Optimal Dark Current Reduction in Quantum Well 9 µm GaAs/AlGaAs Infrared Photodetectors

With Improved Detectivity

Shahram Mohammad Nejad, Maryam Pourmahyabadi, and Ali Asghar AmidianIran Telecommunication Research Center (ITRC)

Optoelectronic and Laser Laboratory, Department of Electrical EngineeringIran University of Science and Technology (IUST)

Tehran, Iranshahramm@iust.ac.ir

Abstract—In this paper, an optimization approach is presented to decrease the dark current in GaAs/AlGaAs QWIPs. Dark current noise, as shown, can be reduced by increasing Al density in barriers, decreasing detector dimensions and increasing the periodic length of the structure. It is also shown that increasing the number of periods can reduce both the dark current and responsivity. Therefore, devices can be optimally designed through judicious choice of these parameters. An optimal photodetector structure is designed and simulated to achieve low dark current (11nA) and detectivity of 1012cm(Hz)1/2/W which is an order of magnitude greater than the present values.

I. INTRODUCTION

With the increasing demand for new optical application of quantum well infrared photodetectors (QWIPs) at a wide variety of wavelength from mid to far infrared, the need for low noise structure has become greater than ever. The absorption of long wavelength light in QW is due to transition from a quasi bound state to the continuum in a narrow well or intersubband transition in a wide well. Also, QWIPs exhibit very fast operation as demonstrated in recent experiments [1]. Their intrinsic high speed is considered as one of the advantages of the QWIPs over standard detectors made of narrow gap semiconductors.

In the meantime, the dark current determines the signal to noise ratio of QWIPs, and therefore, minimizing it, is of utmost important design criterion for their construction. By reducing QWIP dark current, photodetector can respond to weaker optical signal. The three major contributors to the dark current are: Sequential tunneling, Field induced emission and thermionic emission [2-5].

The QWIP under consideration comprises a QW structure sandwiched between the emitter and collector barriers (the extreme barriers of the QW structure) with contact layers heavily doped by donors. The QW structure

includes heavily-doped narrow-gap wells separated by thick undoped wide-gap barriers [6]. It is assumed that the barrier thickness Lb far exceeds the thickness of the wells Lw. The thickness and the depth of the wells are adopted in such a way that the well contains only one bound level and the firstexcited level corresponds to the top of the barriers between the wells. Thus the electron excitation energy is defined by the location of the bound level with respect to the barrier top. This paper is organized as follows: In the next section, the contributors to the dark current and then, the methods applied to the problem of dark current reduction are discussed. Section 3 expresses the QWIP characteristics such as responsivity, gain and detectivity. In section 4, we demonstrated an optimization approach to achieve both low dark current and acceptable responsitivity by judicious choice of the parameters. The photodetector in question is designed and then, the results of numerical analysis and simulation are presented.

II. DARK CURRENT REDUCTION

In this section, we cover the contributors to the dark current and then, the methods applied to the problem of dark current reduction are expressed in details. The first contributor, sequential tunneling is purely electron scattering from states localized in one QW to the next. This process doesn't depend on temperature. At very low temperatures (below 40 K) and sensitivity peak at 50µm, sequential tunneling is the main factor of dark current generation. The second, field induced emission is again scattering mediated by either a phonon or another charge carrier, but this time, it is from upper states of the subband distribution. Although this reduces the “activation energy,” the number of carriers at these energies is smaller, making this process less likely to occur. The final contribution to the dark current is due to thermionic emission. It is simply thermal excitation of the carriers directly out of the top of the QW into the continuum states above the semiconductor

1-4244-0395-2/06/$20.00 ©2006 IEEE. 918

barriers where, they are free to move. Hence, under the influence of the applied bias, constitute a current flow. It is the main factor at high temperatures (over 45 K) and sensitivity peak at 50µm.

