化学蚀刻的硅：IV蚀刻技术

The etching of silicon in HNOg-HF based systems proceeds by a sequen-tial oxidation-followed-by-dissolution process. In those composition regionswhere the solution is very low in HNO3 and rich in HF, the rate-limitingprocess is the oxidation step. Consequently, electron concentration, surfaceorientation, crystal defects, and catalysis by lower oxides of nitrogen play animportant role. In those compositions where HF is in lmited supply, dis-solution of the formed oxide is the rate-controlling step and diffusion of thecomplexing fluoride species is the important factor. Therefore crystal ori-entation and conductivity type independence as well as hydrodynamic controlare the consequences. In order to meaningfully select an etching compositionto solve a specific processing problem, it is necessary to understand this com-position-mechanism interaction, Corollary with the mechanism understandingsample geometry effects have been followed as a function of solution com-position. The HF-HNO-H,O solution composition plane has been characterizedinto various regions where the two basic mechanisms interact and specifcprocesing utilization is shown. Similar results are shown for the system HF-HNO-HC,H;O2. In addition, a number of particular etching problems areposed, and solutions oflered, that make use of these composition characteriza-tions, and show how one can use their information to solve other practicalprocessing problems.

Most of the data on silicon etching that one canfind in the literature involve studies of the mechanismsand kinetics of the dissolution process (1-4). Thereare a few examples of practical applications, but theydeal primarily with crystallographic aspects of theetching,eg, defect delineation (5) or anisotropiccrystal plane etching (6). Because of the sophisticationof present-day silicon technology,it appears to bedesirable to be able to control the geometrical aspectsof a silicon slice, e.g, from sharp, possibly peakedcorners and edges to smooth and rounded edges andcorners, merely by controlling the chemical etchingenvironments.It would, therefore, seem appropriateto present some data on the geometrical effects ob-served on the etching of silicon rectangular parallel-epipeds in solutions of HNOs,HFand HO with andwithout HCHO2.These observations were made atthe same time as the previously published kineticsdata were being obtained (1,2), but it is only recentlythat many requests forgeometric information havemade us aware of theusefulness of disseminatingthese configurational data.In order to make the picture more complete, someof the original iso-etch-rate fgures will also be in-cluded, for comparison purposes.

The technique used for the etching and thicknessmonitoring has been adequately described in previouspublications (1-3).However,in order to make thispaper useful, we will have to repeat some of theexperimental detail, but will stress only the salientfeatures of the sample and solution preparationmethods and how the geometric aspects were followed.

All dice were etched three times, and each etch wasperformed in a fresh portion of the same solution.Since each etching period had been chosen to remove4 to 6 mils from the specimen, it may be assumed thatthe work damage had been removed after the firstetch. The third etch was performed with a few milli-grams of NaNOz added to the solution for possiblecatalytic purposes (1). After each etching step,thesample was rinsed in distilled water, dried,and mea-sured for dimensional changes with a micrometer to士 0.05 mils,and then the geometry of the specimenwas examined under a microscope.

Fig1

The etching was done in a Teflon beaker 1 in. indiameter and 2 in. deep. Agitation was provided byan electric stirrer equipped with a polyethylene paddle.The dice were etched one at a time in 10 cm3 ofsolution,and duplicate results were obtained by re-peating the entire experiment. To quench the reactionat the proper time, a large volume of water was pouredinto the beaker. The etching technique was later mod-ifed when the higher concentration reagents wereused. Here the dice were etched two at a timein 20cm³ of solution, while encased in a small perforatedpolyethylene basket. The basket and dice were agitatedfor the required time in the solution, then immersedin a beaker of cold water to quench the reaction, Theagitating action consisted of approximately 150 ver-tical strokes per minute, with the etching jig rotatedbetween the thum and middle finger as rapidly aspossible during the up and down strokes.