Study on the Mechanism of Silicon Etching in HNO 3-Rich HF/HNO 3Mixtures
M.Steinert,J.Acker,*S.Oswald,and K.Wetzig
Leibniz Institute for Solid State and Material Rearch Dresden (IFW Dresden),P.O.Box 270116,D-01171Dresden,Germany
Recei V ed:September 27,2006;In Final Form:No V ember 8,2006
The wet chemical etching of silicon using HNO 3-rich HF/HNO 3mixtures has been studied.The effect of different parameters on the etch rate of silicon,for example,the HF/HNO 3mixing ratio,the silicon content of the etchant,temperature,and stirring speed in the solutions,has been examined and discusd in light of a previous study on etching in HF-rich HF/HNO 3mixtures.Nitrogen(III)intermediates are generated owing to the dissolution of silicon and the decomposition if the solution is expod to air.The nitrite ion concentration,measured in diluted etchant solution by ion chromatography,acts as a sum parameter for the reactive N(III)species in the concentrated etchant.The etch rate shows two different correlations to the nitrite concentration.In the region of high nitrite concentrations,the etch rate decreas slightly with decreasing nitrite concentration,whereas at lower nitrite concentrations,the etch rate increas linearly with further dec
reasing nitrite concentration.Stirring experiments and the determination of activation energies show that the etching of silicon in HNO 3-rich etchants is controlled by diffusion.X-ray photoelectron spectroscopy measurements of the silicon surface after etching revealed a hydrogen termination independent of the concentration of reactive species and the content of HNO 3in the etchant.Si -O containing surface species were not found.A combined electrochemical (injection of holes into the valence band of silicon)and chemical (Si -Si back-bond breaking by an attack of HF)reaction mechanism of silicon etching without generation of SiO 2is propod.
1.Introduction
The production of clean and smooth surfaces plays an increasingly important role in today’s microelectronic industry for the elimination of defects and impurities that can degrade the electronic properties.The HF/HNO 3etching system perhaps is the most widely ud isotropic etchant for silicon.1In the miconductor industry,HF/HNO 3mixtures were applied in a dipping bath for the removal of contaminations and lattice defects generated by the lapping of silicon wafers.2Further applications are the removal of work damage or roughness (after sawing of ingots)and the texturing of the surface of multi-crystalline silicon wafers for solar cell fabrication.3,4
In the literature,the etching of silicon in HF/HNO 3mixtures is described as a two-step chemical process including (i)the formal oxidation of silicon to SiO 2by nitric acid (eq 1)and (ii)the subquent dissolution of formed SiO 2by HF (eq 2).The overall reaction,as written in eq 3,shows that formally the only reaction products are water,nitrogen monoxide,and hexafluo-rosilicic acid.5-7
The formalized description of the silicon etching chemistry,as stated above,can be traced back to the original studies by Robbins and Schwartz.5-7The step of SiO 2dissolution by HF solutions was widely investigated and out of debate.8-14
However,the crucial and yet unresolved step in this reaction is the oxidation of silicon by nitric acid.The injection of holes into the valence band of the miconductor by the reduction of nitric acid at local cathodic sites and by the dissolution of silicon at local anodic sites shows the electrochemical nature of that process and was first mentioned by Turner 15and reassumed by Kooij et al.1Shih et al.showed that the chemical etching of silicon caus,under certain conditions,the formation of a porous layer that exhibits photoluminescent properties known before only to come from electrochemical etching.The porous layer formation was explained by a mechanism bad on the generation and injection of holes into the valence band of silicon.16,17
Abel et al.18-20suggested,as a mechanistic approach,that veral combined equilibria between different nitrogen oxides lead to nitrous acid as the reactive species in the etching process.The occurrence of an induction period,obrved by a signifi-cantly lower etch rate becau of the etching of silicon in a freshly prepared etch mixture,can be prevented by adding NaNO 2as the catalyst 5,6that yields the formation of the assumed reactive agent nitrous acid.Once the reaction has initiated,HNO 2is assumed to be generated autocatalytically by the etch process itlf.Kelly and co-workers disputed the HNO 2pathway and concluded the incread etch rate on the formation of a stronger oxidizing agent,the nitrosonium ion NO +.16Furthermore,unresolved is the question of the final state of HNO 3reduction after its reaction with silicon.A quantitative analysis of the gas atmosphere during the acidic etching of multi-crystalline silicon with HF -HNO 3-H 2O gave a considerable amount of N 2O as a reaction product as well as NO and NO 2(in the abnce of air or oxygen).21This short lection out of the available literature emphasizes the need for a deeper understanding of
*Address correspondence to this author.Phone:+493514659-694.Fax:+493514659-452.E-mail:j.acker@ifw-dresden.de.
