Characterization of low temperature GaAs antenna array terahertz emitters
M.Awad,a ͒M.Nagel,and H.Kurz
Institut für Halbleitertechnik,RWTH Aachen University,Sommerfeldstr.24,52074Aachen,Germany
J.Herfort and K.Ploog
自由的作文
Paul-Drude-Institut für Festkörperelektronik,Hausvogteiplatz 5-7,10117Berlin,Germany
͑Received 25June 2007;accepted 1October 2007;published online 2November 2007͒
We prent a fabrication concept for photoconductive terahertz antenna arrays bad on substrate-transferred thin films of low-temperature-grown GaAs miconductor material.Adjacent array elements are physically decoupled by removing completely the photoconductive material in between.In contrast to former array devices bad on intrinsic bulk GaAs substrates,this method allows the u of arbitrary carrier substrates with enhanced transmission properties.The emission characteristics of the device are investigated in terms of bandwidth,directivity,and saturation caud by charge-carrier induced field-screening effects.Screening-free operation is experimentally obrved for an average optical power density below 2.2ϫ10−4mW/m 2.©2007American Institute o
f Physics .͓DOI:10.1063/1.2800885͔
Photoconductive metal-miconductor-metal ͑MSM ͒structures such as coplanar strip lines or dipole antennas are still the most common devices for terahertz signal generation bad on femtocond lar oscillator sources with pul en-ergies in the nanojoule range.At high excitation densities,the MSM terahertz sources’efficiency,however,suffers con-siderably from space charge and radiation field screening of the bias field.1A way to improve the efficiency of photocon-ductive terahertz devices is to increa the active area,thereby reducing the excitation density incident onto the de-vice.The large aperture devices,however,require high bias voltages in the kilovolt range.Alternatively,it is pos-sible to fabricate a terahertz emitter with a large effective aperture,consisting of periodically spaced MSM elements,and thus reduce the required bias voltages from kilovolts to tens of volts,depending on the width of MSMs’photocon-ductive gap.Such a device bad on mi-insulating GaAs has recently been demonstrated.2
In this work,we demonstrate an alternative fabrication concept for terahertz antenna arrays bad on substrate-transferred thin films of low-temperature ͑LT ͒grown GaAs miconductor material.The epitaxial lift-off method 3allows the u of LT-GaAs films on arbitrary carrier substrates with enhanced transmission properties at optical and terahertz fre-quencies in comparison to intrinsic Ga
As,which is known to exhibit considerable attenuation and dispersion at terahertz frequencies.4As oppod to mi-insulating ͑SI ͒GaAs,LT GaAs has the very important advantage of short carrier lifetime 5on the order of 150fs allowing the device to be ud interchangeably as an emitter,detector,and under im-pulsive or cw optical excitation.The latter excitation method has recently been discusd for terahertz array devices by Saeedkia et al..6Furthermore,the higher breakdown field for LT GaAs ͑ϳ500kV/cm ͒permits the application of greater electric fields in comparison to SI GaAs ͑10–100kV/cm ͒.7Therefore,LT GaAs is considered to be one of the best ma-terials for photoconductive terahertz devices.8
In the following,the fabrication process of the antenna array devices is described.In the first step,two ts of 14interdigitated 4m wide electrodes are deposited on top of a lattice-matched layer system consisting of 150nm SI GaAs,a 100nm AlAs layer,followed by a 1.3m layer of LT GaAs grown using molecular beam epitaxy onto a ͓100͔ori-ented SI GaAs substrate.The electrode patterning method ud for this step is photolithography using AZ5214photo-resist followed by metal layer deposition ͑Ti/Au,10/200nm ͒and subquent lift-off in acetone.In a further photolithography step,the array structure is lectively masked with a photoresist layer.Next,the expod LT-GaAs areas are etched away by chemical wet etching in a H 2O 2:H 2SO 4:H 2O ͑8:1:1͒solution giving access to the AlAs lay
黑色裤子配什么颜色上衣
er.Finally,the AlAs sacrificial layer is etched off completely in a 10%HF solution yielding a lf-contained,freestanding LT-GaAs chip carrying the antenna structure,shown in Fig.1͑a ͒.The terahertz array chip is then trans-ferred to a 500m thick,optically and terahertz transparent sapphire substrate.As shown in the cross-ctional view in Fig.1͑b ͒,the photoconductive gap between the electrodes comprising each array element is 11m wide,while con-cutive array elements are parated by a distance of 4m.For the proper operation of the device,it is important to
a ͒
Author to whom correspondence should be addresd;electronic mail:
awad@iht.rwth-aachen.de FIG.1.͑Color online ͒͑a ͒Micrograph of terahertz array antenna device.͑b ͒Cross-ctional view of terahertz antenna array.
