Skip to content

THE COURSE / M2 PURE SILICON / M2.7

The metals you can't see

After this lesson you can:

  • Explain what a trace of metal does to the chips built on a wafer, and where in the plant it comes from.
  • Follow the four reads of the quality lab: the surface by a scanned droplet, the layer beneath by vapour etch, the slurry by TXRF, the air by impinger.
  • Explain how the ICP-MS counts metals to parts per quadrillion, and why it needs high resolution to see iron.

Loading the lesson…

Sources

  1. SUMCO, JP2012174964A, analysing metals in a silicon wafer's surface layer (read 9 November 2026). Trace metals on the surface and in the surface layer, where trenches, sources and drains are formed, affect devices: leakage failures, oxide breakdown voltage and lifetime. In heat treatment they diffuse in and form precipitates, dislocations and oxidation-induced stacking faults. The layer is etched in vapour of hydrogen fluoride, nitric acid and nitrogen oxides, about 11 µm in an hour; the wafer heated on a hot plate, its surface at 180 °C or more (180 to 250 °C; 200 °C for 10 minutes in the example), until the white smoke stops; above 250 °C bulk copper may rise to the surface. Then hydrophobised in HF vapour, and about 200 µL of 10% HF and 10% hydrogen peroxide scanned over it, diluted to 1,000 µL, for a double-focusing ICP-MS, Thermo Fisher's ELEMENT 2. Etch depths 3 to 30 µm for device wafers, 3 to 10 µm for epitaxial layers.
  2. Shin-Etsu Handotai, JP2017083334A, recovering metal impurities from a wafer's surface (read 9 November 2026). Metal impurities are one of the causes of poorer device characteristics, so they are recovered from the surface for analysis by AAS, ICP-OES or ICP-MS. A PTFE probe with a slit a little shorter than the wafer's radius, about 0.2 mm above the surface, on NAS Giken's SC2000; 2% HCl and 5% H₂O₂ from Tama Chemicals' TAMAPURE-AA-100, 2,190 µL for a 300 mm wafer; the wafer turned under it so the liquid sweeps the whole surface, about 10 minutes for a 200 mm wafer against about 100 with the older probe, then collected in a vial.
  3. Shin-Etsu Handotai, JP2023133001A, metals in polishing slurry by TXRF (read 9 November 2026). Metal contamination in processing degrades the wafer, so the slurry's metals must be kept low: below 1.7×10¹¹ atoms/cm³ of secondary contamination, devices are unaffected. HF dissolves the colloidal silica; a drop forms a film on a wafer; UV light, 185 nm lamps for 10 to 15 minutes, decomposes the organics; dried at 70 to 100 °C or under 150 Torr; read by total reflection X-ray fluorescence on Technos's TREX630T. Without the UV the film peeled and warped. Compared: microwave wet ashing (PerkinElmer, 1,000 W) and ICP-MS (ELEMENT 2). Elements read: K, Ca, Ti, Cr, Fe, Ni, Cu, Zn, Ce and more.
  4. Shin-Etsu Handotai, JP2022077397A, sampling metals in cleanroom air (read 9 November 2026). Fab chemicals give off gases that corrode equipment and spread metal impurities through the clean room, lowering yield through poor withstand voltage and leakage. Air is drawn at 1 L/min for 60 minutes through a collection tank of acid at pH 0 to 4 (10 mL of nitric acid in a quartz impinger), a neutralising tank (potassium hydroxide, cooled to 20 °C) and an absorbing tank of ultrapure water; the liquid goes in a washed PFA sample cup to the ELEMENT2. Elements: Li, Na, Mg, Al, K, Ca, Ti, Cr, Fe, Ni, Cu, Zn, Mo.
  5. Thermo Fisher Scientific, ELEMENT 2 & ELEMENT XR, high resolution ICP-MS (brochure, read 9 November 2026). Spectral interferences are the main limitation of ICP-MS: the argon plasma, water, acid and the sample combine into ions with the same nominal mass as an analyte. With high mass resolution, iron at m/z 56 is separated from ⁴⁰Ar¹⁶O. Resolutions 300, 4,000 and 10,000; detection power under 1 ppq for non-interfered nuclides; the detector linear over nine orders, from ppq to ppm. Over 500 installed, Micron Technology among them. Thermo Fisher Scientific (Bremen) GmbH.