3Si +4HNO 3f 3SiO 2+4NO +2H 2O
(1)SiO 2+6HF f H 2SiF 6+2H 2O
(2)
3Si +4NHO 3+18HF f 3H 2SiF 6+4NO +8H 2O (3)2133
J.Phys.Chem.C 2007,111,2133-214010.1021/jp066348j CCC:$37.00©2007American Chemical Society
Published on Web 01/12/2007
the yet only formally described step of silicon oxidation in the etching process.公务员题库
One approach to address this issue was given in our previous studies about the impact of N(III)intermediates on the etching of silicon in HF-rich etch solutions.22,23It was obrved that etch solutions can turn color into blue or green during etching in the cold.By means of Raman and UV-vis spectroscopy, N2O3was identified to cau the blue color.Furthermore,a complex N(III)species(3NO+‚NO3-)denoted as[N4O62+]is obrved in the solutions,and from a linear relationship between etch rate and[N4O62+]concentration,NO+is consid-ered a reactive species in the rate-limiting step.23The dilution of an etchant caus the complete conversion of all N(III) interm
ediates into nitrite ions as quantified by eq 4.This underlines the importance of the nitrite concentration as a uful and easily measurable sum parameter to characterize the reactivity of etch mixtures.
The obtained correlation between the etch rate and the nitrite concentration provided the first explanation that an etch solution of given concentrations of HF,HNO3,and dissolved silicon can behave in an entirely different manner and apparently as a function of time.Hence,the induction period is the conquence of the enrichment of the reactive N(III)species formed during the initially slow dissolution in freshly prepared HF/HNO3 mixtures.
The prent paper is devoted to the etching of silicon in HNO3-rich,concentrated HF/HNO3mixtures.A clear dif-ferentiation from the previously investigated HF-rich system is done.The prented correlations between etch rate,nitrite concentration,and stirring speed lead to a new insight into the reaction mechanism of isotropic etching in the kinds of solutions.
2.Experimental Section
Analytical grade nitric acid(65%(w/w),14.45M)and hydrofluoric acid(40%(w/w),22.77M)were ud for all etch mixtures reported herein as a volume percentage(%(v/v)).A volume of50mL of etch solution,p
repared by the mixing of certain amounts of HF and HNO3,was placed into wide-mouthed bottles of HDPE(high-density polyethylene)and thermostated during the ries of experiments to the reaction temperature by using a cryostat(Polystat K12-2,Huber Ka¨lte-maschinen GmbH).
The pre-aging of an etch mixture(in order to avoid an induction period by the generation of a sufficiently high amount of reactive species)with a defined silicon content was performed by slow dissolution of small silicon wafer pieces of ap-proximately70mg(boron doped,thickness675µm,resistivity 24-36Ω‚cm)one after another.Becau the next silicon piece was not added until the first one was completely dissolved,the whole procedure of pre-aging up to1.0-1.4g of silicon took 2h.The advantageous side effects are the minimization of an uncontrollable warming up of the mixture during the dissolution as well as loss of SiF4.During the dissolution and the whole ries of experiments,the reaction vesl was covered by a Teflon cap.
For the determination of the etch rate,a10×10mm silicon sample((111)orientation,boron doped,thickness325µm, resistivity10Ω‚cm)was held between the ends of tweezers and immerd in the etch solution for5-180s(t etch).The reaction was quenched by immersing the sample into a large volume of deionized water with ample rinsing thereafter.To perform this ries of experiments under quasi-constant silicon content,the etch time had to be adjusted to the etch rate so that for every etch
experiment only1-7mg of silicon were dissolved additionally.A typical value for such a ries is the total dissolution of50mg of Si during all etch rate measurements in a pre-aged solution of an initial content of1g of Si.The etch rate r in this paper(eq5)is the quotient of the etched silicon layer(in nanometers)and the immersion time(t etch).The removed silicon thickness(∆d)is calculated from the mass loss obtained by differential weighing(A)1cm2;F(Si))2.33 g‚cm-3).
