DRY ETCHING OF SiC

 

S.J. Pearton

Department of Materials Science and Engineering

University of Florida, Gainesville, FL 32611, USA

 

CHAPTER SCOPE

            In this chapter we discuss wet and dry patterning techniques for SiC and the relative merits of these methods for MEMS processing.  We describe the basic principles involved in etching SiC and problems that can arise because of the binary nature of the lattice and its relatively high bond strength.  Recent developments in the use of high density plasma sources to achieve fast etching rates (in some cases over 1mm∙min-1 for bulk 4H-SiC) are discussed-these sources are likely to play a dominant role for processing of SiC MEMS devices since they are capable of producing etch depths from 0.1 to 100mm with minimal disruption of the SiC surface. 

 

WET ETCHING

            Due to its hardness (H=9+), SiC is one of the most widely used lapping and polishing abrasives for metals, metallic components and semiconductor wafers.  However this very property makes it difficult to etch in typical acid or base solutions.  In its single crystal form, SiC is not attacked by single acids at room temperature.  Indeed the only techniques for etching SiC employ molten salt fluxes, hot gases, electrochemical processes or plasma etching [1-11].  Table 1 shows a list of the molten salt solutions and the temperatures needed for successful etching of SiC.  The disadvantages of these high temperature, corrosive mixtures include the need for expensive Pt beakers and sample holders (which can withstand the molten salt solutions) and the inability to etch masked samples because few masks hold up to these mixtures.  While one can conceivably use Pt masks, the wet etching is isotropic and therefore undercuts the mask.

            Photoelectrochemical etching can be successfully employed for SiC [12].  The dissolution rate of semiconductors may be altered in acid or base solutions by illumination with above bandgap light.  The mechanism for photo-enhanced etching involves the creation of e-h pairs, the subsequent oxidative dissociation of the semiconductor into its component elements (a reaction that consumes the photo-generated holes) and the reduction of the oxidizing agent in the solution by reaction with the photo-generated electrons.  Generally, n-type material is readily etched under these conditions, while p-type material is not due to the requirements for confining photo-generated holes at the semiconductor-electrolyte interface (i.e., the p-surface is depleted of holes because of the band-bending).  This allows for selective removal of n-SiC from an underlying p-SiC layer [12].  Under conditions of no illumination, it is often possible to get the reverse selectivity if the sample is correctly biased, since n-SiC requires photogeneration of carriers for etching to proceed.  Etching over large areas can be achieved using Hg lamps and some degree of anisotropy is obtained because of the shadowing effect of the metal masks (typically Ti) allowing carriers to be generated only in unmasked regions.  Some of the disadvantages of the technique include fairly rough surface morphologies (due to enhanced dissolution rates for areas around crystal defects), inability to pattern very small dimension features and poor uniformity of etch rate.For these reasons,most attention is now focussed on dry etching methods for SiC,most of which have been developed for high power,high temperature electronics in this materials system[14-53]

BASICS OF  PLASMA ETCHING

            A plasma is an ionized gas with equal numbers of free positive and negative charges.  The free charge is produced by the passage of electric current through the discharge.  For most plasma of interest for etching, the extent of ionization is very small.  Typically there is only one charged particle per 100,000 to 1,000,000 neutral atoms and molecules.  The positive charge is mostly in the form of singly ionized neutrals, (i.e., atoms, radicals or molecules) from which a single electron has been stripped (removed).  The majority of negatively charged particles are usually free electrons; although in very electronegative gases such as chlorine, negative ions can be more abundant.

(a) Plasma creation

            For the generation of plasma, a high frequency voltage is applied between the two electrodes.  Free electrons (current flows) are accelerated and collide with neutral gas.  When the collision occur, the energy of the gas molecules becomes high and the molecules can then be dissociated, ionized and excited.

Dissociation      :           e + XY             X + Y + e

Ionization          :           e + XY             XY+ + 2e

Excitation         :           e + XY+           X + Y*                        

* indicates an excited atom or radical.

            In a plasma, the electron and ions densit is equal on average, but less than the density of neutral species .When a plasma is created, electrons and ions will diffuse out of the plasma.  The electrons will reach the surface of material which expose in the plasma before ions due to their (electrons) much greater in velocity.  This causes plasma become more positive, since there is an excess of positive ions left behind. The surfaces of the plasma containment vessel charge up negatively.  This negative charge pushes other electrons away at the same time as attracting positive ions.  In steady state the surface no longer charges up, and thus electrons and positive ions have to arrive at the same rate.  The field near the surface holds the electrons away from the surface, allowing only the most energetic electrons to get there.  The field also accelerates the positive ions toward the surface, and in this way the rates of arrival of electrons and positive ions are made equal.

