Novel Cs-Free GaN Photocathodes

NEERAJ TRIPATHI,1 L. D. BELL,2 SHOULEH NIKZAD,2

MIHIR TUNGARE,1 PUNEET H. SUVARNA,1 andFATEMEH SHAHEDIPOUR SANDVIK1,3

1.—College of Nanoscale Science and Engineering, University at Albany, Albany, NY 12203, USA.2.—Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91109, USA.3.—e-mail: sshahedipour@uamail.albany.edu

We report on a novel GaN photocathode structure that eliminates the use ofCs for photocathode activation. Development of such a photocathode structurepromises reduced cost and complexity of the device, potentially with stableoperation for a longer time. Device simulation studies suggest that depositionof Si delta-doped GaN on p-GaN templates induces sharp downward energyband bending at the surface, assisting in achieving effective negative electronaffinity for GaN photocathodes without the use of Cs. A series of experimentshas been performed to optimize the quality of the Si delta-doped layer tominimize the emission threshold of the device. This report includes significantobservations relating the dependence of device properties such as emissionthreshold, quantum efficiency, and surface morphology on the Si incorporationin the Si delta-doped layer. An optimum Si incorporation has been observed toprovide the minimum emission threshold of 4.1 eV for the discussed Cs-freeGaN photocathodes. Photoemission (PE), atomic force microscopy (AFM), andsecondary-ion mass spectroscopy (SIMS) have been performed to study theeffect of growth conditions on device performance.

Key words: GaN, photocathode, cesiated, detectors, MOCVD, photoemission

INTRODUCTION

High-performance ultraviolet (UV) photocathodesfor low signal detection are critical in astronomy forinvestigation and verification of some of the mostfundamental questions regarding the origin andexpansion of the universe.1 Other applicationsinclude missile plume detection,2 covert communica-tion,3 and image intensifiers.4 Current UV detection/imaging microchannel plates (MCPs) use alkalihalide (KBr, CsI) and Cs2Te photocathodes5 with apeak quantum efficiency (QE) of 50% to 60% whenfreshly installed. These photocathodes are unstableunder exposure to air and/or moisture. Recently ithas been established that the QE of alkali halidephotocathodes, at wavelengths close to the long-wavelength cutoff, degrades with continued expo-sure to UV radiation even in a moisture-free envi-ronment. Furthermore, their visible response

increases by as much as seven orders of magnitudewith continued UV exposure.6 Alkali halide andCs2Te photocathodes have optimized UV responsefor a narrow spectral range, and hence multiplephotocathodes are required for UV spectroscopyover a wide range of wavelengths. This requirementincreases the complexity, mass, and volume of thespectrometer in such applications. The recent pro-gress in high-quantum-efficiency GaN photocath-odes7 and systematic study of temporal emissionresponse with hot carrier relaxation mechanismsinvolved in GaN photocathodes8 has established III-nitride photocathodes as a promising candidate forUV low signal photodetection. The small electronaffinity of GaN allows the development of high-quantum-efficiency negative-electron-affinity (NEA)photocathodes with significant advantages such assolar blindness, radiation hardness, and lownoise.9–11 Activation of state-of-the-art GaN photo-cathodes requires the deposition of a thin layer ofCs.12,13 The high chemical activity of cesiumdemands a vacuum environment for the fabrication,

(Received August 1, 2010; accepted December 27, 2010;published online January 29, 2011)

Journal of ELECTRONIC MATERIALS, Vol. 40, No. 4, 2011

DOI: 10.1007/s11664-010-1507-7� 2011 TMS

382

optimization, and installation of cesiated photo-cathodes, increasing the cost and limiting the rangeof potential applications. Chemical instability anddegradation with time have been observed in suchphotocathodes.13 We report on a novel Cs-free GaNphotocathode that utilizes Si delta-doping close tothe surface for the photocathode activation, elimi-nating the use of Cs and hence the associated limi-tations. We have recently reported on the effect ofpolarization-induced surface charges and n+-GaNcap thickness on the device characteristics of such aphotocathode.14 Herein, we present experimentalresults on the effect of growth parameters suchas Si delta-doping time and SiH4 flow rate ondevice characteristics including surface morphology,emission threshold, and quantum efficiency. Thephotocathode design investigated in this research isenvisioned to be used in reflective mode.

