electrolytic oxidation semiconductor surfaces

Electrocomponent Scienced Technology1977, Vol. ,pp. 225-21

(C) Gordon and Breach Science Publishers Ltd., 1977Printed in Great Britain

ELECTROLYTIC OXIDATION OF SEMICONDUCTORSURFACES

H. L. HARTNAGEL

Department ofElectrical and Electronic Engineering, University ofNewcastle upon Tyne,NE1 7R U. U.K.

(Received December 23, 1976)

Recently a greater understanding of the physics and chemistry of anodic oxide growth on III-V compound semi-conductors has become available. These details are reviewed and critically assessed. With the data avilable now newdevice applications can be considered.

1. INTRODUCTION

The technology of Si is very. advanced. After initialefforts to fabricate SiO2 by electrolytic methods, thiswas abandoned because of the small Faradayicefficiency, particularly when thermal oxidation of Siturned out to be very successful.

Unfortunately, high-temperature processes aredifficult for many of the compound semiconductorswhere a highly volatile component is involved. Indeed,attempts to achieve thermally grown native oxides onsuch materials as GaAs, GaP and InP, have not beensuccessful because decomposition of the semicon-ductor occurs by evaporation of As or P before anyoxide is grown so that the semiconductor surfaces andinterfaces become severely degraded.

It is possible for some of these materials to produceoxide layers by a reasonably-low-temperatureprocess. However, even then the electrical interfaceproperties are not satisfactory. Probably, the semi-conductor surface layer underneath the oxidebecomes severely degraded due to the large differencein self-diffusion coefficients of the componentsinvolved (see pp. 416--421). Some improved resultshave been reported by plasma discharging undersuitable conditions2 when, however, the interfacecharacteristics and the oxide conductivity are stillunsatisfactory for most applications.

The electrolytic oxidation methods were thereforeused on these semiconductors and, under certainconditions, surprisingly high Faradayic efficiencieswere achieved. Further developments since then haveresulted in an oxidation technology with oxidecharacteristics suitable now for many applications.

There are of course numerous applications where a

225

high quality oxide layer is of interest. Firstly one hasthe possibility of replacing the MESFET by aMOSFET4’s in these compound semiconductors forthose areas where the Schottky gate has shownserious disadvantages. MESFETs are easily damagedby high currents applied to the gate, and this isparticularly serious for amplification of higher powerlevels. They are also difficult to operate in integrateddigital circuits because of the bias level requirementto the gate demanding always reverse bias to avoidany damage and because of the relatively high standbypower. This latter point will be specially wellprovided for by normally off inversion-layerMOSFETs.

Secondly, with the new understanding ofcompound-oxide electrolytic growth,6 it is possibleto produce layered structures of high uniformity withcharge injection properties and extremely stable non-volatile memory characteristics. This will be ofinterest for memories of gigabitrate read-out facilitiesby producing MAOS FETs in GaAs13 (Metalaluminium oxide native oxide semiconductorFETs).

Further possibilities exist in the optical field wherethe highly transparent native oxide placed on hetero-junction laser mirrors can be used to act as an anti-reflection coating. Further optical applications willsurely be established by additional experimentation,such as a reduction in any surface recombinationvelocity which is some times the cause for thermalbreakdown of semiconductor lasers.

Native oxides can also be valuable for planarmonolithic circuit production, in a similar manner asSi02 on Si. Or it can be used to passivate any devicesurfaces against corrosion and other long-term effects.

226 H.L. HARTNAGEL

2. THE PHYSICAL PROCESS OF ANODICOXIDATION

A considerable amount of experimental data makes itvery likely that the following process is involved foranodic oxidation. When the semiconductor is insertedinto an aqueous solution with its high dielectricconstant, the surface atoms of the semiconductingmaterial are dissolved into the liquid as their ionicbonds are weakened. The atoms enter the liquid asmultiply positively ionised species, such as Ga /++ orAs //+, leaving on the semiconductor surface animmobile negative charge which has to be neutralisedby mobile holes from the bulk of the semiconductor.After a second electrode is placed as cathode electrodeinto the solution, the semiconductor can be employedas anode with a positive potential which aids the holeflow to the semiconductor surface so that more ionscan be released. These ions travel towards thecathode.