In some cas,a magnetic stirrer(Micro20,H+P Labortech-nik AG)was ud and positioned directly below the reaction vesl(completely immerd into the cryostat cooling liquid). The concentrations of nitrite,nitrate,and fluoride ions in the etch solution were determined by ion chromatography(Deutsche METROHM GmbH&Co.KG)by dilution of an aliquot of 0.5mL of the etch mixture with deionized water to a factor of 1:5000.All species written in squared brackets in the text or in figures denote concentrations in the unit grams per liter.
X-ray photoelectron spectroscopy(XPS)measurements were carried out on a PHI5600CI(Physical Electronics)system using a hemispheric energy analyzer.Typical measurement parameters included an excitation with monochromatic Al K R X-rays at 350W,a measuring area of about800µm in diameter,a pass energy of12eV,and a ba pressure of2×10-8Pa.The samples were measured immediately after drying,however,after
a short(some minutes)transport through air.
3.Results and Discussion
3.1.Etch Experiments.The reliable and preci measurement of the etch rate is only possible in the abnce of an induction period.The addition of a catalyst(NaNO2)may help to avoid the induction period;6,7however,this process does not lead to reproducible concentrations of the N(III)species.A feasible and reproducible way to generate the reactive species for etching is to dissolve a small amount of silicon in a freshly prepared etch solution,an approach that was chon again in the prent study.22The so-called initial etch rates displayed in Figure1
人类文明发展史
([NO
2-])
diluted solution
)([N
2
O
3
]+[NO+]+[HN
2
O
3
+]+
[H
2
NO
2
+]+[HNO
2
])
concentrated solution
(4)
Figure1.Logarithmic view of the initial etch rate as a function of
豁亮the HNO3content.
r)
∆d
t
etch
)
∆m
文化认同感
t
etch
1
A F(Si)
(5)吃什么降血压比较快
2134J.Phys.Chem.C,Vol.111,No.5,2007Steinert et
al.
were determined after 30-50mg of silicon had been dissolved in 50mL of a freshly prepared etchant over a period of 1-5min.
Figure 1shows the plot of the logarithm of the initial etch rate versus the nitric acid content of the etch solutions.The etch rate maximum divides the system into two regions.The region left of the maximum is synonymous to the already mentioned HF-rich etch solutions (dotted curve)that were previously investigated.Here,the etch rate is controlled by the chemical reaction as shown by the strong dependence on the concentration of the intermediate N(III)species.22This is contrary to the finding by Robbins and Schwartz who propod a regime controlled by the diffusion of nitric acid to the silicon surface.5The linear behavior between etch rate and nitric acid content right of the maximum (dashed curve)points toward a diffusion controlled etch regime in good agreement with results by Robbins and Schwartz.5
All studied HNO 3-rich etchants ranging from 60:40to 10:90(%(v/v))HF/HNO 3exhibit a uniform etch behavior with respect to temperature,stirring speed,and concentration of the inter-mediate N(III)species so that all of the solutions are grouped together for further discussion.All results throughout this paper are exemplarily shown for an etch mixture of 50:50(%(v/v))HF/HNO 3.
In the following studies,the nitrite concentration measured in a 1:5000dilution of the original etchant rves as the easily accessible sum parameter reprenting the total content of N(III)intermediates (N 2O 3and NO +)stable only in the concentrated etch solution (eq 4).22,23
The etch solutions ud for the following kinetic studies were prepared by the slow dissolution of a certain amount of silicon in 50mL of the etch mixture.After the last piece had been dissolved,the nitrite concentration was measured and considered as the maximum nitrite concentration,rving as a starting point for the kinetic study.The time-dependent decay of the N(III)intermediates by exposing the etch solution to air,and therefore a decreasing nitrite concentration as a function of time,is ud to investigate the effect of intermediates on the etch rate.The nitrite concentration and the etch rate were determined at the same time in intervals of 1-2h after the starting point.Hence,to follow the proceeding ries of experiments on a time scale,the x -axis of Figure 2has to be read from right to left.