DEVICE STRUCTURE

A schematic of the photocathode structuredesigned and studied here is shown in Fig. 1, beingcomposed of a layer of Mg-doped GaN grown on anundoped GaN template on sapphire. The surface ofthe p-GaN is Si delta-doped followed by depositionof a thin n+-GaN cap. A discontinuous GaN cap isexpected to expose the Si delta-doped layer toatmosphere, making it prone to oxidation that, inturn, will manifest itself as an increase in theemission threshold. The surface conduction band ofthe overall structure is pulled down to close to theFermi level of the p-GaN upon placing a Si delta-doped layer on the Mg-doped GaN. A thin n+-GaNcap is deposited to provide stability to the Sidelta-doped layer. Device design parametersincluding Si delta-doping density and thicknessof the n+-GaN cap layer play critical roles indetermining the emission threshold and quantumefficiency of the device. Furthermore, polarization-induced charges at the device surface influence thedevice characteristics.

EXPERIMENTAL PROCEDURES

Photocathode device structures based on theabove design were grown by metalorganic chemicalvapor deposition (MOCVD) in a Veeco D180 rotatingdisk reactor using trimethylgallium (TMGa) andammonia (NH3) as Ga and nitrogen precursors,respectively, and purified hydrogen as carrier gas.Three-micron-thick undoped GaN was grown on asapphire substrate followed by a 400-nm-thick

Mg-doped GaN layer using bis(cyclopentadienyl)magnesium (Cp2Mg) as the Mg source. The Si delta-doped layer was grown following the deposition ofthe p-GaN layer by switching the Ga and Mg sour-ces to the vent and flowing SiH4. Once the delta-doped layer was completed, TMGa flow wasresumed for a particular amount of time to achievethe desired thickness of n+-GaN cap. Upon comple-tion of the device structure, the TMGa and silanesources were turned off and the growth temperaturewas lowered to allow postgrowth activation of theMg dopants. In situ annealing was performed for30 min under a N2 environment. Photoemissionmeasurements were performed in the opaque modein which the sample was illuminated through then+-GaN layer to obtain the spectral response using aXe arc lamp and a monochromator. The sample wasplaced in a vacuum (P � 10�9 Torr) and illuminatedthrough a quartz viewport. An in situ low-temper-ature anneal at 330�C for 30 min was performedprior to measurements to desorb water and hydro-carbons from the surface. The sample was groun-ded, and emitted electrons were collected by ananode biased at +9 V and suspended above thesample surface. Light output from the monochro-mator was calibrated using a National Institute ofStandards and Technology (NIST)-traceable Siphotodiode within the vacuum chamber.

p-GaN OPTIMIZATION

The quality of the p-GaN layer plays an importantrole in determining the quantum efficiency of thephotocathode. Photoexcited hot electrons generatedin the bulk of the material travel a finite distancetowards the emission surface before escaping thematerial. The quality of the p-GaN determinesthe scattering rate of the hot electrons and hencethe escape length for the hot carriers. Longer scat-tering length leads to higher escape probability. Mgacts as a p-type dopant in GaN with high activationenergy, resulting in low hole concentration. Highhole concentration is required to achieve maximumand steep downward band bending at the surface.

A series of experiments was performed to opti-mize the electrical properties of p-GaN epilayers onundoped GaN (hereinafter referred to as u-GaN).Different growth parameters including growthpressure (P), temperature (T), and Cp2Mg flow wereoptimized to achieve smooth p-GaN layers withacceptable hole concentration. A V-to-III ratio of10,500 was used for the growth of the p-GaN layerswith 6.8 slm NH3, 15 slm H2, and 8 slm N2 flowrates. Samples were grown at the following valuesof temperature (�C)/pressure (Torr): 940/100, 970/100, 990/100, 970/300, 970/500. All five sampleswere grown with a Cp2Mg flow rate of 120 sccm.Hall measurements in van der Pauw geometrywere performed to estimate mobility and carrierconcentration of the p-GaN layers. Measuredvalues are reported in Table I along with AFMFig. 1. Schematic diagram of the photocathode device structure.