            In particular on the powered electrode,electrons are excluded by the positive charge,producing a region above the electrode in which there are fewer collisions of gas molecules with energetic electrons.For this reason,this region appears dark compared to the rest of the plasma region due to the absence of emission from excited molecules.This positively charged region is called a sheath.

(b)        Basic Mechanism of Plasma etching

In plasma, two kinds of active species are produced, neutrals and ions.  Neutrals may be very reactive,while the ions are usually less reactive but their kinetic energy can be controlled by substrate bias.

1.      Sputtering

Positive ions are accelerated through sheath region and strike the substrate with high kinetic energy[54]

            E max   =      eV

where Q is the ionic charge, E0 represents the rf field (kV/cm), M is the mass of the ion and f is the rf frequency (MHz).

  By momentum conservation law, some of this energy is transferred to surface atoms that are then ejected, leading to material removal.  This is mechanical interaction and the sputtering rate is given by[55]

                  R =

where s is etch yield, j is ion flux (mA/cm2), W is atomic weight (g/mol) and r is material density (g/cm3).

Sputtering is unselective etching because the ion energy required to eject material is large compared to differences in surface bond energies and chemical reactivity.  Due to the applied voltage to the substrate, the flux of ions is vertical and this kind of etching is anisotropic.

2.      Chemical reaction

This etching comes about when active species (neutral) from the gas phase are absorbed on the surface material and react with it to form a volatile product.  High product volatility is essential.  The evaporation rate of a material  is given by[55]

mA   =   

where a is the material-dependent efficiency factor, usually between 0.1 and 1.0, M is the molecular weight and P is the vapor pressure

Without volatility the reaction products would coat the surface and prevent gaseous species from reacting it, and cut off the etching reaction.  Chemical etching provides very high selectivity but is non-directional,producing  isotropic etching.

3.      Ion-Assisted Plasma Etching

The substrates are exposed to suitable neutral species in the presence of ion-bombardment.  The combination between sputtering and chemical reaction results in material removal rates exceeding the sum of separate chemical attack and sputtering.  There are two mechanism for ion-assisted etching.

i)        Ion-enhanced energetic etching

This etching, neutral species cause little or no etching without ion bombardment.  Ions damage the substrate material create high roughness which increased the expose surface, dangling the atomic bond of the surface which increase the number of absorption site.  Since ions are accelerated and strike surface vertically, the etching induced is directional.

ii)    Ion-Enhanced Inhibitor

Etching by neutrals is spontaneous so ion bombardment does not cause the etching reaction.  Ion can coat substrate surface and prevent etching reaction from taking place.  The normal going ion flux keeps areas clear of film on the horizontal surfaces, while vertical feature sidewalls are coated with a thin film which inhibits chemical reaction.

c.  Plasma Etching Parameters

1.  Effect of Pressure

Pressure is inversely proportional to the mean free path of particle.  At higher pressure, the mean free path is shorter cause more frequency of collision of electrons.  The electrons will lose their energy and create more reactive neutral species during the collision (generate higher plasma density).  Then the etch mechanism is dominated by chemical reaction rather than physical (sputtering) reaction. 

As pressure is lowered, the characteristic potentials across the sheaths and the voltage applied to a discharge increase sharply.  The rise in potential translates into a higher energy ion flux to substrate surfaces.  Sputtering does not take place until ion energy exceeds the material-ion (molecule) threshold energy.

 
 
2.  Effect of Temperature

Temperature is a function of chemical reaction as e –Ea/RT.  Where Ea is an activation energy, R is gas constant and T is temperature in Kelvin.  Thus, it has a dominant effect on selectivity, etch rates and the degradation on resist mask.

3.  Effect of Loading

The loading effect is the decreasing of etch rate when there are more etchable substrate material placed in a reactor.   The etch rate is usually proportional to etchant concentration, their concentration decreases with the area of etchable surface in the plasma.

d.  Plasma Reactors

1.       Ion milling

Ion milling is a pure physical process.  The commonly gas used is Ar.  The ion energy and ion density is separated control by the filament current and the accelerated voltage adjustment . This process employs high energetic inert ion to erode the surface of material by bombardment, causing high surface damage and degrade the performance of the device.

