Thermal Annealing Of Rf Sputtered Nbn

IEEE TRANSACTIONS ON MAGNETICS, VOL. MAG-23, NO. 2, MARCH 1987

847

THERMAL ANNEALING OF RF SPUTTERED NbN W. L. Carter and E. J. Cukauskas Electronics Technology Division Naval Research Laboratory Washington, DC 20375-5000

temperatures ranging from 150°C to 65OoC have been annealed for times ranging from one sec to 12 hours. The vacuum system and procedures used to deposit the The effects of thermal annealing on RF sputtered films have been described i n . detail el~ewhere.~ In NbCN films and completed NbN-based tunnel junctions brief, preheating, presputtering and sputter etching has been investigated. The starting films were RF steps precede film deposition in a premixed reactive sputtered at temperatures from 150'C to 650°C. After a vacuum anneal at 600°C for six hours the transition Ar/N2/CH4 sputtering gas. The resulting NbCN films have been characterized by measurements of transition temperature reached a maximum of 14.6K and depended temperature Tc and width ATc, resistivity p , and only on the carbon concentration, for those films resistivity ratio RR using standard 4-wire and 2-can deposited at 200°C and 4OO0C. Vacuum annealing for cryostat techniques which have also been described in one hour at temperatures of 900°C or more caused the detail elsewhere.8 X-ray diffractometry has been used formation additional of phases. Rapid thermal to characterize the structure of the films before and annealing, RTA, at temperatures from 600°C to 1200°C for times from one to 100 sec resulted in increased Tc Two typesofthermalannealing afterannealing. treatments were given to the films. with increased annealing temperature. The maximum Tc produced by RTA was 16.6K. The lattice parameter of Vacuum Annealing annealed 6-phase NbCNwasdecreasedtowardthe accepted value of 4.41A. Resistivity of the annealed NbCN films received thermal anneals under vacuum films decreased except when the films cracked during for times ranging from one to 12 hours at temperatures annealing. RTA for 10 sec at 750°C of a completed from 60Q°C to 1200°C. Six and 12 hours anneals at NbN/Si/NbN edge junction increased the electrode gaps 600'C and 10-6 Pa were done in the same vacuum system butforNbN/Si/Nbplanarjunctionsthesumgap decreased. Microshorts also appeared in both types of as were the film depositions. The films annealed at 600'C were deposited on quartz substrates at different junctions after annealing. substrate temperatures and CH4 concentrations in the Introduction sputtering gas in order to determine the effects of lowtemperatureannealingonawidevarietyof Attempts to develop a high quality all refractory differentNbCNstartingmaterials. In additiona high Tc Josephson junction technology based on NbN series of anneals were done at temperatures of 900, represent one of the most active research areas in 1000, 1100 and 1200°C for one hour inUHVa MBE system at pressures of 10-7 Pa or lower. Details of this superconducting microelectronics. Several researchers system can be found elsewhere.9 The film used in this have made considerable progress in this in area recent years.1-5 study was deposited onto a sapphire substrate which At NRLweareinvestigatingNbN/Nband was subsequently broken into four pieces for NbN/NbNjunctionswithsemiconductorbarriersfor annealing. applicationas SIS mixers. Two problemareasin current NbN based all refractory junctions are the width A V of the rise in quasi particle current at the Rapid Thermal Annealing sum of tfe electrode energy gaps and the magnitude of Rsg, the subgap resistance. Large AVg and small hg The second type annealing of which was result from degraded superconductive properties of the investigated was rapid thermal annealing, RTA. In this caseannealsrangingfrom600°Cto1200°Cwere NbN at the junction interfaces as well as from poor barrier material properties. We have also seen performed in one atmosphere of N2 Arorafter a20 min indications of non-ideal superconductive properties in purge for times ranging from one to 100 seconds. RTA the bulk through specific heat measurements made at is used in semiconductor processing, for. example, to Stanford University and in upper critical field, Hc2, anneal out local implant damage without allowing large measurements made at NRL.6 X-ray diffractometry scaleredistributionoftheimplanteddopants. In indicates our films, while single 6-phaseY have a this case it was expected local ordering of the high Tcfcc 6-phase NbN could be accomplished without largerthan expected lattice parameter ao, and thatthe and (200) significantamountsoftheequilibriumhexagonal distorted latticein (111) reflections give a, which differ typically by 1% or c-phase NbN being created. RTA was performed in a more. Since appropriate thermal annealing treatments Heatpulse 210 unit from AG Associates in which high can improve the superconductive properties of temperaturesandfasttemperaturerisetimesare materials, we are investigating the effects of thermal achieved by means of thirteen 1.5 kW quartz halogen annealing on NbNfilmsandNbNbasedjunction lamps. The samples used for RTA were made in four structures. depositions, one at 150°C onto silicon and oxidized silicon substrates, one at400'C onto silicon, one at Thermal Annealing of NbCN 4OOOC onto sapphire and one at 65OoC onto oxidized silicon. The film grown at 150'C was similar to those NbCN films of thickness 250 to 500 nm have been used as counter electrodes in our NbN/NbN junctions deposited by RF sputtering on substrates of sapphire, while the film grown at 650°C was similar to films used as base electrodes in our junctions.10 quartzandoxidizedandunoxidizedsilicon,and subsequently annealed. Films deposited at substrate Abstract

