Improving the performance of headspace sampling using trap-based preconcentration