A. Reducing sequential tunneling with increased barrier widthThe contribution of sequential tunneling to the dark

current can be reduced simply by increasing the width of the barrier separating adjacent wells [7]. From basic semiconductor theories, the current density is J=nqv, where, n is the density, v is the velocity of the carriers. In this model, the electrons scatter a distance equal to the period of MQW Lw+Lbscattering rate, i.e.,

1 1 1LO ee

The velocity, therefore, is given by = (Lw+Lb . Thus,the higher the scattering rate LO or ee, the larger the contribution of this mechanism to the dark current.Examinations proved that the contribution of electron-Lo phonon and electron-electron scattering to the sequential tunneling component of the dark current increases with increasing detection wavelength [4]. Also, as the detection wavelength increases from mid to far infrared values, the barrier required to reduce the sequential tunneling component of the dark current also increases. This is due to the reduced well thickness pushing the electron states toward the top of the barrier, allowing the wave function to delocalize and overlap with the wave function in the adjacent wells. Thus, using standard QWIP designs and pushing them toward the longer wavelengths, will require thicker barriers than their mid infrared counterparts to keep the sequential tunneling to the same values. For every 9.5Åincrease in barrier width of a particular 9.7 µm QWIPs design, the phonon sequential tunneling contribution will be reduced by a factor of e-1 [4]. To reduce this contribution by a factor of 10-3 implies that the barrier width must increase

b=66Å.

B. Reducing sequential tunneling with thermal excitation and thermionic emissionIn this section, we define the model constructed for

evaluating the effects of thermionic emission and thermally-assisted tunneling on the dark current. Calculated results using this model agree well with experiments in a QWIP in which the dominant dark current mechanisms are due to thermionic emission and thermally-assisted tunneling, which has been used until recently. The dark current of a QWIP is given by [5],

*

2 2 1. ( ) ( , )

1 ( / )w DET

p

em A FI f T F dd l F v s

Figure 1. The dark current as a function of bias voltage, for a QWIP at 77 K, with N=50, Lb=500Å, Lw=50Å, 1 f b=137meV,

x=0.1640. (a) ADET2 (b) ADET

2.

where, m*w is the electron effective mass in the QW, A

is the device area, Lp= Lw+Lb is the QWIP period length, µis the electron mobility, F=Vb/NLp is the electric field inside the QWIP, s is the electron saturation velocity, is the Fermi–Dirac distribution function, and is the bias-dependent tunneling current transmission coefficient for a single barrier .

In equation (2), the first parameter which effectively reduces the dark current is the detector dimension. The dark current characteristics as a function of the bias voltage, for ADET

2 and ADET= (2 2 are depicted in Fig.1. As shown, reducing the detection area decreases the dark current. By increasing the bias voltage, the dark current is increased and then saturated.

The second parameter is Al density which reduces the dark current at low level bias voltage. It also influences the effective mass of the barrier m*

b and the barrier height b.Fig.2 shows the dark current characteristics for x=0.25 and x=0.164. It also shows that increase of x in a fixed bias voltage reduces the dark current (Vb<2V), whereas the same change in x does not have any remarkable effect at higher voltages. Equality of the dark current in both curves of the Fig.2 at high level bias voltages is due to stronger bias electric field which in turn, increases the tunneling probability of the upper triangular edge of the barrier in thermal excitation via increase of the barriers number.

Increasing the number of periods decreases the dark current and responsivity, because of decreasing the average electric field through the device [9]. Hence it should be optimally designed through judicious choice. In addition, the dark current density decreases with increasing the height [6].

III. RESPONSIVITY AND DETECTIVITY

The responsivity of a QWIP is a commonly used figure-of-merit for detector performance and is given by,

exp /1 exp /

L Le w cRhcN L Lw c

919

N is the period number, and Lc is decay constant defining the rate of increase of the capture probability Pc with well width Lw [9]. There is a direct proportionality between the responsivityand wavelength and the responsivity increases superlinearily with detection wavelength. Also it slowly decreases with the decrease of average electric field. However, it is valid until the electric field is able to extract the electrons into the collector. If the electric field at the collector tends to zero, the extraction of the electrons from the QW structure decreases and the potential distribution and the electric field at the emitter become insensitive to the electron photoexcitation. It leads to a drastic drop in the responsivity [6].Using the noise model, the noise gain of the device and the peak detectivity are calculated as well [10, 11]. The detectivity can be expressed as

*4DET

d

A fD ReI g

where R is the responsivity, ADET is the device area, is the bandwidth, Id is the dark current, and g is the noise gain, which can be expressed as

0( ) 2/ exp /

I Rdg V

kT c N E kT

here R0 is the dynamic resistance at zero bias, N is the period number, and is the barrier lowering, which is given by,

04eVbE e

l p r

where, Vb is the bias voltage across one period, Lp is one period length [12].