Novel Cs-Free GaN Photocathodes 383

root-mean-square (RMS) roughness values. Basedon these results, higher growth temperaturesreduced the hole concentration, increased the mobil-ity, and increased the surface roughness of thep-GaN epilayers. Higher growth pressures increasedthe hole concentration and reduced mobility up toan optimum pressure, beyond which the hole con-centration started to reduce. In addition, surfaceroughness was observed to reduce with an increasein pressure. Changes in pressure had a greatereffect on hole concentration and surface roughnessas compared with changes in growth temperature.Due to the tradeoff between mobility and holeconcentration in the process of p-type doping,p-GaN layers grown at temperature of 970�C andpressure of 300 Torr were used for the photocath-ode samples discussed in this report. Reasonablep-type conductivity is achieved under this growthcondition. SIMS measurement on a p-GaN samplegrown under such condition suggests Mg incorpo-ration of 8 9 1019 atoms/cm3 in the p-GaN films.Hall measurement shows hole concentration of1.8 9 1017/cm3, similar to the commonly achievedMg activation of only £1%. Higher Cp2Mg flowrates from 120 sccm to 150 sccm and 180 sccm

resulted in layers with lower carrier concentrationand higher resistivity.

EFFECT OF n+-GaN CAP THICKNESSON Cs-FREE GaN PHOTOCATHODE

Four photocathode structures were grown in whichSi delta-doping was introduced with a SiH4 flow rate of5 sccm for 10 s followed by deposition of the n+-GaNcap. Table II lists the details of the different photo-cathode structures along with a summary of the photo-emission results. The thickness of the n+-GaN capwas changed in different structures by changing thegrowth time of the cap layer. As predicted by thetheory of hot carrier scattering, the QE of the devicewas observed to decay exponentially with increasingn+-GaN cap thickness. The n+-GaN cap thicknessplays an important role in limiting the effect ofpolarization-induced surface charges on the photo-cathode emission threshold. A nonintuitive increasein the emission threshold was observed with anincrease in the n+-GaN cap thickness. The lowestemission threshold of 4.48 eV was observed for thedevice structure with the thinnest n+-GaN capthickness with a peak QE value of �2.5% for

Table I. Measured values of hole concentration (cm23), mobility (cm2/V s), and surface roughness (nm) of20 lm 3 20 lm AFM scans for different p-GaN epilayers

Pressure (Torr)

Temperature

T 2 30�C940�C

T970�C

T + 30�C1000�C

100 Holes 1.19 9 1017 Holes 7.72 9 1016 Holes 2.5 9 1016

Mobility 7.2 Mobility 13 Mobility 19RMS 7.3 RMS 11.66 RMS 14

300 – Holes 1.8 9 1017 –Mobility 9.8

RMS 3.26500 – Holes 1.15 9 1017 –

Mobility 8.6RMS 1.49

Table II. Details of the different photocathode structures investigated in this work

DeviceNo.

Silane Flow,Si Delta-Doping (sccm)

Si Delta-DopingTime (s)

n+-GaN CapGrowth Time (s)

Threshold(eV)

Peak QE Wavelength atPeak QE (nm)

1 5 10 20 4.48 0.025 2202 5 10 40 4.58 0.015 2183 5 10 70 4.688 7.0 9 10�3 2164 5 10 130 5.005 1.7 9 10�3 2125 20 10 70 4.846 1.3 9 10�4 2146 40 10 70 4.512 1.2 9 10�3 2167 40 30 70 4.49 3.5 9 10�3 2188 40 60 70 4.16 1.9 9 10�3 2209 40 120 70 4.4 1.5 9 10�4 21810 40 300 70 4.8 2.2 9 10�4 212

Tripathi, Bell, Nikzad, Tungare, Suvarna, and Shahedipour Sandvik384

incident radiation of 220 nm.14 Figure 2 shows thephotoemission spectra of the photocathode samplewith the thinnest n+-GaN cap as a function of inci-dent photon energy.

EFFECT OF Si DELTA-DOPING

The magnitude of the Si delta-doping on thep-GaN layer dictates the degree of downward bend-ing of the GaN photocathode energy bands. Insuffi-cient Si incorporation in the delta-doped layer resultsin high emission threshold, whereas excess Si willalso lead to an increase in the threshold energy due toother reasons. These include formation of a roughsurface caused by excessive SiH4 flow resulting in thepotential formation of discontinuous delta-dopingand n+-GaN cap. Furthermore, degradation of theemission surface results in high carrier scattering,and low emission efficiency, causing a drop in QE.

A series of experiments were performed to optimizethe Si delta-doped layer on the p-GaN surface. In thisset of experiments the flow rate and introduction timeof silane were varied. The resultant devices werecharacterized using photoemission, AFM, SIMS, andx-ray photoelectron spectroscopy (XPS) measure-ments, and correlations were made between variouscharacteristics and growth conditions. The thicknessof the n+-GaN cap was kept similar for all of thestructures, with growth time of 70 s.