            2.  Reactive Ion Etching (RIE)

Reactive Ion Etching technique generates the plasma at a radio frequency of 13.56 MHz between two parallel electrodes in a reactive gas (see Figure 1).  The electrons will be accelerated and collide with gas molecules contribute to sustaining the plasma.

The substrate is placed on the power electrode, not grounded, in this case a large negative dc self-bias develop on the sample and attract ion from plasma which cause damage on the surface.  This results to high etch rate and anisotropic. However, highly energetic ions damage the sample surface and degrade both electrical and optical device performances.

3.  Electron-Cyclotron Resonance (ECR)

Due to the surface damage from high energetic ions, High-density plasma is interesting.  High-density ECR plasmas are formed at low pressures with low plasma potentials and ion energies due to magnetic confinement of electrons in the source region(Figure 2).  Therefore, the surface damage in ECR may less than with the RIE technique if ion energy is the most important parameter in determining damage.In other cases,the higher ion flux may induce more damage than with RIE. 

The frequency of orbital motion of electrons, confined under the action of an external magnetic field (875 Gauss) is equivalent to the drive frequency of 2.45 GHz leading to the occurrence of resonance, called electron cyclotron resonance, if this frequency of power is applied to the plasma.  In this condition, outer shell electrons from gas molecules in the discharge may also be liberated, leading to a very high degree of ionization in the plasma.  ECR provides high ion density (1010-1012 cm-3) compared with RIE (109 cm-3) without inducing high damage on the sample because the plasma potential is much lower .

ECR can also control the ion energy and ion flux independently.  Ion energy is controlled by rf or dc biasing at the substrate holder while the ion and neutral flux is controlled by microwave and gas pressure.

           

 
 
4.  Inductive Coupled-Plasma (ICP)

Inductive coupled-plasma etching offers an alternative high-density plasma technique where plasmas are formed in a dielectric vessel encircled by an inductive coil into which rf power is applied (see figure 3).  A strong magnetic field is induced in the center of the chamber, which generates a high-density plasma (~5 x 1011 cm-3) due to the circular region of the electric field that exists concentric to the coil.  The electrons in circular path will have only a small chance to be lost to the chamber walls resulting in low dc self bias.  At low pressures (£ 20 mTorr), the plasma diffuses from the generation region and drifts to the substrate at relatively low ion energy (<25 eV).  Thus ICP etching is expected to produce low damage while achieving high etch rates.  Anisotropic profiles are obtained by superimposing a rf bias on the sample to independently control ion energy and by using glow pressure conditions to minimize ion scattering and lateral etching.  ICP sources may be easier to scale up than ECR sources and are more economical in terms of cost and power requirements.

 

 Plasma Etching Of Silicon Carbide

In order to etch silicon carbide in a plasma reactor, the chemistry used must be reactive with SiC and the species produced by the chemical reactions must be volatile compounds under the temperature and pressure condition to avoid the residue on the surface.

Several chemistries were examined (see table 2).  The most effective gas is based with fluorine chemistry.  The reaction mechanism of SiC in F2-based chemistry is shown  below.

                              Si + xF             SiFx                  x £ 1-4

                              C + xF             CFx                  x £ 1-2

            J.J. Wang et al[52] showed  from optical emission spectra that ion bombardment plays a role in etch mechanism.

            Various gas addition can have effects on the etch behavior.  Oxygen has often been added to fluorine-based chemistries under RIE conditions to enhance the active fluorine concentration and increase SiC etch rate.  In ECR conditions, there is only little change in the atomic fluorine concentration[48,49].  In contrast, the addition of H2 to the gas mixture reduces the etch rate[13].The introduction of hydrogen into the plasma prevents residue formation through a combination of mechanism, including the formation of volatile alane (AlH3) and the removal of the C-rich surface[13].

The differences in the etch rates are due more to differences in the dangling bond densities and the corresponding reactivities of the crystal faces than to the different crystal structure.  For example, each atom on cubic (001) face has two dangling bonds, whereas only one dangling bond exists on a (111) face or similarly to the (0001) face of hexagonal SiC.

      There is no measurable difference in etch rates between n+ and p+ SiC indicating that Fermi level effects play no role in the etch mechanism under ICP conditions[52].By contrast with the RIE technique, the etch rate increases when the n-type doping increases[24].