If sufficient ions are released they form a denseionic cloud in front of the anode. This is particularlyeasy if either the solubility or the diffusion coeffi-cient of these ions is reduced by the introduction of asuitable liquid into the aqueous solution. It wasfound that some liquids are particularly effective,such as glycol.3

The ions in this cloud will capture OH- ions of theaqueous solution, and, with the simultaneous forma-tion of H2 O, are oxidised. It is likely that most ofthem will form well-known oxides such as Ga2 O3 orAs 03, which will exceed a density saturation levelin this cloud so that they begin to precipitate againstthe semiconductor. An oxide layer begins to getdeposited thus.

Further growth is now only possible if ionictransport takes place across the oxide. Each of theionic species has a threshold field for drift and aspecific value of mobility) With normal currentdensities applied across the semiconductor surfacewhich is immersed into the electrolyte, the fieldsacross the oxide and its interfaces are sufficiently highfor the motion of all ions, namely OH- and thepositively ionised components of the semiconductor.However as the mobility of OH- is much smaller thanthat of the multiply ionised components, the primarygrowth occurs at or near the electrolyte-oxide inter-face. Unfortunately, the current density cannot bereduced beyond a minimum value which is differentfor each material. This value has sometimes beencalled the dissolution current as it seems to agree withthe equivalent cation current due to the dissolutionof the oxide in the electrolyte when no external

current is applied. It can be understood as the flow ofthose cations which, after dissolution from the oxide,successfully leave the anode region. Only when thisminimum current density is exceeded, can growthtake place. This means that with the electrolytesknown so far, native oxides on most III-V CompoundSemiconductors such as GaAs, can only be producedafter exceeding the drift threshold for all the ionicspecies involved.

However, AI and some other metals have a verysmall dissolution current. After depositing a layer ofsuch a metal by evaporation on top of the semi-conductor, this metal will first be oxidised in theelectrolyte. After it has been completely transformed,the oxidation of the semiconductor begins. For ahigh current density, the semiconductor componentsdrift as cations across the Alz 03 and native oxide ofthe semiconductor appears on top of the AI2 O3. Onthe other hand, for a low current density the semi-conductor components cannot become mobile, andit is only OH-which crosses the AI 03. Correspond-ingly, native oxide of the semiconductor appears onlybetween A12 03 and the semiconductor. It is there-fore possible to produce various types of layeredstructures. It is useful to consider this now systema-tically for various types of device applications.

3. QUANTITATIVE ANALYSIS OF OXIDEGROWTH

The oxide growth can oe undertaken either with aconstant-current source or with a constant voltagesupply in series with a sufficiently high seriesresistance in order to limit the initial current density.A voltage supply without such a limiting resistanceis not advisable as the initial current density caneasily become so high that hot spots and otherdetrimental effects can occur during the initial stageof growth so that a deterioration of the oxidequality appears. A constant-current case wouldrequire a voltage measurement between cathode andanode in order to monitor the growth procedure. Avoltage and series-resistance arrangement can enableone to monitor the growth by measuring the currentflowing across the electrolytic cell. For the relativelyhigh final growth voltages involved across most of theoxides of device interest, it is not required to under-take the more refined electrochemical potentialmeasurements as these would then be normally anegligible fraction of the applied voltage.

It is possible to model the growth behaviourrelatively satisfactorily by a simple treatment,

ELECTROLYTIC OXIDATION 227

5o

25O

350

5O

0.15

i00 200 300 400 500 600 700 800 900

t/sec.

FIGURE Normalised anodisation current density ] vs. time t. is initial current density, is mA/cm theparameter is the externally applied voltage in volts.

provided that precautions are taken for a uniformcurrent density across the semiconductor surfacewhich is exposed to the electrolyte. By way ofexample, this is to be outlined here for the constant-voltage plus series-resistance arrangement. The treat-ment of the other case of growth is very similar.

The general current functions measured are shownin Figure 1. A typical experimental set up is given inFigure 2, where it is also indicated that a light source

is needed in order to have sufficient holes availablefor the neutralization of the negative surface chargecreated by the removal of positive cations. Thisillumination is particularly important for n-typematerial where practically no holes would be presentotherwise. Failing to provide a sufficient supply ofholes means that the space charge layer set up by thenegative surface charge caused by the departingcations absorbs a substantial part, if not all, of the

228 H.L. HARTNAGEL

FIGURE 2

YB Rs

Pt cathode

Experimental set up for anodic oxidation

VB bias supply.R series resistance.

