**4. Discussion**

From the analysis of color and reflected light of five di fferent particulate matters according to di fferent liquid additives, chromaticity coordinate values in the diagram were e ffective by measuring the shift amount in the vector under the refractive-index-liquid-added case to identify the non-white color group particulate matters including yellowish pine tree pollen, soil dust and grey household dust from the white color group particulate matters. Soil dust under refractive index liquid and cellophane filters revealed the largest movement in coordinates. Talc and gypsum powder belonging to the intrinsic white color group were identified by comparing the peak at around the reflected light spectrum. For the light spectrum cases, peaks of approximately 420 nm can be regarded as the guideline peak to determine the influence of the filter and additives according to the type of sample. Relative peak intensities of the other peaks were compared with a ratio of the peak intensity at 420 nm. The peak at approximately 720 nm in wavelength was solely observed for talc powder. Water and refractive index liquid could selectively shift the peak at approximately 420 nm and 679 nm to 433 and 690 nm, respectively, only for gypsum powder. More distinct dependency on filter color was observed for gypsum powder rather than household dust, pollen, soil and even similar talc powder. This was in good agreemen<sup>t</sup> with the method called the "browning index" used in other color and light characterization methods using a spectrophotometer [18].

In this study, we were concerned with indoor particulate matters, which possibly can be generated by daily activities such as cooking, vacuum cleaning and installed construction materials, or transported from outdoors by daily activity. It was our intention to find the feasibility of an optical approach to assist in the identification of the source of particulate matters. In cases where a large portion of particulate matters are observed to inflow from outdoor sources, people can be air showered at the front door and contaminants are pre-removed before transporting these into the indoors. In other cases, more indoor sources are found to contribute to indoor air contamination, and mechanical ventilation can be operated with a priority to outflow the indoor air to maintain the air quality level. Our study can at least contribute to this decision making as the first stage at a pre-screening level, but results showed that it has limitations in distinguishing the similar intrinsic color of powders. Light intensity was also vulnerable to the light bulb's source (>6500 k) and power. In addition, other particulate matters such as organic compounds and black carbon, generated from combustion, should be studied to assist in supporting these methods. In future studies, other chemical additives should be tested to reveal di fferent colors after their reaction with white particulate matters. Another approach can be

attempted to combine the presorting method by using a camera according to its color and shape to improve the identification rate by a simple method.
