Ultra-High Sensitivity Surface Photovoltage Measurement of Heavy Metal Contamination in Silicon Wafers with Fast Metal Identification

Abstract

The Surface Photovoltage (SPV) minority carrier diffusion length method is the most sensitive and widely used technique for monitoring heavy metal contamination in silicon wafers starting from crystal growth through various IC processing steps. Relentless IC fabline purity refinement, exhibited by a Fe contamination decrease of about 10 times per decade has been matched by corresponding progress in SPV detection capabilities. An iron contamination limit of mid 107 atoms/cm3 is projected for new IC fablines by the year 2020. For SPV this represents a detection goal for a new generation of metrology. In this paper we discuss this newest version of the SPV technique. The key element is the improvement of accuracy and repeatability of the minority carrier diffusion length determination, demonstrated by the coefficient of variance of 0.01% that is an order of magnitude better than the capability of the previous digital SPV version. This translates directly to an order of magnitude better sensitivity in SPV metal concentration measurements quantified by the change in diffusion length (L) caused by activation of the metal recombination activity. Results are presented for some of the purest state-of-the-art 300mm wafers. They illustrate an ultra-high sensitivity procedure for whole wafer mapping of Fe that includes for the first time a quick Fe identification test. This test is done with very short transient diffusion length measurements made possible by the enhanced precision of the L measurements. The test takes minutes compared to hours or days used in previous SPV measurements. Previous digital SPV metrology incorporated concentration measurements of Fe and Cu. SPV versions for photovoltaics include measurement of light-induced degradation related to boron-oxygen defects. The next generation digital SPV extends this measurement to include other interstitial metals Co, Mn and Cr that at room temperature form stable pairs with substitutional boron acceptors. Identification of the metals and separation of their individual contribution is based on metal-specific features of dissociation and association kinetics of the pairing reaction, Mi+Bs ↔ (Mi Bs), involving interstitial metal donors Mi and substitutional boron acceptors, Bs. Cu is differentiated from the other metals due to different irreversible reactions that change the minority carrier diffusion length. A distinctive feature of the SPV metrology is the use of photo-dissociation of pairs, effective for Fe and Co, but not for Mn and Cr. For the latter two metals, thermal dissociation is used. Thermal dissociation is effective for all pairs, therefore in multi-metal measurement cycles it is used as a post photo-dissociation step. Accordingly, the diffusion length measurement in the initial state, Linit, then after photo-dissociation, LPD, and then after thermal-dissociation, LTD, is used in the example illustrating the case of the simultaneous presence of Fe and Mn. After dissociation, a transient diffusion length measurement is performed which determines the association (pairing) time constant, τp. The value of τp is inversely proportional to metal diffusivity and it increases in the order Co→Fe→Mn→Cr. The digital SPV tool incorporates variable temperature wafer stages facilitating optimization of thermal treatment used in monitoring of dissociation or association kinetics. It is believed that these refinements ensure the latest generation of digital SPV tools will meet the demands of Fe contamination monitoring in IC fablines well beyond the 2020 projection while also enabling for the first time fast heavy metal identification testing.