Fabrication of Semiconductors by Wet Chemical Etch
Total Page:16
File Type:pdf, Size:1020Kb
THE JOURNAL OF UNDERGRADUATE RESEARCH University of Kansas | Summer 2008 Fabrication of Semiconductors by Wet Chemical Etch Selective Etching of GaAs Over InGaP in Dilute H2SO4:H2O2 Fabrication engineering of semi- terial. !e plasma etch process is car- conductor devices has made possi- ried out in a chamber in which a gas ble optoelectronic instruments, laser mixture is partially ionized to create diodes and wireless communica- a plasma, or glow discharge. High en- tion devices among many other mod- ergy ions in the plasma bombard the ern devices. Beginning with Bardeen, semiconductor material and chemi- Brittain and Shockley’s invention of cally reactive components in the gas the transistor in Bell Labs in 1947 and mixture form etch products with the Kilby and Noyce’s introduction of the semiconductor. !e process produces integrated circuit about a decade later, accurately etched features and is one semiconductor devices have dramat- of the primary reasons for the reduc- ically advanced the computing and tion in device size that has made tech- electronics industries. nology such as cell phones and laptop Semiconducting materials, such computers possible. as silicon, germanium, gallium ar- senide, and indium phosphide, are INTRODUCTION neither good insulators nor good con- Semiconductors such as InGaP ductors, but they have intrinsic electri- and InGaAsSb are important for light- cal properties so that by controlled ad- emitting devices as well as communi- dition of impurities, their conductivity cations devices and electronics. Fab- can be altered. With the need to manu- rication of these devices is achieved facture devices at the micro- and nano- by plasma etching in which an ion- scale, the semiconductor industry has ized gas mixture etches the substrate followed “Moore’s Law,” the trend that by both chemical reaction and phys- the number of transistors placed on ical bombardment. In plasma etch- an integrated circuit increases expo- ing for these purposes, indium prod- nentially about every two years. Pro- ucts are not as volatile and are usually duction of the tiny features to create more di#cult to remove than other these integrated circuits is achieved by semiconductor materials. For this ex- plasma etching of semiconductor ma- periment, only wafers with an existing LAURA FRANCOVIGLIA is a senior in chemical engineering at the University of Kansas. 57 epitaxial GaAs cap layer grown over Delta Si Doping 1.1E12/cm2 an underlying InGaP layer were avail- Superlattice Buffer GaAs Buffer able. For the development of an In- S.I. GaAs Substrate ductively Coupled Plasma (ICP) etch process for the InGaP layer, the top Epi Layer Structures of V3339 GaAs layer must "rst be etched o$ to 350 Å GaAs 5E18 Si Doping (Top) expose the underlying InGaP layer. A 400 Å GaAs 1E17 Si Doping common technique used to do this is- 130 Å InGaP not intentionally doped (i) nvolves a selective wet chemical etch Delta Si Doping 4.4E12/cm2 that will remove the GaAs layer with- 30 Å AlGaAs not intentionally doped (i) out etching or damaging the InGaP 130 Å InGaAs not intentionally doped (i) 1 30 Å AlGaAs not intentionally doped (i) layer. Determining the selectivity and Delta Si Doping 1.1E12/cm2 etch rate for removing the GaAs layer Superlattice Buffer are the primary goals of chemical wet GaAs Buffer etch development. Once this has been S.I. GaAs Substrate accomplished, a “recipe” for remov- ing the epitaxial GaAs layer can be cre- ated. EXPERIMENTAL Chemical wet etching selectively Using a 1:8:640 chemical wet etch removes the cap layer of the wafer solution of H2SO4:H2O2:deionized through a series of chemical reactions water, an etch rate of 10 Å/s was ex- in a liquid solution. For this etch pro- pected to remove the 750Å GaAs epi- cess, H2SO4:H2O2:deionized water, a taxial layer and an etch rate of ~ 0 Å/s common solution for removing GaAs, was expected in the InGaP layer. !ese was used in the proportions of 1:8:640. results would be indicative of good se- !e reaction occurs in a sequence of lectivity. steps involving an oxidation reaction of the hydroxide ions when the semi- conductor is immersed in an electro- lyte system to produce Ga2O3 and As2O3. !ese oxides dissolve in the acid of the etchant solution and form soluble salts.2 Wafer samples with a 750Å cap layer of GaAs on top, middle layer of InGaP and thick base layer of GaAs were used. !e layer structure of these wafers is shown below. Later etching will use InGaAsSb wafers. However, because indium is the most di#cult Fig. 1 is a simpli#ed representation of the layer to etch, InGaP is a good starting layers of the wafer. On top of the wafer point. is photoresist. !e gap in between the photoresist has been developed in such way Epi Layer Structures of V3338 that the wafer is ready for chemical wet 350 Å GaAs 5E18 Si Doping (Top) etching in that region. 