APPLIED PHYSICS LETTERS 91,181124͑2007͒
0003-6951/2007/91͑18͒/181124/3/$23.00©2007American Institute of Physics
91,181124-1Downloaded 01 Sep 2011 to 202.115.15.26. Redistribution subject to AIP licen or copyright; e /about/rights_and_permissions
ensure that no miconductor material is left between adja-cent elements,since radiative contributions from the areas would destructively interfere.Also,having the interelement area void from any miconductor material reduces the de-vice dark current by at least 50%.
To measure the emitted terahertz field from a single ar-ray element and a fully illuminated array,the device is placed in a standard confocal terahertz-time-domain spec-troscopy system and excited with optical puls from a mode-locked Ti:sapphire lar,operating at 780nm with a repetition rate of 76MHz and 100fs pul width.The emit-ter is mounted such that the LT-GaAs chip directly faces the optical excitation,while the terahertz radiation is coupled out through the sapphire substrate facing th
e detector.For all measurements,no emitter-side substrate lens is ud.The antenna array is biad with a 1.1kHz,0–40V amplitude square wave,which also provides the reference for lock-in detection.As a detector,a 20m dipole antenna with a 5m photoconductive gap is ud.A 3mm diameter silicon substrate lens placed on the back side of the detector focus the terahertz radiation onto the photoconductive gap.A cur-rent amplifier connected to the dipole antenna converts the induced signal current into a voltage which is detected by a lock-in amplifier.
For the first measurement,the optical beam with 5.5mW power is focud onto a single element at the center of the array illuminating an active area of approximately 95m 2.A cond measurement is performed under the same bias conditions,but with the optical beam fully illuminating the array covering an active area of 4.3ϫ104m 2.In both cas,identical photocurrents of 1.3A are obrved.Hence,without the influence of parasitic effects,identical terahertz field amplitudes should be obrved for single emit-ter and full array excitation since the generated terahertz field amplitude E terahertz ͑t ͒is proportional to ͑ץ/ץt ͒j ͑t ͒,with j be-ing the photocurrent density.Time-resolved scans and their respective amplitude spectra shown in Fig.2,however,re-veal an approximately 30%increa in signal amplitude between the fully illuminated array and single element excitation.
To investigate the origin of the increa in emitted tera-hertz radiation obrved in Fig.2,we measure
the directivity
dependence on the optical excitation area.A more strongly directed terahertz beam will translate into a larger terahertz amplitude at the detector due to a higher amount of radiation that is collected at the finitely sized mirrors ud in the mea-surement tup.In order to determine the directivity for both single element and fully illuminated array excitation cas,the radiation pattern is mapped with a further tup,depicted in the inlay in Fig.3.As shown,the antenna array is placed at the center of a rotation stage.The same terahertz detector as was ud in the previous measurement is coupled to a 1m long optical fiber and mounted on a rotation stage.A grating pair is ud to compensate for the dispersion in the fiber.Two ts of measurements are acquired as before.For the first measurement,the optical beam is focud onto a single array element.The measured radiation pattern,plotted in Fig.3͑dashed line ͒,matches very well with the radiation pattern of a dipole ͑dark solid line ͒.With the entire array illuminated,the radiation pattern,as expected,shows a narrower beam width due to the larger number of excited antenna elements.The light solid line in Fig.3shows the calculated normalized radiation pattern for a 14element array with uniform excita-tion bad on array theory.9Since the element spacing l is much smaller than the center terahertz wavelength l Ӷterahertz ,we do not obrve the high gain and directivity attainable using a typical array configura中期考核个人总结
tion with l in the order of terahertz .However,the measured array radiation pat-tern,Fig.3͑dotted line ͒,fits qualitatively well to the analyti-cal model described by array theory.It should be considered that,the applied excitation spot is not uniform ͑as considered for the calculation ͒.The obrved radiation angles for single element and array emitter are approximately 120°and 90°͑both full width at half maximum ͒,respectively.Accounting for the finitely sized collecting terahertz mirrors ud in the experimental tup,having an aperture angle of 30°,the ob-rved increa of 30%in signal amplitude at low pump power can indeed be largely attributed to the enhanced direc-tivity of the array.
