InP湿法化学蚀刻-去除氧化物

The wet chemical etching of In and its native oxide has been studied in HCl and H SO solution to create oxide-fre surfacesThe (100) InP surface is not etched in <2 M HCl and <6 M HSO4. As the vetch <0.1 nm/min for these concentrations, thcnative oxide ater OFOD treatment can be electively removed without signilcanty etching the surtace, as contirmed by contactangle measurements, ellipsometry and X-ray photoelectron spectroscopy. STM and AFM measurements showed that after OFODtreatment and subsequent oxide removal very smooth surfaces are achieved due to the creation of atomic teraces. The size of thescerraces can be increased up to micron-size in 6 M HSO., Furthermore, it was shown that in presence of oxygen, n-type InP isphotoetched/oxidized dunng wet processing. Last, the anisotropy in etching is discussed for lnP in 2 M HCl and 2 M HSO.

For the past and current technology node, transistors in integratedcircuits are fabricated on silicon substrates, and in some cases, SiGealloys are used. In order to meet the future requirements imposed bythe scaling roadmap, the next generation of transistors (i.e. the 7 nmnode and beyond) will be based on high mobility Ill-V compoundsemiconductors.12 To be cost effective, the Ill-V materials will beintegrated on Si substrates. In part A of this paper, we show a possibleintegration route for such a technology; a Quantum Well based Ill-Vtransistor using InP buffer and InGaAs channel layers.The manufacture of transistors consists of many different pro-cessing steps, which involve multiple exposures to chemical solu-tions. The goal of these wet treatments is to obtain a well-definedsurface, (i.e. free of contaminants and stoichiometric), essentialfor obtaining good interfacial properties and, consequently, deviceperformance.

There are different ways of preparing Ill- V surfaces for subsequent processing steps: ion sputtering followed by high temperatureannealing, atomic hydrogen cleaning, sulfur passivation, and wet-chemical cleaning. Wet-chemical cleaning offers an effective andpractical cleaning method for semiconductor surfaces In the caseof Ill-V compound semiconductors special care should be taken dueto the possible formation of highly toxic hydrides during low pHprocessing. Additionally, anisotropy in etching may be expected,7.8which leads to significant surface roughening. In part A of this pa-per, we propose an oxide formation/oxide dissolution (OFOD) modelthat prevents hydride formation and allows for smooth etching.3 In afirst step the semiconductor is oxidized by a strong oxidizing agent insolution and, subsequently, the (hydr)oxides formed are dissolved bythe acid. Due to the presence of a strong oxidizing agent in solutionno dissolved hydrides are expected. This approach, was developed forcleaning of(100) oriented InP buffer layers after a chemo-mechanical-polishing (CMP) step, and can, in principle, be used for other Ill-Vsemiconductors. After OFOD treatment a native oxide is present. Inorder to prepare the surface for subsequent epitaxial growth of theInGaAs channel layer a subsequent oxide removal step is essential.A systematic study on oxide removal was performed and describedin this work. The focus is on HCl and H,SO4 solution due to favoreddissolution of the Ill component at low pH.’ The etch rates of InPwere studied at the (sub) nm/min range by using inductively coupledplasma mass spectrometry allowing to determine optimum concentra-tion ranges for which the native oxide can be selectively removed (i.e.without etching/roughening the InP surface), as confirmed by ellipsometry, contact angle measurements and -ray photoelectron spectroscopy (XPS).Scanning tunneling microscopy (STM) and atomicforce microscopy (AFM) was used to study the surface morphology after OFOD treatment followed by an oxide removal step. Further more, the inflfluence of conductivity type on the chemical etching and reoxidation of (100) InP surfaces is presented. We conclude with a consideration of the impact of crystal orientation on the etch rate (vetch), especially important for Fin-FET processing.

Experimental

Single side polished bulk 2" InP wafers were obtained from AXTInc. Both the n-type (S doped) and p-type (Zn doped) samples had acarrier concentration of (1E17 cm-3 ). The semi-insulating (SI) typewafers were Fe doped (1E7 cm-3 ). All experiments were performedat room temperature in a cleanroom environment and chemicals ofultra pure quality (Sigma Aldrich) were used.The etching experiments are performed in a dedicated Polyvinylidene fluoride (PVDF) cell. A Kalrez O-ring was used to expose14.52 cm' of the polished surface to the etching solution, preventing the defective back side and edges from contributing to the etchrate. After exposing the wafers to 100 mL of etchant for 15-30 minutes, the etch rate was determined by measuring the total amount ofdissolved 113In and 115In with Inductively Coupled Plasma - MassSpectrometry (ICP-MS, Agilent 7500 cs). This technique enables detection of substrate loss at sub-monolayer level. The accuracy of theetch rate is about 10% under these conditions.The wetting properties were analyzed using a DataPhysics OCA

230L Contact Angle System.A Plasmos SD2000 ellipsometer was used to measure the nativeoxide thickness. The measurements were performed at fixed photonenergy (1.96 eV). The angle of incidence was 70° and the spot size10 pm.

XPS measurements were performed closed coupled. After treatment the samples were rinsed for 3 minutes in ultra pure water (UPW)After N, blow-drying the samples were transported in an N, atmo-sphere to the XPS set-up. The total air exposure was minimized tothe order of a few minutes only. The measurements were carried outa Theta300 system from Thermolnstruments with an exit angle of78 degree as measured from the normal of the sample. The measure.ments were performed using a monochromatized Al Ka X-ray source

(1486.6 eV) and a spot size of 400 microns.Scanning tunneling microscopy was performed in an Omicronlarge sample SPM system with a base pressure of 5 x 10-11 mbarThe depicted images were acquired at a bias of 2 V with currents inthe 100 pA range. The sample preparation procedure was identical tothat used for the  PS measurements.