Manuscript received September30, 1986 0018-9464/87/0300-0847$01.0001987 IEEE

848

Resuits and Discussion

111

10 SEC RTA

Vacuum Annealing Results Films deposited at both 200'C and 4OO0C on quartz substrates have been given thermal anneals at 6OO0C under vacuum in the 10-6 Pa range for periods of six and 12 hours. Fig. 1 shows the Tc dependence of the films before and after annealingon the concentration of CHq in the sputtering gas during deposition. For

I

1 1

I

AS AS DEPOSITED DEPOSITED

2 %%

X 0 X

x-

AA

2bO"C ON Si o 400°C ON Si A

13 0

0 T ~ 2 0 0 " C 1 0 %N2 AT^ 400°C 13% N2

2

31

400

600

800 1000 1200

ANNEALING TEMPERATURE ("C)

15

,.a*---

14

2

200

Figure 2. The RTA temperature dependence of Tc for similar films deposited on different substrates at different temperatures. The error bars represent ATc when larger than the plotting symbol.

dp'-\/

on Auger materialshadsimilarcompositionsbased composition analyses from previous depositions.8 The Tc of the annealed films increased with increasing , I I annealing temperature. The minimum RTA temperature 0 2 that caused Tc to increase was less than 600'C for PERCENTCH, films deposited at150'C but was 8OO0C for films grown at 400°C. The maximum Tc achieved by RTA was 16.6K, as high as the best material we can make by RF bias Figure 1. The transition temperature of vacuum sputteringat 850'C substratetemperatures.3RTA annealed NbCN films as a function of the percentage of above 900°C usually led to increased A Tc. Another CH4 inthesputteringgas.Ts is thesubstrate problem with RTA was that films occasionally did not temperature during deposition. adheretothesubstrateorcrackedduringthe annealing. The films deposited at 150'C on silicon and oxidized silicon were particularly susceptible to films deposited at 2OO0C a six hour anneal caused Tc this problem. This may be due to stress in the film to increase -0.8-1.3K to a maximum of 14.6K. For due to the low deposition temperature. Films thosefilmswhichoriginallyhadunusuallylarge 4OO0CadheredQelltothesubstrate transition widths, annealing sharpened the transition. depositedat during RTA. However, the films deposited on silicon Other films which began with sharp transitions (ATc< at 400'C cracked when annealed at 1000°C or more for .05K) showed little or no increase inATc. A second timesofthreeormoreseconds.Thiswaseasily anneal at 6OO0C for six hours (not plotted) caused determined by an increase in the room temperature additional increases in Tc of up to 0.2K arid generally resistivityofthosefilmsafterannealing.The slight increases inATc-as well. For films deposited resistivity of films which remained intact initially at 4OO0C the Tc after annealing had the same CH4 decreasedandthenwasconstantwithincreasing concentration dependence as for the films deposited at annealingtemperature,reachingabout 90% ofthe 20OoC. Thus the Tc after annealing at 600'C depended original resistivity after RTA at 900°C or more. on the substrate only on the stoichiometry and not temperature. The crystal structure of these films has The lattice parameter and grain size of the film been investigated using X-ray diffractometry. We have respectively. observed previously that our as-deposited films have a after RTA are plotted in Figs. 3a 3band The error bars in Fig. 3a represent the difference in distortedfccstructurewithanunusuallylarge lattice parameter a0.3,8,10 After annealing at 6OO0C . a calculated from the (111) and (200) reflections. The (111) reflection had the larger a0 in almost all all films showed a decrease in a, toward a 'more cases. As can be seen in Fig. 3a, RTA both decreased accepted value for NbCN and also a decrease in the the lattice parameter to the expected value of 4.4U apparent distortion of the fcc lattice. No traces of and decreased or even eliminated the distortion of the additional phases were observed. lattice. In generalfilmsremainedsingle&-phase afterRTA.Thegrainsizeasdeterminedbythe Pieces of another film deposited at 4OO0C were Scherrer formula and plotted in Fig. 3b increased with annealed for one hour at temperatures ranging from 900°C to 1200'C under UHV in the 10-7 Pa range or increasing annealing temperature above about 75OoC. lower.X-raydiffractometryrevealedthatallthe The time dependence of RTA was also investigated. films had multiple phases after annealing. Only the A series of rapid anneals for times ranging from one film annealed at 900°C appeared to remain largely S -phase NbCN with Tc increased to 16.lK. This is in to 100 sec at 1000°C was performed. There was a slight agreement with other results that indicate NbN is increaseinTcandingrainsizewithincreased metastable below about 127OOC and cannot be annealed annealing time. However, the lattice parameter showed higherthan-900°Cwithoutnucleatingequilibrium no dependerice on annealingtime.Oneinteresting point was that the film annealed for one second did phases. 11 12 not crack during annealing while those films annealed RTA Results for longer times were cracked. This may simply mean that the one second annealing time did not actually In Fig. 2 the transition temperatureTc of NbCN allow the film to reach the nominal temperature. The 10 secisillustratedasa filmsafterRTAfor thermal environmentsof the saniple and thermocouple in function of the annealing temperature. The starting the Heatpulse 210 are slightly different. The -AS DEPOSITED ANNEALEDAT 600°C 6 HR. I I I I I I I 4 6 8 10

