Highly precise neutron nuclear data is required in nuclear transmutation research of long life minor actinides (MA) in nuclear waste. It has been difficult to measure neutron-induced reaction cross sections of MAs due to large background of the decay $\gamma$-rays from the radioactive samples. In recent years, with the advent of spallation neutron sources, the qualities of cross section measurements of MAs were significantly improved. The Japanese Spallation Neutron Source (JSNS) in the Japan Proton Accelerator Research Complex (J-PARC) was started in operation in 2008. In order to utilize a high-intensity pulsed neutron beam from JSNS for nuclear data measurement, the Accurate Neutron Nucleus Reaction Measurement Instrument (ANNRI) was built and has been used for the past ten years.

In neutron capture cross section measurement, the number of the incident neutrons is necessary to derive the neutron capture cross section. To normalize the detected $\gamma$ -ray yield to the number of the incident neutrons, the neutron count is usually monitored by detecting the incident neutrons with a neutron detector. However, in measurement with ANNRI, neutron monitoring detection has not been employed and, instead, the number of proton beam pulses injected into the spallation target has been used based on the assumption that the number of proton beam pulses is proportional to the number of incident neutrons. This assumption is mostly plausible but could fail when the conditions of the proton accelerator or the neutron source change. To avoid possible failure of the proton pulse counting method and make measurement with ANNRI more robust, an additional neutron beam monitor is under development.

To develop a neutron beam monitor for ANNRI, there are two issues to overcome. First, very high intensity neutron beam from JSNS requires a fast detector system that can process signals at a high counting rate. Second, $\gamma$ -flash, an intense $\gamma$ -ray burst produced when the proton beam pulse bombards the spallation target, can paralyze a detector. Thus, $\gamma$ -ray sensibility of the neutron monitor should be low. In order to fulfill the requirements, a thin sheet-type plastic scintillator combined with thin $^{6}$Li layer on a Mylar film is adopted for the present neutron monitor. The incident neutrons react with $^{6}$Li and the $^{6}$Li(n,t)$^{4}$He reaction occurs. The emitted particles, tritons and alphas, are detected with the plastic scintillator. The short ranges of tritons and alphas allow for using a thin plastic scintillator film, and the thin detector leads to low $\gamma$ -ray sensibility. Another requirement for fast detection is achieved by the fast response property of plastic scintillator. Simulation studies using Monte Carlo simulation code PHITS was performed to optimize the detector design, especially thickness of the $^{6}$Li layer. A test detector system was built to study the feasibility. Test experiments were carried out at ANNRI. Preliminary results will be given in this contribution.