Effect of Silane Flow Rate

Three photocathode samples were grown in whichthe silane (SiH4) flow rate was changed from 5 sccmto 20 sccm to 40 sccm for the Si delta-doped layerand the GaN cap layer. Delta-doping time and GaNcap growth time for all the three samples were 10 sand 70 s, respectively. The emission threshold cal-culated from the photoemission measurements forthe three samples is shown in Fig. 3. Differenceswere observed in the threshold voltages of the threesamples, but no systematic dependence of emissionthreshold on Si flow rate was seen. AFM measure-ments (Fig. 4) show that the surface morphology

Fig. 2. Photoemission spectrum of photocathode device 1 as afunction of incident photon energy.

Fig. 3. Variation in emission threshold as a function of SiH4 flow.Dotted line is a guide to the eye.

Fig. 4. Large-area AFM scans for photocathode structures with 10 s of Si delta-doped layer and SiH4 flow of (a) 20 sccm and (b) 40 sccm. Thesurface with 5 sccm SiH4 flow looks similar to the p-GaN surface.

Novel Cs-Free GaN Photocathodes 385

degraded for the samples with increased SiH4 flowfrom 5 sccm to 20 sccm to 40 sccm. The peak QE forthe three samples were 0.7%, 0.01%, and 0.1%,respectively, suggesting that higher Si incorpora-tion or lower threshold is not the only parameter toensure high QE. Specifically, the SiH4 flow rate of40 sccm showed the lowest threshold voltage; how-ever, the peak QE for this sample was observed tobe significantly lower than for the sample with

5 sccm SiH4, illustrating the significance of theeffect of scattering on photocathode emission.

Effect of Si Delta-Doping Time

Optimization of the Si delta-doping growth timewas performed in a series of experiments in which aflow rate of 40 sccm of SiH4 was maintained for thedelta-doping, as this flow rate resulted in the lowestemission threshold in the experiments discussedabove. The time for the delta-doped region wasvaried from 10 s, 30 s, 60 s, 120 s, to 300 s in thefive photocathode devices. Figure 5 shows the vari-ation of the threshold voltage as a function of Sidelta-doping growth time. Initially, an increase inthe delta-doping time caused reduction in thethreshold voltage, beyond which the thresholdstarted to increase. The optimum time for delta-doping to cause maximum surface band bendingand the lowest emission threshold was observed tobe 60 s. This structure showed the lowest emissionthreshold of 4.1 eV with peak emission QE of�0.2%. The devices with longer duration of Si delta-doping showed a drop in peak QE by an order ofmagnitude.

Surface morphology of different samples as mea-sured by AFM is shown in Fig. 6. The surface mor-phology points to the presence of different growthmodes for the GaN cap after the Si delta-dopinglayer. Increase in the delta-doping time from 10 s to30 s was observed to increase the surface rough-ness. Further increase in the delta-doping time

Fig. 5. Change in emission threshold as a function of Si delta-dopingtime in the photocathode structures.

Fig. 6. Surface morphology for photocathode samples with: (a) 10 s, (b) 30 s, (c) 60 s, (d) 120 s, and (e) 300 s of SiH4 flow in the delta-dopedlayer.

Tripathi, Bell, Nikzad, Tungare, Suvarna, and Shahedipour Sandvik386

reduced the surface roughness up to 120 s. Growthwith delta-doping time of 300 s resulted in forma-tion of large hillocks.

To further examine the impact of the n+-GaN capthickness on the threshold voltage, a structure wasgrown where the growth time of n+-GaN cap wasreduced to 40 s, for the sample with 40 sccm SiH4

flow rate and 60 s delta-doping time. The PE mea-surement on this device showed an increase inemission threshold to 4.3 eV as compared with4.1 eV for the optimized sample, where the GaN capgrowth time was 70 s.

Secondary-ion mass spectroscopy (SIMS) was alsoperformed to investigate the Si incorporation profilein the photocathode structures as a function ofgrowth time of the delta-doped layer. The Si profilefor the four samples as a function of depth alongwith the details of the SiH4 flow rate are given inFig. 7. As expected, the peak value of Si incorpora-tion was observed to increase with increasing SiH4

flow rate and deposition time of the delta-dopedlayer. The silicon profile for the 5 sccm flow and 10 sdelta-doped layer shows a peak at a depth of about60 A into the sample. Considering the smooth sur-face morphology of this sample, the thickness of then+-GaN cap is estimated to be 6 nm. For the samplewith 40 sccm SiH4 flow rate the Si profile isobserved to be much wider. It is also noted thatsurface roughness has been reported to causebroadening in the SIMS profile.15 In addition,samples with higher SiH4 flow rates show changesin the peak position of the Si profile with the changein delta-doping time.