At higher optical pumping levels,however,screening ef-fects are expected to become increasingly relevant.To inves-tigate this regime,we further compare the emitted terahertz peak amplitudes at incread optical excitation power P NIR for both cas.In Fig.4,the measured terahertz peak ampli-tude,E terahertz ͑P NIR ͒,of the array emitter versus optical exci-tation is plotted.In the ca of negligible screening,a
linear
剪纸轴对称图形FIG.2.͑Color online ͒Time-resolved scans of terahertz pul for array emitter ͑dark line ͒and single element ͑light line ͒and their respective spectra in
inlay.
FIG.3.͑Color online ͒Calculated radiation pattern for dipole ͑black solid line ͒and measured single element radiation pattern ͑dashed line ͒.Array radiation pattern calculated bad on array theory ͑light solid line ͒and mea-surement ͑dotted ͒.Inlay:experimental tup with fiber coupled and disper-sion compensated ͑FC ͒receiver for radiation pattern measurement.
Downloaded 01 Sep 2011 to 202.115.15.26. Redistribution subject to AIP licen or copyright; e /about/rights_and_permissions物业管理培训
学习习惯有哪些
increa of E terahertz ͑P NIR ͒is expected.Recently,Kim and Citrin 1theoretically investigated the saturation behavior of single element GaAs terahertz emitters due to Coulomb screening and back action of the radiated terahertz field.In their work,excitation intensities I NIR below approximately 6ϫ10−4mW/m 2were identified to be desirable to avoid screening related emitter degradation.Here,the single emit-ter is pumped with intensities I NIR ͑or corresponding P NIR ͒between 7.9ϫ10−2mW/m 2͑7.5mW ͒and 2.1mW/m 2͑200mW ͒.Conquently,for the single element,a linear re-gime of E terahertz ͑P NIR ͒cannot be identified in Fig.4͑a ͒.At about 1.63mW/m 2͑155mW ͒,complete saturation occurs.The device finally fails at an excitation intensity of 2.1mW/m
2.Figure 4also shows two further graphs of the emitted terahertz radiation from a fully illuminated array structure with two different illumination spot sizes.The data reprented by Fig.4͑b ͒were acquired using a reference spot size,adjusted to maximize the emitted terahertz signal in the regime of negligible screening at I NIR =8.3ϫ10−5mW/m 2.In this ca the excitation spot area approximately equals the total array area.A linear increa of E terahertz ͑P NIR ͒is ob-rved up to I NIR =2.2ϫ10−4mW/m 2and P NIR =20mW,as indicated by the arrow.A further increa in excitation spot size beyond the reference spot size of Fig.4͑b ͒,in the pump intensity regime where the influence of screening is noticeable ͑I NIR =2ϫ10−3mW/m 2͒,leads to an increa of the generated terahertz radiation.In this latter ca,shown in Fig.4͑c ͒,the device remains longer in a linear regime com-pared to the reference spot excitation in Fig.4͑b ͒,up until P NIR =25mW.The intensity where E terahertz ͑P NIR ͒begins to diverge from linearity is I NIR =2.2ϫ10−4mW/m 2,the same value as determined in the previous ca,shown in Fig.4͑b ͒.
Considering a pul energy density of 2.9fJ/m 2,an absorption coefficient ␣=12000cm −1,and a refractive index n =3.42for LT GaAs,10we infer a critical average photocar-rier density of ϳ1.5ϫ1016cm −3above which screening ef-fects become relevant.For the larger excitation spot size,an incread loss of optical excitation power outside the active area of the device is introduced.