The influence of temperature in the region between -10and 35°C on the etch rate as a function of nitrite concentration is shown in Figure 2.In general,the etch rate increas and the maximum nitrite concentration decreas with rising tempera-
ture.The decrea in temperature yields a stabilization of N(III)intermediates and therefore higher initial nitrite concentrations.The sole exception shown is the ries at -10°C where the maximum nitrite concentration is lower than that at 1°C.This is explained by the relatively low etch rate resulting in a comparably long time for the pre-aging step and the simulta-neously proceeding degradation of the reactive intermediates (e below).The general shape of all curves in Figure 2remains unchanged for all temperatures and can be divided into two regions.Beginning from the maximum nitrite concentration,the etch rate decreas slightly with decreasing nitrite concentration denoted as region 2.At a certain nitrite concentration,different for each reaction temperature,the slope of the function etch rate versus nitrite concentration inverts.In this region of low nitrite concentrations,denoted as region 1,the etch rate increas significantly with a further decreasing nitrite concentration.After the complete decomposition of N(III)intermediates,that is,a nitrite concentration clo to 0g ‚L -1,the etch rate is similar to or even higher than the initial etch rate at the maximum nitrite concentration.
Activation energies in regions 1and 2were determined at constant nitrite concentrations.The activation energy undergoes a significant change from a quite low value of 17kJ ‚mol -1at 25g ‚L -1NO 2-(region 2)to 34kJ ‚mol -1at 2.5g ‚L -1NO 2-(region 1,Table 1).This behavior is explained as a
transition from a regime controlled by diffusion at high nitrite concentra-tions to a regime mainly controlled by the reaction at low nitrite concentrations.As a conquence,the reactive N(III)species are located in excess near the silicon surface and act as a diffusion barrier for the attack of the minority component HF.The results from Figures 2and 3lead to the assumption that the intermediary N(III)species are not the rate-determining species in the studied HNO 3-rich etch mixtures.To clarify this issue,the etch solutions were stirred at different stirring speeds during the etch rate measurement to examine the role of the mass transport on the etch rate.Figure 4shows the effect of stirring on the etch rate as a function of the nitrite content at a constant silicon content ([Si])20g ‚L -1)at 1°C.The lowest curve in Figure 4displays as a reference the etch rate behavior without agitation.For stirring,the general shape of the plotted functions remains unchanged regardless of the applied stirring speed.The etch rate increas with increasing stirring speed
Figure 2.Dependency of the etch rate r from the nitrite concentration at different temperatures (m (Si)dissolved )1.0g,50mL 50%(v/v)HF,50%(v/v)HNO 3
).
Figure 3.Activation energy vs nitrite concentration [m (Si)dissolved )1.0g,50mL 50:50(%(v/v))HF/HNO 3].
TABLE 1:Activation Energies of Etch Rate Determined for Different Nitrite Concentrations
[NO 2-](g ‚L -1)252014107.5 2.5E A (kJ ‚mol -1)17(219(421(423(524(434(4
Silicon Etching in HNO 3-Rich HF/HNO 3Mixtures J.Phys.Chem.C,Vol.111,No.5,20072135
underlining the importance of a forced transport of reactive species to the silicon surface and supporting the model of a process controlled by diffusion.Remarkably,stirring in HF-rich etch solution
s was found to have an opposite effect since the etch rate decread with increasing stirring speed by stirring the reactive N(III)intermediates away from the silicon surface.22 Figure5compares the etch rate as a function of the dissolved silicon content at a maximum nitrite concentration and at a nitrite concentration clo to zero for HF-rich and HNO3-rich etch mixtures.The etch rates decay exponentially with increasing silicon content in the etchant regardless of the nitrite concentra-tion.Figure5exemplarily shows,for a70:30(%(v/v))HF/ HNO3mixture(dotted lines),that the etch rate at a maximum nitrite concentration is considerably higher than that in the abnce of N(III)intermediates.In the ca of HNO3-rich mixtures exemplarily shown for a50:50(%(v/v))HF/HNO3 mixture(dashed lines)in Figure5,the displayed curves for the minimum and maximum nitrite concentration are clo together regardless of the silicon content in the etchant.This underlines the enormous differences in the impact of the N(III)intermedi-ates on the etching of silicon depending on the composition of the etchant.