In order to etch silicon carbide in a plasma reactor, the chemistry used must be reactive with SiC and the species produced by the chemical reactions must be volatile compounds under the operating temperature and pressure conditions to avoid residues on the surface.

            Many plasma chemistries have been examined (see Table 2).  The most effective gases in terms of etch rate are based on fluorine chemistry.  The reaction mechanism of SiC in F2-based chemistry is shown as below.

Si + xF ® SiFx            x£ 1-4 

C + xF ® CFx x£ 1-2

From optical emission spectra it is clear that ion bombardment play a role in the etch mechanism.  When etching silicon atoms with atomic fluorine, a carbon layer is present on the exposed surface and is removed by the ion bombardment.

Various gas additions can have effects on the etch behaviour.  Oxygen has often been added to fluorine-based chemistries under RIE conditions to enhance the active fluorine concentration and increase SiC etch rate.  In high ion density conditions this produces only a small change in the atomic fluorine concentration.  In contrast, the addition of H2 to the gas mixture reduces the etch rate.  The introduction of hydrogen into the plasma prevents residue formation through a combination of mechanisms, including the formation of volatile alane (AlH3) to remove Al sputtered from the reactor and the removal of the C-rich surface.

The differences in the etch rates are due more to differences in the dangling bond densities and the corresponding reactivities of the crystal faces than to the different crystal structures.  For example, each atom on cubic (001) face has two dangling bonds, whereas only one dangling bond exists on a (111) face or similarly on the (0001) face of hexagonal SiC.

Previous results on reactive ion etching of SiC have generally employed F2-based plasmas.Relatively rough surfaces are often observed under these conditions due to sputtering of the electrode material onto the SiC sample, leading to micromasking.

With the advent of high density plasma sources, including Electron Cyclotron Resonance (ECR), Inductively Coupled Plasma (ICP) and Helicon, much higher SiC etch rates have been reported.  The key advantage of these sources is decoupling of ion energy and ion flux, so that relatively low ion energies can be employs.  Schematics of RIE,ECR and ICP reactors are shown in Figures 1-3.  This reduces the electrode sputtering problem and in addition the plasma chemistries for high density sources generally involve gases that do not contain CHx because of the extensive polymer deposition that can occur within the source at high applied powers.  The absence of these two sources of redeposition onto the SiC generally leads to good surface morphologies.

Inductively coupled-plasma etching offers an attractive high-density plasma technique where plasmas are formed in a dielectric vessel encircled by an inductive coil into which rf power is applied.  A strong magnetic field is induced in the center of the chamber, which generates a high-density plasma (~5x1011 cm-3) due to the circular region of the electric field that exists concentric to the coil.  The electrons in this circular path will have only a small chance to be lost to the chamber walls, resulting in low dc self bias.  At low pressures (£20 mTorr), the plasma diffuses from the generation region and drifts to the substrate at relatively low ion energy (<25 eV).  Thus ICP etching is expected to produce low damage while achieving high etch rates.  Anisotropic profiles are obtained by superimposing a rf bias on the sample to independently control ion energy and by using flow pressure conditions to minimize ion scattering and lateral etching.  ICP sources may be easier to scale up the ECR sources and are more economical in terms of cost and power requirements.

In choosing the optimum plasma chemistries for investigation, it is instructive to look at SiC etch rates reported previously in the literature (Table 2).  There are two key points evident in this data.  Firstly the high density reactors do indeed produce faster rates, and secondly F2-based chemistries lead to higher rates than Cl2, F2 or Br2.  This is readily understood by examining the relative volatility of the SiC etch products in F2- or Cl2-based plasmas.  Table 3 shows the boiling points for potential etch products in these plasmas (with the addition of O2 in both cases, although it is reported that O2 itself plays no direct role in SiC etching but rather can influence the etch rate through changing the atomic fluorine neutral density in the discharge).  While it is understood that the high ion flux in ICP discharges can desorb the etch products before they can fully coordinated, the boiling points of the complete molecules do give some indication of relative volatility and hence the trend expected for etch rates in the different chemistries.  From Table 3, it is clear that the fluorinated products are more volatile than their chlorinated counterparts.  Finally, the etching should have a high selectivity over both the mask material and the front-side metal employed as the etch-stop.  The thickness of the SiC substrate enables us to estimate that an etch rate of at least 4000 Å∙min-1 is needed to keep the process time below ~2 hours, which is a rough guess for a practical process.