-___

J

-

a49

4.48

It

- 4.46-

-4

@

n

E

2

s

2UI

0

E 4

T

I

d

-

4.44

4.42 -

o AI203 o Si

DEPOSITED AS

4.40 -

'--...I

go 0

4.38 I

I

I

200400600

I

I

800

I

10001200

ANNEALINGTEMPERATURE

I

("C)

VOLTAGE (mV)

Figure 4. The differential resistance dV/dI of NbN/Si/NbN edge junctions before and after RTA. Note the increased gaps after annealing.

(4

thebarrier.Inallcases , however,cleargap structure was apparent. The minima in the dV/dI-V at the individual electrode gaps have moved to higher 4 voltages after RTA indicating an improvement in the quality of the NbCN within a coherence length of the 160 junction interfaces in both electrodes. The counter 5 electrode gap improved from 1.80 to 1.92 mV and the base electrode gap improved from2.66 to 2.82 mV. The 0 improvement in the base electrode, which was deposited - AS DEPOSITED X underoptimum RF biasconditionsat 85OoC, was / \ unexpected at the 750'C RTA temperature. This indicates that considerable damage was done to the 801 surfaceofthe NbN baseelectrodeinsubsequent processing steps, either in plasma etching the edge or I I I I I I duringsputtercleaningandbarrierandcpunter 0 200 400 600 800 1000 1200 electrodedepositions. Us? ofaplanartechnology wouldalleviatetheformerpossibility.However, ANNEALING TEMPERATURE ("Cl similar RTA of NbN/Si/Nb planar junctions did not produce improved gap values. In fact the sum gap in (b) 4.0 mV to 3.85 mV after this case decreased from annealing. This decrease was apparently in the Nb The RTA temperature dependence of (a) the Figure 3. gap. In both the planar and edge junctions the width lattice parameter and (b) the grain size after 10 sec of the current rise at the sum gap decreased and the anneals. The error bars in (a) represent the difference between .a calculated from the (111) and subgap leakage increased during RTA. We also made (200) reflections. planar NbN/Si/Nb junctions on NbN bases produced in the same depostion but after subjecting one of the 900°C for 10 sec.Thejunction basestoRTAat 4OO0C to 1000°C of the temperature rise time from barriers and counter electrodes were deposited in the samerun.Therewas nosignificantdifferencein thermocouple assembly was two sec. The time constant for thermal equilibrium in the substrate was estimated thesejunctions.Thussurfacedamageduetothe sputter cleaning and barrier and counter electrode 0.1 secat 1000°C. Onesamplewasalso tobe deposition may be the limiting factor in our present 10 sec at 1000°C and annealed in flowing argon for junction fabrication technology. showed properties identical to those of the sample annealed under similar conditionsin nitrogen. Conclusion RTA of Tunnel Junctions Wehaveinvestigatedtheeffectsofthermal annealing of €3 sputtered NbCN films and completed We have investigated the effects of RTA at 750'C NbN-based tunnel junctions. Vacuum annealing at 600'C for 10 sec on several completed junction structures for six hours increased the transition temperature of whosefabricationhasbeendescribedindetail films deposited at 200°C and 4OO0C to a value which elsewhere.13 This temperature was chosen since NbN counterelectrodefilmsdepositedat150°Cshowed depended only on the carbon concentration. For films increased T, after RTA at 75O0C without substantially grown at 400°C vacuum annealing at temperatures of increased grain size. Large increases in grain size 900°C to 12OO0C for one hour caused the onset of would presumably lead to punching through the barrier. second phase formation. However, the T, of the film Annealing at 75OoC would also cause the amorphous annealed at 900°C was significantly improved. This silicon barrier to transform to polysilicsn provided may be of some importancein producing NbN conductors the reaction is not kinetically limited. Fig. with high critical current densities. Rapid thermal 4 shows annealing of similar NbCN films grown at 150°C and dV/dI-V characteristics of NbN/Si/NbN edge junctions both before and after RTA at 750°C for 10 sec. While 400°C resulted in increased T, with increased annealingtemperature.Resistivitydecreasedafter these are not the same junction they are from the same The annealing treatment created microshorts in RTA. The maximum Tc produced by RTA was 16.6K while chip.