Light source

C capillary tube carrying conductor to ohmic contact on back of semiconductor slice.S semiconductor slice deposited by suitable wax or photoresist.

externally applied voltage 6 so that the oxide thicknesswill be much smaller than that predicted by themodel outlined below.

The results of Figure can be understood as anoxide growth rate which is proportional to theanodisation current minus an approximately constantdissolution rate. The Kirchhoff loop for thevoltages gives the following equation:

dVB IR + V +f Edx + IR B + Vs (1)

o

where VB is the supply voltage,/the current flowingthrough the series resistance R s, Vr the rest potentialof the anodisation cell, E the electric field across theoxide of thickness d, RB the resistance of the bath,the semiconducting chip and of any other circuitpart not included yet, and Vs is the potential of thespace charge layer set up by the negative surfacecharge introduced above. Vs could also include anyother space charge layer of the semiconductor, suchas possible with a non-ohmic back contact. Let usassume that Vs 0 and E is constant across theoxide. Particularly the latter point is not alwaysvalid. However, as the agreement between a theore-tical behaviour based on a constant E agrees wellwith the experimental findings when the normalgrowth into a semiconductor without overlaying

metal films is used, we can argue that this assumptiontoo should be valid. It does however seem to breakdown with some of the multiple-layer structure caseswhen probably the ionic charge drifting across theoxide results in a substantial accumulation withstrong field distortion, particularly at the change-overof growth from one type of material to the next one.The voltage changes across the oxide-semiconductor-ohmic contact structure observed then are too largeto be explained by any Schottky-barrier space chargelayer between the overlaying metal and the semi-conductor.

Eq. (1) can then be rewritten:

V V =I(R +R) +ed (2)

where d is determined by the total electronic chargeused to create the oxide thickness, i.e,:

d =A f (j /d)dt (3)o

Here,A is defined as the increase in film thickness perunit charge passed through unit area,/is the currentdensity passing through the oxidising semiconductorsurface up to time t, and/d is the dissolution currentdensity. The differential equation resulting fromEqs. (2) and (3) has the following solution:


VB_V]=

(R +RB)S

where

--Jd] e-tit +Jd

(R + R a)Sr

EA (5)

and S is the surface area of the semiconductor exposedto the electrolyte. Eq. (4) has been found to fitextremely well with experimental findings except forthose cases described above where the constant- Eapproximation does not hold any more.

The inverse value ofE is therefore also the valueof the thickness of oxide produced per voltageapplied. This is a value which is characteristic for aparticular material and changes with current densityemployed for the growth. For GaAs it has forexample been found to vary linearly between 23 ./Vand 19 A/V for low current densities of about0.01 mA/cm: to 15 mA/cm: respectively. This can becompared with the classical model of Mott and Cabrera,where the ionic current density ]i across the oxide isgiven by

ji=a exp ( Q-XE)_k(6)

with a being a constant, Q the activation energy forthe drift of ions, q the ionic charge, X the averagejump distance of the ions from one rest well to thenext and k and T Boltzmann’s constant and theabsolute temperature respectively. The significanceof a is described by various authors (see for example). a contains the density of ions in an energeticallyfavourable position to jump and could possibly beused to derive further useful information on ionictransport across the oxide. The terms of Eq. (6)are primarily given by those ionic species beingresponsible for the largest contribution to ]i. As forOH-both Q and q are small, this equation is onlyaffected by these ions when the current density isvery small, i.e. when E is small. Rewriting Eq. (6), i.e.

ln/_’/= qXE ____Qa kT kT’

one finds from a graph of the experimental growthcurrent density on a logarithmic scale as a function ofE (linear scale) that the above indicated experimentaldependence ofE does indeed follow this relation (seeFigure 15 of a).

This form of Eq. (6) shows also that one canexpect a behaviour as indicated by Figure 3 if oneincludes the recently established case of oxide growth

due to OH- drift only as described above at the endof the previous chapter. There one sees that the slopeis determined by OH- ions when E is small, but thatthe cations with high q and Q are responsible for thelarger slope at high E ranges. Unfortunately detailedexperimental results are so far only available for thehigh-field case.

