the calcination product, after treatment at 1000°C, was revealed to be roughly 1:1:2, which is in agreement with the observation of a single phase by XRD measurements at this temperature.
HSM was performed to gauge sintering behavior under conventional furnace conditions and inform experiment design (Figure 2b). Here, the temperature range in which the sample body's area decreases while the form factor roughly remains stable is taken as the optimal sintering range (1175-1275°C). Based on these results and assuming that sintering and densification temperatures will be significantly lower in SPS, we chose the temperature range of 900-1100°C for SPS.

Conventional Sintering
The phase assemblages in conventionally sintered pellets are indicated in the diffraction patterns given in Figure 4a. All samples sintered below 1250°C contain only LiCrSi 2 O 6 , while at higher temperatures some secondary phases formed (Table 1), which can be attributed to the volatilization of Li and decomposition of LiCrSi 2 O 6 , as described in Section 3.1. The microstructural evolution with sintering temperature is shown in Figure 4c   www.advancedsciencenews.com www.pss-a.com and then decreases again with higher sintering temperatures. The trend for lower permittivity at higher porosity is often reported [8,34,35] and cannot exclusively be related to higher porosity (air) because all the reported values are corrected for porosity of the samples. Hence, it is not completely clear what the mechanism behind this observation is. The sharp decrease of the ε r values at high temperatures can only be partly associated with the formation of SiO 2 side phases (ε r = 4). The excessive presence of Cr 2 O 3 (ε r = 12) should in theory increase the permittivity values. The loss tangent values for the samples decrease from 1175°C, reaching a minimum of 0.0021 (correlating to a quality factor of 35 100 GHz) at 1225°C and then increase again. The optimum at 1225°C can be explained by the combination of rather low porosity and the absence of secondary phases, both of which typically deteriorate the dielectric loss characteristics of ceramics. [5] The effect of phase impurities becomes clear for the samples sintered at 1250 and 1275°C in which the loss tangent values increase again. For the 1275°C sample abnormal grain growth as well as partial liquid phase formation can be observed, which further deteriorate the dielectric loss values (Figure 4g).

SPS
A total of 20 SPS samples, as combinations of the three experimental parameters of peak temperature, heating rate, and dwelling time, have been prepared for this study. An overview containing all parameters and sample properties is given in Table 1.
In Figure 5, typical progresses of the SPS experiments are indicated by the processing data of the samples 950-10-10 and 1000-100-10. The beginning of the sintering process is marked by a steep displacement in the z direction of the SPS crosshead, indicating sample compaction. It can be seen that the sintering starts roughly at the same temperature (815-820°C), independent of the heating rate. This corresponds to a reduction of the sintering temperature of more than 200°C compared to the conventional furnace sintering temperature (Figure 2b).
The phase assemblages of the sintered pellets, examined by means of XRD, are depicted in Figure 6. The target material, LiCrSi 2 O 6 , is the dominant phase in all produced sample pellets. Minor amounts of a secondary LiCrO 2 phase are present in all samples sintered at a peak temperature of 950°C and rapidly heated and briefly dwelled samples 1050-100-1 and 1100-100-1. On the other hand, trace amounts of Cr 2 O 3 are present in the samples 1050-100-10 and 1100-100-10 that were rapidly heated and dwelled for 10 min. All samples sintered at peak temperatures 900 and 1000°C are phase pure LiCrSi 2 O 6 .