400 Å GaAs 1E17 Si Doping 160 Å InGaP not intentionly doped (i) Delta Si Doping 4.4E12/cm2 30 Å AlGaAs not intentionally doped (i) 130 Å InGaAs not intentionally doped (i) 30 Å AlGaAs not intentionally doped (i) 58 Before etching the wafers, a pat- to get a baseline, pre-etch, topograph- tern must "rst be imprinted on the ical pro"le of the photoresist for each wafer surface so that a contrast can sample. In taking this measurement, be seen between the etched and non- it was possible to determine the fea- etched areas. Photolithography, a pro- ture depth prior to etching. It was sus- cess similar to "lm exposure, was used pected that the etch solution would to imprint the pattern by using a mask attack the photoresist as well as the to create an image on the wafers in wafer. !e baseline measurement was a light-sensitive, protective layer of a preparatory step to establish the ini- polymer material called photoresist. tial thickness of photoresist. !e wa- !e photoresist used was Microposit fers were then dipped into a 1:30 so- S1818. !e wafers were exposed using lution of NH4OH:deionized water and a mask aligner, which is a UV exposure agitated for about 45 seconds, rinsed system, and were tested under various in deionized water and blown dry with exposure times until well de"ned fea- nitrogen in order to remove native tures were seen on the wafer surfaces. oxide. An adequate exposure time was 75 sec- !e exposed wafers were then onds for the mask aligner used. (Light cleaved into two or three pieces, so intensity can be controlled to help de- that more data would be available for termine proper exposure time, but for analysis. !e etch solution of 1:8:640 this experiment the intensity poten- of H2SO4:H2O2:deionized water was tiometer was not available, therefore prepared under the hood in manage- a trial-and-error method had to be able proportions of 0.2:1.6:128 mL. used.) After exposure, the wafers were Samples were individually immersed developed in a NaOH-based solution. in the etch solution and gently agi- tated. !ree etch times were used: 35, !e following is a description of 75 and 110 seconds. !e lower and the photolithography and develop- upper ranges from the expected etch ment procedure: time of 75 seconds were used to help verify the etch rate and selectivity. r$MFBWFXBGFSJOUPDNYDNTBNQMFTVTJOH After etching, the wafers were mea- UXFF[FSTBOETDSJCF sured again using the pro"lometer to r$MFBOVTJOHJTPQSPQBOPM BDFUPOFBOEBHBJOJTP- QSPQBOPMJOBCBUIPOBIPUQMBUFBU$GPS compare to the pre-etch results and minutes each estimate etch depth. r#MPXESZXJUIOJUSPHFO r$PBUTBNQMFTXJUIQIPUPSFTJTUVTJOHTQJOOFSBU 3700 rpm for 30 seconds RESULTS AND DISCUSSION r 4PGU CBLF TBNQMFT PO IPUQMBUF GPS NJOVUF BU$ Results were inconclusive after r&YQPTFTBNQMFTVOEFSNBTLBMJHOFSGPSTFD- the "rst round of the etch process. It onds appeared that the wet etch solution r 1PTU CBLF TBNQMFT PO IPUQMBUF GPS NJOVUF did not selectively etch the wafers. BU$ It was hypothesized that the wafers r%FWFMPQTBNQMFTCZBHJUBUJOHJO/B0)CBTFE developer for 1 minute could have been upside down, mean- r 3JOTF XBGFST JO EFJPOJ[FE XBUFS GPS TFD- ing that the thicker, bottom layer of onds GaAs was being etched instead of the r#MPXESZXJUIOJUSPHFO 750Å top layer. In the second round r6TFNJDSPTDPQFUPMPPLGPSIJHIEFàOJUJPOPG of the etch process, care was taken to features patterned onto samples mark the bottom sides of each sam- ple using a scribe. A control wafer was !e developed wafers were mea- also used in the second round, which sured with a pro"lometer (Dektak II) was purposefully %ipped bottom side 59 up for etching. In the second round, a ing the wafers with nitrogen. !e pro- sample from each wafer type was also cess was repeated until all of the pho- used. toresist was removed. !e wafers were Because the etch solution attacked then measured again with the pro- the photoresist, which can be seen in "lometer to determine the actual etch the negative etch rates in Fig. 3 and in rate and depth. Reproducible etch etch depths greater than 750Å in Fig. rates were achieved as shown in Fig. 4, the wafers had to be stripped of the 3. !e etch rate for samples stripped of PR to measure accurate etch depths. PR averaged 8.37Å/s with a standard !is was done by agitating the wafers deviation of 0.92Å/s, which was near in acetone for 2 minutes, rinsing with the expected etch rate of 10Å/s. methanol for 2 minutes, and blow dry- Fig. 2 is an example of the results from the pro#lometer, which gives a cross-sectional view of the wafer.