PLASMA CHEMISTRIES

Figure 4 shows the etch rates (top) and etch yields (bottom) for SiC in ICP discharges of NF3, SF6, BF3 or PF5 at fixed rf chuck power (250 W) and pressure (2 mTorr), as a function of ICP source power.  The yield tends to decrease as the source power is increased, even as etch rate increases with NF3, SF6 and PF5.  This suggests that ion flux is not the limiting factor under these conditions, but rather the supply of fluorine neutrals to the SiC surface limits the etch rate.  The etch rates are significantly higher with NF3 and SF6, which is consistent with the lower bond strength of these molecules compared to PF5 and BF3.  When comparing the relative advantages of NF3 and SF6, the much lower cost of the latter outweighs the faster rates obtained with the former, particularly for long etch times.In these experiments the rf chuck power was held constant at 250 W, corresponding to dc self-biases of roughly –290 V at 250 W source power, to –200 V at 1500 W source power.  Clearly, NF3 and SF6 produce the fastest rates, and this correlates to the relative dissociation of these gases in the ICP source.  Optical emission spectroscopy showed very intense atomic fluorine lines in the range 700-900 nm for both NF3 and SF6, while the intensities of these lines were much lower for BF3 and PF5.  The etch rates are also in good correlation with the average bond energies for the feedstock gases, i.e. BF3 154 kCal/mol[57], PF5 126 kCal/mol[58], SF6 78.3 kCal/mol[59] and NF3 66.4 kCal/mol[60]. The lower the bond energy, the more effective is the dissociation in the ICP source to form atomic fluorine neutrals which are the active etchant species.  The etch products are probably SiFx and CFx species (x would not necessarily have to reach its fully coordinated value of 4 under ion-assisted conditions), although we did not have adequate sensitivity in our OES system to detect them during the etching process.  In the case of BF3 the SiC etch rate decreases slightly at high source powers, which might be related to the fall-off in ion energy under those conditions or to desorption of the reactant fluorine before it can form etch products with the SiC.

Figure 5 shows the rf power dependence of SiC etch rate at a fixed ICP source power of 750 W.  The dc self-bias increases almost linearly with chuck power, as shown in the lower part of the Figure.  In NF3, SF6 and PF5 there is a general trend for increasing etch rate as rf chuck power is increased.  This could be related to increased Si-C bond-breaking efficiency at higher ion energies, allowing more etch products to form.  In the case of BF3 the etch-limiting step is probably the supply of atomic fluorine because of the lower efficiency in dissociating this gas.

Polished SiC surfaces often have relatively rough morphologies due to residual mechanical damage. After dry etching with any of the different plasma chemistries, the surface roughness improves to values in the range 0.6-2.0 nm.  This smoothing of initially rough surfaces is commonly observed in ion-driven etch processes and originates in the angular dependence of ion mill rates.  This leads to faster removal rates for high aspect ratio features and creates a smoother morphology.  Figure 6 shows the dependence of RMS surface roughness in ICP source power in the four different plasma chemistries.  Under virtually all conditions the etched surfaces are smoother than the unetched control samples.

  Mask Materials

Standard conditions of 750 W source power 250 W rf chuck power and 2 mTorr for NF3 discharges, and addition of O2 to the chemistry were examined for their  effects on etch selectivity of SiC to the different mask materials.  Figure 7 shows the etch rates (top) and resultant selectivities for SiC over the masks (bottom) for NF3/O2 discharges, as a function of NF3 percentage of the gas load (15 sccm).  The etch rates increase with NF3 composition for SiC and the mask materials.  At high O concentrations there is actually net deposition on Al as it oxidizes, so that the SiC selectivity over Al is infinite.

However the requirement for via hole etching is that the SiC etch rate be > 4000 Å-min-1.  Maximum selectivities were > 20 over Ni and ~7 over Al.  Note that there were unacceptably low selectivities for SiC over ITO.

Figure 8 shows scanning electron micrographs of features etched ~ 60mm (top) or 100mm (bottom) into SiC substrates.  The top micrograph shows the effect of feature diameter in etch depth – the smaller diameter features (~30mm) are shallower by ~15% than the larger openings, which gives the magnitude of the aspect ratio dependent etch rate.  The bottom micrograph shows features etched all the way through 100mm thick SiC substrates mounted on sapphire substrates.

In situations in which only a mesa etch is required, it is desirable that the pattern transfer process not degrade the electrical properties of the SiC. If higher rates are desirable, then the majority of the etching can be performed at higher dc self-biases and this latter parameter can be decreased toward the end of the process.