Therefore,the increa fac-tor for the linear regime is considerably reduced for the larger excitation spot plotted in Fig.4͑c ͒in comparison to Fig.4͑b ͒.However,at higher pump power,this loss of exci-tation beyond the array area is compensated for by the lower excitation density leading to incread emission efficiency.Due to power limitations of our lar source,we were not
able to identify the power range where complete saturation of the fully illuminated array occurs.The maximum excita-tion power applied was 350mW corresponding to an inten-sity of 2.8ϫ10−3mW/m 2,which is almost three orders of magnitude below the intensity that was needed to completely saturate the single emitter.At 350mW excitation power,the terahertz array antenna achieves a fivefold field amplitude increa in comparison to the maximum terahertz field gen-erated by the single emitter.
Interestingly,terahertz radiation emitted from SI-GaAs bad array structures 11exhibited a strong blueshift of the generated terahertz spectrum for decreasing excitation spot sizes,with maximum emission for terahertz wavelengths in the range of the excitation spot diameter.In our ca,how-ever,we do not obrve any significant frequency depen-dence on illumination spot size.Referring back to Fig.2͑a ͒,the amplitude spectra of single emitter and array both peak at approximately 750GHz,even in the ca of the single emit-ter with an excitation spot diameter d =10m ͑array:d =3
00m ͒,where d is much smaller than the generated tera-hertz wavelength terahertz .We attribute this behavior to the improved electrodynamic decoupling of our array elements in comparison with SI-GaAs structures in Ref.11resulting from the lack of a metallic masking layer and miconductor material in between elements.Therefore,in contrast to an aperture-like irradiation the behavior of our array is mainly determined by the radiation property of each array-embedded single element.Our results indicate also that the propod antenna arrays can be further enhanced in terms of directivity and terahertz field amplitude by increasing the number of array elements and element paration distance.
In summary,we prented a terahertz antenna array de-vice fabricated on LT GaAs using an epitaxial lift-off tech-nique.Epitaxial lift-off allows added versatility in the choice of substrate,or even direct bonding onto integrated terahertz systems.We demonstrated that the terahertz antenna array can generate drastically incread terahertz field amplitudes compared to a single element and that even for high fluence,the array remains in an unsaturated region of operation,whereas the single element device is well beyond its failure limit.
The authors would like to acknowledge the Innovation Department of North Rhine Westphalia and the Federal Ministry of Education and Rearch ͑BMBF ͒for financial support.
1
D.S.Kim and D.S.Citrin,Appl.Phys.Lett.88,161117͑2006͒.
2A.Dreyhaupt,S.Winnerl,T.Dekorsy,and M.Helm,Appl.Phys.Lett.86,121114͑2005͒.3
E.Yablonovitch,D.M.Hwang,T.J.Gmitter,L.T.Florez,and J.P.Harbison,Appl.Phys.Lett.56,2419͑1990͒.4
C.J.Johnson,G.H.Sherman,and R.Weil,Appl.Opt.8,1667͑1969͒.5
G.Segschneider,T.Dekorsy,H.Kurz,R.Hey,and K.Ploog,Appl.Phys.Lett.71,2779͑1997͒.6
qq说说文案D.Saeedkia,R.R.Mansour,and S.Safavi-Naeini,IEEE Trans.Antennas Propag.53,4044͑2005͒.7
J.K.Luo,H.Thomas,D.V .Morgan,D.Westwood,and R.H.Williams,Semicond.Sci.Technol.9,2199͑1994͒.8
M.Tani,S.Matsuura,K.Sakai,and S.Nakashima,Appl.Opt.36,7853͑1997͒.9
C.Balanis,Antenna Theory ,3rd ed.͑Wiley-Interscience,Hoboken,NJ,2005͒.10
P.Grenier and J.F.Whitaker,Appl.Phys.Lett.70,1998͑1997͒.11
肚子疼是啥原因A.Dreyhaupt,S.Winnerl,M.Helm,and T.Dekorsy,Opt.Lett.31,1546͑2006͒
.
FIG.4.͑Color online ͒Saturation curve for ͑a ͒single element ͑focud optical spot,X marks failure point of
the device ͒,͑b ͒reference excitation spot ͑spot size adjusted to maximize terahertz amplitude at low excitation power ͒,and ͑c ͒array with large excitation spot compared to reference spot ͑spot size adjusted to maximize terahertz amplitude at high excitation power ͒.
Downloaded 01 Sep 2011 to 202.115.15.26. Redistribution subject to AIP licen or copyright; e /about/rights_and_permissions