CONCLUSIONS

To better understand the impact of Si dopingand n+-GaN cap thickness in a delta-doped GaN

photocathode design on its characteristics, a seriesof experiments with varying growth conditions havebeen performed. The emission threshold of thedevice was observed to increase with increasingn+-GaN cap thickness. Such behavior has beensuccessfully modeled and is attributed to localelectric fields caused by negative polarization-induced surface charges. The quantum efficiency ofthe devices follows an exponential decay withincreasing n+-GaN cap thickness. Increase in Siincorporation in the delta-doped layer, up to anoptimum value, has been shown to improve devicecharacteristics, including emission threshold andQE. Further increase in Si incorporation causedincrease in emission threshold and degradation ofthe photocathode surface. The lowest emissionthreshold of 4.1 eV was achieved without use of Cs,demonstrating the potential of this novel approach.Surface morphology of the photocathode structureshas been observed to be strongly dependent on theSi incorporation. Although we observed formation ofhillocks on the surface of the device structure at thehighest SiH4 flow rate and smooth morphology forlower values, no obvious dependence between thesurface morphology and emission threshold of thedevice could be identified.

REFERENCES

1. M.P. Ulmer, Proc. SPIE 7222, 722210 (2009).2. R.M.E. Illing, N.H. Zaun, and R.L. Bybee, Proc. SPIE 932,

246 (1988).3. G.A. Shaw, A.M. Siegel, J. Model, and D. Greisokh, Proc.

SPIE 5796, 214 (2005).4. J.W. Glesener, A.M. Dabiran, and J.P. Estrera, Proc. SPIE

7339, 73390S1 (2009).5. S.A. Stern, D.C. Slater, J. Scherrer, J. Stone, G. Dirks,

M. Versteeg, M. Davis, G.R. Gladstone, J.W. Parker, L.A.Young, and O.H.W. Siegmund, Space Sci. Rev. 140, 155(2008).

6. A.S. Tremsin and O.H.W. Siegmund, Proc. SPIE 5920,59200I (2005).

7. S. Uchiyama, Y. Takagi, M. Niigaki, H. Kan, and H. Kondoh,Appl. Phys. Lett. 86, 103511 (2005).

8. I.V. Bazarov, B.M. Dunham, X. Liu, M. Virgo, A.M. Dabiran,F. Hannon, and H. Sayed, J. Appl. Phys. 105, 083715(2009).

9. S. Fuke, M. Sumiya, T. Nihashi, M. Hagino, M. Matsumoto,Y. Kamo, M. Sato, and K. Ohtsuka, Proc. SPIE 6894, 68941F(2008).

10. I. Mizuno, T. Nihashi, T. Nagai, M. Niigaki, Y. Shimizu, K.Shimano, K. Katoh, T. Ihara, K. Okano, M. Matsumoto, andM. Tachino, Proc. SPIE 6945, 69451N (2008).

11. O.H.W. Siegmund, A.S. Tremsin, J.V. Vallerga, J.B.McPhate, J.S. Hull, J. Malloy, and A.M. Dabiran, Proc.SPIE 7021, 70211B (2008).

12. Y. Qian, B. Chang, J. Qiao, Y. Zhang, R. Fu, and Y. Qiu,Proc. SPIE 7481, 74810H (2009).

13. O. Siegmund, J. Vallerga, J. McPhate, J. Malloy, A. Tremsin,A. Martin, M. Ulmer, and B. Wessels, Nucl. Instrum. Meth-ods Phys. Res. A 567, 89 (2006).

14. N. Tripathi, L.D. Bell, S. Nikzad, and F. Shahedipour-Sandvik, Appl. Phys. Lett. 97, 052107 (2010).

15. B. Fares, B. Gautier, P. Holliger, N. Baboux, G. Prudon, andJ.C. Dupuy, Appl. Surf. Sci. 253, 2662 (2006).

Fig. 7. Depth profile of Si atoms into the photocathode samples asmeasured using SIMS.

Novel Cs-Free GaN Photocathodes 387