It is also desirable that there is high selectivity for etching SiC over the mask material (and also the front-side metallization in the case of via holes).  Figure 9 shows the dependence of SiC etch rate (top) and selectivity for SiC over Al (bottom) as a function of O2 percentage (by flow) in 500 W source power, 150 W rf chuck power, SF6/O2 discharges.  Note that the SiC etch rate initially increases as O2 is added to the SF6.  This is probably due to the increase of atomic fluorine neutrals present at low O2 percentages, a feature well established for CF4/O2 and SF6/O2 plasma chemistries.  The etch rate falls off at higher O2 percentages because atomic oxygen does not appear to play an active or direct role in etching of SiC.  However, the etch selectivity over Al increases rapidly with O2 addition, since the Al oxidizes and does not etch beyond ~ 40% O2 addition to the SF6.

The fact that ion energy is a key factor in determining the SiC etch rate is evident in the data of Figure 10 .At fixed source power, the incident ion energy is controlled by the sum of this dc self-bias and the plasma potential (-20-25 V in this particular tool).  The etch rates are always slightly higher with SF6/O2 (25% O2 by flow in this case) compared to pure SF6 and the rates begin to saturate beyond ~ 350 V where Si-C bond breaking is no longer the limiting step.  We should also mention that passing hot gases such as Cl2, F2, H­2 and HCl over SiC at high temperatures (> 1200oC) will etch the surface and this process is often employed to clean SiC substrates prior to epitaxial gro

 

RECENT DEVELOPMENTS AND FUTURE TRENDS

It has also been recently shown that use of UV illumination during plasma etching in Cl2-based gas chemistries can enhance the etch rates of SiC, probably through photo-excitation of the chlorinated etch products.  This process does not produce any increase in etch rates with F2-based gas chemistries, because the etch products are already quite volatile.

The achievement of high etch rates for SiC in the various high density plasma sources has now placed the emphasis on developing mask materials that can withstand long plasma exposures, such as needed during via hole formation.  The Al masks described earlier work well most of the time provided the residual stress in the metal is minimized.  However to pattern smaller features, one would ideally like to avoid thick metal masks and use more convenient materials such as photoresists or dielectrics.  Unfortunately these materials etch more rapidly than SiC in F2-based plasmas, limiting their application to the etching of shallow features.

Since it is clear that more dissociated plasmas with separate control of ion energy produce the fastest etch rates for SiC, it is likely that even higher source powers will be employed in future.  Most of the etching to date has been carried out at source powers £ 1500 W, but reactors are available with powers of 3-5 kW.  The higher ion fluxes in these systems will place even greater demands on the durability of mask materials.

      An ICP process based on SF6 or NF3 provides practical etch rates for deep patterning of SiC.  The use of the former gas is probably favored due to its much lower cost and the simpler, less expensive regulators required.  Other F2-based plasma chemistries involving PF5 or BF3 do not produce adequate SiC etch rates.  Through-wafer vias have been demonstrated using the ICP SF6 process, as well as low damage conditions for etching of mesas.  More conventional RIE techniques can also be employed in most situation in device processing, but suffer from lower etch rates and poorer surface morphologies.

 

SUMMARY

            The etching of very deep features for MEMS structures in SiC substrates in practical time frames appears feasible, using the combination of ICP NF3 or SF6 discharges and thick metal masks.  Addition of O2 to the plasma chemistry increases etch selectivity for SiC over Al under some conditions, due to oxidation of the Al. In contrast, with very low O2 concentrations, Ni shows better mask performance.  The selectivity for SiC over Ni under this condition is up to 20.  Etch rates in excess of 8,000 Å-mim-1 have been achieved for 5x5 mm2 samples of SiC, with ~50% of this area exposed to the plasma.  The etch rates for larger samples will be less due to loading effects.  Based on our experience with other materials, the fall-off is likely to be of the order of 20-30% when scaling to 3” diameter wafers with ~10% of the area exposed to the plasma.  The SiC etch rates with PF5 and BF3 are much lower than with NF3 and SF6, which is a result of their lower dissociation efficiency in the ICP source.

            The main results of etch rate enhancement with UV illumination  may be summarized as follows:

(i)                  SiC etch rates in ICP Cl2/Ar discharges can be increased by UV illumination.

