Arsenic speciation: limitations with direct hydride analysis