盐酸等溶液中GaAs和InP的选择性湿蚀

Experimental  

Materials used for the etching studies were Fe-doped InP  and Cr-doped GaAs substrates, and GaInP grown on GaAs  substrates by organometallic vapor phase epitaxy  (OMVPE). The GaInP layers were 500- nm thick, silicon  doped (n = 1 x 1017 cm-3), and nearly lattice-matched to  GaAs (nominal composition of Ga0.~2In0.48P. The samples  were patterned photolithographically with 100 ~m squares  using AZ5214 photoresist in image-reversal mode. When  adequately hard-baked, the resist served as an etch mask  and resisted attack well in most etchants. Etchant solutions  were prepared by mixing standard solutions of HC1 [36  weight percent (w/o)], H2Q (30 w/o), and glacial acetic acid  (CH3COOH 99.9%) in various proportions. Typically, the  HC1 and CH~COOH were premixed, and the H2Q was  added seconds before emersing the samples in the solution.  The samples were typically etched at 20~ for 1 min without stirring or agitation, and were subsequently quenched  in running deionized water and blown dry with nitrogen.  The etch depth was measured using a Sloan Dektak  profilometer, and the smoothness of the etched surface was  evaluated by optical and scanning electron microscopy.

Results and Discussion  

The effect of an oxidizing agent on the etching characteristics was examined by varying the amount of H202 in a mixture of 1 HCI:20 CH3COOH:xH2Q, with 0 -< x --< 5. The  etch rates in Fig. 1 represent the first minute after H2Q is  added to the mixture. When no oxidizing agent is present,  both GaInP and InP are etched selectively with respect to  GaAs. The etch rates are approximately 4000, 700, and less  than 50 A/min for InR GaInP, and GaAs, respectively. 

Fig1

To evaluate further the usefulness of these etchants for a  gate recess process, the stability of a mixture of 1 HCI:40  CH3COOH:I H2Q was evaluated by measuring the change  in etch rate with the age of the solution. Fig. 5 shows that  the etch rate for all materials increases strongly over the  first 40 rain after mixing, levels out, and then slowly decays.  Initially, the relative etch rates vary as InP > GaInP > GaAs.  However, within 20 rain all materials etch at comparable  rates s,howing only slight selectivity with relative etch rates  GaAs > InP > GaInP. The surface morphology remained  relatively smooth with increasing age of solution for all  three materials.

The change in the relative etch rates suggests that a  change in the dominant reaction mechanism occurs over  time. One can gain insight into the changing chemical  mechanisms of this system by considering the thermodynamics of chloride oxidation by H202. In aqueous solutions  Cl- is oxidized by H~O2, and the reactions listed in Table I  can occur. The standard Gibb's free energy change, AG ~ for  each reaction has been calculated from standard electrochemical potentials. 13 When considering the value of  AG~ mole of H202 consumed, the reaction producing  Cl2(g) is the most favorable. Other reactions which result in  chemical species containing chlorine in higher oxidation  states are also highly exothermic.  

An idea] contact does not exhibit any interfaces. In fact,  practical contacts differ from that ideal behavior mainly  due to (i) differences in the work functions of metal and  semiconductor yielding potential barriers, (it) surface  states of the semiconductor, and (iii) poor adhesion.