The mechanism for the enhancement is still the subject of investigation.  By analogy with past results on Cl2/Ar etching of Cu, it is possible that the UV light is absorbed by SiClx and CClx species on the SiC surface, promoting more complete coordination, i.e. x ® 4.  We rule out any change in chlorine radical concentration in the plasma because optical emission spectroscopy showed that the intensity of these emission lines was unchanged by the UV illumination.  Similarly we do not believe that sample heating explains the results, because of the absence of any enhancement with SF6/Ar discharges and the stability of the photoresist masks.  The surface morphologies were similar to those obtained without UV illumination.

(ii)        There was no effect on SiC etch rates in SF6/Ar ICP discharges.  This may be because the etch products are already quite volatile in this case, and desorption of these species is not the rate-limiting step.  Rather it is likely that either the initial bond-breaking that must precede etch product formation or the supply of reactive fluorine radicals to the SiC surface are the limiting factors, depending in the exact plasma conditions.  The etched surfaces were in general smoother with UV illumination, which may be a result of more uniform desorption of the etch products.  In this case the effect of the UV photons may be to increase surface mobility of the adsorbed fluorine.

REFERENCES 

[1]               CRC Handbook of Metal Etchants, ed. P. Walker and W.H. Tarn (CRC Press, Boca Raton, FL. 1991) pp. 1092-1101.

[2]               G.L. Harris, in Properties of SiC, Ed. G.L. Harris, EMIS Data Review No. 13 9INSPEC, London, UK, 19950, pp. 134-135.

[3]               D. Buckley, J. Vac. Sci. Technol. A (3) 762 (1985).

[4]               T.L. Chu and R.B. Campbell, J. Electrochem. Soc. 112 955 (1965).

[5]               W.K. Liebmann, J. Electrochem. Soc. 12 885 (1964).

[6]               E.D. Wolley, J. Appl. Phys. 37 1588 (1966).

[7]               L.B. Griffith, J. Phys. Chem. Sol. 27 257 (1966).

[8]               R.W. Brander, J. Electrochem. Soc. 12 881 (1964).

[9]               P. Nordquist, H. Lessoff, R.J. Gorman and M.L. Gripe, Springer. Proc. Phys. 43 119 (1989).

[10]           P. Pirouz, C.M. Chorey and J.A. Powell, Appl. Phys. Lett. 50 221 (1987).

[11]           M.W. Jepps and T.F. Page, J. Microscop. 124 227 (1981).

[12]           J.S. Shor in ref. [2], pp. 141-149.

[13]           P.H. Yih and A.J. Steckl, J. Electrochem. Soc. 142 312 (1996).

[14]           J. Wu, J.D. Parsons and D.R. Evans, J. Electrochem. Soc. 142 669 (1995).

[15]           J. Casady, E.D. Luckowski, M. Bozack, B. Sheridan, R.W. Johnson and J.R. Williams, J. Electrochem. Soc. 143 1750 (1996).

[16]           J.R. Flemish, K. Xie and J. Zhao, Appl. Phys. Lett. 64 2315 (1994).

[17]           J.R. Flemish and K. Xie, J. Electrochem. Soc. 143 2620 (1996).

[18]           J. Hong, R.J. Shul, L. Zhang, L.F. Lester, H. Cho, Y.B. Hahn, D.C. Hays, K.B. Jung, S.J. Pearton, C.-M. Zetterling and M. Ostling, J. Electron. Mater. 28 196 (1999).

[19]           F. Lanois, P. Lassagne, D. Planson and M.L. Locatelli, Appl. Phys. lett. 69 236 (1996).

[20]           J.J. Wang, E.S. Lambers, S.J. Pearton, M. Ostling, C.-M. Zetterling, J.M. Grow and F. Ren, Solid-State Electron. 42 743 (1998).

[21]           J.J. Wang, E.S. Lambers, S.J. Pearton, M. Ostling, C.-M. Zetterling, J.M. Grow, F. Ren and R.J. Shul, J. Vac. Sci. Technol. A16 2204 (1998).

[22]           P. Leerungnawarat, D.C. Hays, H. Cho, S.J. Pearton, R.M. Strong, C.-M. Zetterling and M. Ostling, J. Vac. Si. Technol. B17 2050 (1999).

[23]           F.A. Khan and I. Adesida, Appl. Phys. Lett. 75 2268 (1999).

[24]           P. Chabert, N. Proust, J. Perrin and R.W. Boswell, Appl. Phys. Lett. 76 2310 (2000).

