GaN has been emerging as the next technological semiconductor after silicon and has already secured its superiority in the niche of power electronics with a rapidly growing interest since the advent of electric vehicles. To enable GaN to concur more microelectronic applications, one major barrier has yet to be surpassed. Rather limited control has so far been achieved over the density of GaN interface states. To achieve it, native and process-induced surface states need to be fully identified. Photoluminescence (PL) has been the most commonly used method in the study of GaN defects, and it typically shows a wide sub-bandgap peak centered at 2.2 eV that has been called the yellow luminescence (YL) band. Despite 30 years of studies, the debate as to whether this is a bulk or surface defect has not been settled. Present deep level characterization methods have not provided clear enough evidence. Theoretical studies attempted to associate this luminescence with various bulk defects, e.g., Ga-vacancies, O or O-C complexes,[1] while various optical spectroscopy studies, point to a surface origin.[2]Here, we report results of using an optical spectroscopy method based on surface photovoltage before and after a mild anneal in vacuum. The method is capable of showing the energy distribution of charge density trapped in surface states and thus of monitoring changes thereof. Using the method in air on untreated GaN reveals two peaks over the same ranges of the yellow and blue peaks observed by PL. After a mild anneal at 170 C for 24 hrs in vacuum, the “yellow” charge density peak was absent. Exposure to air was then observed to fully restore the “yellow” charge density peak. This observation clearly points to the involvement of an airborne molecule, a constituent of air, as the source for charge in the yellow-luminescence-related state. If this deep level were a bulk defect, it would not be able to interact with air. This observed interaction with air thus provides a first direct evidence that the yellow luminescence-related defect is a surface state. Experiments of controlled exposure to the various air constituents to identify the related airborne molecule are currently ongoing. The same method is now used in a search for passivation method for this surface state.The method we use is based on the following model. Illumination using sub-band gap photon energies excites electrons from the surface states over the surface potential barrier, consequently, reducing the surface band bending, and this change is observed as photovoltage (Fig. 1). We have already shown how this photovoltage can be used for calculation of the surface charge distribution in GaN HEMT.[3] In the case of bulk crystal, the model is slightly different, and a quantitative relation may be obtained between the measured photovoltage and the surface state charge distribution. Quantitative calculation requires knowledge of the surface band bending in equilibrium, which may not always be reliably obtained. However, when the spectrum is acquired using low photon flux, assuring the change in the band bending is small compared with the equilibrium band bending the surface charge density is practically relative to the derivative of the photovoltage, and thus, qualitative distribution may be obtained.[4]Figure 2 exemplifies the use of the model presented here to obtain a qualitative charge density spectrum. It shows surface charge density obtained using the model from the photovoltage spectrum shown as inset.This method is then used in Fig. 3 to compare three spectra of the same sample: In air (red), in vacuum (green) - the yellow peak is slightly reduced, and after heating up to 450 K for 24 h in vacuum (blue) - the yellow peak is practically gone.These results suggest that the charge in the yellow-luminescence-related state is contributed by a non-inert constituent of air, and this interaction may not be possible unless this state is a surface state. In the talk, preliminary results on the chemical identity of this air constituent will be presented.[1] Van de Walle et al., Microscopic origins of surface states on nitride surfaces, J. Appl. Phys. 101, 081704 (2007).[2] Shalish et al., Yellow luminescence and related deep levels in unintentionally doped GaN films, Phys. Rev. B 59, 9748 (1999).[3] Turkulets et el., Surface states in AlGaN/GaN high electron mobility transistors: Quantitative energetic profiles and dynamics of the surface Fermi level, Appl. Phys. Lett. 115, 023502 (2019).[4] Turkulets et al. The GaN(0001) yellow-luminescence-related surface state and its interaction with air. Surfaces and Interfaces 38, 102834 (2023). Figure 1