Effect of Temperature and Heating Profile
Micrographs of polished and thermally etched samples heated with the profile 100-10 are exemplarily shown in Figure 7a-e to demonstrate the effect of peak temperature on the final microstructure. The relative density values increase monotonously with increasing sintering temperatures: 61.1% (900°C) ! 88.0% www.advancedsciencenews.com www.pss-a.com Table 1. Overview of the properties of the samples produced for this study (conventional and SPS). The sample names indicate the experimental parameters used: peak temperature for the conventionally prepared ones and peak temperature (°C)-heating rate (°C min À1 )-holding time (min), e.g., the sample 950-10-10 was heated to 950°C with 10°C min À1 and sintered at this temperature for 10 min. The main and secondary phases tabulated here are all crystalline. Small amounts of X-ray amorphous quantities are possible and expected especially for the samples sintered above 1000°C (as will be discussed later).  ( Figure 7f,g), whereby the increased holding time leads to an increased density of the sample (from 65.2% to 98.6%). The trend of increased grain size and densities with higher sintering temperatures and holding times in SPS experiments is expected and has been reported previously for other silicate materials. [21] Interestingly, these authors reported the formation of a significant amount of glassy phase during the low-temperature SPS, leading to a mechanical reinforcement of their samples. In the present study, we do not find evidence for substantial amounts of amorphous phases after SPS in any of our samples, while still being mechanically stable and denser. We can therefore not report on any benefits of the formation of glassy phases, contrary to what was reported earlier. The heating rate (10 and 100°C min À1 ) in our study does not have a profound effect on the microstructure and densities of the samples produced. This is in contrast to what was reported in the landmark study by Shen et al. [36] It has to be taken into account though, that only at very high heating rates exceeding 350°C min À1 , significant differences to low heating rates were observed. The reason for this could be that very rapid heating may minimize surface diffusion during heating in SPS, as proposed by other authors. [37] We can therefore presume, that for ceramics in general, that as long as the heating rates in SPS experiments are moderate, the effect on microstructure and density will be minor. From the dielectric property measurements (Table 1) it can be observed that the permittivity and the loss tangent values fluctuate over a rather broad regime for the samples sintered for only 1 min which does not allow the identification of a trend here. The reason for this is most probably an insufficient sintering process, resulting in porous microstructures, as described above. The greatest runaway values in terms of dielectric loss (tan δ) are those that exhibit very high porosity values  www.advancedsciencenews.com www.pss-a.com (namely, 950-100-1, 1100-100-1, 900-10-1, and 1050-10-1). This behavior is understandable because high porosity values are a known detrimental factor for dielectric loss in ceramics. [5] Samples sintered for 10 min appear to be producible with better control of permittivity and loss characteristics. With the exception of sample 1000-10-10 (which can be considered an outlier, as its porosity is abnormally high compared to the other samples in Figure 8), two clear trends are identifiable. First, relative permittivity values are all very close to the theoretical value for LiCrSi 2 O 6 (%7), as assessed via the method proposed by Roberts. [38] Small deviations could be caused by closed porosity or small impurity phases, both of which we cannot correct for. Second, for both heating rates (100-10 and 100-10) loss tangent values decrease from 900°C and then increase again toward higher temperatures. An optimum is either reached at 1000 or 950°C with tan δ = 0.0017. When comparing these results to the conventionally sintered samples, we can see that for the best performing samples the grain sizes are similarly small and their standard deviations are low. This observation indicates that in order to achieve good dielectric loss values, small and ideally uniform grain size distributions should be targeted.