Provided that a particular material satisfies theabove assumptions, especially regarding the constancyof E, the decay of current as shown by Figure ischaracteristic of a given material, i.e. measured valuesofE and A of Eq. (5) give evidence on thecomposition of the material. This is firstly of interestwhere for example by liquid epitaxy ternarycompound semiconductors are produced. E and Agive easy evidence on the ratio of the materialsinvolved. With caution, this can also be employed tomeasure interdiffusion of thin films on semi-conductors,a Here one of course has to be awarethat a particular metal might form hardly any oxideelectrolytically (e.g. In) possibly due to a very highdissolution rate, whereas a mixture of it with someother material does generate a beautiful oxide (forexample InP does generate useful native oxides,9 inspite of the fact that the Pourbaix diagram o forphosphorous does not indicate any area where anoxide formation should occur as all the P is used toform some solution like HaPO4).

The growth rate of the oxide (e.g. around20 A/Volt for GaAs) and the consumption rate ofthe material (e.g. 14 A/V for GaAs) give a clearmeasure of the thickness of a particular layer. Thecurrent decay or voltage increase (depending onwhether the constant voltage cum series resistancearrangement is used, or the constant current source)will remain the characteristic function until all thematerial of a particular layer has been consumed andsome new material with a different characteristicfunction arrives. The change from one type offunction to another one is often very marked sothat the voltage absorbed by the oxide produced givesa clear indication of both the new oxide thickness andthe thickness of the layer consumed.

An estimate of the density of the oxide producedis useful. This is now possible with GaAs-oxide byemploying the data available. The total electroniccharge Q per surface area employed to grow an oxidethickness defined by the applied voltage VB (using theabove quoted growth rate of 20 A/V) is given by anintegration of Eq. (4), i.e.:

230 H.L. HARTNAGEL

FIGURE 3 Theoretically expected ion current density versus oxide field. E kT/eh of Equation (6),e -electronic charge. For high E, the larger value ofq for cations causes the current to rise steeper than for small E.

where EA 3.2 x 102 cm2 F-1 as obtained from suchcurves as shown in Figure 1. This gives the number ofelectrons per surface area and per voltage used,namely n 1.65 x 1016 cm-2 V-1. For the consump-tion rate of GaAs, 14 A/V, there are, on the otherhand, 30.9 x 1014 Ga atoms per surface area and pervoltage applied, and the same amount of As atoms.Assuming that both types of atoms are tripley ionisedthis would mean that 1.85 x 1016 electrons/cm2Vwere freed, a number which is about 12% higher onlythan the total number of electrons forming thecurrent causing this anodic process. This smalldifference is likely to be caused by experimentalinaccuracies.

Some useful data was obtained recently byVerphnke and Tijburg on the composition of thenative GaAs oxide produced by the glycol solutiondescribed here, and on the amount of Ga and As lostinto the electrolyte. These results were obtained byradioactive tracer techniques after GaAs was exposedto a neutron flux for two hours so that the materialbecame radioactive. When the anodic growth wasundertaken subsequently, the radioactive intensities

of the electrolyte after growth and of an etch solutionfor partial removal of the native oxide could bemeasured by a Ge (Li) detector and simihr facilities.The recorded activities are then analysed to give thetotal numbers of Ga and As in these solutions.

It was then found that 25% of Ga and 5% of As islost into solution. They also found that the Asconcentration in the oxide is larger than the Gaconcentration, except near the surface with theelectrolyte where Ga predominates. This seems toindicate that the mobility of Ga ions is higher thanthose of As so that the latter is ionised by combiningwith OH-somewhere away from the surface. It islikely that this method can be used to determine theionic mobilities for the anodic processes consideredhere.

Using the remaining Ga and As in the oxide andassuming that the oxide consists of Ga20a andAs20a, the density of it can be calcuhted to be4.23 g/cma where the oxide thickness value usedfor this calculation is rehtively reliable as it is basedon thickness profile measurements with a SloanDektak.


The loss data of Verplanke et al indicates also thatthe current efficiency for the anodic oxidation is85% for an average of both cationic species. This isonly slightly smaller than the previously publishedefficiency3 which was based on very rough estimatesof data which was not yet as accurately availablethen as nowadays.