The surface morphology of each etched sample was deter-mined for a3×3mm area,and the arithmetic average roughness value(R a)was obtained.The surface of initially polished silicon samples(R a)0.7nm)at1°C became,owing to etching,significantly rougher regardless of the composition of the etchant.In HF-rich etch mixtures,an evolution of the surface morphology,in depen
dence on the nitrite concentration, was obrved.23At a nitrite concentration above9g‚L-1,the etched slices exhibit a smooth morphology with an R a value of about0.08µm.Below9g‚L-1of[NO2-],the etch rate of a HF-rich mixture diminishes with further decreasing nitrite content,and the evolution of etch pits results in a roughening of the surface up to0.4µm.In contrast,no changes in surface morphology of etched silicon were obtained in HNO3-rich mixtures yielding R a values lower than0.1µm independent of the nitrite concentration.This finding underlines the polishing property of HNO3-rich etch mixtures.Stirring enhances the etch rate by a forced transport of the reactive species to the silicon surface as shown in Figure4,but the R a value remains comparable to etching in a non-stirred solution.This fact is consistent with the suppod diffusional control of the dissolu-tion mechanism in HNO3-rich etch solutions.
3.2.Decomposition of Nitrite.The N(III)generated during the dissolution of silicon is not stable and decompos by exposing the solution to air.This becomes easily visible since the blue or green-blue colored solutions fade out beginning at the liquid-air interface.22,23As in the previous experiments, the nitrite concentration measured after the last piece of silicon had been dissolved is t as the starting point at zero time for the kinetic measurements.In Figure6,the influence of tem-perature on the nitrite decay for a HNO3-rich etch mixture is shown at a constant silicon content.The higher the t
emperature, the lower is the initial nitrite concentration at zero time,and the decomposition of the N(III)intermediates proceeds faster. The degradation follows a first-order kinetics as shown by the linear slopes for each temperature in the ln[NO2-]versus time plot in Figure6.The calculation of an activation energy from an Arrhenius diagram provided a value of16.6(2.2kJ‚mol-1 which is comparable to the value for the HF-rich system of 20.1(0.9kJ‚mol-1.22
Remarkably,N(III)intermediates exhibit a considerably higher stability in HNO3-rich etch solutions than in HF-rich solutions.Table2compares the rate constants k and half-life time t1/2for the decomposition at different temperatures for a mixture consisting of70%(v/v)HF and30%(v/v)HNO3(A) and a mixture of50%(v/v)HF and50%(v/v)HNO3(B)at constant silicon content.
To emphasize the obrvations,the decay of the nitrite concentration was investigated as a function of the etch bath composition at1°C.As it can be en in Figure7,the mixing ratio of HF and HNO3has a significant influence on the lifetime of the intermediate N(III)species yielding a decrea of the
Figure4.Effect of stirring speed on the etch rate r(ϑ)1°C,50 mL,50%(v/v)HF,50%(v/v)HNO3,m(Si)dissolved)1.0g).
Figure5.Variation of the etch rate r at different nitrite concentrations of the solution in dependence on
the Si content(ϑ)1°C,50mL of etch solution).Figure6.Influence of temperature on the decomposition of nitrite (m(Si)dissolved)1.0g,50mL,50%(v/v)HF,50%(v/v)HNO3).
2136J.Phys.Chem.C,Vol.111,No.5,2007Steinert et
al.
nitrite decomposition rate k(right plot)with increasing nitric acid content.As a conquence,HNO3-rich solutions remain their color for veral days in contrast to the fast decoloration, within1day,for a HF-rich solution.
The intermediate N(III)species were found to degrade by oxidation at the liquid-air interface under the formation of nitrate ions.The formation of nitrate ions follows a first-order kinetics with exactly the same rate constant as the nitrite concentration decays.The molar ratios between the formed nitrate a
nd the decompod nitrite for given time intervals amount to values ranging from0.9to clo to unity pointing to an apparent direct oxidation of the N(III)intermediates.Molar ratios below1indicate the existence of a marginal,parallel pathway of nitrite decay,either by disproportionation or by outgassing of nitric oxides into the gas pha.22
The results in Table3show that the same mechanism is valid for the decay of the N(III)intermediates in the HNO3-rich solution.Stirring the etch solution at500rpm increas the area of the liquid-gas interface,and a higher amount of oxygen accelerates the nitrite decay compared to the nonstirred system by a factor of2.If the oxygen supply to the etch solution is prevented by a tight aling of the reaction vesl,an interruption of the nitrite decay occurs.Such an etch solution holds its color at1°C up to veral weeks.