200

I

fi

Ig

f

~

I

-

'

850

the maximum Tc after vacuum annealing was 16.1K. W e 10. attribute thisdifference to the lack of second phase formation due to kinetic limitations during RTA. The lattice parameter of both vacuum annealed and rapid thermal annealed&-phase NbCN was decreased toward the accepted value of 4.418. The distortion of the fcc 11. structure seen in our films was decreased and often eliminated by annealing. RTA of completed NbN/Si/NbN tunnel junctions at 75OoC for 10 sec increased both electrode gaps, but for NbN/Si/Nb junctions the sum In both cases, however, gap was decreased. 12. microshorts appeared after annealing. RTA at 900°C for 10 sec of a NbN base electrode deposited at 400'C before completion of the junction did not increase the sum gap. This indication an is that our 13. barrierlcounterelectrodedepositiontechniqueis limiting our junction quality. Further RTA studies are in progress. Acknowledgements The authors wish to thank John Claassen for the high temperature vacuum anneals, Carl Vold for use of his x-ray diffractometer, Phillip Thompson and Harold Hughes for use of the rapid thermal annealer and MartinNisenoffformanyusefuldiscussions.This work was partially supported by the Office of Naval Research. References 1.

A. Shoji, M. Aoyagi, S. Kosaka, F. Shinoki and H. Hayakawa, "Niobium Nitride Josephson Tunnel Junctions with Magnesium Oxide Barriers," Appl. 1098-1100, 1985. Phys. Lett, 5,

2.

J. C. Villegier, L. Vieux-Rochaz, M. Goniche, P. Renard and M. Vabre, "NbN Tunnel Junctions, IEEE Trans. Magn., MAG-21, 498-504, 1985. "

3.

E. J. Cukauskas and W. L. Carter, "Superconducting and Structural Properties of RF Magnetron Sputtered Niobium Nitride Josephson for Junctions," Adv.CryogenicEng., 3 2 , 643-650, 1986.

4. J.

Talvacchio and A. I. Braginski, "Tunnel Junctions Fabricated from Coherent NbN/MgO/NbN and NbN/Al203/NbN Structures," this volume.

5.

H.G. LeDuc, J. Stern, S. Thakoor and S. K. Khanna, "All Refractory NbN/MgO/NbN Tunnel Junctions," this volume.

6.

W. L.

7.

E. J. Cukauskas, "The Effects of Methane in the DepositionofSuperconductingNiobiumNitride ThinFilmsatAmbientSubstrateTemperature," J. Appl. Phys, 54, 1013-1017, 1983.

8.

W. L. Carter, E. J. Cukauskas, S. B. Qadri, A. S. Lewis and R. J. Mattauch, "Effects ofRF Bias on the Superconducting and Structural Properties of RF Magnetron Sputtered NbN," J. Appl. Phys., 2, 2905-2907, 1986.

9.

S. A. Wolf, S. B. Qadri, J. H. Claassen, T.L. Francavillaand B. J. Dalrymple,"Epitaxial Growth of Superconducting Niobium Thin Films by J. Vac. Sci. UltrahighVacuumEvaporation," Technol., &, 524-527, 1986.

Carter, to be published.

E. J. Cukauskas, W. L. Carter and S. B. Qadri, "Superconducting and Structure Properties of Niobium Nitride Prepared by RF Magnetron Sputtering," J. Appl. Phys., E, 2538-2542, 1985. A. M. Cuculo, L. Maritato, A. Saggeseand Vaglio, "Properties of Niobium Nitride-based Josephson Tunnel Junctions," Cryogenics 45-47, 1984.

R.

6,

E. K.

Storms, "Inorganic Chemistry Series One, SolidStateChemistry,"Butterworths,London, Vol. 10, p. 37, 1972.

E. J. Cukauskasand W. L. Carter,"Niobium Nitride Based Josephson Junctions with Unoxidized Silicon Barriers," this volume.