Figure 2. The dark current as a function of bias voltage, for a QWIP at 77 K, with N=50, Lb=500Å, Lw=50Å, 1 f b=137meV,

ADET m)2 (a) x=0.1640, (b) x=0.25.

IV. THE DESIGN OF OPTIMAL PHOTODETECTOR

Following the above mentioned remarks, the design objective is the minimization of the dark current at 77 K.QWIP design steps are expressed in details here after.

C. Determination of quantum well dimensions1) Well depth

The depth of the quantum well is determined by considering the energy of the absorbed photons.

0hcV p

where, V0wavelength of responsivity peak and p is the energy of absorbed photons, then V0 p 1 where 1 is the energy of the first confined level of potential well, therefore, V0 1.

2) Well thicknessThe results of Workman simulation revealed that the

number of subbands increases with well widening. However, by considering the quantization conditionand the energy levels of wells [9], for m*

w =0.067m0,Lw=50Å is obtained.

3) Barrier widthTo reduce the effects of tunneling on the dark

current, the barrier width should be increased as much as possible. A suitable criterion for barrier width isLb=10Lw.

D. Determination of Al density and base subband energyIt is clear that, increasing Al density will decrease the

dark current. To calculate Al density, x=V0(meV)/835.5 issuitable for AlxGa1-xAs structure [9]. Substitution of V0=150meV, leads to x=0.18. To modify the dark current and to optimize the well depth, x=0.25 is assumed, henceV0=209meV 1> 70meV. In addition the effective mass of the carriers in the barriers is given bym*

b=(0.067+0.083x)m0, then, m*b=0.0878m0. Also

defined as [13],

*02 / 2wwl m V

The well can only have an even base state for =0.9.Hence, the wave number is given by,

2 /( )evenwk lw .

and the energy associated to even base state is,

2 *1 ( ) / 2even

wwk m

Therefore, , kw 1 are obtained 1.5038, 3.6×10-2/ Å and 75meV respectively. In addition to performing precise

920

design steps for detection wavelength, the well depth is also enhanced. Increasing the well depth increases the trapping of thermionic emission carriers and hence, the dark current is more reduced. In other words, an energy filter is designedto prevent transfer of the dark current carriers. These carriers generally have less energy than photoelectrons.

E. Determination of the optimum period numberIncreasing the number of periods reduces both

responsivity and the dark current. On the other hand, detection coefficient increases with increasing the number of periods. Considering the discussed factors, and assuming N=50, the dark current is constrained to 10nA.

V. THE RESULTS OF ULTRA LOW DARK CURRENT PHOTODETECTOR

The simulation results of the optimized device are summarized in table I. It is revealed excellent dectetivity and low dark current of designed photodetector compared to the results presented by Gunapala et al. [11] and Hsu et al.[14] which are specified by design 1 and design 2 in the table, respectively.

Fig.3 shows the dark current as a function of the bias voltage in the optimal QWIP. The value of dark current is approximately 11nA for voltages more than 1V in temperature 77 K. The responsivity value was calculated by assuming a constant quantum efficiency of 10% and a quantum well captures probability of 6% which is typical of such devices.

VI. CONCLUTION

We have improved the QWIP devices performance by an order of magnitude in detectivity and dark current. Investigation of different contributors to the dark current in GaAs/AlGaAs QWIPs proved that optimal determination of physical dimensions of the photodetector, the periodic length of the structure, Al density and the number of periods, will reduced the dark current by an order of magnitude 11nA and thereby significantly increase the detectivity of 1012cm(Hz)1/2/W.

Figure 3. The dark current as a function of bias voltage in optimal QWIP.