Although the mechanism of the apparent oxidation from the N(III)intermediates to nitrate ions is yet unresolved,the decomposition of dissolved N2O3(aq)with the generation of NO-(aq)and NO2(aq)(eq6)might be assumed((aq)denotes that all compounds are prent as dissolved species in aqueous media).Nitrogen dioxide or its dimer N2O4(aq)(eq7)decom-pos,by disproportionation in the reaction with water,into nitric and nitrous acids(eq8).Nitrous acid,only stable in a highly diluted solution and at low temperatures and furthermore not detected by Raman spectroscopy,23decompo
s by dispro-portionation into nitric acid and NO(aq)(eq9).Finally,the fraction of dissolved NO(aq)is oxidized by dissolved oxygen near the interface to NO2(aq)(eq10).Hence,the decomposition of the N(III)species can be understood as a formal oxidation of N2O3to nitric acid(eq11).
Figure8summarizes the relationship between the etch rate, that is,the rate of the N(III)species generation,and the decomposition rate for the species.Each point in Figure8 reprents a single experiment in which a certain amount of silicon was dissolved at1°C and the nitrite concentration was measured thereafter.It should be noted that,as a conquence of the different etch rates,the individual points do not correlate to a uniform time scale.Starting with a freshly prepared acid mixture,silicon is quickly dissolved,and a high concentration is established since the decomposition of the intermediates proceeds much slower than the dissolution of silicon.The more silicon an etch solution contains,the lower is the etch rate (Figure5).Then,the maximum in the curves shown in Figure 8is the conquence of a now faster proceeding decomposition than silicon dissolution.With a further increasing silicon content, the maximum nitrite concentration drops becau of a further decreasing etch rate so that,in the same time interval,more nitrite is consumed than generated.
3.3.XPS Study of Etched Silicon.The composition of the etchant has a distinct influence on the etch rate(Figure1)and
TABLE2:Rate Constants(k(min-1))of Nitrite Decomposition(First-Order Reaction)and Half-Life Time a
mixture A mixture B temperature(°C)k(min-1)t1/2(min)k(min-1)t1/2(min) -100.00116300.0008866
10.0012578
80.0020347
150.00242890.0016433
250.00342040.0017408
350.0027257
a A:50mL,70%(v/v)HF,30%(v/v)HNO3,m(Si)dissolved)1.4g. B:50mL,50%(v/v)HF,50%(v/v)HNO3,m(Si)dissolved)1.0g.
Figure7.Decomposition of nitrite in various etch mixtures(50mL etchant,m(Si)dissolved)1.0g,1°C).
TABLE3:Rate Constants k of Nitrite Decomposition (First-Order Reaction)and Half-Life Time for Different Reaction Conditions[50mL,50:50(%(v/v))HF/HNO3,[Si] )20g‚L-1]
reaction vesl covered by Teflon cap跳水兔的做法
stirred solution
(500rpm)
covered by Teflon cap
tightly aled
血泪之路vesl
k(min-1)0.00110.0019 2.596×10-5 t1/2(min)63036526704Figure8.Maximum nitrite concentration in dependence on the etch bath composition and the silicon content(50mL,1°C).
Silicon Etching in HNO3-Rich HF/HNO3Mixtures J.Phys.Chem.C,Vol.111,No.5,2007
2137
on the surface morphology of the silicon wafer after the etching process (structuring or polishing).Mo
reover,if one assumes that the propod model by Robbins and Schwartz of a two-step dissolution mechanism for Si (oxidation followed by dissolution of generated SiO 2)is valid,then the Si surface should be hydrogen terminated after etching in a HF-rich solution and oxygen terminated after etching in a HNO 3-rich solution or for a high N(III)intermediate concentration.Etched silicon slices were studied by XPS,and a complete hydrogen termination independent of the etchant composition was revealed (Figure 9).The investigation of etched Si in a HNO 3-rich mixture [50:50(%(v/v))HF/HNO 3]during the decomposition of N(III)showed also that the surface coverage with Si -H bonds is always prent independently of the nitrite concentration (Figure 10).The same result was received in a former study on HF-rich etch mixtures.23The fact that neither Si -O bonds nor Si -F or Si -O -F bonds (103.3-105eV)could be detected even if nitric acid was prent in considerable excess is surprising.If Si -F bonds were formed during the etching of silicon,replace-ment with Si -OH bonds would occur during water rinsing,and a change in the Si 2p peak should occur.24,25Only a complete hydrogen termination of the silicon surface and its resistivity against oxidation explains why no native oxide was obrved in the Si 2p spectra.