Discussion
Silicate ceramics are of growing interest as mmW dielectrics, and have been the subject of numerous recent studies, aimed at gauging their applicability toward low-loss components. In the first study of its kind, we examine here the high-frequency dielectric performance of a lithic pyroxene, namely, Li-kosmochlor. The selection of this material was based on its expected densification and polarization behavior, which should be amenable to the obtainment of a co-firable low-permittivity ceramic, materials which are of growing value in the design of high-frequency systems. In particular low-loss LTCCs are expected to play an important role in certain emerging 5G technologies including multiple input multiple output (MIMO) antennae, substrate integrated waveguides, filters, and substrates. We show here that LiCrSi 2 O 6 does indeed exhibit the expected level of permittivity while sintering at temperatures that are among the lowest for silicate ceramics, natural consequences of the selected material composition. Through field-assisted sintering, densification can be achieved at yet lower temperature, alongside particularly low levels of dielectric loss not generally found for such low-temperature sinterable ceramics.
Specifically, LiCrSi 2 O 6 samples prepared via the conventional sintering route show decent loss characteristics (Q f = 35 000 GHz for the sample sintered at 1225°C). However, several mechanisms limit the performance of traditionally fabricated dielectric ceramics of LiCrSi 2 O 6 compared to SPS samples. 1) Relative density values are significantly lower; 2) the peritectic decomposition of LiCrSi 2 O 6 leads to the formation of secondary phases in samples processed at high sintering temperatures, which do not revert to the equilibrium composition upon cooling. Although the peritectic was reported to be at 1283 AE 8°C (Figure 9), the decomposition in our samples is already observable at 1250°C (Figure 4a). The decomposition is most probably also accelerated due to volatilization of Li at these temperatures. 3) Above the eutectic point at 1032°C (Figure 9), very small amounts of melt will form that probably leads to a very small amorphous layer at grain boundaries upon cooling. All of those three factors discussed above are known to be detrimental to the dielectric loss of ceramics and explain the reason for the comparably worse performance of the conventionally prepared LiCrSi 2 O 6 ceramics.
By employing the SPS technique, the dielectric loss characteristics can be improved (Q f = 80 700 GHz for sample 950-10-10) while simultaneously lowering the sintering temperature by more than 200°C, a greater reduction than that found in previous studies with steatite silicates. [20] The sintering temperatures for the best performing samples were well below the eutectic and Figure 8. a) Permittivity and b) tan δ plotted against relative density values for all samples (SPS in green squares and conventional red circles) produced in this study. Two clear trends can be observed: (a) with increasing relative density permittivities reach closer to the theoretical value of %7, due to closed porosity that cannot be corrected for. And (b) dielectric loss decreases with improved densification. The marked outliers are all sintered for only one minute showing that the SPS process is hard to control with this short sintering duration. The outlier for the conventionally sintered sample 1275 is not predominantly LiCrSi 2 O 6 anymore.
www.advancedsciencenews.com www.pss-a.com peritectic temperatures discussed above, allowing improved phase purity. Compared to other silicate ceramics, and other inosilicates in particular, the results of the present study stand unique ( Figure 9). While peak values for quality factors in other single-phase inosilicate ceramics such as CaMgSi 2 O 6 and Mg 2 Si 2 O 6 can reach up to 120 000 GHz, [7,8] their required processing temperature is in the range of 1300-1380°C, which is up to 400°C more than the required temperature in this study. A survey of reported performance data for high-frequency dielectric ceramics shows that the materials fabricated here exhibit the best dielectric loss performance of any known single-phase ceramic processed at such low temperatures. The relative permittivity values for dense and single-phase samples formed here were typically very close to the calculated value of %7, demonstrating the usefulness of the theoretical assessment of permittivity values based on ionic polarizabilties that was set forth by Roberts. [38] The greatest deviations from the theoretical values can be observed for SPS samples sintered for only 1 min. Here, the values are typically lower, indicating that a residual closed porosity remains in the samples, which is not measurable by Archimedes' method and therefore remains unconsidered in the correction of the relative permittivity values. This further reinforces the observation that very short sintering durations should be avoided for the production of controllable high quality dielectrics via SPS.
SPS processing leads here to samples with relative density values greater than 97%. From the few lithic silicate materials that have been studied toward high-frequency dielectrics, none of those sintered by conventional solid state can reach this level of densification. [12,39,40aÀc] Considering that inclusions in the form of porosity are typically detrimental to dielectric performance of ceramics, there is a great opportunity of improving density and thus performance values of the lithic silicate materials in the literature that have so far only been produced by standard procedures.
The results obtained here reveal new possibilities for the production of high-frequency dielectric materials. So far the only single-phase silicate materials with comparably low processing temperatures are BaCo 2 Si 2 O 7 (1020°C [41] ), Ba 0.8 Sr 0.2 CuSi 2 O 6 (1000°C [42] ), and Li 2 Ca 0.96 Mg 0.04 SiO 4 (925°C, [39] Figure 10). However, with Q f values between 4200 and 58 000 GHz, none of those materials perform as well as LiCrSi 2 O 6 prepared here with SPS. The processing applied here motivates further investigations for the production of LTCC materials with ultrahigh-quality factors. By sacrificing some of the excellent loss characteristics by the addition of sintering aids in combination with SPS could potentially even allow for the co-firing with cheap base metal electrodes such as Ni or Cu. [43] Another potential avenue for optimizing the SPS process for dielectric silicate ceramics could be the omission of the calcination step, forcing reaction sintering during SPS. Although this has so far only been tested with conventionally prepared silicate ceramics, the results were promising. [32]

Conclusion
Li-kosmochlor was found to perform remarkably well as a lowloss, low-temperature sinterable dielectric ceramic. Performance levels in terms of dielectric loss at high frequencies were shown to be better than any known single-phase system that can be processed below 1000°C.
SPS is shown to be an effective approach toward enhancing the low-temperature densification characteristics of these silicate dielectric ceramics for mmW applications. Among silicate ceramics, which are in general promising candidates for highfrequency dielectrics, Li-kosmochlor is of particular promise as it offers both low-temperature densification alongside low levels of dielectric loss. The processing of this compound via a solgel approach based on the Pechini method was effective in producing near phase pure materials which were readily densified at  www.advancedsciencenews.com www.pss-a.com low temperatures via SPS. A minimum sintering duration of 10 min is found to be necessary toward achieving consistent dielectric behavior. The use of SPS is found to reduce required sintering temperatures by more than 200°C and avoid phase segregation in the LiO 2 -SiO 2 -Cr 2 O 2 ternary system. While no significant influence of the heating rates of 10 and 100°C min À1 could be found, moderate peak temperatures of 950 and 1000°C and a sintering duration of 10 min were identified to be optimal. The LiCrSi 2 O 6 sample heated with 10°C min À1 to 950°C and held for 10 min showed the best performance with tan δ = 0.0017 (Q f = 80 700 GHz) and ε r = 7.5. This combination of ultralow loss and extremely low processing temperatures means a great advance in the pursuit of cheap and high performing high-frequency dielectrics.