稀释HF清洗过程中硅表面颗粒沉积的机理

The mechanism of particle deposition onto silicon wafer suraces during dilute HF cleans is discussed. Direct surface forcemeasurement using an atomic force microscone showed that partice redeposition on a slicon surace is due to the dominant vander Waals interaction between the particle and the wafer surface. The addition of surfactants can affect the clean effectiveness ofa dilute HF solution by changing the surface interaction forces between a particle and a wafer.We show that there is a simplecorrelation between particle deposition and the product of the zeta-potentials of the wafer and particle. This correlation can becxplained by introducing the lincar superposition approximation to the derivaton of clectrical double-layer interactions. Iheaddition of surfactant will also decrease dispersion attraction,introduce steric repulsion, and eliminate adhesion force, as indicatedby the results of surface force measurements.

With the semiconductor industry moving to smaller and smallerdimensions, removing particulate contamination has become extremely critical.' Silicon wafer ultra-cleaning is thus a vital area insemiconductor processing. The most widely used cleaning method isthe RCA-based wet chemical clean, which removes organic resi-dues, particles, and metal ions on wafer surfaces. An RCA-cleanedwafer has a chemical oxide layer about 1-2 nm in thickness on thesurface. The dilute HF clean is employed to remove this layer and toobtain a hydrogen-passivated bare silicon surface, which has highstability against oxidation34 with low density of surface states.However during HF cleaning, particles in the bath tend to redepositonto the bare silicon surface.6,7

The mechanism of bare silicon wafer contamination by particleshas been discussed previously6-8 It was postulated that the electrostatic interaction between particles and wafers played a key role inparticle deposition/removal. To simulate a 0.5% HF environmentexperiments have been conducted at pH 3.3 using dilute HCl.6,However, the simulated pH value may be higher than the actualvalue. It was found that the zeta-potential of bare silicon at pH 3.3was negative (about -23 mV) while most particles were positive atthis pH. Therefore, the susceptibility of the bare silicon wafer tocontamination by particles was explained based on the attractiveelectrical interaction between the silicon wafer and particles. Inmore general discussions on particle deposition Spielman et al.' andRiley et al.10 considered the deposition as diffusion of particles fromthe bulk solution to the surface under a potential caused by theelectrical double-layer interaction and van der Waals attraction.More recently, Vos et al. postulated that a repulsive hydration forceexists in the interaction between a silicon surface and a silicon nitride particle.

In this paper the mechanism of dilute HF cleaning is studiedusing an atomic force microscope (AFM). To our knowledge this isthe first instance to quantitatively elucidate the mechanism for DHFcleans by this method. Direct measurements on forces between thesilicon surface and the silicon nitride surface in different solutionswere conducted. These measurements provide detailed informationon the repulsive or attractive forces between the particles and thewafer surface. The experimental results are explained with modifiedDLVO theory (Derjaguin-Landau-Verway-Overbeek).12,13

Experimental

All ofthe experiments were conducted in room temperature. The0.5% HF solutions for each experiment were freshly prepared froman electronic grade 49% HF solution (Ashland Chemical). In these experiments, silicon nitride particles were chosen as model particlesThe zeta potentials of the silicon nitride particles in 0.5 wt % HFsolutions containing 200 ppm silicon nitride particles (Alfa Aesar)with 0.3 um mean diameter were measured using a BrookhavenZeta Plus. For accuracy, each zeta potential measurement was con-ducted ten times. The zeta potential of silicon was determined bymeasuring 5 um silicon powders (obtained from Atlantic EquipmentEngineers) dispersed in solutions at the same conditions as siliconnitride particles. The pH was measured using a HF resistant pHprobe from Fisher Scientific. Surface force measurements were con-ducted on a Digital Instruments SPM Nanoscope Ill. The liquid celland the silicon nitride tips were all obtained from Digital Instru-ments. For the AFM measurements, the 2 in. n-type silicon(100)Czochralski (Cz) wafers with 1-10  cm resistivity (Silicon QuestInternational) were cut into pieces ( 1.3 X 1.3 cm2 in size). Thesesamples were subjected to a SC-1 clean to remove particulate con-taminations and then a deionized water (DIW) rinse. followed by a10 min dip in 0.5% HF solution to remove the chemical oxide on thesurface. The measurement of silicon-silicon nitride tip interaction inair was conducted right after the sample was taken out from diluteHF solution and blow dried with nitrogen. The silicon nitride tip wasdipped into a 0.5 wt % HF solution for 12 h to remove the goldcoating then DIW rinsed before mounting on the liquid cell. Thesolutions for the measurement of interaction forces in liquid mediawere prepared with DIW and the pHs were adjusted with HCl to1.87-1.90. The solutions also contained 0.1% HF in order to preventoxide formation on the silicon surface.

The particle deposition experiments were conducted as followsSilicon wafers were cleaned in SC-1 solution before being dippedinto 05 wt % HF solutions containing 200 ppm silicon nitride par-ticles (Alfa Aesar) with a 0.3 um mean diameter and 1 wt % surfactant for 10 min Different cationic, anionic, amphoteric, or nonionic surfactants were added, respectively, into dilute HF solution tochange the surface charges. After a DI water rinse and nitrogen blowdry the wafers were analyzed using a Jeol 6400 scanning electronmicroscope (SEM). At least four random positions were examinedon each sample, and 60 and 6000 times magnification photographswere taken at each position. The surface particles were counted sta-tistically from the photographs.