[25]           C.E. Weitzel, J.W. Palmour, C.H. Carter, Jr.K. Moore, K.J. Nordquist, S. Allen, C. Thero and M. Bhatanagar, IEEE Trans. Electron. Dev. 43 1732 (1996).

[26]           K.Z. Xie, J.H. Zhao, J.R. Flemish, T. Burke, W.R. Buchwald, G. Lorenzo and H. Singh, IEEE Electron. Dev. Lett. 17 142 (1996).

[27]           B.J. Baliga, IEEE Trans. Electron. Dev. 43 1717 (1996).

[28]           A.K. Agarwal, G. Augustine, V. Balakrishna, C.D. Brandt, A.A. Burke, L.S. Chen, R.C. Clarke, P.M. Esker, H.M. Hobgood, R.H. Hopkins, A.W. Morse, L.B. Rowland, S. Seshadri, R.R. Siergiej, T.J. Smith, Jr. and S. siriam, Tech. Dig. Inst. Electron. Dev. Meeting pp 9.1.1-9.1.6, Dec. 1996.

[29]           D.M. Brown, E. Downey, M. Ghezzo, J. Kretchmer, V. Krishnamurthy, W. Hennessy and G. Michon, Solid State Electron. 39 1531 (1996).

[30]           V.E. Chelnokov, Mat. Sci. Eng. B11 103 (1992).

[31]           T.P. Chow and M. Ghezzo, Mat. Res. Soc. Symp. Proc. 423 9 (1996).

[32]           S. Siriam, R.C. Clarke, A.A. Burk, H.M. Hobgood, Jr., P.G. McMullin, P.A. Orphanos, R.R. Siergiej, T.J. Smith, C.D. Brandt, M.C. Driver and R.H. Hopkins, IEEE Electron. Dev. Lett. 15 458 (1994).

[33]           K.E. Moore, C.E. Weitzel, K.J. Nordquist, L.L. Pond III, J.W. Palmour, S. Allen and C.H. Carter, Jr., IEEE Electron. Dev. Lett. 18 69 (1997).

[34]           J.W. Palmour, J.A. Edmond, H.S. Kong and C.H. Carter, Jr., Physica B 185 461 (1993).

[35]           J.B. Casady, D.C. Sheridan, W.C. Dillard and R.W. Johnson, Mat. Res. Soc. Symp. Proc. 423 105 (1996).

[36]           S.T. Sheppard, M.R. Melloch and J.A. Cooper, IEEE Trans. Electron. Dev. 41 1257 (1994).

[37]           J.B. Casady, J.D. Cressler, W.C. Dillard, R.W. Johnson, A.K. Agarwal and R.R. Siergiej, Solid State Electron. 39 777 (1996).

[38]           J.N. Shenoy, J.A. Cooper, Jr. and M.R. Melloch, IEEE Electron. Dev. Lett. 18 93 (1997).

[39]           A.K. Agarwal, R.R. Siergiej, M.H. White, P.G. McMullin, A.A. Burk, L.B. Rowland, C.D. Brandt and R.H. Hopkins, Mat. Res. Soc. Symp. Proc. 423 87 (1996).

[40]           P.G. Neudeck, D.J. Larkin, C.S. Salupo, J.A. Powell and L.G. Matus, Inst. Phys. Conf. Ser. 137 475 (1994).

[41]           O. Kordina, J.P. Bergman, A. Henry, E. Fanzen, S. Savage, J. Andre, L.P. Ramberg, U. Lindfelt, W. Hermanasson and K. Bergman, Appl. Phys. Lett. 67 1561 (1995).

[42]           J.B. Casady and R.W. Johnson, Solid State Electron. 39 1409 (1996).

[43]           P.H. Yih and A.J. Steckl, J. Electrochem. Soc. 140 1813 (1993).

[44]           B.P. Luther, J. Ruzyllo and D.L. Miller, Appl. Phys. Lett. 63 171 (1993).

[45]           J. Wu, J.D. Parsons and D.R. Evans, J. Electrochem. Soc. 142 669 (1995).

[46]           F. Lanois, P. Lassagne, D. Planson and M.L. Locatelli, Appl. Phys. Lett. 69 236 (1996).

[47]           J.R. Flemish, K. Xie and J. Zhao, Appl. Phys. Lett. 64 2315 (1994).

[48]           J.R. Flemish and K. Xie, J. Electrochem. Soc. 143 2620 (1996).

[49]           G.F. McDaniel,