Large quantities of African dust are carried across the Atlantic to the Caribbean Basin and southern United States where it plays an important role in the biogeochemistry of soils and waters and in air quality. Dusts' elemental and isotopic composition was comprehensively characterized in Barbados during the summers of 2013 and 2014, the season of maximum dust transport. Although total suspended insoluble particulate matter (TSIP) mass concentrations varied significantly daily and between the two summers, the abundances (μg element/g TSIP) of 50 elements during “high‐dust days” (HDD) were similar. Aerosols were regularly enriched in Na, Cu, Zn, As, Se, Mo, Cd, Sn, Sb, and W relative to the upper continental crust. Enrichment of these elements, many of which are anthropogenically emitted, was significantly reduced during HDD, attributed to mixing and dilution with desert dust over source regions. Generally, Ti/Al, Si/Al, Ca/Al, Ti/Fe, Si/Fe, and Ca/Fe ratios during HDD differed from their respective values in hypothesized North African source regions. Nd isotope composition was relatively invariant for “low‐dust days” (LDD) and HDD. In contrast, HDD‐aerosols were more radiogenic exhibiting higher 87Sr/86Sr, 206Pb/204Pb, 207Pb/204Pb, and 208Pb/204Pb ratios compared to LDD. Generally, Barbados aerosols' composition ranged within narrow limits and was much more homogeneous than that of hypothesized African source soils. Our results suggest that summertime Barbados aerosols are dominated by a mixture of particles originating from sources in the Sahara‐Sahel regions. The Bodélé Depression, long suspected as a major source, appears to be an insignificant contributor of summertime western Atlantic dust.

Every summer, large-scale dust plumes emerge from the west coast of North Africa, cross the Atlantic, enter the Caribbean Basin, and on occasion pass over the southern United States. We show that such dust events significantly increase the concentration and alter the chemical composition of ambient fine particulate matter (PM 2.5 ) in southern Texas. Between 1 August and 13 September 2014, PM 2.5 was sampled daily at Clinton Drive, an industrial location in Houston, and at Galveston located on the Gulf of Mexico coast that is not directly/significantly impacted by local and regional pollution sources. Between 17-25 August, average PM 2.5 concentrations nearly doubled when a large dust-laden African air mass moved through the region. Using ground-based measurements, satellite observations, and modeling tools, we demonstrate that this massive influx of mineral dust greatly reduced the relative abundance of pollutants in Houston where aerosols are otherwise dominated by local and regional emissions. Chemical mass balance modeling estimated that African dust contributed 19-48% of PM 2.5 during the 9-day dust episode. Moreover, African dust was present throughout the measurement period even on days outside the main dust event contributing an estimated~8% of PM 2.5 which suggests that it may be a normal summertime component in this region. These measurements agree closely with inferences of "fine dust" based on a generic soil composition in the IMPROVE (National Interagency Monitoring of Protected Visual Environments) network. Our detailed protocol supports the conclusion that African dust affects a wide region of the South, Southeast, and East Coast United States.

Various metallization schemes are used to provide conducting layers in IC fabrication. X-ray fluorescence spectrometry has been utilized to examine the chlorine and fluorine levels in some of these metallization schemes after patterning and etching. The metallization systems investigated include A1-Si and A1-Si-Cu, with and without TiW. The corrosion characteristics of these metallizations are found to be predominantly a function of chlorine levels. The galvanic cell effects of the Cu and TiW with A1 are not significant within the range of etch processes investigated. The effect of different in situ passivations as well as various postetch treatments are presented. The residual chlorine levels at various stages of the pattern definition process are also presented.As device geometries continue to shrink, the use of A1-Si-Cu combined with TiW is fast becoming the metallization system of choice to meet required reliability and electromigration specifications. The pattern definition of A1-Si-Cu alloys is currently done by reactive ion etching or plasma etching in chlorine containing plasmas. One of the primary concerns with this process is postetch corrosion.Postetch corrosion is believed to be due to high levels of AIC13 on the metal. AIC13 is hygroscopic and in a humid environment will react with atmospheric moisture to form HC1. This is a self-perpetuating reaction as the HC1 product can further attack A1 and produce more A1CI~, with the rate depending on the moisture content and the residual C1 level (1, 2). Chloride ions have also been reported to accelerate intergranular or pitting corrosion in bulk metals (3,4).The introduction of Cu or TiW to the A1-Si metallization system has been reported to enhance the corrosion of A1 (5-7). The suggested mechanisms are the increase in C1 levels due to the presence of the TiW or Cu and/or the galvanic cell effects of TiW or Cu with A1. It has also been suggested that the C1 and moisture levels needed to induce corrosion are much lower when Cu is present (1).In the past Auger analysis (6) has been used for the analysis of chlorine. In this paper, wavelength dispersive x-ray fluorescence spectrometry (XRF) was used to determine the chlorine and fluorine levels. XRF is an established analytical technique (8) which is used in various industries requiring sensitivities of 1014 atoms/cm 2. XRF provides the advantage of nondestructive inline measurements, with sensitivities for C1 and F in the 1013 atoms/cm 2 range (if on surface) and 10 TM atoms/cm 2 (if within the layer) for the metallization thickness considered.This paper discusses the effects of chlorine levels on corrosion for an Al-l%Si-2%Cu on 10%Ti-90%W metallization scheme used in a double metal submicron CMOS process. It also correlates the magnitude of the galvanic cell corrosive effects of both the TiW and the Cu. Finally, it presents an etch passivation strategy to minimize corrosion and suggests an allowable range of chlorine levels which can be tolerated on the wafer.

No abstract