Multiband superconductivity and a deep gap minimum from the specific heat in ${\mathrm{KCa}}_2{\left({\mathrm{Fe}}_{1-x}{\mathrm{Ni}}_x\right)}_4{\mathrm{As}}_4{F}_2$ $(x=0, 0.05, 0.13)$

Specific heat can explore low-energy quasiparticle excitations of superconductors, so it is a powerful tool for bulk measurement on the superconducting gap structure and pairing symmetry. Here, we report an in-depth investigation on the specific heat of the multiband superconductors ${\mathrm{KCa}}_2{\left({\mathrm{Fe}}_{1-x}{\mathrm{Ni}}_x\right)}_4{\mathrm{As}}_4{F}_2$ $(x=0, 0.05, 0.13)$ single crystals and the overdoped nonsuperconducting one with $x=0.17$. Clear specific heat anomalies can be observed at the superconducting transition temperature of 33.6 K and 28.8 K for the samples with $x=0$ and $x=0.05$, respectively. For the two samples, the magnetic-field-induced specific heat coefficient $\Delta \gamma \left(H\right)$ in the low-temperature limit increases rapidly below 2 T, then it rises slowly above 2 T. Using the nonsuperconducting sample with $x=0.17$ as a reference, the specific heat of phonon background for various superconducting samples can be obtained and subtracted, which allows us to extract the electronic specific heat of the superconducting samples. Through comparative analyses, it is found that the energy gap structure including two $s$-wave gaps and an extended $s$-wave gap with large anisotropy can reasonably describe the electronic specific heat data. According to these results, we suggest that at least one anisotropic superconducting gap with a deep gap minimum should exist in this multiband system. With the doping of Ni, the superconducting transition temperature of the sample decreases along with the decrease of the large $s$-wave gap, but the extended $s$-wave gap increases due to the enlarged electron pockets via adding more electrons. Despite these changes, the general properties of the gap structure remain unchanged when doping Ni. In addition, the calculation of condensation energy of the parent and doped samples shows the rough consistency with the correlation of $U_0\propto {T_c}^n$ with $n=3–4$, which is beyond the understanding of the BCS theory.