4. CONCLUSIONS

The anodic oxidation methods available now and thelevel of understanding of the observed effects make itpossible to envisage applications in numerous fields ofsemiconductor electronics. Further studies are ofcourse needed in order to obtain really optimisedconditions, such as a study of the drift mobility forthe relevant ionic spedes.

ACKNOWLEDGMENTS

The author is grateful to R. Tijburg of the Philips ResearchLaboratories in Eindhoven for permission to quote theirexperimental results before their publication has appeared.Gratitude is also expressed to both the United KingdomScience Research Coundl and the European Research Officefor financial support for the various projects on anodicoxidation under way in the author’s laboratories. Finally hewould like to thank those of his research colleagues fornumerous useful discussions who are working in this field.

REFERENCES

1. H.C. Casey, Jr., "Diffusion in the III-V compoundsemi-conductors", Chapter 6 in Atomic Diffusion in

Semiconductors, editor D. Shaw, Plenum Press, London,1973.

2. R. P. H. Chang, A. K. Sinha., "Plasma oxidation ofGaAs",Appl. Phys. L., 29, July 1976, p. 56.

3. H. Hasegawa, H. L. Hartnagel., "Anodic oxidation ofGaAs in mixed solutions of glycol and water", J.Electrochem. Soc., 123, No. 5, 1976, pp 713-723.

4. B. Bayraktaroglu, E. Kohn, H. L. Hartnagel., "Firstanodic-oxide GaAs MOSFETs based on easy techno-logical processes",Electronics Letters, 12, No. 2, 1976,pp. 53-54.

5. B. Weiss, E. Kohn, B. Bayraktaroglu, H. L. Hartnagel,"Native oxides on GaAs for MOSFETs Annealingeffects and inversion-layer mobilities", 1976 Interna-tional Symposium on Gallium Arsenide and RelatedCompounds, Edinburgh, September 1976, Proc. to bepublished.

6. A. Colquhoun, H. L. Hartnagel., "Studies of n-typeGaAs material properties by anodic current behaviour",So#d-State Electronics, 19, 1976, pp. 819-826.

7. D. A. Vermilyea, Kinetics orion Motion in AnodicOxide Films, p. 328, Conference on Non-CrystallineSolids, N.Y. September 1958, Proceedings by J. Wiley,N.Y. 1960, Editor" V. D. Fr6chette.

8. A. F. A. B. E1-Safti, B. L. Weiss, H. L. Hartnagel,"Analysis of multicomponent thin films on GaAs byanodic processes", Electronics Letters, 12, No. 13,1976, pp. 322-324.

9. A. Colquhoun, H. L. Hartnagel., "Anodic oxidation o findium phosphide in glycol based solutions", to appearin Surface Technology, 1977.

10. M. Pourbaix, Atlas ofElectrochemical Equilibria inAqueous Solutions, Oxford, Pergamon, 1966.

11. J. C. Verplanke, R. P. Tijburg, "The analysis of anodicoxide films on GaAs and GaP by means of radio activetracer techniques," Private Communication.

12. B. Bayraktaroglu, S. Hannah, H. L. Hartnagel., "Controlof AI O3 position in anodic GaAs native oxide",to appear in J. Electrochem. Soc., 1977.

13. B. Bayraktaroglu, S. Hannah, H. L. Hartnagel., "Stablecharge storage of MAOS diodes on GaAs by new anodicoxidation", Electronics Letters, 13, No. 2, 1977,pp. 45-46.

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttp://www.hindawi.com Volume 2010

RoboticsJournal of

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014


Active and Passive Electronic Components

Control Scienceand Engineering

Journal of



RotatingMachinery


Hindawi Publishing Corporation http://www.hindawi.com

Journal ofEngineeringVolume 2014

Submit your manuscripts athttp://www.hindawi.com

VLSI Design



Shock and Vibration


Civil EngineeringAdvances in

Acoustics and VibrationAdvances in



Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation http://www.hindawi.com

Volume 2014

The Scientific World JournalHindawi Publishing Corporation http://www.hindawi.com Volume 2014

SensorsJournal of


Modelling & Simulation in EngineeringHindawi Publishing Corporation http://www.hindawi.com Volume 2014


Chemical EngineeringInternational Journal of Antennas and

Propagation




Navigation and Observation



DistributedSensor Networks


electrolytic oxidation semiconductor surfaces

Documents