TABLE I. THE ACHIEVED PERFORMANCE DATA OF DESIGNED PHOTODETECTOR RESULTS

QWIPParameter

Design 1 Design 2 Optimized Design

T (K) 75 77 77

R (mA/W) 105 168 166

D* (cmHz1/2/W) 1.5×1010 1011 1.42×1012

Id 0.2-10µA 10nA-1µA 10-11.5nA

REFERENCES

[1] V. Ryzhii, I. Khmyrova, and M. Ryzhii, “Impact of Transit Time and Capture Effects on High Frequency Performance of Multiple Quantum -Well Infrared Photodetectors,” J. IEEE, Electron Devices. , vol. 45, No. 1, January 1998.

[2] S. D. Gunapala, J.S. Park, G. Sarusi, L. True_Lon, J.K. Liu, and P.D. Maker,“15 µm 128×128 GaAs/AlxGa1-xAs Quantum Well Infrared Photodetector Focal Plane Array Camera,” Transactions on Electron Devices, IEEE, vol. 44, No. 1 , pp. 45-50, January 1997.

[3] D. C. Wang, G. Bosman, and S. S. Li, “Current Conduction in Quantum Well Infrared Photodetectors under Low Bias,” Appl. Phys. Lett, vol. 68, pp. 2532-2534, 1996.

[4] N. E. L. Ettech and P. Harrison, “Carrier Scattering Approach to the Origins of Dark Current in Mid-and Far Infrared (Terahertz) Quantum Well Intersubband Photodetectors (QWIPs),” Journal ofQuantum Electronics, IEEE, vol. 37, No. 5, pp. 672 – 675, May 2001.

[5] B. F. Levine, C. G. Bethea, G. Hansian, V. O. Shen, E. Pelve, R. R.

GaAs Quantum Well Infrared Photodetectors,” Appl. Phys. Lett, vol. 56, pp. 851-853, 1990.

[6] V. Ryzhi, “Theory of Quantum Well IR Photoconductors with Tunneling Electron Injection,” Proceeding on Optoelectron, IEE, Vol. 44, No. 5, pp. 343-349, 1997.

[7] S. D. Gunapala, G. Surasi, J. Park, T. Lin, and B. F.Levine, “Infrared Detectors Reach New Length,” Phys. World, vol. 7, p. 35, 1994.

[8] H. C. Liu, A. G. Steele, M. Buchnan, and Z. R. Wasilewski, “Dark Current in Quantum Well Infrared Photodetector,” Appl. Phys., vol. 73, p. 2029, 1993.

[9] M. A. Gadir, and P. Harison, “Responsivity of Quantum Well Infrared Photodetectors (QWIPs) at Terahertz Wavelength,” J. Appl. Phys. Lett, vol. 91, pp. 5820-5825, 2002.

[10] X. Jians, S. S. Li, and M.Z. Tidrow, “Investigation of a Multistack Voltage Tunable for Color Quantum Well Infrared Photodetectors for Mid and Long Wavelength Infrared Detection,” Transactions on Quantum Electronics, IEEE, vol. 35, No. 11, pp. 1685-1693, 1999.

[11] S.D. Gunapala, S.V. Bandara, A. Singh, J. K. Liu, and B. Rafol, “640×486 Long Wavelength Two color GaAs/AlxGa1-xAs quantum Well Infrared Photodetector (QWIP) focal plane array camera,” Transactions on Electron Devices, IEEE, vol. 47. No 5. , pp. 963-971, May 2000.

[12] M. Ershov, and H. C. Liu, “Low-frequancy noise gain and photocurrent gain in quantum well infrared photodetectors,” J. Appl. Phys. vol.86, No 11, pp 6580-6585, 1999.

[13] John P. Loher “Physics of Strained Quantum Well Lasers,” Kluwer Academic Publisher, 1998.

[14] M. C. Hsu, Y. H. Hsu, and C. H. Kuan, “Design and Characterization of superlattice Infrared Photodetector Operating at Low Bias Voltage,” IEEE, Journal of Electron Devices, vol 47, No. 5. , May 2000.

921