4.Conclusions
The etching of silicon using HF/HNO 3etch mixtures is characterized by an etch rate maximum at 45
%(v/v)HNO 3(Figure 1).The etching process at the HF-rich side is controlled by the chemical reaction by means of the concentration of the reactive N(III)intermediates.22,23The linear dependence between the logarithm of the etch rate and the composition of the HNO 3-rich etch mixture with HNO 3contents higher than 45%(v/v)indicates a diffusional control of the etch process (Figure 1).Contrary to HF-rich etch mixtures,the etch rate is widely independent of the concentration of reactive nitrogen intermedi-ates,expresd by the nitrite concentration (Figure 2).However,at nitrite concentrations lower than 5-7g ‚L -1,a strong increa of the etch rate is obrved.The activation energy ris gradually with decreasing nitrite concentration and increas suddenly when the nitrite concentration is below a critical limit of 5-7g ‚L -1(Table 1,Figure 3).This obrvation is not considered as an indication for a change in the reaction mechanism but rather for a transition from the diffusional controlled etch regime at high concentrations of N(III)[E A )17(2kJ ‚mol -1]to a more and more reactionally controlled dissolution of silicon at lower N(III)concentrations [E A )34(4kJ ‚mol -1].Compa-rable values of the activation energy were determined earlier for the reaction controlled regime of HF-rich etch mixtures as 41(1kJ ‚mol -1for high nitrite concentrations and 44(6kJ ‚mol -1for low nitrite concentrations.22
Stirring experiments support this hypothesis becau the etch rate increas with enhanced agitatio
n of the etch solution (Figure 4).As a conquence,the reactive N(III)intermediates are not involved in the rate-determining step as it was obrved for HF-rich etch solutions.22,23Moreover,they act as inhibitors in HNO 3-rich etch mixtures.Since N(III)intermediates are generated in an electrochemical reaction between Si and HNO 3on the silicon surface,it is very likely to assume that their concentrations are enriched clo to the silicon surface.There-fore,the N(III)species prevent the attack of hydrofluoric acid.Indeed the attack of nitric acid molecules should also be hindered,but this has presumably no significant effect on the reaction rate since the N(III)intermediates and HNO 3(N(V))are both very reactive species.This confirms the results by Schwartz and Robbins for the HNO 3-rich etch regime.7
端午祝福
A remarkable outcome of the prent study is the discovery of the increa of the lifetime of N(III)intermediates with increasing nitric acid content (Figures 6and 7),that is,their dramatically enhanced stability against oxidation by oxygen from air,against disproportionation,and against outgassing of NO x .Although the reason for this behavior is still ambiguous,it can be assumed that the intermediates are stabilized by the formation of complex solution species.The existence of [3NO +‚NO 3-]in HF-rich mixtures and its role as a stable 26and etch-rate determining intermediate 23support this hypothesis.From the prented results,conclusions for an improved industrial etch pro
cess control can be drawn.The behavior of an HNO 3-rich etch bath is quite constant insofar as the silicon content does not change significantly during the etching and the concentration of the N(III)intermediates is higher than approximately 7g ‚L -1NO 2-.If the etch solution enriches with silicon during wafer processing,the etch rate decreas as the acids are consumed.However,the etch behavior remains basically the same within one etch batch as long as the nitrite concentration is above the limit.If it decreas,either by outgassing due to a high bath temperature during etching or by a long stagnation of the production process,the subquent etch process might turn into a less controllable regime.Becau of
Figure 9.XPS Si 2p spectra of etched Si(111)samples as a function of etchant composition.
Figure 10.(a)Correlation among etch rate r and nitrite concentration,and (b)related XPS Si 2p spectra of etched Si(111)samples [ϑ)1°C,50mL,50:50(%(v/v))HF/HNO 3,m (Si)dissolved )1.0g].Selected samples are numerated.
2138J.Phys.Chem.C,Vol.111,No.5,2007
Steinert et
al.
